SimSci

APC

User Guide
Version 2016 SP2

September 2017
 2017 Schneider Electric Software, LLC. All rights reserved.
No part of this documentation shall be reproduced, stored in a retrieval system, or transmitted
by any means, electronic, mechanical, photocopying, recording, or otherwise, without the
prior written permission of Schneider Electric Software, LLC. No liability is assumed with
respect to the use of the information contained herein.
Although precaution has been taken in the preparation of this documentation, Schneider
Electric Software, LLC assumes no responsibility for errors or omissions. The information in
this documentation is subject to change without notice and does not represent a commitment
on the part of Schneider Electric Software, LLC. The software described in this
documentation is furnished under a license agreement. This software may be used or copied
only in accordance with the terms of such license agreement.
ArchestrA, Avantis, DYNSIM, eDNA, EYESIM, Foxboro, Foxboro Evo, I/A Series, InBatch,
InduSoft, InStep, IntelaTrac, InTouch, PIPEPHASE, PRiSM, PRO/II, PROVISION, ROMeo,
Schneider Electric, SIM4ME, SimCentral, SimSci, Skelta, SmartGlance, Spiral Software,
VISUAL FLARE, WindowMaker, WindowViewer, and Wonderware are trademarks of
Schneider Electric SE, its subsidiaries, and affiliated companies. An extensive listing of
Schneider Electric Software, LLC trademarks can be found at: http://software.schneider-
electric.com/legal/trademarks/. All other brands may be trademarks of their respective
owners.
Schneider Electric Software, LLC
26561 Rancho Parkway South
Lake Forest, CA 92630 U.S.A.
(949) 727-3200
http://software.schneider-electric.com/
Publication date: 9/18/2017
Contact Us
Contact Schneider Electric Software Technical Support
Avantis Technical Support
Listing of regional and local country contacts: http://software.schneider-
electric.com/support/avantis/
Technical support: http://softwaresupport.schneider-electric.com/
For Avantis.PRO, Avantis Approvals, Avantis.DSS, and Condition Manager:
Email: [email protected]
Phone (8:30 a.m. to 5:00 p.m. Monday to Friday, Eastern Time):
o Toll-Free U.S. and Canada: 1-888-262-7111
o Toll-Free EMEA: 1-800-4670-6466
o Direct dial: 1-905-632-0635
For Avantis.XA:
Email: [email protected]
Phone (8:30 a.m. to 5:00 p.m. Monday to Friday, Eastern Time):
o Toll-Free U.S. and Canada: 1-800-991-8000
o Toll-Free EMEA: 1-800-4670-6466
o Direct dial: 1-905-632-4122
SimSci Technical Support
Listing of regional and local country contacts: http://software.schneider-
electric.com/support/simsci/
Technical support: http://softwaresupport.schneider-electric.com/
Email U.S. and Canada: [email protected]
Phone (USA 8:00 a.m. to 5:00 p.m. Central Time)
o Toll-Free U.S. and Canada: 1-800-746-7241
Skelta Technical Support
Email: [email protected]
Phone:
o U.S.: 1-678-306-4110 Option 3
o India: 91-80-4060-2600 Option 3
Wonderware Technical Support
Listing of regional and local country contacts: https://www.wonderware.com/contact/contact-
support/
Technical support: http://softwaresupport.schneider-electric.com/
Priority email for Customer FIRST Members: [email protected]
Email for customers without a support agreement: [email protected]
Phone
o Toll-Free North America and Latin America: 1-800-966-3371
o Direct dial: 1-949-639-8500
InStep Technical Support
Contact page: http://www.instepsoftware.com/contact-us
Technical support: https://support.instepsoftware.com/
Email: [email protected]
Phone (USA 8:00 a.m. to 5:00 p.m. Central Time)
o 1-312-894-7870
Schneider Electric Smart Water Software Technical Support
Help desk email: [email protected]
Help desk telephone hotline: +45 88 30 20 77 (09:00 to 16:00 Monday to Thursday,
Friday 09:00 to 15:00, Central European Time)
Contact Schneider Electric Software Learning Services
Contact Software Learning Services for assistance regarding classes, schedule, offerings,
frequently asked questions, tuition, policies, and more.
Email: [email protected]
Toll-Free U.S. and Canada: 1-866-998-7246
Direct: 1-949-639-8508
Fax: 1-949-639-1847
Acknowledgements
IronPython
The SimSci APC Director uses the IronPython programming language for all Director scripts.
IronPython is an open-source implementation of the Python programming language for
Microsoft .NET Framework. It is free software available under the Apache License 2.0
(http://www.apache.org/licenses/LICENSE-2.0.html), as published by The Apache Software
Foundation.
The source code for IronPython is available at https://github.com/IronLanguages/main.
 SimSci APC User Guide

Contents
Acknowledgements .................................................................................................................. 5
Chapter 1 Getting Started with SimSci APC ..................................................................... 15
Overview ................................................................................................................................ 15
User Interface ......................................................................................................................... 16
SimSci APC User Interface Badging ......................................................................................... 17
Diagnostic Messages .............................................................................................................. 18
Managing Projects .................................................................................................................. 19
Importing Data ........................................................................................................................ 20
Importing Aspen DMCplus Dat a ............................................................................................... 23
Importing .csv Files ................................................................................................................. 25
Format for .csv Files ................................................................................................................ 26
Changing the View for the Tables............................................................................................. 28
Trends and Strips.................................................................................................................... 28
Automatically Generating a Trend ...................................................................................... 30
Creating a Trend ............................................................................................................... 32
Managing Trends .............................................................................................................. 34
Managing Strips ................................................................................................................ 36
Adjusting the Scaling for a Strip .......................................................................................... 39
Managing the Trends in the Trend Manager ........................................................................ 41
Viewing Future Values for Trends ....................................................................................... 41
Limiting the Data wit hin the APC Project ................................................................................... 42
Scanning ................................................................................................................................ 43
Running SimSci APC as a Service ........................................................................................... 43
Enabling Auto Save for an APC Project .................................................................................... 44
Engineering Units and Normalized Units ................................................................................... 44

Chapter 2 Connecting to Data............................................................................................. 45

Overview ................................................................................................................................ 45
Connecting to an OPC Server .................................................................................................. 45
Connecting to an AIM*AP I Server ............................................................................................ 46
Adding a Group to a Connected OPC Server ............................................................................ 47
Editing a Group....................................................................................................................... 47
Removing a Group from a Connected OPC Server .................................................................... 48
Removing a Connected Server ................................................................................................ 48
Tags ...................................................................................................................................... 48

Version 2016 SP2 7

SimSci APC User Guide Contents

Manually Adding Tags ....................................................................................................... 49

Auto Discovery .................................................................................................................. 52
Managing Rulesets ...................................................................................................... 52
Rules .......................................................................................................................... 53
Managing Rules .......................................................................................................... 54
Applying a Rules et....................................................................................................... 55
Example of Automatically Adding Tags ......................................................................... 55
Configurations for the Selection Criteria ........................................................................ 56
Moving Tags to Different Groups ........................................................................................ 58
Removing Tags ................................................................................................................. 58
The Data Table ....................................................................................................................... 58
Replacing and Mapping Tags ............................................................................................. 59
Editing Data ...................................................................................................................... 60
Creating a Tag Configuration File from the Data Table ......................................................... 61

Chapter 3 Process Response Testing ............................................................................... 63

Overview ................................................................................................................................ 63
Readback Signals ................................................................................................................... 64
PRBS Generators ................................................................................................................... 64
Creating a PRBS Generator..................................................................................................... 65
Configuring a P RBS Generator ................................................................................................ 65
Adding Tags to a PRBS Generator ..................................................................................... 66
Cont rol Interval .................................................................................................................. 66
Random Interval ................................................................................................................ 67
Configuration Parameters for Output Tags in a PRBS Generator .......................................... 67
Auto............................................................................................................................ 67
PRBS Length .............................................................................................................. 68
Mean .......................................................................................................................... 69
Increment.................................................................................................................... 69
Amplitude.................................................................................................................... 69
Low Limit .................................................................................................................... 70
High Limit.................................................................................................................... 70
Operating a PRBS Generator .................................................................................................. 70
State of a PRBS Generator ................................................................................................ 70
Initializing a PRBS Generator............................................................................................. 71
Activating a PRBS Generator ............................................................................................. 71
Performing a Manual Step Test .......................................................................................... 72
Performing a P RBS Test.................................................................................................... 72
Viewing Trends from a P RBS Generator ............................................................................. 72
Deactivating a PRBS Generator ......................................................................................... 73
Deleting a PRBS Generator ..................................................................................................... 73
A vailable PRBSs ..................................................................................................................... 74

Chapter 4 Data Processing ................................................................................................. 77

Overview ................................................................................................................................ 77
The Translation Table ............................................................................................................. 77
Creating SimSci APC Tag Data and Comma Delimited Files ...................................................... 78
Creating a Tag Configuration File from the Data Table .............................................................. 79
Managing Custom Tags .......................................................................................................... 79

8 Version 2016 SP2

Contents SimSci APC User Guide

Managing Tag Collections ....................................................................................................... 80

Tag Data Manipulation ............................................................................................................ 82
Default Equations .............................................................................................................. 82
ADD ........................................................................................................................... 83
SUB............................................................................................................................ 83
MUL ........................................................................................................................... 83
DIV ............................................................................................................................. 83
LOG ........................................................................................................................... 83
E XP ............................................................................................................................ 83
SQRT ......................................................................................................................... 83
DIFF ........................................................................................................................... 83
LOG10........................................................................................................................ 84
LOW PASS FILTER..................................................................................................... 84
AVERAGE .................................................................................................................. 84
SMOOTH.................................................................................................................... 84
REMOVE SPIKES ....................................................................................................... 84
SHIFT LEFT................................................................................................................ 85
SHIFT RIGHT ............................................................................................................. 85
XY Lookup .................................................................................................................. 85
Customized Equations ....................................................................................................... 85
Managing Customized Equations ....................................................................................... 86
Creating a Calculation for a Custom Tag ............................................................................. 87
Configuring a Calculation for a Custom Tag ........................................................................ 88
Running Calc ulations for the Custom Tags ......................................................................... 88
Deleting Calculations for the Custom Tags .......................................................................... 89
Correlation Analysis ................................................................................................................ 89
Correlation Coefficients ..................................................................................................... 89
Autocorrelation Function .................................................................................................... 90
Cross-correlation Function ................................................................................................. 90
Differencing ...................................................................................................................... 91
Caus e Tags and Effect Tags .............................................................................................. 92
Managing Tags in the Caus e Tags and Effect Tags Collections ............................................ 92
Creating a Correlation A nalysis Calculation ......................................................................... 93
Running Correlation Analysis Calculations .......................................................................... 93
Deleting Correlation Analysis Calculations .......................................................................... 94
Generating Trends for the Correlation Analysis Calculations ................................................ 95
Characteristics of the Trends for the Correlation Analysis Calculations ................................. 95
Displaying the Statistical Results for the Correlation Analysis Calculations ............................ 97
PeakMaxNegativeV alueTime ....................................................................................... 97
PeakMaxNegativeV alue ............................................................................................... 98
PeakMaxPositiveV alueTime ......................................................................................... 98
PeakMaxPositiveV alue ................................................................................................ 98
Granger Index ............................................................................................................. 98
Suggested Prediction Interval ....................................................................................... 99
Dead Time .................................................................................................................. 99
Suggestion.................................................................................................................. 99

Modeling ................................................................................................................................ 101

Overview .............................................................................................................................. 101
Linear Models ....................................................................................................................... 101
Creating a Linear Model ................................................................................................... 102
Creating an Integrating Linear Model ................................................................................ 102
Copying a Linear Model ................................................................................................... 102
Deleting a Linear Model ................................................................................................... 103

Version 2016 SP2 9

SimSci APC User Guide Contents

Renaming a Linear Model ................................................................................................ 103

Changing the Order of the Linear Models .......................................................................... 103
Exporting the Data for a Linear Model ............................................................................... 104
Configuring the Structure of a Linear Model ...................................................................... 104
Adding Notes to a Model .................................................................................................. 106
Scenarios for a Linear Model ................................................................................................. 106
Managing Scenarios ............................................................................................................. 106
Selecting Data for Scenarios .................................................................................................. 108
Scenario Parameters............................................................................................................. 110
ID ................................................................................................................................... 111
Order.............................................................................................................................. 111
Least Squares Type ........................................................................................................ 111
Scaling ........................................................................................................................... 112
Prediction Interval ........................................................................................................... 112
Model Length .................................................................................................................. 112
Number of Cycles............................................................................................................ 113
ARX Mask ...................................................................................................................... 113
Trans fer Functions ................................................................................................................ 113
Trans fer Function Parameters .......................................................................................... 113
Managing Transfer Functions ........................................................................................... 114
Trans forming Transfer Functions ...................................................................................... 116
Identification Algorithms ........................................................................................................ 117
Generaliz ed Equation ...................................................................................................... 117
Order of Dynamics..................................................................................................... 118
Prediction Interval Ratio ............................................................................................. 118
Discrete Time Instants ............................................................................................... 118
Vectors for the Process Outputs ................................................................................. 118
Vectors for the Process Inputs .................................................................................... 119
Steady-State Properties ............................................................................................. 121
Examples of Typical Model Structures ........................................................................ 122
Incremental Equation....................................................................................................... 123
Recursive Least Squares (RLS ) Algorit hm ........................................................................ 124
Unbiased Least Squares ............................................................................................ 126
Implementation.......................................................................................................... 126
Normalization ............................................................................................................ 128
Ordinary Least Squares (OLS) and Partial Least Squares (PLS) Algorithms ........................ 129
Results of a Linear Model ...................................................................................................... 129
Model Coefficients ........................................................................................................... 130
Viewing the Sum Model Coefficients ........................................................................... 130
Viewing the Impulse Model Coefficients ...................................................................... 130
Copying and Pasting Model Coefficients ........................................................................... 131
Viewing Names or Descriptions of Tags ............................................................................ 131
Example of Model Coefficients on the Result Tab .............................................................. 131
Modifying the Model Coefficients ...................................................................................... 133
Curve Smoothing for the Model Coefficients...................................................................... 136
Gains ............................................................................................................................. 138
Residuals ....................................................................................................................... 138
Covariance Trace ............................................................................................................ 140
Steady-state Responses .................................................................................................. 140
Expanding Sparklines ...................................................................................................... 141
Model Data Comparisons ................................................................................................ 141

10 Version 2016 SP2

Contents SimSci APC User Guide

Composite Linear Model ........................................................................................................ 142

Creating a Composite Linear Model .................................................................................. 142
Effect of Prediction Intervals ............................................................................................. 143
Viewing the Composite Linear Model ................................................................................ 143
Configuring the Composite Linear Model .......................................................................... 144

Controller Design.................................................................................................................. 147

Overview .............................................................................................................................. 147
Final Linear Model ................................................................................................................ 147
Creating a Final Linear Model .......................................................................................... 148
Updating and Viewing the Final Linear Model .................................................................... 148
Configuration for the Final Linear Model ............................................................................ 148
Prediction Interval for the Final Linear Model ............................................................... 149
Order of Dynamics for the Final Linear Model .............................................................. 149
Model Length of the Final Linear Model ....................................................................... 149
Impulse Model Coefficients for the Final Linear Model .................................................. 150
Example of Impulse Model Coefficients for the Final Linear Model ................................ 151
Autoregressive Coefficients for the Final Linear Model ................................................. 152
Example of Autoregressive Coefficients for the Final Linear Model ............................... 152
Steady-state Responses for the Final Linear Model ..................................................... 153
Gains for the Final Linear Model ................................................................................. 154
Deleting the Final Linear Model ........................................................................................ 154
Cont roller Assembly .............................................................................................................. 155
Creating the MVC Controller ............................................................................................ 155
Configuring the MV C Controller ........................................................................................ 155
Cont roller Name ........................................................................................................ 156
Cont rol Interval .......................................................................................................... 156
Coupling ................................................................................................................... 156
Cont roller Model Order .............................................................................................. 157
Long Range Prediction .............................................................................................. 157
Blocking Width .......................................................................................................... 157
Blocking Horizon ....................................................................................................... 157
Cont rol Horizon ......................................................................................................... 157
Predictions ................................................................................................................ 158
Cont rol Modelset ....................................................................................................... 158
Watchdog From DCS ................................................................................................. 158
Watchdog To DCS ..................................................................................................... 159
Fail Time................................................................................................................... 159
Recover Time............................................................................................................ 159
MVC/LP Modes ......................................................................................................... 159
Simulation Modelset .................................................................................................. 159
Parameters for the Variables in an MV C Controller ............................................................ 160
Initial Value ............................................................................................................... 162
Vital .......................................................................................................................... 162
Switch....................................................................................................................... 162
Constraint Mode ........................................................................................................ 163
Track Setpoint ........................................................................................................... 163
Increment.................................................................................................................. 164
Low Constraint .......................................................................................................... 164
High Constraint ......................................................................................................... 164
Low Limit .................................................................................................................. 165
High Limit.................................................................................................................. 165
Low Validity............................................................................................................... 165
High Validity .............................................................................................................. 166

Version 2016 SP2 11

SimSci APC User Guide Contents

Scales ...................................................................................................................... 166

Error Factor............................................................................................................... 166
Deadzone ................................................................................................................. 167
Cushion .................................................................................................................... 167
Priority ...................................................................................................................... 167
MV Priority ................................................................................................................ 167
Low Importance......................................................................................................... 168
High Importance ........................................................................................................ 168
Lr Horizon ................................................................................................................. 168
Readback ................................................................................................................. 169
DCS Readback ......................................................................................................... 169
DCS Status ............................................................................................................... 169
Linearizing Function................................................................................................... 171
Define....................................................................................................................... 171
Rate of Change ......................................................................................................... 172
Frozen Time.............................................................................................................. 172
Infer Time ................................................................................................................. 172
Data Valid Time......................................................................................................... 172
Display ..................................................................................................................... 173
Page Number ............................................................................................................ 173
Subcontrollers ................................................................................................................. 173
Managing Subcontrollers ................................................................................................. 173
Configuring Subcontrollers ............................................................................................... 174
Vital .......................................................................................................................... 175
Switch (Subctrlr) ........................................................................................................ 175
Modelsets ....................................................................................................................... 175
Managing Modelsets ....................................................................................................... 176
Creating the Optimizer ..................................................................................................... 178
Configuring the Optimizer ................................................................................................ 178
Optimizer Name ........................................................................................................ 178
Cont rol Interval .......................................................................................................... 179
Optimize Int erval ....................................................................................................... 179
Origin Sourc e ............................................................................................................ 179
Iteration .................................................................................................................... 179
Cont rol GainSet ......................................................................................................... 180
Parameters for the Variables in the Optimizer ................................................................... 180
Cost.......................................................................................................................... 181
Increment.................................................................................................................. 181
Low Importance......................................................................................................... 182
High Importance ........................................................................................................ 182
Integrating................................................................................................................. 182
Integrating Scale ....................................................................................................... 183
Vital .......................................................................................................................... 184
Switch....................................................................................................................... 184
Low Constraint .......................................................................................................... 185
High Constraint ......................................................................................................... 185
Low Limit .................................................................................................................. 186
High Limit.................................................................................................................. 186
Low Validity............................................................................................................... 187
High Validity .............................................................................................................. 187
Cushion .................................................................................................................... 187
Linearizing Function................................................................................................... 188
Rate of Change ......................................................................................................... 188
Frozen Time.............................................................................................................. 189
Infer Time ................................................................................................................. 189
Data Valid Time......................................................................................................... 189

12 Version 2016 SP2

Contents SimSci APC User Guide

Gainsets ......................................................................................................................... 190

Managing Gains ets ......................................................................................................... 190
Updating the MVC Controller and Optimizer ...................................................................... 191
Results of Updating the MVC Controller and Optimizer ...................................................... 191
Piecewise Nonlinear Transforms ...................................................................................... 193
Simulation Mode ................................................................................................................... 194
Cost Function ....................................................................................................................... 195
Infer Time ....................................................................................................................... 196
Design Weights for the Cost Function ............................................................................... 197
Cont rol Algorithm ............................................................................................................ 198
Example of a Single-Input, Single-Out put (SIS O) Cont roller Structure ........................... 199
Example of a Multiple-Input, Multiple-Output (MIMO) Controller Structure ..................... 200
Profit Function ...................................................................................................................... 201
Normalization for the Profit Function ................................................................................. 202
Normalization of Variables for the Profit Function ......................................................... 202
Normalization of the Cost Coefficients for the Profit Function ........................................ 203
Incremental Form ............................................................................................................ 203
Initial Steady-state Values................................................................................................ 204
Steady-state Model ......................................................................................................... 205

Enabling Controllers............................................................................................................. 207

Overview .............................................................................................................................. 207
Enabling the MV C Controller.................................................................................................. 207
State of the MV C Controller ................................................................................................... 208
Online Configuration Parameters for the MVC Controller ......................................................... 209
Simulation Mode ............................................................................................................. 209
Condition Number Calculation .......................................................................................... 210
Cont roller Mode .............................................................................................................. 210
Current Status................................................................................................................. 210
Online Parameters for the Variables in the MV C Controller ...................................................... 210
Set Point......................................................................................................................... 211
Opt Target ...................................................................................................................... 211
Vital................................................................................................................................ 212
Switch ............................................................................................................................ 212
Current Value.................................................................................................................. 212
Trend ............................................................................................................................. 213
Low Constraint ................................................................................................................ 213
High Constraint ............................................................................................................... 214
Move .............................................................................................................................. 214
CV Weight ...................................................................................................................... 214
MV Weight ...................................................................................................................... 215
Target Weight ................................................................................................................. 215
Status............................................................................................................................. 216
Increment ....................................................................................................................... 216
Tracking ......................................................................................................................... 216
Changing the Parameters for the Enabled MV C Controller ....................................................... 216
Managing Trends for the MV C Controller ................................................................................ 218
Creating Radial Plots for the MVC Cont roller .......................................................................... 219
Enabling the Optimizer .......................................................................................................... 221

Version 2016 SP2 13

SimSci APC User Guide Contents

State of the Optimizer............................................................................................................ 222

Online Configuration Parameters for the Optimizer .................................................................. 223
Cont roller Mode .............................................................................................................. 223
Current Status................................................................................................................. 223
Online Parameters for the Variables in the Optimizer ............................................................... 223
Current Value.................................................................................................................. 224
Trend ............................................................................................................................. 224
Optimum......................................................................................................................... 225
Status............................................................................................................................. 225
Vital................................................................................................................................ 225
Origin ............................................................................................................................. 226
Changing the Parameters for an Enabled Optimizer ................................................................ 226

Monitoring Controllers ......................................................................................................... 229

SimSci APC Reports ............................................................................................................. 229
Tag Data Reports.................................................................................................................. 229
Managing Reports................................................................................................................. 230

Director Calculations............................................................................................................ 233

Overview .............................................................................................................................. 233
Script Editor.......................................................................................................................... 233
Managing Scripts .................................................................................................................. 234
Managing Sequences ............................................................................................................ 235

Glossary of Terms ................................................................................................................ 239

14 Version 2016 SP2

 SimSci APC User Guide

C HAPTER 1
Getting Started with SimSci APC
In This Chapter
Overview...................................................................................................................................... 15
User Interface............................................................................................................................... 16
SimSci APC User Interface Badging .............................................................................................. 17
Diagnostic Messages .................................................................................................................... 18
Managing Projects ........................................................................................................................ 19
Importing Data .............................................................................................................................. 20
Importing Aspen DMCplus Dat a..................................................................................................... 23
Importing .csv Files ....................................................................................................................... 25
Format for .csv Files ..................................................................................................................... 26
Changing the View for the Tables .................................................................................................. 28
Trends and Strips ......................................................................................................................... 28
Limiting the Data wit hin the APC Project ........................................................................................ 42
Scanning...................................................................................................................................... 43
Running SimSci APC as a Service ................................................................................................. 43
Enabling Auto Save for an APC Project .......................................................................................... 44
Engineering Units and Normalized Units ........................................................................................ 44

Overview
SimSci APC is a powerful software package that provides the integrated capability to
develop, implement, and support model-predictive control and optimization strategies. These
strategies tighten the control of the key variables in a process and push the process as a
whole closer to the economic optimum.
The principal functions of the SimSci APC software are:
Establishing the real-time distributed control system (DCS)/PLC communications
Acquiring the real-time or historical plant data
Performing the process response tests
Processing the imported plant data and the results of the process response tests
Performing the real-time calculations and manipulations for imported data
Modeling the plant data
Designing the controllers

Version 2016 SP2 15

SimSci APC User Guide Getting Started with SimSci APC

Implementing the final model-predictive constraint controller

Implementing the LP-constrained optimizer
Performing inferential predictions for any data tags with which the SimSci APC software
has lost communications.

User Interface
You can dock, undock, float, or hide any main window in the SimSci APC user interface. You
can also change the layout of the windows and your preferences are automatically saved and
displayed the next time you load the SimSci APC software.
The SimSci APC user interface includes the following:
SimSci APC Button (A)
Application Ribbon (B)
APC Explorer (C)
Page View (D)
Diagnostics Window (E)
Status Bar (F)

16 Version 2016 SP2

Getting Started with SimSci APC SimSci APC User Guide

You can use the SimSci APC Button to manage projects and to import data. See Managing
Projects on page 19 and Importing Data on page 20 for more information.
You can use the tabs found on the Application Ribbon to:
Manage trends, diagnostics, and diagnostic logging levels.
Configure servers, system times, and graph limits.
Start scanning.
Use tools such as the Theme selection and the Director Script Editor.
Display product Help.
You can access the majority of the SimSci APC features by using the APC Explorer, which
contains an hierarchical view of the project features. You can expand items in the tree and
click on a feature to open the corresponding tab in the Page View.
You can use the tabs in the Page View to view and edit data for the various components of
an APC Project, such as OPC server connections, PRBS Generators, models, and
controllers. You can also float, dock, hide, and unhide the individual tabs in the Page View.
You can view all logged Error, Warning, and Information messages related to the project in
the Diagnostics Window. See Diagnostic Messages on page 18 for more information.
The list in the Theme area on the Home tab controls the color scheme of the UI. If you do not
like the Dark theme, you can change it to the Light theme.

SimSci APC User Interface Badging

To convey information in an intuitive way close to the source of the problem, the SimSci APC
software dynamically badges parts of the user interface.
As a rule of thumb, green badges signify a good state or quality, yellow badges indicate a
warning state, and red badges mean that there is an error.
The following table lists the variety of badges used in the SimSci APC software, and what
each badge means.

Badge or Attribute Applies To Where Found Description

Connected OPC or Interface tab, Data The item has a good

AIM*API servers, Table tab, Script quality or is currently
tag quality, and Editor running.
running Director
scripts and
sequences

Connected OPC or Interface tab, Data The item had a bad or

AIM*API servers, Table tab, Script unknown quality or is in
tag quality, and Editor error.
running Director
scripts and
sequences

Diagnostic Diagnostics The diagnostic message is

messages Window an informational message.

Version 2016 SP2 17

SimSci APC User Guide Getting Started with SimSci APC

Diagnostic Diagnostics The diagnostic message is

messages Window a warning message.

Diagnostic Diagnostics The diagnostic message is

messages Window an error message.

Parameters Most tabs There is an input error.

Missing data must be
supplied, invalid dat a must
be corrected, or an action
is required.

Table rows Most tabs The row cont ains invalid or

missing data.

Current Value Cont roller The current value of the

slider Overview tab, variable is within the
Optimizer specified upper and lower
Overview tab constraints.

Current Value Cont roller The current value of the

slider Overview tab, variable is within a 5%
Optimizer margin of the upper or
Overview tab lower constraint.

Current Value Cont roller The current value of the

slider Overview tab, variable is outside of the
Optimizer upper or lower constraint
Overview tab by a margin greater than
5%.

Diagnostic Messages
Diagnostic messages are informational messages that are displayed in the Diagnostics
Window. You can hide the Diagnostics Window by clearing the Diagnostic Window check
box on the Home tab, in the Diagnostics group.
You can change the number of diagnostic messages that appear in the Diagnostics Window
by changing the value of the Messages variable on the Home tab, in the Diagnostics group.
If the number of logged diagnostic messages is greater than the number of diagnostic
messages that can be displayed, the SimSci APC software stores the excess diagnostics
messages in the logfile.txt file for the APC Project. The SimSci APC software stores the
oldest diagnostic messages first. You can find the logfile.txt file in the
...\APC2016\User\Projects\<Project> folder, where <Project> is the name of the APC
Project.
There are three types of diagnostic messages:
Information: Information diagnostic messages detail the actions that the SimSci APC
software performs during a SimSci APC session, such as connecting to an OPC server or
creating a model.
Warning: Warning diagnostic messages detail any potentially erroneous data,
connections, or configurations.

18 Version 2016 SP2

Getting Started with SimSci APC SimSci APC User Guide

Error: Error diagnostic messages detail any errors that occur during the APC session.
Each diagnostic message has four parameters:
Time Stamp: This is the time at which the SimSci APC software logs the diagnostic
message.
Type: This is the type for the diagnostic message.
Area: This is the SimSci APC area from which the diagnostic message originated.
Message: This is the content of the diagnostic message.
The SimSci APC software stores all the logged diagnostic messages in the APC Project file. It
reloads the logged diagnostic messages when you reopen an APC Project.

You can filter the diagnostic messages based on the logging level. You can choose from one
of three logging levels:
High: Logs all diagnostic messages.
Medium: Logs only relatively important diagnostic messages.
Low: Logs only vital diagnostic messages.
You can change the logging level by clicking the desired logging level on the Home tab, in
the Logging Levels group. The current logging level appears in the status bar at the bottom
of the SimSci APC window.
The logging level applies to all types of diagnostic messages. You cannot specify individual
logging levels for the different types of diagnostic messages.
By default, the Diagnostics Window jumps to the most current diagnostic message when the
SimSci APC software logs a new diagnostic message. You can freeze the Diagnostics
Window at the current location in the message log by selecting the Freeze Diagnostic check
box on the Home tab, in the Diagnostics group. If you select the Freeze Diagnostic check
box, the Diagnostics Window no longer jumps to the latest diagnostic message when the
SimSci APC software logs a new diagnostic message. Note that you can still scroll through
the Diagnostics Window.

Managing Projects
The SimSci APC software stores all APC Projects in an APC Project (.apc) file. An .apc file
contains the logged data and configuration information for the following:
OPC server connections
OPC tags
Saved trends
PRBS generators
Custom tags
Models

Version 2016 SP2 19

SimSci APC User Guide Getting Started with SimSci APC

Controllers
Director calculations
You can create, open, and save APC Projects by using commands found under the SimSci
APC Button. New APC Projects are initially titled untitled. You can rename a new APC
Project by saving it for the first time. We recommend that you save your APC Projects to the
...\APC2016\User\Projects folder in your SimSci APC install directory to allow for quick
access to your saved APC Projects. Note that this is the default save location for APC
Projects.

Acti vity Directions

Creating a new project Click the SimSci APC Button, and then click New Project.

Saving a project Click the SimSci APC Button, and then click Save Project. When
you save a project for the first time, the Save As dialog box
appears. In the Save As dialog box, in the File name box, type a
name for the project, and then click Save.

Saving a project under a new Click the SimSci APC Button, and then click Save As Project. In
name the File name box, type the new name for the project, and then
click Save.

Opening an existing project Click the SimSci APC Button, and then click Open Project.
Browse to and select the APC Project that you want to open, and
then click Open.

Importing Data
You can import the following data files by using commands under the SimSci APC Button, on
the Import menu:
SimSci APC tag data (.tagData) files
Connoisseur tag data (.fdax) files
SimSci APC tag configuration (.tagconfig files)
SimSci APC model data (.model) files
Connoisseur model data (.dat) files
SimSci APC transfer function model (.TFModel) files
SimSci APC or Connoisseur controller data (.dat) files
Note: Connoisseur is a powerful software package offered by SimSci; you can consider it as
the predecessor of the SimSci APC software.
If you import a SimSci APC or Connoisseur tag data (.tagData or .fdax) file while an APC
Project is open, the SimSci APC software closes the open APC Project, and then creates a
new APC Project for the imported tag data file. You can choose the tags that you want to
import from the tag data file. You can then use the data from the tag data file to create
models and design controllers.

20 Version 2016 SP2

Getting Started with SimSci APC SimSci APC User Guide

You can import an OPC tag (.tag) file from the Manual page by right-clicking on the Group
tab and selecting Import.
You can import the model data (.model or .dat) files and controller data (.dat) files only into an
open APC Project. You can use these data files to import one of the following:
A SimSci APC model
A Connoisseur model
A SimSci APC controller
A Connoisseur controller
There are two types of controller data (.dat) files, one for the QP controller (that is, the MVC
controller) and one for the LP controller (that is, the Optimizer). You can import both the
controller data files or only the QP controller data file. At the minimum, you must include both
the model data file and the QP controller data file in the import to provide all the required
information. If you import a SimSci APC or Connoisseur model, the SimSci APC software
creates only a new linear model during the import. If you import a SimSci APC or
Connoisseur controller, the SimSci APC software creates the following during the import:
A new linear model with identified model coefficients, gains, and steady state responses
A new composite linear model
A new final linear model
A new MVC controller and an associated modelset
A new Optimizer with the same tag structure as the new MVC controller and an
associated gainset
If you do not include the LP controller data file in the import, the new Optimizer is blank. If you
include the LP controller data file in the import, the new Optimizer contains configuration data
from the data file.
Note: In the Connoisseur software, you can create the Connoisseur model data files and
Connoisseur controller data files by clicking Write to File on the Controller Configuration
page for a QP or LP controller. The Write to File button is available in the Connoisseur 15.5
software and later Connoisseur versions. The Connoisseur QP controller Write to File
function writes a .txt file to the directory that the specification file was read from. It writes four
.dat files (QP controller, QP model, LP controller, and LP model) to the Connoisseur/bin
directory. Only the .dat files are required. Please see the Connoisseur documentation for
more information.

Version 2016 SP2 21

SimSci APC User Guide Getting Started with SimSci APC

When you import data files, the SimSci APC software adds all the Connoisseur signals or
SimSci APC tags in the data files to the APC Project as tags. Initially, it adds these tags to an
OPC group named ImportedGroup<#>, where <#> is the number of times you have imported
data files. In the APC Project, this Group exists within a pseudo OPC server (PseudoServer)
on a pseudo OPC server host (PseudoMachine). You can view this setup on the Interface
tab, in the Connected Servers tree:

To use the data in the imported data files for an online application, you must move the
imported Group to an actual OPC server by dragging the Group into the actual OPC server in
the Connected Servers tree. After you move the OPC Group, you must map the tags to
existing tags in the OPC server to connect to process data. Typically, you map an imported
tag to the same tag to which the original Connoisseur signal or SimSci APC tag connected.
See Connecting to an OPC Server on page 45 and Replacing and Mapping Tags on page 59
for more information.

Acti vity Directions

Importing a SimSci APC tag Under the SimSci APC Button, on the Import menu, on the Data
data file menu, click Import File. You must browse to and select a
.tagData file created from an APC Project to successfully import
data.
Note that you may change the Tag Type for tags in .tagData files
during import.

Importing a SimSci APC model Under the SimSci APC Button, on the Import menu, on the
Model menu, click Import Model. You must browse to and select
the .model file created from a SimSci APC linear modeling
scenario to successfully import data.

Importing a SimSci APC Under the SimSci APC Button, on the Import menu, on the
controller Controller menu, click Import Controller. You must browse to
and select both the model and controller .dat files created from a
SimSci APC controller to successfully import data. This type of
import adds tags to the PseudoServer group.

Importing a SimSci APC tag Under the SimSci APC Button, on the Import menu, on the Data
configuration file menu, click Import Tag Configurations. You must browse to
and select a .tagconfig file created from an AP C project to
successfully import data.

22 Version 2016 SP2

Getting Started with SimSci APC SimSci APC User Guide

Importing a Connoisseur tag Under the SimSci APC Button, on the Import menu, on the Data
data file menu, click Connoi sseur Data Files. You must browse to and
select an .fdax file creat ed from the Connoisseur software to
successfully import data.
Note that you may change the Tag Type for tags in .fdax files
during import.

Importing a Connoisseur Under the SimSci APC Button, on the Import menu, on the
model Model menu, click Connoisseur Model. You must browse to and
select both the model and controller .dat files creat ed from a
Connoisseur controller to successfully import data.

Importing a Connoisseur Under the SimSci APC Button, on the Import menu, click Import
controller Controller. You must browse to and select both the model and
controller .dat files created from a Connoisseur controller to
successfully import data.

Example: Importing a Connoisseur Tag Data File

1. Click the SimSci APC Button.
2. On the Import menu, on the Data menu, click Connoisseur Data Files. The Open dialog
box appears.
3. Browse to and select the .fdax file, and then click Open. The Import Data dialog box
appears.
4. Select the corresponding check boxes for the tags that you want to import.
5. If desired, change the tag type for any tag by selecting Input or Output from the
dropdown menu in the Tag Type column, and then click Import Data.

Importing Aspen DMCplus Data

You can import the following data files by using commands under the SimSci APC Button, on
the Import menu:
Aspen DMCplus model data (.mdl) files
Aspen DMCplus controller data (.ccf) files
Note: Aspen DMCplus is an advanced process control (APC) software package offered by
AspenTech. Aspen Technologies, Inc. produces licenses, and distributes the Aspen DMCplus
software.
You can import the Aspen DMCplus model data (.mdl) and controller data (.ccf) files only into
an open APC Project. You can use these data files to import one of the following:
An Aspen DMCplus model
An Aspen DMCplus controller
When you import an Aspen DMCplus model, you should include only the .mdl file in the
import. The SimSci APC software creates only a new linear model during the import.
When you import an Aspen DMCplus controller, you should include both the .ccf file and its
associated .mdl file in the import. If you are importing a controller that is connected to a
Foxboro I/A Series DCS, you can also import the associated .map file. The SimSci APC
software creates the following during the import:

Version 2016 SP2 23

SimSci APC User Guide Getting Started with SimSci APC

A new linear model with identified model coefficients, gains, and steady state responses
A new composite linear model
A new final linear model
A new MVC controller and an associated modelset
A new Optimizer with the same tag structure as the new MVC controller and an
associated gainset
Any subcontrollers that are present in the Aspen DMCplus controller data (.ccf) file
When you import Aspen DMCplus data files, the SimSci APC software adds all the Aspen
DMCplus tags in the data files to the APC Project. Initially, it adds these tags to an OPC
group named ImportedGroup<#>, where <#> is the number of times you have imported data
files. In the APC Project, this Group exists within a pseudo OPC server (PseudoServer) on a
pseudo OPC server host (PseudoMachine). You can view this setup on the Interface tab, in
the Connected Servers tree:

To use the data in the imported data files for an online application, you must move the
imported Group to an actual OPC server by dragging the Group into the actual OPC server in
the Connected Servers tree. After you move the OPC Group, you must map the tags to
existing tags in the OPC server to connect to process data. Typically, you map an imported
tag to the same tag to which the original Aspen DMCplus tag connected. See Connecting to
an OPC Server on page 45 and Replacing and Mapping Tags on page 59 for more
information.

Acti vity Directions

Importing an Aspen DMCplus Under the SimSci APC Button, on the Import menu, on the
model Model menu, click Import DMC+ Model. You must browse to
and select a .mdl file created from Aspen DMCplus to
successfully import data.

Importing an Aspen DMCplus Under the SimSci APC Button, on the Import menu, on the
controller Controller menu, click Import DMC+ Controller. You must
browse to and select both the model data (.mdl)and controller
data (.ccf) files creat ed from Aspen DMCplus to successfully
import data.

Example: Importing an Aspen DMCplus Controller

1. Click the SimSci APC Button.
2. On the Import menu, on the Controller menu, click Import DMC+ Controller. The
Import Controller Data dialog box appears.
24 Version 2016 SP2
Getting Started with SimSci APC SimSci APC User Guide

3. Browse to and select the .mdl and .ccf files, and then click Open. The SimSci APC
software imports the controller data.

4. Check the Linear Modeling, Composite Linear Model, Final Linear Model, and
Controller Details tabs to verify that the imported data is correct.

Importing .csv Files

A .csv file provides a convenient format to manually edit tag data produced by the
Connoisseur software or other APC software. The Connoisseur software produces .fdax files,
which are ASCII comma-delimited text files. These files may be difficult to read in a text
editor. You can change the file extension for an .fdax file to .csv and then open the .csv file in
Microsoft Excel. Microsoft Excel parses the data in the file and then presents the data in
columns and rows that you can easily read and modify.
You can import two types of .csv files into an APC Project, a .csv file that you create from an
.fdax file or a .csv file that you manually create from a set of tag data. Note that the SimSci
APC software cannot directly import .csv files. You must change the file extension for the .csv
file to .fdax before you import the file into an APC Project. If you want to import a .csv file into
an APC Project, the .csv file must adhere to a specific format. See Format for .csv Files on
page 26 for more information.
If you import an .fdax file while an APC Project is open, the SimSci APC software closes the
open APC Project, and then creates a new APC Project for the imported tag data file. You
can choose the tags that you want to import from the tag data file. You can then use the data
from the tag data file to create models and design controllers.

Version 2016 SP2 25

SimSci APC User Guide Getting Started with SimSci APC

To import a .csv file:

1. Create a .csv file that adheres to the required format. See Format for .csv Files on page
26 for more information.
2. Rename the .csv file so that it uses the .fdax file extension.
3. On the SimSci APC GUI, click the SimSci APC Button.
4. On the Import menu, on the Data menu, click Connoisseur Data Files. The Open dialog
box appears.
5. Browse to and select the .fdax file, and then click Open. The Import Data dialog box
appears.
6. Select the corresponding check boxes for the tags that you want to import.
7. If desired, change the tag type for any tag by selecting Input or Output from the
dropdown menu in the Tag Type column, and then click Import Data.

Format for .csv Files

The first row of the .csv file is the header row. It must contain the following:
The number of data points in the file
The number of tags in the file
The date of the last data point in the file, in dd/mm/yyyy format
The time of the last data point in the file, in hh mm ss format
The scan interval (that is, the amount of time between data points), in seconds

Important: The SimSci APC software uses the header row to populate the timestamp for the
imported data points. That is, when the SimSci APC software imports the data, it ignores the
values in the Date and Time column. Instead, it uses the specified number of data points, the
final date, the final time, and the scan interval to calculate the time range for the data points
and assign a timestamp to each data point. Thus, it removes any time gaps in the file by
overwriting the value in the Date and Time column with the timestamp that it calculates from
the header row values. For this reason you should not have any nonconsecutive data points
in any .fdax file (or .csv file) that you import.
The second row of the .csv file is a list of the A-Signals and M-Signals in Connoisseur. Each
signal should have its own column with the first column reserved for timestamp information.
The SimSci APC software imports A -Signals as output tags and M-Signals as input tags.

26 Version 2016 SP2

Getting Started with SimSci APC SimSci APC User Guide

The third row of the .csv file is a list of tag names that are associated with the A -Signals and
M-Signals in the second row. Each tag should have its own column with the first column
reserved for timestamp information. Each tag should appear under its associated A-Signal or
M-Signal. Note that the names of the imported tags in the APC Project are the same as these
tag names and not the names of the associated A-Signals and M-Signals.

The fourth row of the .csv file is a list of descriptions that are associated with the A-Signals
and M-Signals in the second row. Each description should have its own column with the first
column reserved for timestamp information. Each description should appear under its
associated A-Signal or M-Signal.

You should not leave the descriptions blank in the .csv file. A blank value in the .csv file
results in a string of commas in the resulting .fdax file. The SimSci APC software ignores
strings of commas when you import the .fdax file. If you leave a description blank in the .csv
file, the SimSci APC software cannot parse the resulting .fdax file correctly and gives errors
when you import it into an APC Project. You must fill each description with at least one
character. We suggest that you type a space or hyphen ("-") for each description that you
want to leave blank.
The fifth row of th.csv file is a list of units of measure (UOMs) that are associated with the A-
Signals and M-Signals in the second row. Each UOM should have its own column with the
first column reserved for timestamp information. Each description should appear under its
associated A-Signal or M-Signal.

You should not leave the UOMs blank in the .csv file. A blank value in the .csv file results in a
string of commas in the resulting .fdax file. The SimSci APC software ignores strings of
commas when you import the .fdax file. If you leave a description blank in the .csv file, the
SimSci APC software cannot parse the resulting .fdax file correctly and gives errors when you
import it into an APC Project. You must fill each description with at least one character. We
suggest that you type a space or hyphen ("-") for each description that you want to leave
blank.
The remaining rows of the .csv file are the data points that you want to import into an APC
Project. The first column should be a timestamp for the data point. The timestamp can be in
any format, since the SimSci APC software does not use this timestamp when you import the
resulting .fdax file into an APC Project.

Version 2016 SP2 27

SimSci APC User Guide Getting Started with SimSci APC

The following figure shows an example .csv file that you can convert into an .fdax file and
then import into an APC Project. See Importing .csv Files on page 25 for more information.

Changing the View for the Tables

You can hide the columns in the various data tables by using the Field Chooser. Columns
always appear in the same order. Hiding and subsequently displaying a column does not
change the position of the column in relation to the other displayed columns.
You can change the position of a column in relation to the other displayed columns by
dragging the column heading to the position in which you want it to appear.
To change the view for a table:

1. In the upper-left corner of the table, click Field Chooser to open the Field Chooser.
2. In the Field Chooser, clear the check box for a column to hide it in the table.
3. In the Field Chooser, select the check box for a column to display it in the table.
4. Repeat steps 2 and 3 until the table contains only the columns that you want to display.
5. Close the Field Chooser.
6. Drag a column heading (for example, Description) to the position in which you want it to
appear.
7. Repeat step 6 until the columns appear in the desired order.

Trends and Strips

In the SimSci APC documentation, a trend is a collection of time-dependent visualizations of
the variation in a tag or tags. Each time-dependent visualization is known as a Strip. The
SimSci APC software logs time-dependent data for each tag that you add to the APC Project.
You can use this data to display Strips for a single tag or for multiple tags at one time. That is,
there are two types of Strips:

28 Version 2016 SP2

Getting Started with SimSci APC SimSci APC User Guide

Single-tag Strips: Use this type of Strip to plot data for tag parameters with independent
y-axis values. Time appears on a common x-axis.

Version 2016 SP2 29

SimSci APC User Guide Getting Started with SimSci APC

Superimposed Strips: Use this type of Strip to plot data for comparative analysis of tag
parameters. By default, the tag data appears on the same y-axis.You can change this
default axis setup. See Adjusting the Scaling for a Strip on page 39 for more information.

Concepts
A Strip displays the time-dependent behavior of a tag.
A trend is a collection of Strips.
You can choose between single-tag Strips and superimposed Strips.
Actions
Generate a trend.
Create a trend.
Add Strips to a trend.
Delete Strips from a trend.
Change the line colors for a Strip.
Change the scaling for a Strip.
Save trends to the Trend Manager.
Reopen saved trends.

Automatically Generating a Trend

You can automatically generate a trend by using the right-click commands in the following
tabs:

30 Version 2016 SP2

Getting Started with SimSci APC SimSci APC User Guide

Data Table: Generate single-tag Strips or a superimposed Strip on the trend depending
on which right-click command you use to generate the trend. Note that when you
generate a trend from this tab, you can generate a trend for only input tags or for only
output tags at one time.
Operation: Generate single-tag Strips that compare the input tags and the output tags for
the PRBS generator. See Viewing Trends from a PRBS Generator on page 72 for more
information.
All Tags: Generate single-tag Strips or a superimposed Strip depending on the right-click
command you use to generate the trend. Note that when you generate a trend from this
tab, you can generate a trend for only raw tags or for only custom tags at one time.
Result: Generate a trend for the coefficient results of linear modeling. See Expanding
Sparklines on page 141 for more information.
You can automatically generate a trend by clicking buttons in the following tabs:
Data Manipulation: Generate single-tag Strips that compare the custom tags to the tags
from which the SimSci APC software calculates the custom tags. See Running
Calculations for the Custom Tags on page 88 for more information.
Data Mining: Generate single-tag Strips for the results of the correlation analysis. See
Generating Trends for the Correlation Analysis Calculations on page 95 for more
information.
You can also automatically generate a number of specialized trends for an MVC controller by
using the Controller Overview and Sub Controller Summary tabs. See Managing Trends
for the MVC Controller on page 218 and Creating Radial Plots for the MVC Controller on
page 219 for more information.

Acti vity Directions

Generating single-tag Strips On the Input Tags tab, the Output Tags tab, the Custom Input
from the Data Table tab Tags tab, or the Custom Output Tags tab, select the tags for
which you want to generate single-t ag Strips. Right-click the
selection, and then click Add to Stripped Trends.

Generating a superimposed On the Input Tags tab, the Output Tags tab, the Custom Input
Strip from the Data Table tab Tags tab, or the Custom Output Tags tab, select the tags for
which you want to generate a superimposed Strip. Right-click the
selection, and then click Add to SuperImposed Trends.

Generating single-tag Strips In either the Raw Tag(s) area or the Custom Tag(s) area, select
from the All Tags tab the tags for whic h you want to generate single -tag Strips. Right-
click the selection, and then click Generate Stripped Trend(s).

Generating a superimposed In either the Raw Tag(s) area or the Custom Tag(s) area, select
Strip from the All Tags tab the tags for whic h you want to generate a superimposed Strip.
Right -click the selection, and then click Generate Superimposed
Trend(s).

See the preceding references for more information on generating a trend from tabs other than the Data Table and All Tags tabs.

Version 2016 SP2 31

SimSci APC User Guide Getting Started with SimSci APC

Creating a Trend
You can create a fully customized trend by clicking Add Trend on the Home tab, in the
Trends group. A trend created in this manner contains no Strips, unlike a trend that you
automatically generate from a tab. You can add multiple Strips to the trend. Each Strip can be
a single-tag Strip or a superimposed Strip.
Note that a Strip can contain visualizations for more than one tag parameter for a single tag.
Although the visualized data relates to a single tag, the SimSci APC documentation still
classifies this type of Strip as a superimposed Strip, because more than one line appears on
the Strip.
You can add tag data to a Strip by using general tag collections, by using controller-specific
tag collections, or by using linear model-specific tag collections.
If you add tag data to a Strip by using a general tag collection, you can add only the Current
Value parameter to the Strip.
If you add tag data to a Strip by using controller-specific tag collections, you can add the
Current Value parameter as well as the following controller-specific tag parameters to the
Strip:
High Constraint (CVs and MVs)
Low Constraint (CVs and MVs)
SetPoint (CVs only)
Opt Target (MVs only)
Optimum (CVs and MVs)
Model Prediction (CVs only)
FiveStep Prediction (CVs only)
LongRangeModel Prediction (CVs only)
Switch Status (CVs and MVs)
QP Constraint Status (CVs and MVs)
LP Constraint Status (CVs and MVs)
If you add tag data to a Strip by using model-specific tag collections, you can add the Current
Value parameter as well as the following model data parameters to the Strip:
Prediction
IncrementError
ResidualError
You can also add the Controller Status and Optimizer Status of the controller to the Strip.
Note that the model data parameters are specific to a linear modeling scenario within a
specified linear model. They include data only for the time interval in which you identify the
associated linear modeling scenario. See Model Data Comparisons on page 141 for more
information.

32 Version 2016 SP2

Getting Started with SimSci APC SimSci APC User Guide

WARNING: If you add tag data to a Strip by using the model-specific tag collections, you
cannot use the controller-specific or general tag collections to add tag data to a Strip or to
create new Strips. The reverse is also true.
After you add Strips to the trend, you can change the order of the Strips, the scaling for a
Strip, and the line color for each visualization within a Strip. See Managing Strips on page 36
for more information.
To create a trend:
1. On the Home tab, in the Trends group, click Add Trend. A blank trend appears.
1. On the trend, enter the Trend Name.

2. Click on the Trend Settings button and enter the Trend Range.
2. Click Add an empty chart. A new Strip appears in the trend.
3. To add tag data to a Strip by using a general tag collection, expand the Tag List area.
4. In the Collection Name list, select the tag collection that contains the tags you want to
add to the Strip.
5. In the list of tags, select the tags you want to add to the Strip, and then drag the selection
onto the Strip. Note that you can create a new Strip by dragging the selection onto an
area outside of any existing Strips.
6. To add tag data to a Strip by using controller-specific tag collections, expand the
Controller Tags area.
7. In the Controller Name list, select the controller containing the tags you want to add to
the Strip.
8. In the Parameters list, select the check box for each tag parameter you want to add to
the Strip. Note that the SimSci APC software always adds the Current Value parameter
to the Strip.
9. In the list of tags on the Control variables tab, the Manipulated variables tab, or the
Feed Forward variables tab, select the tags you want to add to the Strip, and then drag
the selection onto the Strip.Note that you can create a new Strip by dragging the selection
onto an area outside of any existing Strips.
10. To add tag data to a Strip by using linear model-specific tag collections, expand the
Models area.
11. In the Model Name list, select the linear model that contains the tags you want to add to
the Strip.
12. In the Scenario Number list, select the scenario that contains the model data you want to
add to the Strip.
13. In the Model Data Parameters list, select the check box for each tag parameter you want
to add to the Strip. Note that the SimSci APC software always adds the Current Value
parameter to the Strip.
14. In the Controlled Variable list, select the tags you want to add to the Strip, and then drag
the selection onto the Strip.Note that you can create a new Strip by dragging the selection
onto an area outside of any existing Strips.
15. Repeat steps 3-15 as desired.

Version 2016 SP2 33

SimSci APC User Guide Getting Started with SimSci APC

Managing Trends
You can add Strips to a trend. Newly added Strips appear at the bottom of the list of Strips.
You can move a Strip up or down in the list.
You can also delete Strips from the trend.
You can save a trend by storing it in the Trend Manager. After you save a trend, you can
reopen the trend from the Trend Manager. A reopened trend contains the same displayed line
series, line colors, and scaling as when you last saved it. You can open the Trend Manager
by clicking Trend Manager on the Home tab, in the Trends group. See Managing the Trends
in the Trend Manager on page 41 for more information.
You can rename a trend by using the Trend Name parameter on the trend.
You can export the data in a trend. The exported trend data includes only the data that
appears in the trend at the time of the export. Exported trend data is saved as a .csv file. You
can then open the .csv file in Microsoft Excel.
While the APC Project is scanning, the Strips in a trend update with the most recent
information.
Only a specified time range appears on the trend. You can set the time range by using the
Trend Range and Trend Offset parameters on the trend. You can also set the Trend Range
parameter on the Configure tab, in the Trend Properties group.The Trend Range
parameter on the Configure tab does not apply to existing trends; that is, it applies to newly
created or generated trends only. The Trend Range parameter determines the total amount
of time that trend spans. The Trend Offset allows you to shift the trend view back in time so
that you can focus on the data that you are interested in. When you navigate through trends,
the range and offset from the first trend is preserved.
If you want to stop the Strips from updating with new data while t he APC Project is scanning,
you can click Stop on the trend. If you want to restart the Strips so that they update with new
data, you can click Start on the trend. Note that the Start and Stop buttons appear in the
same location.

Acti vity Directions

Adding a blank Strip to a trend On the trend, click Add an empty chart. A new Strip appears on
the trend.

Adding tag data to a Strip by Expand the Tag Li st area. In the Collection Name list, select the
using a general tag collection tag collection that contains the tags you want to add to the Strip.
In the list of tags, select the tags you want to add to the Strip, and
then drag the selection onto the Strip. Not e that you can create a
new Strip by dragging the selection onto an ar ea outside of any
existing Strips.

34 Version 2016 SP2

Getting Started with SimSci APC SimSci APC User Guide

Adding tag data to a Strip by Expand the Controller Tags area. In the Controller Name list,
using controller-specific tag select the controller that contains the tags you want to add to the
collections Strip. In the Parameters list, select the check box for each tag
parameter you want to add to the Strip. Note that the SimSci APC
software always adds the Current Value parameter to the Strip.
In the list of tags, select the tags you want to add to the Strip, and
then drag the selection onto the Strip. Not e that you can create a
new Strip by dragging the selection onto an area outside of any
existing Strips.

Adding tag data to a Strip by Expand the Models area. In the Model Name list, select the
using linear model -specific tag linear model that contains the tags you want to add to the Strip. In
collections the Scenario Number list, select the scenario that contains the
model data you want to add to the Strip. In the Model Data
Parameters list, select the check box for each tag parameter you
want to add to the Strip. Note that the SimSci APC software
always adds the Current Value parameter to the Strip. In the
Controlled Variable list, select the tags you want to add to the
Strip, and then drag the selection onto the Strip.Note that you can
create a new Strip by dragging the selection onto an area outside
of any existing Strips.

Deleting Strips from a trend In the Tag Details area for the Strips you want to remove, select
the check box at the top of the Tag Details area, and then click

Delete Trends .

Moving a Strip up in the list of In the Tag Details area for the Strip you want to move, select the
Strips check box at the top of the Tag Details area, and then click

Move Strip up . Note that you can move only one Strip at a
time.

Moving a Strip down in the list In the Tag Details area for the Strip you want to move, select the
of Strips check box at the top of the Tag Details area, and then click

Move Strip down . Note that you can move only one Strip at
a time.

Renaming a trend On the trend, click the Trend Name box. Type the new name for
the trend, and then press Enter.

Saving the trend

On the trend, click Save to Trend Manager . The trend
appears in the Trend Manager under the name you specified in
the Trend Name parameter. Note that if the trend already exists
in the Trend Manager, you can click Save to Trend Manager

again to save any changes you make to the trend.

Exporting trend data

On the trend, click Export Data . The Export Trends Data
dialog box appears. Browse to the folder in which you want to
save the .cs v file. In the File name box, type the name of the .csv
file, and then click Save.

Version 2016 SP2 35

SimSci APC User Guide Getting Started with SimSci APC

Setting the time range and

offset On the trend, click the Trend Settings box . Type the new
Trend Range, and set the UOM for the new value. Type the new
Trend Offset, set the UOM for the new value, and then press
Enter.

Stopping the trend updat es

On the trend, click Stop . Note that you can click this button
only when the APC Project is scanning.

Restarting the trend updates

On the trend, click Start . Note that you can click this button
only when the APC Project is scanning.

Displaying statistics for tags in

the trend On the trend, click Display Stati stics . Note that you can
click anywhere else on the trend to once again hide the statistics.

Stepwise zooming in and out

on a trend On the trend, click the Trend Settings box . Type the new

Zoom %, and then press Enter. Press the Zoom In button

or the Zoom Out button to zoom in or out by the specified

percentage.

Zooming in on a trend by On the trend, double-click the left side of the area you want to
selection a region zoom in on, and then double click the right side of the area you
want to zoom in on.

Resetting the zoom on a trend

On the trend, click the Reset Zoom button to reset the
range and offset back to the original settings.

Example: Exporting Trend Data

1. On the trend, click Export Data . The Export Trends Data dialog box appears.
2. Browse to the folder in which you want to save the .csv file.
3. In the File name box, type the name of the .csv file, and then click Save.

Managing Strips
Strips appear in a trend. You can add tag data to a Strip by using general tag collections, by
using controller-specific tag collections, or by using linear model-specific tag collections.
WARNING: If you add tag data to a Strip by using the model-specific tag collections, you
cannot use the controller-specific or general tag collections to add tag data to a Strip or to
create new Strips. The reverse is also true.
If you add tag data to a Strip by using a general tag collection, you can add only the Current
Value parameter to the Strip.
If you add tag data to a Strip by using controller-specific tag collections, you can add the
Current Value parameter as well as the following controller-specific tag parameters to the
Strip:

36 Version 2016 SP2

Getting Started with SimSci APC SimSci APC User Guide

High Constraint (CVs and MVs)

Low Constraint (CVs and MVs)
SetPoint (CVs only)
Opt Target (MVs only)
Optimum (CVs and MVs)
Model Prediction (CVs only)
FiveStep Prediction (CVs only)
LongRangeModel Prediction (CVs only)
If you add tag data to a Strip by using linear model-specific tag collections, you can add the
Current Value parameter as well as the following model data parameters to the Strip:
Prediction
IncrementError
ResidualError
Note: The values of the preceding parameters appear in the Value column in the Tag Details
area for a Strip. The values automatically update when you point to specific points on the line
series for a Strip.
You can remove tag data from a Strip. For a superimposed Strip, the line for the removed tag
data no longer appears in the superimposed Strip. Note that this may affect the scaling of the
superimposed Strip. For a single-tag Strip, the line for the removed tag data no longer
appears in the single-tag Strip. That is, the Strip is blank.
You can change the color of the line series for a Strip to any color in the color palette. If you
do not want to use any of the colors in the color palette, you can provide your own advanced
color selections on an RGB, HSL, or CMYK basis.
You can adjust the scaling for a Strip. See Adjusting the Scaling for a Strip on page 39 for
more information.
You can hide the Tag Details areas for the Strips to increase the area in which the Strips
appear on a trend. Note that you cannot hide the Tag Details area for a single Strip. That is,
when you hide or display the Tag Details area, the change applies to all Strips on the trend.

Acti vity Directions

Adding tag data to a Strip by Expand the Tag Li st area. In the Collection Name list, select the
using a general tag collection tag collection that contains the tags you want to add to the
Strip.In the list of tags, select the tags you want to add to the
Strip, and then drag the selection onto the Strip. Note that you
can create a new Strip by dragging the selection onto an area
outside of any existing Strips.

Version 2016 SP2 37

SimSci APC User Guide Getting Started with SimSci APC

Adding tag data to a Strip by Expand the Controller Tags area. In the Controller Name list,
using controller-specific tag select the controller that contains the tags you want to add to the
collections Strip. In the Parameters list, select the check box for each tag
parameter you want to add to the Strip. Note that the SimSci APC
software always adds the Current Value parameter to the Strip.
In the list of tags, select the tags you want to add to the Strip, and
then drag the selection onto the Strip. Not e that you can create a
new Strip by dragging the selection onto an area outside of any
existing Strips.

Adding tag data to a Strip by Expand the Models area. In the Model Name list, select the
using linear model -specific tag linear model that contains the tags you want to add to the Strip. In
collections the Scenario Number list, select the scenario that contains the
model data you want to add to the Strip. In the Model Data
Parameters list, select the check box for each tag parameter you
want to add to the Strip. Note that the SimSci APC software
always adds the Current Value parameter to the Strip. In the
Controlled Variable list, select the tags you want to add to the
Strip, and then drag the selection onto the Strip.Note that you can
create a new Strip by dragging the selection onto an area outside
of any existing Strips.

Deleting tag data from a Strip In the Tag Details area for the Strip, select the tags or tag

parameters you want to delete. Click Delete .

Changing the color of a line

series for a Strip In the Tag Details area for the Strip, click Configure to
expand the configuration area. On the Variables tab, in the Color
list for the line series you want to change, click any color in the
color palette to use that color for the line series or click
Advanced for more color selections. You must enter advanced
color selections on an RGB, HSL, or CMYK basis.

Hiding the Tag Details area

for the Strips on the trend At the top of the trend, click Click here to hide tag The Tag
Details area no longer appears for the Strips on the trend.

Displaying the Configure

button for the Strips on the At the top of the trend, click Click here to show tag . The
trend Tag Details area now appears for the Strips on the trend.

Example: Changing the Color of a Line Series for a Trend

1. On the trend, on the Strip, click Configure . The configuration area appears.
2. On the Variables tab, click the Color list for the line series you want to change.
3. If you want to use a color on the color palette, click the color. Otherwise, skip to the next
step.
4. If you want to use your own color, click Advanced. The Advanced dialog box appears.
5. To specify the RGB color code for your color, on the RGB tab, enter the Red, Green,
Blue, and Alpha values for your color.

38 Version 2016 SP2

Getting Started with SimSci APC SimSci APC User Guide

Note: The Alpha value represents the opacity of the color. A higher Alpha value results
in a more opaque color. A lower Alpha value results in a more transparent color. You may
find transparent colors useful for Strips that contain intersecting lines, such as
superimposed Strips and the trend results for linear model identification scenarios.
6. To specify the HSL color code for your color, on the HSL tab, enter the Hue, Saturation,
Lightness, and Alpha values for your color.
7. To specify the CMYK color code for your color, on the CMYK tab, enter the Cyan,
Magenta, Yellow, Black, and Alpha values for your color.
Note: You need to specify only one type of color code (RGB, HSL, or CMYK) for your
color.
8. After you specify the color code, click OK.

Adjusting the Scaling for a Strip

You can adjust the scaling of a Strip by using the controls in the Tag Details area for the
Strip. Note that you can hide or display this area by using controls at the top of the trend.
You can automatically scale a Strip or you can provide your own scaling for a Strip. When
you automatically scale a Strip, the scale of the y-axis changes automatically to provide the
best fit for the data currently displayed on the Strip. When you provide your own scaling for a
Strip, the scale of the y-axis does not change unless you change the value of the High Scale
parameter or the Low Scale parameter.
You can also change which end of the Strip displays the y-axis for each line series. For
example, you can display the y-axis for one line series on the right end of the Strip and the y-
axis for another line series on the left end of the Strip. The y-axis on the left end of the Strip is
known as the primary y-axis and the y-axis on the right end of the Strip is known as the
secondary y-axis.
If a line series appears by itself on one of the y-axes, the y-axis uses the auto-scaling for that
line series. If a line series appears with other line series on one of the y-axes, the y-axis uses
the auto-scaling that encompasses all the values for all the line series on that y-axis. If you do
not want to automatically scale the y-axes, you can provide your own scaling for both the
primary y-axis and the secondary y-axis.
If you want to compare data for tags that have significantly different ranges of values, you
may want to configure the associated line series to appear on different y-axes. For example,
a tag with values on the scale of 101 appears as a poorly detailed line when it appears on the
same y-axis as a tag with values on the scale of 102.

Acti vity Directions

Hiding the Tag Details area

for the Strips on the trend At the top of the trend, click Click here to hide tag . The Tag
Details area no longer appears for the Strips on the trend.

Displaying the Tag Details

area for the Strips on the trend At the top of the trend, click Click here to show tag . The
Tag Details area now appears for the Strips on the trend.

Version 2016 SP2 39

SimSci APC User Guide Getting Started with SimSci APC

Changing the y-axis on which

a line series appears In the Tag Details area for the Strip, click Configure to
expand the configuration area. On the Variables tab, in the Y-
Axis list for the line series, select L to display the y-axis on the
left end of the Strip. Select R to display the y-axis on the right end
of the Strip.

Automatically scaling the

primary y-axis In the Tag Details area for the Strip, click Configure to
expand the configuration area. On the Scales tab, select the
Auto Scale check box for the Primary Y-Axi s.

Providing your own scaling for

the primary y-axis In the Tag Details area for the Strip, click Configure to
expand the configuration area. On the Scales tab, clear the Auto
Scale check box for the Primary Y-Axi s. Change the value of the
High Scale parameter to adjust the great est value on the primary
y-axis. Change the value of the Low Scale parameter to adjust
the least value on the primary y-axis.

Automatically scaling the

secondary y-axis In the Tag Details area for the Strip, click Configure to
expand the configuration area. On the Scales tab, select the
Auto Scale check box for the Secondary Y-Axis.

Providing your own scaling for

the secondary y-axis In the Tag Details area for the Strip, click Configure to
expand the configuration area. On the Scales tab, clear the Auto
Scale check box for the Secondary Y-Axis. Change the value of
the High Scale paramet er to adjust the greatest value on the
secondary y-axis. Change the value of the Low Scale parameter
to adjust the least value on the secondary y-axis.

Example: Providing Your Own Scaling for the Primary y-Axis

1. In the Tag Details area for the Strip, click Configure . The configuration area
appears.
2. On the Scales tab, clear the Auto Scale check box for the primary y-axis.
3. Change the value of the High Scale parameter to adjust the greatest value on the primary
y-axis.
4. Change the value of the Low Scale parameter to adjust the least value on the primary y-
axis.
Example: Providing Your Own Scaling for the Secondary y-Axis

1. In the Tag Details area for the Strip, click Configure . The configuration area
appears.
2. On the Scales tab, clear the Auto Scale check box for the secondary y-axis.
3. Change the value of the High Scale parameter to adjust the greatest value on the
secondary y-axis.
4. Change the value of the Low Scale parameter to adjust the least value on the secondary
y-axis.
40 Version 2016 SP2
Getting Started with SimSci APC SimSci APC User Guide

Managing the Trends in the Trend Manager

You can use the Trend Manager to do the following:
Save a trend, including its configured Strips.
Quickly view saved trends.
When you view a saved trend, the reopened trend contains the same displayed tags, line
colors, and scaling as when you last saved it. You can browse through the saved trends by
using the controls at the top of the trend.

Acti vity Directions

Saving a trend to the Trend After you generate or create a trend, on the trend, click Save to
Manager
Trend Manager . The trend appears in the Trend Manager
under the name you specified in the Trend Name parameter.
Note that if the trend already exists in the Trend Manager, you

can click Save to Trend Manager again to save any

changes you make to the trend.

Viewing saved trends In the Trend Manager, select the trends you want to view, and
then click View.

Navigating to the next trend in

the Trend Manager On the trend, click Show next trend from trend manager .
Note that you cannot click this button unless you open and view
the trend from the Trend Manager.

Navigating to the previous On the trend, click Show previous trend from trend manager
trend in the Trend Manager
. Note that you cannot click this button unless you open and
view the trend from the Trend Manager.

Deleting trends from the Trend In the Trend Manager, select the trends you want to delet e, and
Manager
then click Delete Trend Item .

Viewing Future Values for Trends

You can view future values for controller tags. The future values include the next steps that
the MVC controller predicts for the variables within the controller. This includes predictions for
MVs and CVs. The Long Range Prediction parameter determines how many future values
that the MVC controller predicts. See Long Range Prediction on page 157 for more
information.
Note that if you add a tag to a Strip that is not predicted by an MVC controller (that is, not
included in an MVC controller), the future values for that tag are blank.
You can turn the future values on or off for a trend.

Version 2016 SP2 41

SimSci APC User Guide Getting Started with SimSci APC

To turn on the future values for a trend:

On the trend, select the Future Chart check box.
To turn off the future values for a trend:
On the trend, clear the Future Chart check box.

Limiting the Data within the APC Project

In some case, the APC Project may contain too much data to function in an efficient manner.
Typically, APC Projects that contain a large amount (on the scale of days or months worth) of
data are slow to load when you open them and it may take a long time to generate the
various page views and saved trends in the APC Project. You can limit the amount of data
that is stored in the APC Project by using the Data Management features. You can also
remove all data from the APC Project by using the Purge All Data feature.
You can turn Data Management on or off. You can specify the maximum number of days
worth of data that you want to store in the APC Project at one time. The SimSci APC software
removes any data older than the specified number of days. You must keep at least 2 days
worth of data in the APC Project at one time, but you can increase the number of days worth
of data that you want to store in the APC Project.
WARNING: The SimSci APC software does not create backup copies of the APC Project
data when it removes data based on the Data Management settings or when you use the
Purge All Data feature. If you do not want to lose this data, you should set up tag data reports
on the Process Monitor Configuration tab. See Tag Data Reports on page 229 for more
information.
The Data Management feature is an online operation. Because the removal of data from the
APC Project is likely to remove the data from which you generated the linear models and the
modelsets and gainsets in the MVC controllers, you should ensure that you have an offline
copy of your APC Project that contains the required modeling data before you turn on Data
Management.

Acti vity Directions

Turning on Data Management On the Configure tab, in the Data Management group, click On.
Note that you can click On only when the AP C Project is
scanning.

Specifying the amount of data On the Configure tab, in the Data Management group, in the
to store in the APC Project Recent Hi story box, type the number of days worth of data that
you want to store in the APC Project. The SimSci APC software
removes any data older than the specified number of days.

Removing all data stored in On the Home tab, in the Purge group, click Purge All Data. The
the APC Project SimSci APC software removes all data stored in the AP C Project.

42 Version 2016 SP2

Getting Started with SimSci APC SimSci APC User Guide

Turning off Data Management On the Configure tab, in the Data Management group, click Off.
Note that you can click Off only when the APC Project is
scanning.

Scanning
Scanning is an online operation that contacts the OPC servers, loads the current data for the
input tags into the APC Project, and writes the current values for the output tags in the APC
Project to the OPC servers.
You can activate a controller only when you are scanning. If you stop scanning, the SimSci
APC software sets all the controllers to the Manual state.
You can start scanning by clicking On on the Home tab, in the Scan group.
You can stop scanning by clicking Off on the on the Home tab, in the Scan group.

Running SimSci APC as a Service

You can run the SimSci APC software as a Windows service, which means APC Projects can
be scanning without having the user interface open. When you use this functionality and
begin scanning, the SimSci APC software binds your current APC Project to the service and
continues scanning until you turn scanning off. If you lose power or restart your computer,
your APC Project automatically opens when you reopen the SimSci APC user interface. It
continues scanning and the service functionality is still on after the computer restarts.
Controller Function While Running SimSci APC as a Service
You can run your controllers in the Manual, Ready, or Active state when you run the SimSci
APC software as a service. If you lose power or restart your computer, your controller
continues to run in the mode in which you set it after the computer restarts, which you can
see when you reopen the SimSci APC user interface.
If your controllers are running in the Active state, they may drop back to the Ready state
when a vital MV or CV goes to a bad state. You can run Director scripts or sequences to
automatically bring the controllers back to the Active state when the MVs and CVs return to a
good state. See Director Calculations (see "Overview" on page 233) for more information.
If your controllers are running in the Active or Ready state and you stop scanning, they drop
back to the Manual state.
To run SimSci APC as a service:
1. On the Home tab, in the APC as Service group, click On.
2. The following message appears. Click Yes to proceed.

Version 2016 SP2 43

SimSci APC User Guide Getting Started with SimSci APC

While the SimSci APC software runs as a service, you can close the user interface and the
software continues to run.
To stop running SimSci APC as a service:
1. Open the SimSci APC user interface.
2. On the Home tab, in the APC as Service group, click Off.

Enabling Auto Save for an APC Project

You can automatically save an APC Project by using the Auto Save feature. When you
enable Auto Save, the SimSci APC software automatically saves the APC Project at the
specified frequency, even while the APC Project is scanning. You can set the frequency to
any time length greater than or equal to 5 minutes. Note that you cannot change the
frequency while Auto Save is enabled.
To enable Auto Save:
On the Home tab, in the Auto Save group, in the Frequency box, type the frequency at
which you want to automatically save the APC Project, and then click On.

Engineering Units and Normalized Units

In the SimSci APC software, you can view various variables in terms of engineering units and
normalized units. The SimSci APC software uses engineering units for raw plant data and for
variables that it calculates by using raw plant data. It uses normalized units for normalized
plant data and for variables that it calculates by using normalized plant data. See
Normalization on page 128 for more information.

44 Version 2016 SP2

 SimSci APC User Guide

C HAPTER 2
Connecting to Data

In This Chapter
Overview...................................................................................................................................... 45
Connecting to an OPC Server........................................................................................................ 45
Connecting to an AIM*AP I Server .................................................................................................. 46
Adding a Group to a Connected OPC Server .................................................................................. 47
Editing a Group ............................................................................................................................ 47
Removing a Group from a Connected OPC Server ......................................................................... 48
Removing a Connected Server ...................................................................................................... 48
Tags ............................................................................................................................................ 48
The Data Table............................................................................................................................. 58

Overview
The first step in building a SimSci APC Project is to connect to data from the DCS.
The SimSci APC software acts as an OPC server client. It can connect to local or remote
OPC servers, which in turn communicate with the DCS. It can also connect to local or remote
AIM*API servers, which in turn communicate with the DCS by using the APC Bridge software.

Connecting to an OPC Server

You can connect to an OPC server by using the Interface tab. To open the Interface tab,
expand the Connect node in the APC Explorer, and then click Interface. The Interface tab
opens in the Page View.
You can use the Explore Servers tree to locate available OPC servers on local or remote
machines.
After you successfully connect to an OPC server, the OPC server appears in the Connected
Servers tree, under the host machines node. An indicator badge appears next to the name
of the OPC server in the Connected Servers tree. If the indicator is green, the connection to
the OPC server is good. If the indicator is red, the connection to the OPC server is bad.
To connect to an OPC server:
1. In the ExploreServers list, select OPC.
2. If the server is hosted locally, in the Explore Servers tree, expand the Local Host node,
and then browse to the OPC server to which you want to connect.

Version 2016 SP2 45

SimSci APC User Guide Connecting to Data

3. If the server is hosted on a remote machine and you know the IP address of that machine,
do the following:
a. Expand the Manual Host node.
b. Double-click the resulting node.
c. Type the IP address, and then press Enter.
d. Browse to the OPC server to which you want to connect.
4. If the server is hosted on a remote machine, expand the Remote Host node to explore
the machines on your network, and then browse to the OPC server to which you want to
connect.
5. Do one of the following:
o In the Explore Servers tree, right-click the OPC server, and then click Connect to
Server.
o In the Explore Servers tree, double-click the OPC server.
o Drag the OPC server from the Explore Servers tree into the Connected Servers
tree.

Connecting to an AIM*API Server

You can connect to an AIM*API server by using the Interface tab. To open the Interface tab,
expand the Connect node in the APC Explorer, and then click Interface. The Interface tab
opens in the Page View.
You must connect to an AIM*API server to use the APC Bridge software. This allows the APC
Project to connect to Foxboro software, including the following:
AOS
I/A Series
Foxboro Evo
FoxView
You can use the Explore Servers tree to locate available AIM*API servers on local or remote
machines.
After you successfully connect to an AIM*API server, the AIM*API server appears in the
Connected Servers tree, under the host machines node. An indicator badge appears next to
the name of the AIM*API server in the Connected Servers tree. If the indicator is green, the
connection to the AIM*API server is good. If the indicator is red, the connection to the
AIM*API server is bad.
To connect to an AIM*API server
1. In the ExploreServers list, select AIM API.
2. If the server is hosted locally, in the Explore Servers tree, expand the Local Host node,
and then browse to the AIM*API server to which you want to connect.
3. If the server is hosted on a remote machine and you know the IP address of that machine,
do the following:

46 Version 2016 SP2

Connecting to Data SimSci APC User Guide

a. Expand the Manual Host node.

b. Double-click the resulting node.
c. Type the IP address, and then press Enter.
d. Browse to the AIM*API server to which you want to connect.
4. Do one of the following:
o In the Explore Servers tree, right-click the AIM*API server, and then click Connect to
Server.
o In the Explore Servers tree, double-click the AIM*API server.
o Drag the AIM*API server from the Explore Servers tree into the Connected Servers
tree.

Adding a Group to a Connected OPC Server

You must add at least one Group to a connected OPC server to communicate with the tags
on the server.
A Group appears in the Connected Servers tree, under the OPC server to which you add it.
The selected Read Mode option for the Group appears in gray, parenthetical text next to the
name of the Group.
To add a Group to a connected OPC server:
1. On the Interface tab, in the Connected Servers tree, right-click the connected OPC
server, and then click Add Group. The Add Group dialog box appears.
2. In the Group Name box, type the name of the Group that you want to add.
3. In the Update Interval box, type the update interval, in seconds, for the Group.
4. Click the desired Read Mode option:
o Click Device Read to read tags directly from the DCS device.
o Click Cache Read to read tags from the temporary tag values stored on the OPC
server.

5. Click Add .

Editing a Group
You can edit a Group by changing the Read Mode option or the Update Interval value. You
cannot change the Group Name value. Thus, you cannot rename a Group after you add it to
the OPC server.
You can also change the Read Mode option by using features available on the Application
Ribbon.

Version 2016 SP2 47

SimSci APC User Guide Connecting to Data

Acti vity Directions

Editing a Group On the Interface tab, in the Connected Servers tree, right-click
the Group, and then click Edit Group. On the the Update Group
dialog box, change the values for the Update Interval parameter
and the Read Mode option as desired, and then click Update

Changing the Read Mode On the Interface tab, in the Connected Servers tree, select the
option for a Group desired Group. On the Configure tab, in the OP C DA group, click
Read Mode, and then click Device to read tags directly from the
DCS device or click Cache to read tags from the temporary tag
values stored on the OP C server.

Removing a Group from a Connected OPC Server

You can remove a Group from a connected OPC server by using the commands found on the
Application Ribbon. You cannot remove a Group from a connected OPC server if the Group
contains any tags.
To remove a Group from an OPC server:
1. On the Interface tab, in the Connected Servers tree, select the Group.
2. On the Configure tab, in the OPC DA group, click Remove Group. You can click
Remove Group regardless of which tab is currently displayed or opened in the Page
View. This command always removes the Group currently selected in the Connected
Servers tree.
3. A confirmation message appears. Click Yes.

Removing a Connected Server

You can remove a connected OPC or AIM*API server from an APC Project by using the
commands found on the Application Ribbon. You cannot remove an OPC server from the
APC Project if the OPC server contains any Groups.
To remove an OPC or AIM*API server:
1. On the Interface tab, in the Connected Servers tree, select the OPC or AIM*API server.
2. On the Configure tab, in the OPC DA group, click Remove Server. You can click
Remove Server regardless of which tab is currently displayed or opened in the Page
View. This command always removes the OPC or AIM*API server currently selected in
the Connected Servers tree.
3. A confirmation message appears. Click Yes.

Tags
Tags are variables in the APC Project that represent parameters in the DCS. After you add
an OPC Group to a connected OPC server, you can add tags to the project. You can do this
manually or automatically. After you add tags, you can move them between Groups or
remove them from the APC Project.
There are two types of tags:

48 Version 2016 SP2

Connecting to Data SimSci APC User Guide

Input tags: The plant measurements that the controller uses to determine the appropriate
actuation values for the plant. Typically, input tags are the CVs and FVs.
Output tags: The actuation values that the SimSci APC software sends to the plant.
Typically, output tags are the MVs.
Concepts
An OPC Group contains tags.
Tags correspond to DCS parameters.
Tags are either input tags or output tags.
Actions
Add tags manually or automatically.
Remove tags from the APC Project.
Map imported tags to existing tags in an OPC server.

Manually Adding Tags

You can manually add tags to the APC Project by using the Manual tab. To open the Manual
tab, click Manual in the APC Explorer, under the Connect node.
If you know the exact name of the tag that you want to add, you can use the Manual Tag
Entry area on the Manual tab to add the tag. After you add the tag, it appears in the Tag
Data Grid area.
To add a tag by using the Manual Tag Entry area:
1. Enter the name of the tag in the Tag ID box.
2. Click Input Tag if the tag is an input for the APC Project. Click Output Tag if the tag is an
output of the APC Project.

3. Click the Add button . Alternatively, press Enter.

If you do not know the exact name of the tag that you want to add, you can use the Tag
Entry: Filter and Browse area to locate the tag in the OPC server and add it to the APC
Project. There are two available browsing methods, Flat Browsing and Hierarchical
Browsing.

Version 2016 SP2 49

SimSci APC User Guide Connecting to Data

You can click Flat Browsing to browse the tags on the OPC server in a single list. The
SimSci APC software lists the tags alphabetically.

50 Version 2016 SP2

Connecting to Data SimSci APC User Guide

You can click Hierarchical Browsing to browse the tags on the OPC server in a tree
structure based on the hierarchical location of the tags. You can expand or collapse nodes as
desired. Click a node in the tree to view the tags contained in the node. The SimSci APC
software lists the tags in the node alphabetically.

You can filter tags based on their type and name:

To display available input tags, click Input Tag.
To display available output tags, click Output Tag.
To apply a filter, type the filter string into the Tag Filter box, and then click the Refresh

button . You can use wildcards to narrow or broaden the filter. When you use the
Hierarchical Browsing method, the SimSci APC software applies the filter to the
currently selected node in the tree, that is, the node outlined in green.
The SimSci APC software applies filters to all levels of the hierarchical tree, regardless of a
node's expanded or collapsed state. That is, it applies the search string at the node level and
again for all the associated tags for that node. If the associated tags do not contain the
search string you use to filter the nodes, you may need to change the search string to see the
tags in those nodes.
You can select multiple tags by using the Shift key or the Ctrl key. You can then add the
selected tags to the Group at the same time by using the same methods you use to add a
single tag to the Group. Note that you must hold the Shift key or Control key while you add
the selected tags.

Version 2016 SP2 51

SimSci APC User Guide Connecting to Data

To add a tag to a Group, do one of the following:

Right-click the tag, and then click Add Tag.
Double-click the tag.
Drag the tag into the Tag Data Grid area.

Auto Discovery
An OPC server may contain a very large number of tags. In some cases, manually adding
tags consumes a large amount of time and may be cumbersome to users. For such cases,
you can use Auto Discovery to streamline the process of adding tags to the APC Project.
Auto Discovery automatically adds tags to the APC Project based on user-specified tag
selection criteria for a particular make of DCS. Currently, only selection criteria for Foxboro
I/A Series DCSs are configured by default. See Configurations for the Selection Criteria on
page 56 for more information on configuring selection criteria for DCSs other than Foxboro
I/A Series.
You can access the Auto Discovery features on the AutoDiscovery tab. To open the
AutoDisovery tab, click Auto Discovery in the APC Explorer, under the Connect node.
There are three tabs on the AutoDiscovery tab:
Rule Book: Use this tab to configure and apply the selection criteria for tag addition.
Input Tags: Use this tab to view summaries of the input tags that you add to the APC
Project by using Auto Discovery.
Output Tags: Use this tab to view summaries of the output tags that you add to the APC
Project by using Auto Discovery.
The Rule Book tab contains the Rule Book, which you can use to manage Rulesets. A
Ruleset is a collection of rules that filter the tags based on their associated block type and tag
type; that is, a rule contains the selection criteria that you use to automatically add tags to an
APC Project.

Managing Rulesets
You can add, delete, edit, or apply a Ruleset by using the Rule Book on the AutoDiscovery
tab. You can edit a Ruleset by adding, deleting, or editing its rules.

Acti vity Directions

Adding a Ruleset
At the top of the Rule Book, click Add Ruleset . In the
RuleSet Name box, type the name for the new Ruleset, and then
click Add. The name must be unique to add the new Ruleset.

Deleting a Ruleset In the Rule Book, select the Ruleset that you want to delete, and

then click Delete Ruleset .

52 Version 2016 SP2

Connecting to Data SimSci APC User Guide

Editing a Ruleset In the Rule Book, select the Ruleset that you want to edit, and

then click Edit Rule set . A Ruleset window appears, in which

you can manage the rules in the Ruleset. See Rules on page 53
and Managing Rules on page 54 for more information.

Applying a Rules et In the Rule Book, select the Ruleset that you want to apply, and

then click Apply Ruleset . An Apply Rules window appears,

in which you can select the rules that you want to apply. See
Applying a Rules et on page 55 for more information.

Example: Deleting a Ruleset

1. In the Rule Book, select the Ruleset.

2. At the top of the Rule Book, click Delete Ruleset .

3. A confirmation message appears. Click Yes.

Rules
A rule is a set of selection criteria that you can use to filter tags in an OPC server. The SimSci
APC software maintains block input and output values in individual block parameters. It
organizes the blocks into groups, or compounds, which are contained within a single control
processor (CP). The structured hierarchical name for a CP data value, that is, a tag, is thus:
<Compound>:<Block>.<Parameter>
A typical tag example might be:
UNIT1:FIC1000.MEAS
You can filter these kinds of tags by using the rules. Each rule contains the following user-
specified selection criteria:
Rule Name: A user-specified name for the rule.
Tag Type: The type of tags that you want to add. You can select INPUT or OUTPUT.
Block Type: The type of Block that contains the tags. You can select any available input
or output block in the DCS. The selections in this list change based on the Tag Type
selection.
Block Name Filter: A filter that Auto Discovery applies to the Block portion of the tag
name. The SimSci APC software adds only tags with names that contain the specified
filter string.
DCS Parameter: The DCS Parameter that contains the value of interest. The SimSci
APC software adds tags for the selected DCS Parameter. The selections in this list
change based on the Block Type selection.
Description: The DCS Parameter that contains the description for a Block. The SimSci
APC software uses this DCS PArameter in conjunction with the Description Filter. The
selections in this list change based on the Block Type selection.

Version 2016 SP2 53

SimSci APC User Guide Connecting to Data

Description Filter: A filter that Auto Discovery applies to the value of the description DCS
Parameter for a Block. The SimSci APC software includes only Blocks that have a
description DCS Parameter that contains the filter string.
Units: The DCS Parameter that contains the engineering unit specification for the Block.
The selections in this list change based on the Block Type selection.
The SimSci APC software adds only tags that match all the specified selection criteria to the
APC Project.

Managing Rules
You can manage the rules in a Ruleset by using the Ruleset window. Rules appear in the
rules grid. Each rule is unique. No two rules can have the same name or the same Tag Type,
Block Type, and DCS Tag combination.
You can export the rules currently listed in the rules grid to a .Rules file. A .Rules file contains
information for all the selection criteria associated with the rules that you exported. You can
import the rules in a .Rules file into any APC Project, into any Ruleset.
When you import the rules in the .Rules file, the SimSci APC software adds the rules to the
rules grid (that is, the Ruleset), except in the following situations:
The rule in the .Rules file and an existing rule in the Ruleset have the same name. The
existing rule does not change during the import; that is, the SimSci APC software does
not overwrite the rule with the selection criteria contained in the .Rules file.
The Tag Type, Block Type, and DCS Tag selections for a rule in the .Rules file are the
same as the Tag Type, Block Type, and DCS Tag selections for another rule in the
Ruleset.

Acti vity Directions

Adding a rule In the pane on the left, select the desired selection criteria, and

then click Add Rule .

Editing a rule In the pane on the right, in the rules grid, select the rule that you

want to edit, and then click Edit Rule . In the pane on the
left, you can change the selection criteria for the rule. You must

click Save Rule to update the rule in the rules grid.

Deleting rules In the pane on the right, in the rules grid, select the rules that you

want to delete, and then click Delete Rule .

Exporting rules
In the pane on the right, click Export Rule s . You can save
the .Rules file to any location accessible by your machine. In the
File name box, type a name for the .Rules file, and then click
Save.

Importing rules
In the pane on the right, click Import Rules . Browse to and
select a previously saved .Rules file, and then click Open.

54 Version 2016 SP2

Connecting to Data SimSci APC User Guide

Example: Editing a Rule

1. In the pane on the right of the Ruleset window, in the rules grid, select the rule, and then

click Edit Rule . The selection criteria for the rule appears in the left pane.
2. In the left pane, change the selection criteria as desired.

3. In the pane on the right, click Save Rule to update the currently selected rule. This
overwrites all the selection criteria for the selected rule with the selection criteria currently
specified in the left pane, including the Rule Name selection criterion.

Applying a Ruleset
After you add rules to a Ruleset, you can apply the Ruleset to add tags to the APC Project.
You can selectively apply the rules in the Ruleset to any connected OPC server.
The SimSci APC software adds the tags to the AutoDiscoveryGroup Group under the select
OPC server. If the AutoDiscoverGroup Group does not exist, the SimSci APC software
creates it. You can view all the added input tags by using the Input Tags tab on the
AutoDiscovery tab. You can view all of the added output tags by using the Output Tags tab
on the AutoDiscovery tab.
To apply a Ruleset:
1. In the Rule Book, select the Ruleset.

2. At the top of the Rule Book, click Apply Ruleset . The Apply Rules dialog box
appears.
3. In the Connected OPC Server list, select the OPC server that contains the tags that you
want to add to the APC Project.
4. In the rules list, select the Select check box for each rule that you want to apply.
5. Click Discover to start the Auto Discovery process, which adds tags matching the
selection criteria for each applied rule to the APC Project.

Example of Automatically Adding Tags

In this example, we want to add all temperature-related tags found in the following block
types:
AIN
AOUT
PID
PIDA
To do this:
1. On the AutoDiscovery tab, in the Rule Book, add a Ruleset named Temperature Tags.

Version 2016 SP2 55

SimSci APC User Guide Connecting to Data

2. In the Rule Book, select the Temperature Tags Ruleset, and then click Edit Ruleset .
The Temperature Tags dialog box appears.
3. Add the following rules:

Rule Name Input1 Input2 Output1 Output2

Tag Type INP UT INP UT OUTP UT OUTP UT

Block Type AIN AOUT PID PIDA

Block Name Filter * * * *

DCS Parameter PNT MEAS OUT OUT

Description DESCRP DESCRP DESCRP DESCRP

Description Filter *TEMP* *TEMP* *TEMP* *TEMP*

Units EO1 EI1 EO1 EO1

4. Close the Temperature Tags dialog box.

5. In the Rule Book, select the Temperature Tags Ruleset, and then click Apply Ruleset

. The Apply Rules dialog box appears.

6. Click Select All, and then click Discover.
7. On the Input Tags tab, view the tags that you added to the APC Project by using Auto
Discovery.
8. On the Output Tags tab, view the tags that you added to the APC Project by using Auto
Discovery.

Configurations for the Selection Criteria

In the Ruleset dialog box, the SimSci APC software automatically loads the available
selections for each selection criterion from the Block_info.xml file located in the
<dir>\APC2016\Displays\ModuleData\APCGuiControls folder, where <dir> is the SimSci
APC installation directory. By default, the Block_info.xml file already includes configured
selection criteria for Foxboro I/A Series DCSs.
You can add block information (that is, available selections for a selection criterion) to the
Block_info.xml file to support tags from other DCSs or to support changes to the I/A Series
DCS blocks. For example, if a new attribute is introduced in the AIN block of an I/A Series
DCS, you can edit the Block_info.xml file to include that parameter attribute.
An example of the structure of the Block_info.xml file is:
<Block Type="AIN" TagType="INPUT">
<DCSTag>
<Param Selection="Default">PNT</Param>
</DCSTag>
<Description>
<Param Selection="Default">DESCRP</Param>

56 Version 2016 SP2

Connecting to Data SimSci APC User Guide

</Description>
<Units>
<Param Selection="Default">EO1</Param>
</Units>
<Interface0P>
<Param>LSCO1</Param>
</Interface0P>
<Interface100P>
<Param>HSCO1</Param>
</Interface100P>
<Engg0P>
<Param>LSCO1</Param>
</Engg0P>
<Engg100P>
<Param>HSCO1</Param>
</Engg100P>
</Block>
This structure directly corresponds to the selections available for a rule with a Tag Type of
INPUT and a Block Type of AIN.

Version 2016 SP2 57

SimSci APC User Guide Connecting to Data

Moving Tags to Different Groups

You can move single or multiple tags from one OPC Group to another by using the Manual
tab. This is especially useful for organizing tags that you add by using Auto Discovery.
To move tags to a different Group:
1. On the Manual tab, select the Group tab that contains the tags that you want to move.
2. In the Tag Data Grid area, select the tags that you want to move, and then right-click the
tag selection.
3. In the Move to Group submenu, click the Group into which you want to move the tags.

Removing Tags
You can remove single or multiple tags from the APC Project by using the Manual tab. You
can also remove tags by using a right-click command on the Data Table tab.
To remove tags from the APC Project:
1. On the Manual tab, select the Group tab that contains the tags that you want to remove.
2. In the Tag Data Grid area, select the tags that you want to remove, and then right-click
the tag selection.
3. Click Remove Tag.
4. A confirmation message appears. Click Yes.

The Data Table

The Data Table is a grid on the Data Table tab. It contains various attributes and values for
all the tags that you added to the APC Project. You can use the Data Table to:
Monitor all the tags in a single location.
Edit configuration data for the tags.
Define readback signals for output tags.
Replace tags with other tags in the same connected OPC server.
Map tags from an imported Connoisseur or SimSci APC file to existing tags in an OPC
server.
Export tags to a .tagconfig file.
Changes in the Data Table propagate throughout the APC Project.
Concepts
The Data Table provides a consolidated view of all the tags in the APC Project.
You can configure tags, including their associated readback signals, on the Data Table.
Actions
Monitor tags.
Change tag configurations.

58 Version 2016 SP2

Connecting to Data SimSci APC User Guide

Define readback signals for use in process response testing and controllers.
Replace tags.
Map imported tags to existing tags in an OPC server.
Export tags.

Replacing and Mapping Tags

You can replace a configured tag with a different tag in the same OPC server by using the
Data Table. When the replacement is successful, references to the replaced tag throughout
the APC Project update to use the new tag. This includes all tag references in:
Tag collections
PRBS Generators
Linear models
MVC controllers
Optimizers
Note: Hard-coded tag names in Director scripts do not update when you replace tags.

WARNING: When you replace a tag, all archived data for the tag is lost.
You can also map a tag that you imported from a set of model data and controller data (.dat)
files to an existing tag in an OPC server. If you want your imported tags to connect to current
process data, you must map the imported tags to existing tags. Before you map the imported
tags to existing tags, you must move the imported OPC Group that contains the imported
tags to an existing, connected OPC server. See Importing Data on page 20 for more
information. When the mapping is successful, references to the imported tag throughout the
APC Project update to use the existing tag. This includes all tag references in:
Tag collections
PRBS Generators
Linear models
MVC controllers
Optimizers
Note: You can generate the model data and controller data (.dat) files from either the
Connoisseur software or the SimSci APC software. For Connoisseur data files, the imported
tags have the same names as the Connoisseur signals, which do not have a direct naming
correlation to existing tags in an OPC server. You may need to refer to the signal
configurations within the Connoisseur software to determine the best mapping for the
imported tags.
You can replace or map only one tag at a time.
To replace or map a tag:
1. In the Data Table, select the tag.
2. Right-click the tag, and then click Map Tag. A dialog box appears.

Version 2016 SP2 59

SimSci APC User Guide Connecting to Data

3. If you know the name of the tag with which you want to replace or map the current tag,
select the first option, and then type the name of that tag.
4. If you do not know the name of the tag with which you want to replace or map the current
tag, select the second option. In the associated list, select the name of the tag with which
you want to replace or map the current tag.
5. Click Map.

Editing Data
For input tags, you can change the values for any column in the Data Table except the
following columns:
Quality
Tag
Value
Trend
OPC Group
OPC Server
Machine
For output tags, you can change the values for any column in the Data Table except the
following columns:
Quality
Tag
Trend
OPC Group
OPC Server
Machine
For custom input tags, you can change the values for any column in the Data Table except
the following columns:
Quality
Tag
Value
Trend
For custom output tags, you can change the values for any column in the Data Table except
the following columns:
Quality
Tag
Trend

60 Version 2016 SP2

Connecting to Data SimSci APC User Guide

To edit the data in a column:

1. Double-click the column.
2. Type the new value.
3. Press Enter.

Creating a Tag Configuration File from the Data Table

You can create SimSci APC tag configuration (.tagconfig) files from the Data Table by using
the Export Tag Configurations button.
For ease of use, we recommend that you create the file in the ...\APC2016\User\Data folder,
which is the default folder. The SimSci APC software automatically opens to this folder when
you import any data files.
To create a tag configuration (.tagconfig) file:

1. Click on the Export Tag Configurations button .

2. Name the file and click Save.

Version 2016 SP2 61

 SimSci APC User Guide

C HAPTER 3
Process Response Testing

In This Chapter
Overview...................................................................................................................................... 63
Readback Signals ......................................................................................................................... 64
PRBS Generators ......................................................................................................................... 64
Creating a PRBS Generator .......................................................................................................... 65
Configuring a P RBS Generator ...................................................................................................... 65
Operating a PRBS Generator ........................................................................................................ 70
Deleting a PRBS Generator........................................................................................................... 73
A vailable PRBSs .......................................................................................................................... 74

Overview
You can generate process response data in two ways, both of which are available with the
SimSci APC software:
Manual step tests applied to process Manipulated Variables (MVs)
Statistical sequence tests, such as Pseudo-Random Binary Sequence (PRBS) tests
The statistical sequence test method is the more efficient method of generating suitable data
with minimum plant disturbance. You can perform both manual step tests and PRBS tests by
using PRBS Generators.
When a response test is complete, you can then process the process response data. See
Overview on page 77 for more information.
You must specify readback signals for output tags that you add to a PRBS Generator. See
Readback Signals on page 64 for more information.
The following tasks are performed to generate process response data:
1. Create a PRBS Generator.
2. Configure the PRBS Generator.
3. Activate the PRBS Generator to start the response test.
4. Monitor the response test.
5. Deactivate the PRBS Generator to terminate the response test.

Version 2016 SP2 63

SimSci APC User Guide Process Response Testing

Readback Signals
A readback signal is an input tag that monitors the actual value of a field point that is updated
by its associated output tag. Typically, you use readback signals to implement tracking and
initialization functions. You can set a readback signal for each output tag by using the
ReadBack Signal list on the Data Table tab, on the Output Tags tab.
You can use raw input tags or custom input tags as readback signals.
A PRBS Generator does not initialize past a Waiting state if an output tag in the PRBS
Generator does not have a set readback signal.
Concepts
Readback signals are required for process response testing.
Readback signals are input tags that are associated with output tags. They monitor the
actual plant values for output tags.
Readback signals are defined in the Data Table.
Actions
Set the readback signal for an output tag.
Use an output tag with a defined readback signal in a PRBS Generator.

PRBS Generators
A PRBS Generator is a specialized controller that does not execute any form of control.
Instead, it deliberately moves its associated MVs and monitors the response of selected CVS
to the changes in the MVs. The MVs are the inputs to the process that is being controlled and
the CVs are the outputs of the process that is being controlled.

You can use a PRBS Generator to perform a manual step test or a PRBS test. In the case of
a manual step test, you are required to manually change values for the MVs. In the case of a
PRBS test, the PRBS Generator automatically changes the MVs between two specified
values. See Creating a PRBS Generator on page 65 and Configuring a PRBS Generator on
page 65 for more information.
Concepts
PRBS Generators are specialized controllers you can use to perform process response
tests.
You can perform manual step test or PRBS tests by using PRBS Generators.

64 Version 2016 SP2

Process Response Testing SimSci APC User Guide

Actions
Create a PRBS Generator.
Configure a PRBS Generator.
Activate a PRBS Generator.
Deactivate a PRBS Generator.

Creating a PRBS Generator

You can create a PRBS Generator by using the Configuration tab. To open the
Configuration tab, click Configuration in the APC Explorer, under Process Test.
To create a PRBS Generator:

1. In the PRBS Generators area, click Add PRBS Generator .

2. In the PRBS Generator column, type the name of the PRBS Generator.
3. In the Control Interval column, type the control interval. See Control Interval on page 66
for more information.
4. In the Random Interval column, type the random interval. See Random Interval on page
67 for more information.
5. Press Enter.

Configuring a PRBS Generator

You can configure a PRBS Generator by using the Configuration tab. To open the
Configuration tab, click Configuration in the APC Explorer, under Process Test.
Each PRBS Generator has the following configuration components:
A set of input tags (that is, a set of associated CVs)
A set of output tags (that is, a set of associated MVs)
A control interval
A random interval
A set of configuration parameters for each output tag
You can change these components at any time. If the APC Project is scanning, the changes
you make to a PRBS Generator configuration immediately take effect.
To completely configure a PRBS Generator:
1. Add input tags and output tags to the PRBS Generator.
2. If desired, change the control interval and random interval for the PRBS Generator.
3. In the tags area, on the Output Tags tab, configure the output tags for the PRBS
Generator. This includes setting the following parameters for each output tag:
o Auto on page 67
o PRBS Length on page 68

Version 2016 SP2 65

SimSci APC User Guide Process Response Testing

o Mean on page 69
o Increment on page 69
o Amplitude on page 69
o Low Limit on page 70
o High Limit on page 70

Adding Tags to a PRBS Generator

You can add any number of input tags and output tags to a PRBS Generator. You must add
at least one output tag to a PRBS Generator to operate it.

Acti vity Directions

Adding input tags to a PRBS In the Tag Li st area, on the Input Tags tab, select the tags you
Generat or want to add, and then drag the selection int o the Generator tags
area, on the Input Tags tab. Alternatively, you can right-click the
selection, and then click the desired P RBS Generator in the Add
To Generator submenu to add the tags to that PRBS Generator.

Adding output tags to a PRBS In the Tag Li st area, on the Output Tags tab, select the tags you
Generat or want to add, and then drag the selection int o the Generator tags
area, on the Output Tags tab. Alternatively, you can right-click
the selection, and then click the desired PRBS Generator in the
Add To Generator submenu to add the tags to that PRBS
Generat or.

Example: Adding Input Tags to a PRBS Generator

1. In the Tag List area, on the Input Tags tab, select the tags you want to add.
2. Do one of the following:
o Drag the selection into the Generator tags area, on the Input Tags tab.
o Right-click the selection, and then click the desired PRBS Generator in the Add To
Generator submenu.

Control Interval
The control interval is the frequency at which the PRBS Generator computes the next move
for its associated output tags (that is, its associated MVs).
The control interval must be a multiple of the scan interval. You can set the control interval by
using the Configuration tab. See Configuring a PRBS Generator on page 65 for more
information.
This interval applies to all the tags in the PRBS Generator. It is used only for PRBS tests and
does not apply to manual step tests.

66 Version 2016 SP2

Process Response Testing SimSci APC User Guide

Random Interval
The random interval (T) is the least number of seconds for which the MVs hold at a particular
value. The duration that the MVs hold at a particular value is a pseudo-random multiple of the
random interval.

The random interval must be a multiple of the control interval. You can set the random
interval by using the Configuration tab. See Configuring a PRBS Generator on page 65 for
more information.
This interval applies to all the tags in the PRBS Generator. It is used only for PRBS tests and
does not apply to manual step tests.

Configuration Parameters for Output Tags in a PRBS Generator

Output tags represent the MVs of the process. You can configure the output tags by using the
Output Tags tab on the Configuration tab, in the Generator tags area.
The Output Tags tab contains a grid displaying the following configuration parameters for
each output tag:
Auto on page 67
PRBS Length on page 68
Mean on page 69
Increment on page 69
Amplitude on page 69
Low Limit on page 70
High Limit on page 70

Auto
The Auto check box specifies whether a manual step test or a PRBS test is performed for the
output tags. You can select the Auto check box to perform a PRBS test for the respective
output tag. By default, the Auto check box is cleared for each newly added output tag.

Version 2016 SP2 67

SimSci APC User Guide Process Response Testing

PRBS Length
The PRBS Length value is an input to the algorithm that generates the pseudo-random
binary sequence (PRBS). The SimSci APC software uses this value as the "seed" of the
random number generator. Thus, the PRBS Length value determines the pattern of the
PRBS. That is, it defines the number of random intervals that pass before the sequence
pattern of random changes repeats exactly. The SimSci APC software calculates this number
according to the following equations:
2n 1
N 2n1 1

t T 2n 1 1
where
n is the specified PRBS length
is the number of times the variable value changes in the sequence pattern
N is the number of random intervals in the sequence pattern
T is the random interval
t is the total time that passes for each complete sequence pattern
For example, a PRBS Length value of 2 results in the following sequence pattern:

Note that is 3 and N is 7. If the random interval is 4 seconds, the sequence pattern
repeats every 28 seconds. The resulting PRBS looks similar to the following figure:

You can set the PRBS Length value to any integer value between 0 and 32. Typically, you
set this value to integer values greater than 9, because the statistical properties of the
resulting PRBS approximate those of white noise.

68 Version 2016 SP2

Process Response Testing SimSci APC User Guide

You should set the PRBS Length for each output tag to a unique value. PRBSs that have
different PRBS Length values are hardly correlated with each other. When you carry out
PRBS tests with a number of output tags, it is important that the PRBSs for those outputs are
not correlated. Otherwise, when you subsequently analyze the PRBS data, the SimSci APC
software cannot separate out the various causes and effects. See Correlation Analysis on
page 89 for more information.

Mean
The Mean value is used to define the high and low values for the PRBS. The high value of
the PRBS is greater than the Mean value by an amount that is half the Amplitude value. The
low value of the PRBS is less than the Mean value by an amount that is half the Amplitude
value.

Increment
The Increment value is the total amount by which an output tag value can change for any
one move of an MV.
For manual step tests, the total change in an output tag does not exceed the Increment
value. If you attempt to apply a step change that is greater than the Increment value, the
current value changes by only the Increment value.
For PRBS tests, a step change of the full Increment value is usually applied when there is a
significant difference between the current value and the Mean value for the output tag. The
step change is applied every random interval until the current value approaches the Mean
value. After the current value approaches the Mean value, the specified Amplitude value
determines the step changes to the current value.

Amplitude
The Amplitude value is the amount by which an output tag value changes about the Mean
value for one move of an output tag value.

The Amplitude value cannot be greater than the Increment value.

Version 2016 SP2 69

SimSci APC User Guide Process Response Testing

Low Limit
The Low Limit value is the lower constraint on the output tag value. A PRBS Generator
cannot move an output tag to a value less than the Low Limit value. If the PRBS for an
output tag is configured to have a low value that is less than the Low Limit value, the Low
Limit value overrides the configured low value for the PRBS.

High Limit
The High Limit value is the upper constraint on the output tag value. A PRBS Generator
cannot move an output tag to a value greater than the High Limit value. If the PRBS for an
output tag is configured to have a high value that is greater than the High Limit value, the
High Limit value overrides the configured high value for the PRBS.

Operating a PRBS Generator

You can operate a PRBS Generator by using the Operation tab. To open the Operation tab,
click Operation in the APC Explorer, under Process Test.
The Operation tab contains tabs for each PRBS Generator configured in the Configuration
tab. Each PRBS Generator tab contains two areas:
Summary: The summary area contains information about the control interval, random
interval, and current status for the PRBS Generator. It also contains a list box and a
Change Status button, which you can use to change the status of the PRBS Generator.
Tags: The tags area is divided into two tabs, Input Tags and Output Tags.
The Input Tags tab displays the parameters for the input tags used in the PRBS testing.
These parameters are read-only.
The Output Tags tab displays the parameters for the output tags used in the PRBS
testing. This includes the mean, amplitude, and increment for each output tag, which you
should set before you operate the PRBS Generator.
The typical workflow for operating a PRBS is as follows:
1. Start scanning. See Scanning on page 43 for more information.
2. Initialize the PRBS Generator
3. Activate the PRBS Generator
4. View Trends
5. Deactivate the PRBS Generator

State of a PRBS Generator

The state of a PRBS Generator is the current mode of operation for the PRBS Generator. A
PRBS Generator can be in one of five states:
Manual: The PRBS Generator does not perform any calculations and does not write
values to the output tags.
Waiting: The PRBS Generator performs steps to transition to an initialized state.

70 Version 2016 SP2

Process Response Testing SimSci APC User Guide

Initialize: The PRBS Generator loads initial values for the output tags based on values
read from the readback signals specified on the Data Table tab, on the Output Tags tab.
Ready: The PRBS Generator writes values to the output tags based on values read from
the readback signals. It also calculates the Mean value based on the values read from the
readback signals. The Increment and Amplitude values do not apply.
Active: The PRBS Generator writes values to the output tags based on its configuration.
See Configuring a PRBS Generator on page 65 for more information.
If you configure the output tags in a PRBS Generator for manual step tests, you can
change the Current Value value manually to change the output values.
If you configure the output tags in a PRBS Generator for PRBS tests, the PRBS
Generator automatically changes the output values such that it creates a PRBS for the
output tags. The configuration for the PRBS Generator defines the PRBS for each output
tag. You can change the Mean, Amplitude, and Increment values for an output tag to
change the output values that the PRBS Generator automatically generates. Note that
you can use a combination of manual step tests and PRBS tests in a single PRBS
Generator.
You can use the Current Status field on the Operation tab, on the desired PRBS Generator
tab, to determine the current state of a PRBS Generator.

Initializing a PRBS Generator

You can initialize a PRBS Generator by using the Operation tab. When you initialize a PRBS
Generator, it transitions through the following states:
Waiting
Initialize
Ready
Once a PRBS Generator reaches the Ready state, you can activate it.
Note that the APC Project must be scanning before you can initialize a PRBS Generator.
To initialize a PRBS Generator:
1. On the Operation tab, on the desired PRBS Generator tab, select Initialize in the status
list box.

2. Click Change Status .

Activating a PRBS Generator

You can activate a PRBS Generator by using the Operation tab. You can activate only PRBS
Generators that are in the Ready state. When you activate the PRBS Generator, it transitions
to the Active state. In this state, the PRBS Generator writes values to the output tags based
on its configuration. You can perform manual step tests and PRBS tests for only PRBS
Generators that are in an Active state.
To activate a PRBS Generator:
1. On the Operation tab, on the desired PRBS Generator tab, select Active in the status list
box.

Version 2016 SP2 71

SimSci APC User Guide Process Response Testing

2. Click Change Status .

Performing a Manual Step Test

A manual step test is a method of generating process response data in which you are
required to manually change the value of the process MV, that is, the output tag.
The total change in the values of the output tags cannot exceed the Increment value. If you
attempt to apply a step change that is greater than the Increment value, the current values of
the output tags change by only the Increment value. See Configuration Parameters for
Output Tags in a PRBS Generator on page 67 for more information.
To perform a manual step test for an output tag:
1. Activate the PRBS Generator.
2. On the Configuration tab, on the Output Tags tab, clear the Auto check box for the
output tag.
3. On the Operation tab, on the Output Tags tab, click the Current Value value.
4. Type a new value for the Current Value value, and then press Enter to change the value
of the output tag.
5. Repeat steps 3 and 4 numerous times until adequate response data is collected.

Performing a PRBS Test

A PRBS test is a method of generating process response data in which a PRBS Generator
automatically changes the value of the process MV, that is, the output tag, between two
specified values.
The pattern of the PRBS is determined by the PRBS Length value. The Mean and
Amplitude values determine the high and low values for the PRBS. A step change of the full
Increment value is usually applied when there is a significant difference between the current
value and the Mean value for the output tag. The step change is applied for every random
interval until the current value approaches the Mean value. See Configuration Parameters for
Output Tags in a PRBS Generator on page 67 for more information.
To perform a PRBS test for an output tag:
1. Activate the PRBS Generator.
2. On the Configuration tab, on the Output Tags tab, select the Auto check box for the
output tag.
3. Change the value of the following configuration parameters as desired to change the
PRBS:
o PRBS Length
o Mean
o Amplitude

Viewing Trends from a PRBS Generator

You can view sparklines for the tag values in the Trend column on the Operation tab, on
either the Input Tags tab or the Output Tags tab. For further analysis, you can view trends
for the input tags or the output tags.The trends open in a Trend window.

72 Version 2016 SP2

Process Response Testing SimSci APC User Guide

On the Input Tags tab, you can select only one input tag for which to view trends. When you
view the trends for an input tag in a PRBS Generator, the following trends are added to the
Trend window:
The trend for the selected input tag
The trends for all the output tags in the PRBS Generator
On the Output Tags tab, you can select only one output tag for which to view trends. When
you view the trends for an output tag in a PRBS Generator, the following trends are added to
the Trend window:
The trend for the selected output tag
The trends for all the input tags in the PRBS Generator
Alternatively, you can use the Data Table tab to view trends for the input tags and output
tags. This is especially useful if you want to see multiple input tags and output tags together
in the same Trend window. See Trends and Strips on page 28 for more information.

Acti vity Directions

Viewing the trends for an input On the Operation tab, on the Input Tags tab, select the input tag
tag for which you want to view trends. Right-click the Trend column
for the tag, and then click Generate Trend(s).

Viewing the trends for an On the Operation tab, on the Output Tags tab, select the output
output tag tag for which you want to view trends. Right-click the Trend
column for the tag, and then click Generate Trend(s).

Deactivating a PRBS Generator

You can deactivate a PRBs Generator by using the Operation tab. Alternatively, you can
deactivate the PRBS Generator, as well as any other active or initialized controllers, by
stopping scanning. When you stop scanning, the PRBS Generator is automatically set to the
Manual state.
To deactivate a PRBS Generator:
1. On the Operation tab, on the desired PRBS Generator tab, select one of the following in
the status list box:
o Ready: Returns the PRBS Generator to the Ready state. The PRBS Generator still
writes values for the output tags.
o Initialize: Reinitializes the PRBS Generator and returns it to the Ready state. The
PRBS Generator still writes values for the output tags.
o Manual: Returns the PRBS Generator to the Manual state. The PRBS Generator
does not write values for the output tags.

2. Click Change Status .

Deleting a PRBS Generator

You can delete a PRBS Generator by using the Configuration tab. To open the
Configuration tab, click Configuration in the APC Explorer, under Process Test.

Version 2016 SP2 73

SimSci APC User Guide Process Response Testing

To delete a PRBS Generator:

1. In the PRBS Generators area, select the PRBS Generator you want to delete.

2. Click Delete PRBS Generator .

3. A confirmation message appears. Click Yes.

Available PRBSs
When you use PRBS tests to generate process response data, you can specify a PRBS
Length value for each output tag that you add to a PRBS Generator. The value of the PRBS
Length determines the pattern of the PRBS; the random interval of the PRBS Generator
determines the length of the PRBS. See Random Interval on page 67 and PRBS Length on
page 68 for more information.
The following table displays the patterns for the available PRBSs. The purpose of this table is
to give you an idea of what the PRBS patterns look like so that you can better select the
PRBS Length values that fit your needs. The graphs that appear in this table encompass
approximately 5 minutes of data. The PRBS Generator that generated this data has a
Control Interval of 1.00 and a Random Interval of 2.00. Note that for PRBS Length values
greater than 6, the full pattern of the PRBS does not complete within the time shown. The
PRBSs that you see in your APC Project may not look exactly the same as the graphs in the
table, but they should look similar.

PRBS Pattern
Length

74 Version 2016 SP2

Process Response Testing SimSci APC User Guide

10

11

12

13

14

15

Version 2016 SP2 75

 SimSci APC User Guide

C HAPTER 4
Data Processing

In This Chapter
Overview...................................................................................................................................... 77
The Translation Table ................................................................................................................... 77
Creating SimSci APC Tag Data and Comma Delimited Files ........................................................... 78
Creating a Tag Configuration File from the Data Table .................................................................... 79
Managing Custom Tags ................................................................................................................ 79
Managing Tag Collections ............................................................................................................. 80
Tag Data Manipulation .................................................................................................................. 82
Correlation Analysis ...................................................................................................................... 89

Overview
Data processing is an offline tool that is used to configure and manipulate data. You can also
use the data processing features to determine mathematical correlations in the data.
Typically, the tag correlation analysis is performed by using data obtained from process
response tests. This can include manipulated data or raw data. See Process Response
Testing (see "Overview" on page 63) for more information on process response tests.
The typical workflow for data manipulation is as follows:
1. Create the custom tags used to archive the manipulated data.
2. Create and manage the tag collections used to quickly access tags.
3. Define the calculations used to manipulate the data.
4. Run the calculations and view the resulting data with respect to the raw data.

The Translation Table

The Translation Table is a list of translations between the OPC server and the APC project.
Each translation converts a specified tag value in an OPC server to a corresponding tag
value in the APC Project. Typically, you use the Translation Table to translate string values in
an OPC server to numerical values in the APC project. For this reason, the translations are
also known as string translations.
You can open the Translation Table by clicking Translation Table on the Configure tab, in
the OPC DA group.
The string translations in the Translation Table apply to all values downloaded into the APC
Project and all values uploaded to the OPC server. That is, the translations apply to both
input tags and output tags.

Version 2016 SP2 77

SimSci APC User Guide Data Processing

Acti vity Directions

Adding a string translation to

the Translation Table In the Translation Table, click Add String Translation .

Editing a string translation in In the Translation Table, double-click either the OP C Server
the Translation Table Value column or the AP C Value column for the string translation
you want to edit. Type the new value, and then press Enter.

Deleting string translations In the Translation Table, select the string translations you want to
from the Translation Table delete, and then click Delete String Translation .

Example: Editing a String Translation in the Translation Table

1. In the Translation Table, double-click the OPC Server Value column for the string
translation you want to edit.
2. Type the OPC server value you want to translate, and then press Enter.
3. Double-click the APC Value column for the string translation you want to edit.
4. Type the value to which you want to translate the OPC server value, and then press
Enter.

Creating SimSci APC Tag Data and Comma Delimited Files

You can create SimSci APC tag data (.tagData) and comma delimited (.fdax) files by using
the All Tags tab. These files include only the tags you select. You cannot include both raw
tags and custom tags in the same file. After you create a tag data or .fdax file, you can import
the file into any APC Project to start working with the archived data. See Importing Data on
page 20 for more information.
For ease of use, we recommend that you create the file in the ...\APC2016\User\Data folder,
which is the default folder. The SimSci APC software automatically opens to this folder when
you import any data files.
To create a tag data or comma delimited file:
1. On the All Tags tab, in either the Raw Tag(s) area or the Custom Tag(s) area, select the
tags you want to include in the file.
2. Right-click the selection, and then click Export. The Export Data dialog box appears.
3. In the Start Date box, type the starting time of the data range that you want to export.

Note that you can click the down arrow in the box to select and change the date based
on a calendar view.
4. In the End Date box, type the ending time of the data range that you want to export. Note

that you can click the down arrow in the box to select and change the date based on a
calendar view.
5. In the File Type box, click the down arrow to select either a .tagData file or .fdax file type.
6. Click Export.The Export Data dialog box appears.
7. Browse to the folder in which you want to create the tag data file.
78 Version 2016 SP2
Data Processing SimSci APC User Guide

8. In the File name box, type the name for the tag data file, and then click Save.
Note: You can export tags in an .fdax with different start and end times. Any data gaps will be
filled with a value of -999.

Creating a Tag Configuration File from the Data Table

You can create SimSci APC tag configuration (.tagconfig) files from the Data Table by using
the Export Tag Configurations button.
For ease of use, we recommend that you create the file in the ...\APC2016\User\Data folder,
which is the default folder. The SimSci APC software automatically opens to this folder when
you import any data files.
To create a tag configuration (.tagconfig) file:

1. Click on the Export Tag Configurations button .

2. Name the file and click Save.

Managing Custom Tags

Custom tags are user-defined tags that are not added to the project from a connected OPC
server or from an imported data file. They are created within the APC Project and remain
internal to the APC Project. Typically, they have values that are calculated from input tags or
output tags. A calculated custom tag value is defined by using data manipulation or Director
calculations. See Tag Data Manipulation on page 82 and Director Calculations (see
"Overview" on page 233) for more information.
Note: You may create a custom tag while scanning and then check the box in the Calculate
Per Scan column to start updating the tag. See Creating a Calculation for a Custom Tag on
page 87 for more information.
You can manage custom tags by using the Custom Tag(s) area on the All Tags tab. To
open the All Tags tab, click All Tags in the APC Explorer, under Data Processing.

Acti vity Directions

Creating a custom tag

In the Custom Tag(s) area, click Add Custom Tags .

Deleting a custom tag In the Custom Tag(s) area, select the custom tag you want to
delete, and then click

Delete Custom Tag Collections .

Renaming a custom tag In the Custom Tag(s) area, double-click the Custom Tag
column for the tag, type a new name for the tag, and then press
Enter.

Entering a description for a In the Custom Tag(s) area, double-click the Description column
custom tag for the tag, type a description for the tag, and then press Enter.

Version 2016 SP2 79

SimSci APC User Guide Data Processing

Setting the UOM for a custom In the Custom Tag(s) area, double-click the Unit column for the
tag tag, type a unit of measure (UOM) for the tag, and then press
Enter.

Changing the tag type for a In the Custom Tag(s) area, in the Tag Type column, select
custom tag either Input or Output to specify the custom tag as an input tag
or as an output tag, respectively.

Converting a custom tag into a Open the Data Table tab under the Connect node in the APC
raw tag Explorer. On the Custom Input Tags or Custom Output Tags
tab, right-click on the custom tag(s ) you want to convert to a raw
tag and select Convert To RawTag.

Note: If a custom tag is being used in a custom calculation, it

cannot be converted to a raw tag.

Example: Deleting a Custom Tag

1. In the Custom Tag(s) area, select the custom tag you want to delete.

2. Click Delete Custom Tag Collections .

3. A confirmation message appears. Click Yes.

Managing Tag Collections

Tag collections are collections of input and output tags that you can use in various tabs of the
APC Project. Tag collections are useful for limiting tag displays and lists to only the tags of
interest. For example, you can create a tag collection that contains only the tags in a certain
process area of the plant to easily locate tags that you want to manipulate or correlate.
You can manage tag collections by using the All Tags tab. To open the All Tags tab, click
All Tags in the APC Explorer, under Data Processing.
The All Tags tab contains five areas:
Raw Tag(s): Contains a list of tags that have been added to the APC Project from either
connected OPC servers or imported data files.
Custom Tag(s): Contains a list of tags that have been created to hold manipulated data
values.
Tag Collection: Contains a list of the collections that have been created in the APC
Project.
Tag List: Contains a list of the tags contained in the collection currently selected in the
Tag Collection area.
Controller Collection: Contains a list of the controllers that have been created in the
APC project.
You can export all log channel data for a controller from the Controller Collection area.
Right-click on the controller and select Export. You can then save the controller data as a
.tagData or .fdax file.
You can create or delete a tag collection by using the Tag Collection area on the All Tags
tab. You cannot change or delete the Default Collection, which contains all the raw and
custom tags in the APC Project.

80 Version 2016 SP2

Data Processing SimSci APC User Guide

You can add tags in the Raw Tag(s) area and in the Custom Tag(s) area to a tag collection.
You can remove tags from a tag collection by using the Tag List area on the All Tags tab.

Acti vity Directions

Creating a tag collection In the Tag Collection area, click Add Tag Collections .

Renaming a tag collection In the Tag Collection area, double-click the Collection Name
column for the tag collection, type the new name for the tag
collection, and then press Enter.

Adding a description for a tag In the Tag Collection area, double-click the De scription column
collection for the tag collection, type a description for the tag collection, and
then press Enter.

Adding tags to a tag collection In either the Raw Tag(s) area or the Custom Tag(s) area, select
the tags you want to add, and then drag the selection into the
Tag List area. Alternatively, you can right-click the selection, and
then click the desired tag collection in the Add to Collection
submenu to add the tags to that tag collection.

Removing tags from a tag In the Tag Collection area, select the tag collection from which
collection you want to remove tags.Iin the Tag Li st area, select the tags
that you want to remove. Right-click the selection, and then click
Delete Tag(s).

Exporting the tag dat a for a tag In the Tag Collection area, right-click the tag collection, and then
collection click Export. You can then save the tag data for all the tags in the
tag collection as a SimSci APC tag data (.tagData) file. See
Creating SimSci APC Tag Data and Comma Delimited Files on
page 78 for more information.

Deleting a tag collection In the Tag Collection area, select the tag collection you want to

delete, and then click Delete Tag Collections . Alternatively,

you can right-click the tag collection, and then click Delete Tag
Collection.

Example: Adding Tags to a Tag Collection

1. In the Raw Tag(s) area, select the tags you want to add, and then do one of the following:
o Drag the selection into the Tag List area.
o Right-click the selection, and then click the desired tag collection in the Add to
Collection submenu.
2. In the Custom Tag(s) area, select the tags you want to add, and then do one of the
following:
o Drag the selection into the Tag List area.
o Right-click the selection, and then click the desired tag collection in the Add to
Collection submenu.

Version 2016 SP2 81

SimSci APC User Guide Data Processing

Tag Data Manipulation

Typically, you perform data manipulation after you create and configure the custom tags and
the tag collections. You use the custom tags to archive the manipulated data. You use the tag
collections to quickly access the custom tags, input tags, and output tags that you want to use
in the data manipulation.
You can manipulate tag data by using the Data Manipulation tab. To open the Data
Manipulation tab, click Data Manipulation in the APC Explorer, under Data Processing.
The Data Manipulation tab has two areas:
Formula Editor: Used to create and manage the customized equations.
Custom tags list: Used to create and configure the calculations for the custom tags.

Default Equations
A default equation is a predefined equation that you can use to manipulate raw data and
calculate the data that is stored in a custom tag.
Default equations are defined in the Formula Editor. You cannot edit or delete these
equations from the APC Project. These equations are not included in the exported .csv file
when you export the customized equations in the Formula Editor. See Managing Customized
Equations on page 86 for more information.
There are 15 default equations that are automatically included in an APC Project:
ADD on page 83
SUB on page 83
MUL on page 83
DIV on page 83
LOG on page 83
EXP on page 83
SQRT on page 83
DIFF on page 83
LOG10 on page 84
LOW PASS FILTER on page 84
AVERAGE on page 84
SMOOTH on page 84
REMOVE SPIKES on page 84
SHIFT LEFT on page 85
SHIFT RIGHT on page 85
XY Lookup on page 85

82 Version 2016 SP2

Data Processing SimSci APC User Guide

ADD
The ADD equation uses two variables, X and Y. The value of the X variable is added to the
value of the Y variable for each sampled instant in the entire analysis range. The resulting
values are used for the values of the custom tag.

SUB
The SUB equation uses two variables, X and Y. The value of the Y variable is subtracted
from the value of the X variable for each sampled instant in the entire analysis range. The
resulting values are used for the values of the custom tag.

MUL
The MUL equation uses two variables, X and Y. The value of the X variable is multiplied by
the value of the Y variable for each sampled instant in the entire analysis range. The resulting
values are used for the values of the custom tag.

DIV
The DIV equation uses two variables, X and Y. The value of the X variable is divided by the
value of the Y variable for each sampled instant in the entire analysis range. The resulting
values are used for the values of the custom tag.

LOG
The LOG equation uses one variable, X. It calculates the natural logarithm of the value of the
X variable for each sampled instant in the entire analysis range. The resulting values are
used for the values of the custom tag.

EXP
The EXP equation uses one variable, X. It calculates the natural exponential function for the
value of the X variable for each sampled instant in the entire analysis range. The resulting
values are used for the values of the custom tag.

SQRT
The SQRT equation uses one variable, X. It calculates the square root of the value of the X
variable for each sampled instant in the entire analysis range. The resulting values are used
for the values of the custom tag.

DIFF
The DIFF equation uses one variable, X. The previous value of the X variable is subtracted
from the current value of the X variable for each sampled instant in the entire analysis range.
The resulting values are used for the values of the custom tag.
The first value of the custom tag is set to zero to indicate that a previous value of the X
variable does not exist in the data set for this data point.

Version 2016 SP2 83

SimSci APC User Guide Data Processing

LOG10
The LOG10 equation uses one variable, X. It calculates the base-10 logarithm of the value of
the X variable for each sampled instant in the entire analysis range. The resulting values are
used for the values of the custom tag.

LOW PASS FILTER

The LOW PASS FILTER equation uses one variable, Tag, and one constant, TC. It applies a
low pass digital filter to the values of the Tag variable and passes the results to the custom
tag. This equation reduces any measurement noise that may be obscuring any significant
underlying variations in the data. The following filter algorithm is used for the calculations:

xkf xkf1 xk e
T
TC
xk
where
T is the frequency, in seconds, at which the data for the Tag variable is sampled
TC is the filter time constant, TC, in seconds
xk
is the value of the Tag variable at instant k
xkf
is the value of the custom tag, that is, the filtered value, at instant k

AVERAGE
The AVERAGE equation uses one variable, Tag, and one constant, NoS. It calculates the
moving average of the value of the Tag variable for each set of sampled instants in the entire
analysis range. The maximum number of data points in each set of sampled instants is equal
to the specified NoS value. The resulting values are used for the values of the custom tag.

SMOOTH
The SMOOTH equation uses one variable, Tag, and one constant, NoS. It calculates the
average of the values of the Tag variable for each set of sampled instants in the entire
analysis range. The maximum number of data points in each set of sampled instants is equal
to the specified NoS value. Each set of sampled instants contains:
A number of past values of the Tag variable
The current value of the Tag variable
A number of future values of the Tag variable.
The number of past values of the Tag variable is always equal to the number of future values
of the Tag variable. Therefore, you must specify an odd number for the NoS value.

REMOVE SPIKES
The REMOVE SPIKES equation uses three variables, Tag, LL, and HL. It applies a lower
limit, LL, and an upper limit, HL, on the values of the Tag variable to remove any spikes that
are present in the data. The resulting values are used for the values of the custom tag.
For each sampled instant in the entire analysis range, the following logic determines the
value of the custom tag:

84 Version 2016 SP2

Data Processing SimSci APC User Guide

If the value of the Tag variable is less than the LL value, the value of the custom tag is set
equal to the LL value.
If the value of the Tag variable is greater than the HL value, the value of the custom tag is
set equal to the HL value.
If the value of the Tag variable is greater than the LL value and less than the HL value,
the value of the custom tag is set equal to the current value of the Tag variable.

SHIFT LEFT
The SHIFT LEFT equation uses two variables, Tag and Shift. It shifts the values of the Tag
variable to the left, and then copies the values of the Tag variable to the values of the custom
tag. The Shift value specifies the time, in seconds, by which the values of the Tag variable
are shifted to the left.
A shift to the left is equivalent to shifting the time reference of the data points for the Tag
variable so that they correspond to an earlier instant in time. A number of values in the
custom tag do not have corresponding values of the Tag variable due to the time shift in the
data range. The SHIFT LEFT equation sets these values equal to zero.

SHIFT RIGHT
The SHIFT RIGHT equation uses two variables, Tag and Shift. It shifts the values of the Tag
variable to the right, and then copies the values of the Tag variable to the values of the
custom tag. The Shift value specifies the time, in seconds, by which the values of the Tag
variable are shifted to the right.
A shift to the right is equivalent to shifting the time reference of the data points for the Tag
variable so that they correspond to a later instant in time. A number of values in the custom
tag do not have corresponding values of the Tag variable due to the time shift in the data
range. The SHIFT RIGHT equation sets these values equal to zero.

XY Lookup
The XY Lookup equation uses one variable, Tag, and a lookup table that defines a series of
X and Y values used to transform the Tag data values. The XY Lookup equation applies a
piecewise nonlinear transform function to the Tag data values and passes the results to the
custom tag.
See Piecewise Nonlinear Transforms on page 193 for an explanation of how SimSci APC
software fits the Tag data values to the defined X and Y values.

Customized Equations
A customized equation is a user-defined equation that you can use to manipulate data. It is
used to calculate the data that is stored in a custom tag.
Customized equations are defined by using the Formula Editor. The Formula Editor includes
a number of default equations that you can use to create and configure the calculations for
the custom tags. See Default Equations on page 82 for more information. If you cannot find
an appropriate equation among the default equations, you can create a customized equation
by using any number of variables and available functions.
A custom tag contains the same number of data points as the input tags and output tags that
you use in the customized equation to calculate the values for the custom tag.

Version 2016 SP2 85

SimSci APC User Guide Data Processing

Managing Customized Equations

You can manage the customized equations by using the Formula Editor in the Data
Manipulation tab.
When you create a customized equation, you can use the buttons below the Formula box to
quickly include the following functions and notations:
Addition
Subtraction
Multiplication
Division
Open Parenthesis
Close Parenthesis

You can change the Formula for a customized equation by using the Formula Editor. You can
edit only customized equations that are not used in a calculation for a custom tag. You cannot
change the name of a customized equation after you create it.
You can export a set of customized equations for later use or for use in another APC Project.
Exported customized equations are saved to a .csv file.
You can import a set of customized equations that are stored as a .csv file into an APC
Project. Imported customized equations are added at the bottom of the equation list in the
Formula Editor. If an imported customized equation already exists in the APC Project, a copy
of the customized equation is added to the equation list.

You can delete only customized equations that are not used in a calculation for a custom tag.

Acti vity Directions

Creating a customized In the Formula Editor, in the Name box, type the name of the
equation customized equation. In the Formula box, type the equation, and

then click Add Formula .

86 Version 2016 SP2

Data Processing SimSci APC User Guide

Editing a customized equation In the Formula Editor, in the equation list, double-click the
Equation column for the customized equation, type the new
formula for the customized equation, and then press Enter.

Exporting customized In the Formula Editor, select the Select Formula check box for
equations each equation that you want to export, and then click Export

Formula(s) . You can then save the customized equations to

a .csv file.

Importing customized
equations In the Formula Editor, click Import Formula(s) . Browse to
and select the .csv file cont aining the customized equations, and
then click Open.

Deleting customized equations In the Formula Editor, select the Select Formula check box for
each equation that you want to delete, and then click Delete

Formula . Alternatively, you can press Delete instead of

clicking Delete Formula.

Example: Exporting Customized Equations

1. In the Data Manipulation tab, in the Formula Editor, select the Select Formula check
box for each equation that you want to export.

2. Click Export Formula(s) . The Export Formulas dialog box appears.

3. Browse to the folder in which you want to save the .csv file containing the customized
equations.
4. In the File name box, type the name of the .csv file.
5. Click Save.

Creating a Calculation for a Custom Tag

You can create a calculation for a custom tag by using the custom tag list area in the Data
Manipulation tab.
To create a calculation for a custom tag:

1. In the calculations grid, click Add Calculation .

2. In the Custom Tag column, select the custom tag that you want to calculate.
3. In the Formula Name column, select the customized equation that you want to use for
the calculation.

4. In the Define column, click the Define button to configure the calculation. See
Configuring a Calculation for a Custom Tag on page 88 for more information.

Version 2016 SP2 87

SimSci APC User Guide Data Processing

5. In the Calculate Per Scan column, select the check box to automatically recalculate the
custom tag values at every scan internal while the APC Project is scanning. You should
select this check box if you want to use the custom tag in an MVC controller and
Optimizer.
6. In the Comments column, type any comments you may have about the equation.
7. Press Enter.

Configuring a Calculation for a Custom Tag

You can configure a calculation for a custom tag by using the custom tag list area on the
Data Manipulation tab.
To configure a calculation for a custom tag:

1. In the calculations grid, in the Define column, click the Define button . The Equation
dialog box appears.
2. In the Collection Name list, select the tag collection that contains the tags that you want
to map to the variables in the equation.
3. If you want to map a variable to a tag value, do the following:
a. In the process tags grid, select the tag that you want to map to a variable in the
equation.
b. Right-click the selected tag, and then click the variable to which you want to map the
tag.
c. Check the parameters grid or the Formula Equation field to ensure the mapping is
successful.
4. If you want to set a constant value for a variable, in the parameters grid, in the Tag/Value
column, type the value to which you want to set the variable.
5. Repeat steps 3 and 4 as desired until all the variables in the equation are mapped to tags
or set to constant values.
6. Click OK.

Running Calculations for the Custom Tags

You can run the calculations for the custom tags by using the custom tags list area on the
Data Manipulation tab. The results appear in a trend. One trend appears for each custom
tag that is calculated. The Strips for the following tags are added to each trend:
The calculated custom tag
The input tags used in the calculation for the custom tag
The output tags used in the calculation for the custom tag

You can export the trend data by clicking Export Data on the trend. The exported data is
saved as a .csv file. You can then open the .csv file in Microsoft Excel for further analysis of
the data.

88 Version 2016 SP2

Data Processing SimSci APC User Guide

To run calculations for the custom tags:

1. In the custom tags list area, in the Select column, select the check box for each
calculation that you want to run.

2. Click Run Calculation(s) .

Deleting Calculations for the Custom Tags

You can delete a calculation for a custom tag by using the custom tag list area on the Data
Manipulation tab.
To delete calculations for the custom tags:
1. In the custom tags list area, in the Select column, select the check box for each
calculation that you want to delete.

2. Click Delete Calculation .

3. A confirmation message appears. Click Yes.

Correlation Analysis
Correlation analysis is a statistical tool used to analyze the dependencies in the dynamic
variation of two signals. The two tags analyzed are termed cause tags and effect tags.
The correlation may be of two types:
Autocorrelation: The same tag is defined as both the cause tag and the effect tag. The
result reflects the statistical properties of that tag alone. See Correlation Coefficients on
page 89 and Autocorrelation Function on page 90 for more information.
Cross-correlation: Two different tags are defined as the cause tag and the effect tag.
The result quantifies their statistical dependency. See Correlation Coefficients on page 89
and Cross-correlation Function on page 90 for more information.
You can perform a correlation analysis by using the Data Mining tab. To open the Data
Mining tab, click Data Mining in the APC Explorer, under Data Processing.

Correlation Coefficients
A correlation coefficient, x, y , is a measure of the overall deviation around the linear fit,
where x and y are the two variables in question. This measure is also known as Pearsons
correlation coefficient, or Pearsons r. The correlation coefficient is calculated as follows:

x x y
i i y
x, y i

x x y y
2 2
i i
i i

where
x is the mean value of the variable x
y is the mean value of the variable y
The correlation coefficient can take any value between +1 and 1:

Version 2016 SP2 89

SimSci APC User Guide Data Processing

A correlation coefficient of +1 indicates a perfect, positive linear correlation. All of the data
points lie exactly on a line, which has a positive slope.
A correlation coefficient of 1 indicates a perfect, negative linear correlation. All of the
data points lie exactly on a line, which has a negative slope.
A correlation coefficient of 0 indicates that no correlation is present. The data points are
randomly distributed around a line.
Correlation coefficients of a value other than +1, 1, or 0 indicate varying strengths of
correlation.
The SimSci APC software calculates the value of the correlation coefficient for a range of
different offsets between the cause and effect variation. This range is 10% of the current
analysis range, that is, the total number of data points archived for the tags. It then plots the
result as a graph of the value of the correlation constant versus the offset. See Generating
Trends for the Correlation Analysis Calculations on page 95 for more information.

Autocorrelation Function
The autocorrelation function of a tag indicates the extent to which the value of the tag at one
instant in time is related to the values of the same tag at any other subsequent instant. It
provides a rough measure of the time constant of the dynamics that are associated with a
tag. Therefore, you can use it as an aid to selecting the interval at which the tag should be
monitored for control purposes.
Rxx
The discrete autocorrelation function, , can be expressed as follows:
Rxx xt , xt , N , ,1,0,1, , N
where
a, b is the correlation between the variable a and b .

xt
is the variable of interest

is the same variable with a time shift of time steps applied

xt

N is the number of data points in the analysis range.

Cross-correlation Function
The cross-correlation function, which relates one tag to another, indicates the extent to which
the value of the cause tag, at one instant in time, determines the value of the effect tag at any
subsequent instant in time. It measures whether there is any significant correlation between
the two tags.
The cross-correlation function also represents the impulse response for a single-input, single-
output (SISO) dynamic system that involves the cause and effect tags if the following
statements are true:
The data collected for the cause tag has properties that approximate those of white noise.
One example of white noise is a tag manipulated as a PRBS. See Process Response
Testing (see "Overview" on page 63) for more information.
The dynamics of the effect tag are approximately linear.

90 Version 2016 SP2

Data Processing SimSci APC User Guide

Rxy
A discrete cross-correlation function, , can be expressed as follows:
Rxy xt , yt , N , ,1,0,1,, N
where
a, b is the correlation between the variable a and b .

xt
is the cause variable with no time shift applied.

is the effect variable shifted by time steps

yt

N is the number of data points in the analysis range.

Differencing
In the SimSci APC software, differencing is the act of using the increments from one data
point to the next to perform calculations instead of the absolute tag values.
By default, the correlation function is calculated by using the absolute values of the cause
and effect tags, that is, the values expressed in normal engineering units. However, in certain
situations, this absolute form of correlation does not accurately represent the behavior of the
system and could lead to unsatisfactory or misleading results for the correlation analysis. This
is particularly true for integrating systems, where the system response may be a varying rate
of change. In such situations, you can perform the correlation analysis by using differencing
for the tags to get a more accurate and satisfactory result. You can use differencing for both
the cause tag and the effect tag or for either the cause tag or effect tag individually.
Differencing does not alter the original tag data. When you use differencing for a tag, you
change the correlation function such that it uses the difference between the previous archived
tag value and the current tag value for each value in the data set. You cannot apply
differencing to the first archived tag value in the data set because a previous archived value
does not exist for the first archived tag value. The number of data points that are used in the
correlation analysis calculation therefore decreases by 1.
The cross-correlation function with differencing applied to the cause variable can be
expressed as follows:
Rxy xt , yt , N ,,1,0,1,, N
where
a, b is the correlation between the variable a and b .
x is the cause variable
xt xk xk 1 , k 1, , N 1

is the effect variable shifted by time steps

yt

N is the number of data points in the analysis range.

Version 2016 SP2 91

SimSci APC User Guide Data Processing

Cause Tags and Effect Tags

A cause tag is a variable that causes a change in other variables when its value changes.
Typically, cause tags are manipulated variables (MVs) and feedforward variables (FVs). You
can specify a tag as a cause tag by adding the tag to the Cause Tags collection in the Data
Mining tab. See Managing Tags in the Cause Tags and Effect Tags Collections on page 92
for more information.
An effect tag is a variable that changes due to changes in other variables. Typically, effect
tags are controlled variables (CVs). You can specify a tag as an effect tag by adding the tag
to the Effect Tags collection in the Data Mining tab. See Managing Tags in the Cause Tags
and Effect Tags Collections on page 92 for more information.

Managing Tags in the Cause Tags and Effect Tags Collections

You can manage the tags in the Cause Tags and Effect Tags collections by using the Data
Mining tab. You can add tags of all variable classes (that is, MVs, FVs, and CVs) to both the
Cause Tags collection and the Effect Tags collection. If you want to perform an
autocorrelation analysis for a tag, you must add the tag to both the Cause Tags collection
and the Effect Tags collection.
The Cause Tags and Effect Tags collections contain the tags that can be specified as cause
tags and effect tags in the Correlation Analysis area, respectively. You can use only the
tags that are included in either the Cause Tags collection or the Effect Tags collection as the
tag selections in the Correlation Analysis area.

Acti vity Directions

Adding tags to the Cause In the Collection Name list box, select the collection that
Tags collection contains the tags of interest. In the Proce ss Tag column, select
the tags that you want to add to the collection, and then drag the
selection into the Cause Tags list. Alternatively, you can right-
click the selection, and then click Cause Tags in the Add To
Collection submenu to add the selection to the Cause Tags list.

Adding tags to the Effect Tags In the Collection Name list box, select the collection that
collection contains the tags of interest. In the Proce ss Tag column, select
the tags that you want to add to the collection, and then drag the
selection into the Effect Tags list. Alternatively, you can right-
click the selection, and then click Effect Tags in the Add To
Collection submenu to add the selection to the Effect Tags list.

Removing tags from the In the Cause Tags list, select the tags you want to remove from
Cause Tags collection the collection. Right-click the selection, and then click Delete.

Removing tags from the Effect In the Cause Tags list, select the tags you want to remove from
Tags collection the collection. Right-click the selection, and then click Delete.

Example: Adding Tags to the Cause Tags Collection

1. In the Data Mining tab, in the Collection Name list, select the collection that contains the
tags of interest.
2. In the Process Tag column, select the tags that you want to add to the collection.

92 Version 2016 SP2

Data Processing SimSci APC User Guide

3. Do one of the following:

o Right-click the selection, and then click CauseTags in the Add To Collection
submenu.
o Drag the selection into the Cause Tags list.

Creating a Correlation Analysis Calculation

A correlation analysis calculation is a SimSci APC object containing a specified cause tag
and effect tag pairing for which a correlation analysis is performed. It also contains the
configuration for the differencing that is applied to its tags. You can create a correlation
analysis calculation by using the Correlation Analysis area in the Data Mining tab. Each
row in the grid in this area is a correlation analysis calculation.
In the Correlation Analysis area, the entries in the Cause Tags column link to the Cause
Tags collection. You can select any tag in the Cause Tags collection as the cause tag for a
correlation analysis calculation. The entries in the Effect Tags column link to the Effect Tags
collection. You can select any tag in the Effect Tags collection as the effect tag for a
correlation analysis.
If you want to perform an autocorrelation analysis for a tag, you must select the tag in both
the Cause Tags column and the Effect Tags column for the correlation analysis calculation.
See Correlation Coefficients on page 89 and Autocorrelation Function on page 90 for more
information on the calculations that are performed.
To create a correlation analysis calculation:

1. In the Correlation Analysis area, click Add Data Mining .

2. In the Cause Tags column, select the cause tag you want to use for the correlation
analysis.
3. In the Effect Tags column, select the effect tag you want to use for the correlation
analysis.
4. In the Difference Option column, select the differencing to apply to the tags:
o None: Do not apply differencing.
o Both: Apply differencing to both the cause tag and the effect tag.
o Cause: Apply differencing to the cause tag only.
o Effect: Apply differencing to the effect tag only.
5. Press Enter.

Running Correlation Analysis Calculations

You can run correlation analysis calculations by using the Correlation Analysis area on the
Data Mining tab.
You cannot run a correlation analysis calculation if it has an analysis rangethat is, a start
time and end time for its data setthat is different from the analysis range for the other
correlation analysis calculations defined in the Correlation Analysis area.

Version 2016 SP2 93

SimSci APC User Guide Data Processing

The results of the run for a correlation analysis calculation are displayed as sparklines in the
Correlation Analysis area, in the Result column. These sparklines are miniatures of the
trends for the correlation analysis calculations. See Generating Trends for the Correlation
Analysis Calculations on page 95 for more information.
To run the correlation analysis calculations:
1. In the Correlation Analysis area, in the Select column, select the check box for each
correlation analysis calculation that you want to run.

2. Click Run .

Deleting Correlation Analysis Calculations

You can delete correlation analysis equations by using the Correlation Analysis area in the
Data Mining tab.
To delete correlation analysis calculations:
1. In the Correlation Analysis area, click the area to the left of the Select column for a
correlation analysis calculation to select the correlation analysis calculation.
2. If desired, select more than one correlation analysis calculation by using the Shift and Ctrl
keys.

3. Click Delete Data Mining . Alternatively, you can press Delete.

4. A confirmation message appears. Click Yes.

94 Version 2016 SP2

Data Processing SimSci APC User Guide

Generating Trends for the Correlation Analysis Calculations

You can generate trends for the correlation analysis calculations by using the Correlation
Analysis area on the Data Mining tab. These trends are the correlation functions calculated
during the runs for the correlation analysis calculations (see Autocorrelation Function on page
90 and Cross-correlation Function on page 90). Each trend is named based on its associated
cause tag and effect tag. They display in a Trend window as stripped trends. You can add
this Trend window to the Trend Manager.

Each time you generate a trend for a correlation analysis calculation, the SimSci APC
software performs a run for the correlation analysis calculation. See Running Correlation
Analysis Calculations on page 93 for more information.
To generate a trend for the correlation analysis calculations:
1. In the Correlation Analysis section, in the Select column, select the check box for each
correlation analysis calculation for which you want to generate a trend.

2. Click Display Trends .

Characteristics of the Trends for the Correlation Analysis

Calculations
The y-axis of the trends has a maximum range of -1 to +1. Its effective range is always
automatically scaled to provide the best view of the correlation function. The values on the y-
axis correspond to the values for the correlation function.

Version 2016 SP2 95

SimSci APC User Guide Data Processing

The x-axis of the trends is centered on zero and has a range of N 10 to N 10 , where N
is the number of data points in the analysis range. The positive half of the x-axis represents a
positive time shift ( ) for the tag data, that is, the cause tag leads, or drives, the effect tag.
The negative half of the x-axis represents a negative time shift for the tag data, that is, the
effect tag leads, or drives, the cause tag. The negative side is useful for assessing whether
there is any significance to be read from the correlation function on the positive side of the x-
axis. If the correlation function on the positive half of the x- axis does not stand out from the
correlation function on the negative half of the x-axis, either there is no significant relationship
between the two tags, or the data analyzed is not excited enough to provide a basis for
determining the required information.

If you generate multiple trends for the correlation analysis calculations, the x-axes for each
trend are the same as the x-axis for the most recently run correlation analysis calculation.
The values on the x-axis are measured in units of time shift and not in units of time. If you
want to calculate the elapsed time at a certain point on the x-axis, multiply the value on the x-
axis by the frequency at which the data points were taken. For example, if a data set is
sampled at a frequency of 2 seconds, the elapsed time at an x-value of 15 (in absolute terms)
is 30 seconds; that is, the fifteenth data point in the set is taken after 30 seconds have
elapsed.

96 Version 2016 SP2

Data Processing SimSci APC User Guide

You can use the peak values of the correlation function as useful measures of the
interdependence between the tags. The closer the peak is to 1, the more confidence there is
that the correlation is significant. The time at which the peak occurs indicates the time that
elapses before the impact of the one tag on the other is at its greatest. As a rule-of-thumb,
you can use peak values that are greater than 0.2 (in absolute terms) to indicate a significant
correlation. However, your interpretation of the results for the correlation analysis calculation
should also consider the Granger index calculated for the run. See Displaying the Statistical
Results for the Correlation Analysis Calculations on page 97 for more information.

Displaying the Statistical Results for the Correlation Analysis

Calculations
You can display the statistical results for the correlation analysis calculations by using the
Correlation Analysis area in the Data Mining tab.
The statistical results include the following parameters for the correlation analysis calculation:
PeakMaxNegativeValueTime on page 97
PeakMaxNegativeValue on page 98
PeakMaxPositiveValueTime on page 98
PeakMaxPositiveValue on page 98
Granger Index on page 98
Suggested Prediction Interval on page 99
Dead Time on page 99
Suggestion on page 99
You must perform a run for a correlation analysis calculation in order to display its statistical
results. The statistical results are displayed in the Correlation Statistics Table window.

To display the statistical results for the correlation analysis calculations:

1. In the Correlation Analysis area, in the Select column, select the check box for each
correlation analysis calculation for which you want to display the statistical results.

2. Click Display Statistics .

PeakMaxNegativeValueTime
The PeakMaxNegativeValueTime value is the time at which the greatest peak occurs for the
negative half of the correlation function. It indicates the time that elapses before the impact of
the effect tag on the cause tag is at its greatest.

Version 2016 SP2 97

SimSci APC User Guide Data Processing

PeakMaxNegativeValue
The PeakMaxNegativeValue value is the value of the correlation function at the greatest
peak for the negative half of the correlation function. It provides a useful measure of the
interdependence between the tags. The closer the peak is to 1, the more confidence there is
that the correlation is significant. A PeakMaxNegativeValue value that is greater than the
Granger Index value indicates a significant correlation.
A PeakMaxNegativeValue value that is greater than the PeakMaxPositiveValue value
indicates that there is a stronger correlation between the variables when the specified cause
tag acts as the effect variable and the specified effect tag acts as the cause variable. If this is
the case, we recommend that you swap the cause and effect tags.

PeakMaxPositiveValueTime
The PeakMaxPositiveValueTime value is the time at which the greatest peak occurs for the
positive half of the correlation function. It indicates the time that elapses before the impact of
the cause tag on the effect tag is at its greatest.

PeakMaxPositiveValue
The PeakMaxPositiveValue value is the value of the correlation function at the greatest peak
for the positive half of the correlation function. It provides a useful measure of the
interdependence between the tags. The closer the peak is to 1, the more confidence there is
that the correlation is significant. A PeakMaxPositiveValue value that is greater than the
Granger Index value indicates a significant correlation.
A PeakMaxPositiveValue value that is greater than the PeakMaxNegativeValue value
indicates that there is a stronger correlation between the variables when the specified cause
tag acts as the cause variable and the specified effect tag acts as the effect variable. If this is
the case, we recommend that you keep the cause and effect tags in their current
configuration.

Granger Index
The Granger Index value is the threshold value above which we consider the cause tag to be
significant to the effect tag. It provides a quantitative method of excluding small correlation
peaks from consideration and is estimated as:
2
GrangerIndex
Nc 2
where
Nc
is the number of data points used in the correlation analysis.
Any value of the correlation function that is greater (in absolute terms) than the Granger
Index value is statistically significant at a 95 % confidence level. Alternatively, you can
consider any value of the correlation function that is less (in absolute terms) than the
Granger Index value to be effectively zero, with 95% confidence.

98 Version 2016 SP2

Data Processing SimSci APC User Guide

Suggested Prediction Interval

The Suggested Prediction Interval value is the recommended maximum Prediction
Interval value for a linear model containing the cause tag and effect tag for the correlation
analysis calculation. The Suggested Prediction Interval value is calculated as follows:
t1, RHS
SuggestedPredictionInterval
4
where
t1, RHS
is the time at which the earliest significant peak on the positive half of the correlation
function occurs.

Dead Time
The Dead Time value is the time at which the earliest significant peak on the positive half of
the correlation function occurs.
If both the cause and effect tags are differenced (that is, you select Both for the Difference
Option selection for the correlation analysis calculation), the Dead Time value is the same as
or similar to the pure dead time of the cause and effect relationship. That is, it indicates the
time that elapses before the impact of the cause tag on the effect tag is at its greatest.
If both the cause and effect tags are not differenced (that is, you select Cause, Effect, or
None for the Difference Option selection for the correlation analysis calculation), the Dead
Time value is set to zero and does not represent the pure dead time of the cause and effect
relationship.

Suggestion
The Suggestion value is a message that indicates whether or not the cause tag and the
effect tag for the correlation analysis calculation should be changed. If it is blank, we
recommend that you do not change the cause tag and the effect tag. If it contains the
"CAUSE and EFFECT tags should be swapped" message, we recommend that you change
the cause tag to be the current effect tag and change the effect tag to be the current cause
tag.
The Suggestion value is set based on which half of the correlation function, positive or
negative, contains the earliest significant peakthat is, the peak with the least x-value in
absolute terms. If the earliest significant peak occurs on the positive half of the correlation
function, the Suggestion value is left blank. If the earliest significant peak occurs on the
negative half of the correlation function, the Suggestion value is set to the "CAUSE and
EFFECT tags should be swapped" message.

Version 2016 SP2 99

 SimSci APC User Guide

C HAPTER 5
Modeling

In This Chapter
Overview.................................................................................................................................... 101
Linear Models ............................................................................................................................. 101
Scenarios for a Linear Model ....................................................................................................... 106
Managing Scenarios ................................................................................................................... 106
Selecting Data for Scenarios ....................................................................................................... 108
Scenario Parameters .................................................................................................................. 110
Trans fer Functions ...................................................................................................................... 113
Identification Algorithms .............................................................................................................. 117
Results of a Linear Model ............................................................................................................ 129
Composite Linear Model ............................................................................................................. 142

Overview
In the SimSci APC software, a model is a mathematical representation of the dynamics of the
process plant that provides a basis for predicting the process behavior. Models are used to
configure model-based predictive control strategies for the process plant in question. You can
establish a model by applying statistical identification procedures to plant-sampled
information. Typically, you collect such information by scanning tag data, but you may also
attain this information by importing a Connoisseur tag data (.fdax) file or a SimSci APC tag
data (.tagData) file. See Importing Data on page 20 for more information.
Currently, you can create only linear models in an APC Project.

Linear Models
A linear model is a mathematical representation of the dynamics of the process plant. It uses
linear modeling techniques to predict the process behavior. The cause and effect
relationships between the tags within a linear model are assumed to be linear or can be
approximated as linear. Linear models include the following:
Autoregressive models with exogenous inputs (ARX models)
Fixed Impulse Response (FIR) models
Transfer function models
State space models

Version 2016 SP2 101

SimSci APC User Guide Modeling

Currently, you can use only ARX models, FIR models, and transfer function models in an
APC Project. See Identification Algorithms on page 117 and Transfer Functions on page 113
for more information.
You can perform linear modeling by using the Linear Modeling tab. To open the Linear
Modeling tab, click Linear Modeling in the APC Explorer, under Modeling.

Creating a Linear Model

You can create a linear model by using the model tabs on the Linear Modeling tab.
The first time that you open the Linear Modeling tab, the SimSci APC software automatically
creates a linear model named Model1. You can use this model for the linear modeling. You
can also create your own models in addition to the initial linear model, Model1.
To create a new model:
1. On the Linear Modeling tab, select any model tab.
2. Right-click the selected model tab, and then click Add Model.

Creating an Integrating Linear Model

You can create an integrating linear model by using the model tabs on the Linear Modeling
tab.
An integrating linear model is the same as a linear model except for the following aspects:
All scenarios in and integrating model have an Order of 1. You cannot change this value.
The ARX coefficient for all the CVs is initially -1 in the results of the scenarios
identification.
Note that the identification results of an integrating model are different from the results of a
non-integrating model for the same configured scenarios.
If a linear model is an integrating linear model, the "Integrating Model" identifier appears at
the top of the Variables area on the associated model tab.
To create a new integrating linear model:
1. On the Linear Modeling tab, select any model tab.
2. Right-click the selected model tab, and then click Add Integrating Model.

Copying a Linear Model

You can create a copy of a linear model by using the associated model tab on the Linear
Modeling tab. You cannot copy a linear model that does not have at least one MV, FV, or CV
within its tag structure.
A copy of a linear model contains the same tag structure and scenarios as the original linear
model. A copy of a linear model is named Copy(<CopyNumber>) of <ModelName>, where
<CopyNumber> is the identification number of the copy with regards to the other copies of the
original linear model and <ModelName> is the name of the original linear model.
For example, if you make three copies of a linear model named Model1, the SimSci APC
uses the following naming strategy:

102 Version 2016 SP2

Modeling SimSci APC User Guide

The first copy is named Copy of Model1.

The second copy is named Copy(1) of Model1.
The third copy is named Copy(2) of Model1.

To copy a linear model:

1. On the Linear Modeling tab, select the associated model tab for the linear model that
you want to copy.
2. Right-click the selected model tab, and then click Copy Model.

Deleting a Linear Model

You can delete a linear model by using the associated model tab on the Linear Modeling
tab.
To delete a linear model:
1. On the Linear Modeling tab, select the associated model tab for the linear model that
you want to delete.
2. Right-click the selected model tab, and then click Delete Model.

Renaming a Linear Model

You can rename a linear model by using the associated model tab on the Linear Modeling
tab.
You cannot leave the model name blank or change the name of the model to be the same as
another existing model. If the SimSci APC software cannot complete the renaming, it prints
an Information message in the Diagnostics Window that indicates the error or discrepancy.
To rename a model:
1. On the Linear Modeling tab, double-click the title of the associated model tab.
2. Type the new name for the model.
3. Press Enter.

Changing the Order of the Linear Models

You can change the order in which the linear models appear in the Linear Modeling and
Results tabs by moving the associated model tabs on the Linear Modeling tab. When you
change the order of the linear models on the Linear Modeling tab, you also change the order
of the linear models on the Results tab. You can quickly access the linear models that are
the most important or the most interesting to you by moving those linear models to the front of
the list.

Version 2016 SP2 103

SimSci APC User Guide Modeling

To change the order of the linear models:

1. In the Linear Modeling tab, drag the model tab that you want to move onto another
model tab. The model tab that you move always appears to the immediate left of the
linear model onto which you drag it.
2. Repeat step 1 as desired.

Exporting the Data for a Linear Model

You can export the data for a linear model by using the associated model tab on the Linear
Modeling tab.
The SimSci APC software saves the exported data for a linear model in a model data
(.ModData) file.
To export the data for a linear model:
1. On the Linear Modeling tab, select the model tab for the linear model for which you want
to export data.
2. Right-click the selected model tab, and then click Export Model Data. The Export Model
Data dialog box appears.
3. Browse to the folder in which you want to save the model data file.
4. In the File name box, type the name of the model data file.
5. Click Save.

Configuring the Structure of a Linear Model

The structure of a linear model is the setup for the input and output variables for the model.
The input and output variables for a model are divided into three categories:
Controlled Variables (CVs): These are the output variables of both the plant process
and the model. The purpose of the model is to predict the future variation of the CVs so
that the SimSci APC software can ultimately control the CVs. The CVs are either raw
input tags or custom input tags.
Manipulated Variables (MVs): These are the input variables to both the plant process
and the model. You can use the SimSci APC software to deliberately modify the MVs.
Typically, the MVs are setpoints of regulatory PID controllers. They are either raw output
tags or custom output tags.
Feedforward Variables (FVs): These are the input variables to both the plant process
and the model. You cannot use the SimSci APC software to deliberately modify the FVs.
The FVs are also commonly referred to as disturbance variables. They can be raw input
tags, raw output tags, custom input tags, or custom output tags.
You can add any number of CVs, MVs, and FVs to a linear model. You can add variables to a
model by using the pane on the left of the Linear Modeling tab. The SimSci APC software
always adds variables to the model whose tab is currently selected in the pane on t he right of
the Linear Modeling tab. The variables appear in the Variables area after you successfully
add them to the model.

104 Version 2016 SP2

Modeling SimSci APC User Guide

You can change the order in which the variables appear in the Variables area. When you
change the order of the variables on the Linear Modeling tab, you also change the order of
the variables on the Results tab. You can quickly access the variables that are the most
important or the most interesting to you by moving those variables to the top of the list. You
can change the order of only one variable at a time.
You can also delete any number of CVs, MVs, or FVs from a linear model. Note that you
cannot delete different types of variables at the same time. For example, if you want to delete
a set of CVs and MVs from a linear model, you should first delete the CVs and then delete
the MVs.
To configure the structure of a linear model:
1. On the Linear Modeling tab, select the tab for the linear model.
2. In the Tags List list, select the collection that contains the tags of interest.
3. In the Input Tags area, select the tags that you want to specify as CVs for the model.
4. Drag the selection into the Control Variables area. Alternatively, you can right-click the
selection, and then click Add to Control Variables.
5. If desired, in the Input Tags area, select the input tags that you want to specify as FVs for
the model.
6. Drag the selection into the Feed Forward Variables area. Alternatively, you can right-
click the selection, and then click Add to FeedForward Variables.
7. In the Output Tags area, select the tags that you want to specify as MVs for the model.
8. Drag the selection into the Manipulated Variables area. Alternatively, you can right-click
the selection, and then click Add to Manipulated Variables.
9. If desired, in the Output Tags area, select the output tags that you want to specify as FVs
for the model.
10. Drag the selection into the Feed Forward Variables area. Alternatively, you can right-
click the selection, and then click Add to FeedForward Variables.
To change the order of a variable in the linear model:
1. In the Linear Modeling tab, on the tab for the associated linear model, in the Variables
area, select the variable that you want to move.
2. If you want to move the selected variable up in the list, right-click the variable, and then
click Move Up.
3. If you want to move the selected variable down in the list, right-click the variable, and then
click Move Down.
4. If you want to move the selected variable to the top of the list, right-click the variable, and
then click Move To Top.
5. If you want to move the selected variable to the bottom of the list, right-click the variable,
and then click Move To Bottom.
To delete variables from a linear model:
1. On the Linear Modeling tab, on the tab for the associated linear model, in the Variables
area, select the variables that you want to delete. Note that you cannot select different
types of variables at the same time.
2. Right-click the selection, and then click Delete Variable(s).
Version 2016 SP2 105
SimSci APC User Guide Modeling

3. A confirmation message appears. Click Yes.

Adding Notes to a Model

You can add notes to a linear model by using either the Linear Modeling tab or the Result
tab.
When you add notes by using the Linear Modeling tab, the notes take the form of a floating
window in which you can type text. The notes window is visible regardless of which tab is
open in the Page View.
When you add notes by using the Result tab, the notes take the form of a Notes tab on the
Result tab. If you edit the notes for the model by using the Notes tab while the notes window
is open, the notes revert to the text that appears in the notes window after you close the
notes window.
The timestamp that appears on a note is the time at which you last closed the notes window.
If you have never opened and subsequently closed the notes window for a model, the
timestamp that appears on the note is the time at which you last opened the APC Project
The SimSci APC software automatically saves notes when you close the notes window or
when you navigate away from the Notes tab.
To add notes to a model:
1. On the Linear Modeling tab, right-click the tab for the model.
2. Click Add/Edit Note. The note window appears.
3. Type the note.
4. Close the note window.

Scenarios for a Linear Model

A scenario for a linear model is a set of parameters that identifies the algorithms that the
SimSci APC software should use to calculate the model coefficients, the residuals, the
process gains, and the steady-state responses for a model.

Managing Scenarios
You can manage the scenarios for a linear model by using the scenarios grid on the Linear
Modeling tab, on the associated model tab.
When you create a scenario, the new scenario uses the default values for all the scenario
parameters. See Scenario Parameters on page 110 for more information.
When you copy scenarios, the SimSci APC software copies the scenarios in the order in
which you selected them.
When you export scenarios, you can export only one scenario at a time. The SimSci APC
software saves the exported scenario as a .model file. You can then import then import the
.model file into another APC Project.

106 Version 2016 SP2

Modeling SimSci APC User Guide

When you identify scenarios, you run the least square algorithm for each of the selected
scenarios. That is, you run the calculations for the model coefficients, residuals, process
gains, and steady-state responses for each scenario. You can filter out bad data or data that
is not related to the identification of the scenarios by using the Model Data Selector. You can
also choose which data you want to include in the identification calculations. See Selecting
Data for Scenarios on page 108 for more information.
If you modify any of the model coefficients for a scenario after you identify the scenario, you
should regenerate residuals for those scenarios. You can regenerate residuals for the entire
linear model or for a single MV or FV within the linear model. When you regenerate residuals
for the scenario, the SimSci APC software re-identifies the scenario by using the modified
coefficients and creates a cohesive model for the linear modeling scenario. See Modifying the
Model Coefficients on page 133 for more information.
Note that you can regenerate residuals for unidentified scenarios, such as those that you
import from a SimSci APC or Connoisseur model or those that you build by copying and
pasting model coefficients from other scenarios. See Importing Data on page 20 and Copying
and Pasting Model Coefficients on page 131 for more information.

Acti vity Directions

Creating a scenario On the associated model tab, click Add Modeling Scenario

. If desired, change the default entries for the scenario

parameters.

Copying scenarios On the associated model tab, in the scenarios grid, in the Select
column, select the check box for each scenario that you want to

copy, and then click Copy .

Exporting a scenario On the desired model tab, in the scenarios grid, right-click the
scenario, and then click Export Scenario. Save the .model file in
the desired location.

Identifying scenarios On the associated model tab, in the scenarios grid, in the Select
column, select the check box for each scenario that you want to

identify, and then click Identify .

Regenerating residuals for On the associated model tab, in the scenarios grid, in the Select
scenarios column, select the check box for each scenario for which you
want to regenerate residuals, and then click Regenerating

Residual s .

Regenerating residuals for a On the associated model tab, in the scenarios grid, in the Select
single MV or FV column, select the check box for each scenario for which you
want to regenerate residuals. In the Variables area, right-click
the MV or FV for which you want to regenerate residuals, and
then click Regenerate Residuals.

Version 2016 SP2 107

SimSci APC User Guide Modeling

Deleting scenarios On the associated model tab, in the scenarios grid, in the Select
column, select the check box for each scenario that you want to

delete, and then click Delete Modeling Scenario .

Alternatively, you can press Delete.

Example: Exporting a Scenario

1. On the associated model tab, in the scenarios grid, right-click the scenario, and then click
Export Scenario. The Export Scenario dialog box appears.
2. Browse to the folder in which you want to save the .model file.
3. In the File Name box, type the name of the .model file.
4. Click Save.
Example: Deleting Scenarios
1. On the associated model tab, in the scenarios grid, in the Select column, select the check
box for each scenario that you want to delete.
2. Do one of the following:

o Click Delete Modeling Scenario .

o Press Delete.
3. A confirmation message appears. Click Yes.

Selecting Data for Scenarios

You can choose which data you use to identify scenarios by using the Model Data Selector.

You can open the Model Data Selector by clicking Model Data Selector on the
associated model tab.
The Model Data Selector appears as a trend in which you can select sets of data. It displays
a Strip for each tag in the linear model. You can adjust the scaling for these Strips by using
the same procedures to adjust the scaling for Strips in any other trend. You can also add tags
to a Strip by using the general tag collections. See Adjusting the Scaling for a Strip on page
39 and Managing Strips on page 36 for more information.
There are two views in the Model Data Selector, Bad Data and Model Data.
On the Bad Data view, you select the bad quality data that you want to filter out of the
scenario identification. The bad data selections apply to all scenarios within the associated
linear model. The SimSci APC software does not include the selected data in any
identification calculations when it identifies the scenarios. When you select a set of bad data
points in a Strip, the SimSci APC software highlights the bad data in blue. In all other Stips in
the Model Data Selector, it highlights the bad data in red. That is, it highlights the bad data in
blue only in the Strip in which you select the bad data.

108 Version 2016 SP2

Modeling SimSci APC User Guide

On the Model Data view, you select the data that you want to include in the scenario
identification. The model data selections apply to a single scenario. Each selection is known
as a data slice. If you do not add any data slices to a scenario, the SimSci APC software
includes all the model data (that is, all the data that you have not selected as bad data) in the
identification calculations when it identifies the scenario. If you add data slices to a scenario,
the SimSci APC software includes only the data slices in the identification calculations when it
identifies the scenario. If a data slice includes a bad data selection, the SimSci APC software
excludes that portion of the data slice when it identifies the scenario. Note that you can copy
a data slice configuration in one scenario to another scenario. When you add a data slice to a
Strip, the SimSci APC software highlights the data slice in green.
The Bad Data view and the Model Data view always display the same highlighting. Only the
data slices for the current scenario (that is, the scenario you select in the Selected Scenario
list on the Model Data view) appear in these views.
In both the Bad Data view and the Model Data view, you must double-click a point on the
trend to select data. When you first double-click the trend, a highlighted spike appears on the
trend to indicate the starting point for the selection. You must double-click another point on
the trend to select a set of data points. That is, you cannot select only a single data point. If
you select a set of data points that overlaps an existing set of selected data points, the
SimSci APC software removes the existing set of selected data points. On the Model Data
view, you can resize your set of data points by using the Start Time and End Time
parameters for a data slice or by dragging the ends of the data slice on the strip.

Acti vity Directions

Opening the Model Dat a On the Linear Modeling tab, on the associated model tab, click
Selector
Model Data Selector .

Switching to the Bad Data On the upper-left corner of the Model Data Selector, click Bad
view Data.

Switching to the Model Data On the upper-left corner of the Model Data Selector, click
view
Model Data.

Selecting a set of bad data On the Bad Data view, double-click a point on a Strip. Double-
points click another point on the Strip to select all dat a points bet ween
the data points on which you double-clicked.

Zooming in on an area of bad On the Bad Data view, at the bottom of the list of Strips, drag the
data points zoom bar. Drag the leftmost end of the zoom bar to zoom in on
the rightmost side of the Strips. Drag the rightmost end of the
zoom bar to zoom in on the leftmost side of the Strips.

Removing a set of selected On the Bad Data view, right-click a set of selected data points,
bad data points and then click Remove.

Adding a data slice On the Model Data view, in the Selected Scenario list, select
the scenario to which you want to add the data slice. Double -click
a point on a Strip. Double-click another point on the Strip to select
all data points between the data points on which you double-
clicked.

Version 2016 SP2 109

SimSci APC User Guide Modeling

Zooming in on an area of On the Model Data view, select the Enable Zoom check box. At
model data points the bottom of the list of Strips, drag the zoom bar. Drag the
leftmost end of the zoom bar to zoom in on the rightmost side of
the Strips. Drag the rightmost end of the zoom bar to zoom in on
the rightmost side of the Strips. Alternatively, select an area on a
Strip to zoom in on that area.

Renaming a dat a slice On the Model Data view, in the Model Data Selections area,
double-click the Data Slice value for the data slice that you want
to rename. Type the new name, and then press Enter.

Changing the start time for a On the Model Data view, in the Model Data Selections area,
data slice double-click the Start Time parameter for the data slice. Type the
new start time for the data slice, in mm/dd/yyyy hh:mm:ss format,
and then press Enter.

Changing the end time for a On the Model Data view, in the Model Data Selections area,
data slice double-click the End Time parameter for a data slice. Type the
new end time for the dat a slice, in mm/dd/yyyy hh:mm:ss format,
and then press Enter.

Resizing a data slice by On the Model Data view, clear the Enable Zoom check box. To
dragging the ends of the dat a change the start time of the data slice, on any Strip, drag the
slice leftmost end of the data slice to the position at which you want
the data slice to start. To change the end time of the data slice,
on any Strip, drag the rightmost end of the d ata slice to the
position at which you want the data slice to end.

Copying dat a slices to another On the Model Data view, in the Copy to Scenario(s) list, select
scenario the scenario to which you want to copy the data slices. In the
Model Data Selections area, click the area immediately to the
left of the Data Slice column for a data slice to select the data
slice. Select the data slices that you want to copy, and then click

Copy .
The selected data slices appear highlighted in green.

Removing a data slice On the Model Data view, right-click a set of selected dat a points,
and then click Remove. Alternatively, on the Model Data view, in
the Model Data Selections area, select the data slices you want

to remove, and then click Delete model data selection .

Using director scripts to mark On the Tool s tab, select Director Calc to open the Script
bad data slices Editor. Type in the SetBadDataSlicesForTag function and fill
out the required arguments, and then run the script.
Bad data slices will be marked for the tag you specified.
For more information, see the Director Functions Guide.

Scenario Parameters
Each scenario has the following parameters:
ID on page 111
110 Version 2016 SP2
Modeling SimSci APC User Guide

Order on page 111

Least Squares Type on page 111
Scaling on page 112
Prediction Interval on page 112
Model Length on page 112
Number of Cycles on page 113
ARX Mask on page 113

ID
The ID value is the identification number of the scenario. The SimSci APC software gives
each scenario a sequential ID value that corresponds to the order in which you create the
scenarios. The SimSci APC software references these ID values in all the modeling results
available on the Result tab.You cannot change the ID value for a scenario.

Order
The Order value is the order of the dynamics of the model. It determines the number of auto-
regressive terms associated with each CV prediction. The specified order is applied to each
CV in the model. You can set the Order value to any integer value greater than or equal to
zero and less than or equal to 3.

Least Squares Type

The Least Square Type selection is the type of least squares algorithm that is used to
identify a scenario. You can choose from one of three types of least squares identification
methods:
Ordinary: This identification algorithm considers all plant data simultaneously. A set of
model coefficients are obtained to minimize the square of the model prediction error.
Partial: This identification algorithm is the same as the Ordinary selection except that it
uses more sophisticated mathematics to deal with insufficient or bad data that would
otherwise not reveal any significant relationships.
Recursive: This identification algorithm uses a recursive approach. The identification
algorithm considers the plant data point-by-point from the beginning to the end of the
analysis range until it converges on a fixed set of model coefficients that minimize the
square of the model prediction error.
By default, the Least Square Type selection is initially set to Ordinary.
The Ordinary type provides a faster identification than the Recursive type when applied to
the same set of scenario parameters. However, the Recursive type requires less memory
and is compatible with real-time adaptive implementation. Both types give the same results
for the model coefficients and process gains when applied to the same data set for the same
Order value.
If the Order value is set to a nonzero value, you must use the Recursive type. If you use the
Recursive type for a scenario, you cannot change the Least Square Type selection for the
scenario back to Ordinary or Partial.

Version 2016 SP2 111

SimSci APC User Guide Modeling

Scaling
The Scaling option allows you to change the normalization scaling for the contribution of
each MV, CV, and FV to the model. You can choose from one of two selections:
Auto: You cannot change the scaling for any variable. The Scale value in the Variables
area for each variable is calculated when you identify the scenario.
Predefined: You can change the scaling for any variable. That is, you can change the
Scale value in the Variables area for any variable in the model.
Each scenario has its own set of Scale values. There is one Scale value for each variable.
By default, the Scaling option for each scenario is initially set to Auto and the Scale value for
each variable is initially set to one.
The Predefined selection is typically used for specialized circumstances in which it is
inappropriate to automatically derive the Scale values. For example, you may want to make a
side-by-side comparison of two models that are identified by using different data sets. For the
model parameters to be directly comparable, the models should have the same normalization
scaling. Otherwise, you must compare the engineering form of the model parameters.
See Normalization on page 128 for more information.

Prediction Interval
The Prediction Interval value is the time step at which new predictions of the CVs are
generated. It is also the execution interval of the model-based predictive controller derived
from the model.
By default, the Prediction Interval value is initially set to the scan interval. The Prediction
Interval value must be a multiple of the scan interval.
If you have performed correlation analyses for your data, you should set the Prediction
Interval value to 25% of the time to the earliest cross-correlation peak obtained for all of the
possible correlation analysis calculations.
If you have not performed correlation analyses for your data, you can set the Prediction
Interval value to 5% of the shortest time to steady state exhibited by the system.
We recommend that you select a single prediction interval for all of the model scenarios at
the beginning of the model identification to avoid problems when you develop the composite
linear model and the final linear model. See Effect of Prediction Intervals on page 143 for
more information.

Model Length
The Model Length value is the minimum number of past values for the input variables (that
is, the MVs and FVs) that the identification algorithm considers when you identify a model.
Both the Order value and the Model Length value determine the total number of past values
that are considered. See Vectors for the Process Inputs on page 119 for more information.
The Model Length value determines the number of past values for the input variables that
the model-based predictive controller considers. See Model Length of the Final Linear Model
on page 149 for more information.
By default, the Model Length value is initially set to the scan interval. You can set the Model
Length value to any integer value greater than or equal to zero.

112 Version 2016 SP2

Modeling SimSci APC User Guide

Number of Cycles
The Number of Cycles value is the number of cycles required to go through the data. It is
used in conjunction with the Least Squares Type selection. It can have any integer value
greater than or equal to zero.
The default value for the Number of Cycles value is determined by the Least Squares Type
selection:

Least Squares Type Default Number of Cycles

Ordinary 0

Partial 25

Recursive 3

If you set the Number of Cycles value for a scenario and then change the Least Squares
Type selection for that same scenario, the Number of Cycles value changes to the default
value for the selected type.

ARX Mask
The ARX Mask check box determines whether the ARX coefficients for the scenario change
when you re-identify the scenario or regenerate residuals for the scenario. If you select this
check box, the SimSci APC software does not update the ARX coefficients for the scenario.
Note that this check box is available only for first-order scenarios.

Transfer Functions
Transfer functions are Laplace transforms that you can use to model the dynamics of the
process plant. Each transfer function corresponds to a variable relationship, that is, a CV and
MV pair or a CV and FV pair. The SimSci APC software uses transfer function models to
manage the transfer functions in the APC Project. The transfer function models act as
alternative simulation models to the ARX and FIR models that you obtain by identifying
scenarios in a linear model.
The SimSci APC software currently offers the following types of transfer functions:
First Order
Second Order Time Constant
Second Order Damping Ratio
Integrating

Transfer Function Parameters

The following table lists the available transfer function parameters:

Parameter Description

K The gain for the variable relationship. You use this parameter for all the
available types of transfer function.

Version 2016 SP2 113

SimSci APC User Guide Modeling

Parameter Description

D The time delay for the variable relationship. You use this parameter for all the
available types of transfer function.

0 The numerator time constant for the variable relationship. You use this
parameter for transfer functions that are second order time constants, second
order damping ratios, or integrating.

1 The numerator time constant for the variable relationship. You use this
parameter for transfer functions that are first order or second order time
constants.

2 The numerator time constant for the variable relationship. You use this
parameter for transfer functions that are second order time constants or
integrating.

n The natural frequency for the variable relationship. You use this paramet er
only for trans fer functions that are second order damping ratios.

The damping ratio for the variable relationship. You use this parameter only for
transfer functions that are second order damping ratios.

The following table shows the mathematical formula for each type of transfer function.

Type Formula

First order Ke D s
Gs
1 1s
Integrating K 1 0 s e D s
Gs
s1 2 s

Second Order Damping K 1 0 e D s n2

Gs

Ratios
s 2 2n s n2

Second Order Time K 1 0 s e D s

Constants Gs
1 1s 1 2 s

Managing Transfer Functions

You can manage transfer functions by using the variable relationships grid on the Transfer
Function tab, on the associated transfer function model tab, on the Configuration tab.
Transfer function models contain transfer functions. Each transfer function corresponds to a
variable relationship, that is, a CV and MV pair or a CV and FV pair. The SimSci APC
software automatically creates a transfer function for each CV and MV pair and each CV and
FV pair that you add to transfer function model.

114 Version 2016 SP2

Modeling SimSci APC User Guide

The first time that you open the Transfer Function tab, the SimSci APC software
automatically creates a transfer function model named TF Model1. You can use this transfer
function model to create and manage transfer functions for variable relationships. You can
also create your own transfer function models in addition to the initial transfer function model,
TF Model1.
You can import and export transfer function models. The SimSci APC software saves transfer
function models as .tfmodel files.

Acti vity Directions

Creating a trans fer function On the Transfer Function tab, right-click a tab for a transfer
model function, and then click Add Model.

Copying a transfer function On the Transfer Function tab, right-click a tab for a transfer
model function, and then click Copy Model.

Adding CVs to a transfer On the tab for the associated transfer function model, on the
function model Configuration tab, in the Input Tags area, select the tags that
you want to specify as CVs for the trans fer function model. Drag
the selection into the Control Variables area. Alternatively, you
can right -click the selection, and then click Add to Control
Variables.

Adding MVs to a transfer On the tab for the associated transfer function model, on the
function model Configuration tab, in the Output Tags area, select the tags that
you want to specify as MVs for the transfer function model. Drag
the selection into the Manipulated Variables area. Alternatively,
you can right-click the selection, and then click Add to
Manipulated Variables.

Adding FVs to a transfer On the tab for the associated transfer function model, on the
function model Configuration tab, in the Input Tags area, select the tags that
you want to specify as FVs for the transfer function model. Drag
the selection into the Feed Forward Variables area. In the
Output Tags area, select the tags that you want to specify as
FVs for the transfer function model. Drag the selection into the
Feed Forward Variables area. Alternatively, you can right-click
the selection, and then click Add to FeedForward Variables.

Configuring a transfer function On the tab for the associated transfer function model, on the
Configuration tab, in the variable relationships grid, in the
Transfer Function list, select the type of trans fer function that
you want to use for the variable relationship. Change the trans fer
function parameters as desired. See Transfer Function
Parameters on page 113 for more information.

Modifying transfer function On the tab for the associated transfer function model, on the
parameters from the Transfer Transfer Function tab, double-click a transfer function. A dialog
Function tab box appears. In the boxes, type the values for the corresponding
parameters. See Transfer Function Parameters on page 113 for
more information.

Version 2016 SP2 115

SimSci APC User Guide Modeling

Exporting transfer functions On the Transfer Function tab, right-click the tab for the transfer
function, and then click Export Model. The Export Transfer
Function Model dialog box appears. Browse to the folder in
which you want to save the file. In the File name box, type the
name of the file, and then click Save.

Importing trans fer function Click the SimSci APC Button.On the Import menu, on the Model
menu, click Import Transfer Function Model. The Open dialog
box appears. Browse to and select the desired .tfmodel file, and
then click Open.

Renaming a transfer function Double-click the tab for the associated transfer function model.
model Type the new name, and then press Enter.

Deleting transfer functions On the Transfer Function tab, right-click a tab for a transfer
function, and then click Delete Model.

Transforming Transfer Functions

You can transform a transfer function model into a scenario within a linear model.
You can also transform scenarios in a linear model into transfer function models. The SimSci
APC software creates a new transfer function model for each scenario that you transform.
Note that you must identify the scenarios before you transform them into transfer function
models.
To transform a transfer function model into a linear modeling scenario:
1. On the Transfer Function tab, on the tab for the associated transfer function model, on
the Configuration tab, click Transform to ARX/FIR . The Transform To ARX/FIR
Model dialog box appears.
2. For a zero-order model, select FIR. For a first-order model, select ARX.
3. In the Linear Model list, select New Model to create a new linear model or select an
existing linear model to which you want to add the transformed transfer function. Note that
the tag structure of the existing linear model must match the tag structure of the transfer
function model.
4. In the Scenario list, select New Scenario to create a new scenario in the linear model or
select an existing scenario to overwrite that scenario with the transformed transfer
function.
5. If you selected ARX in step 2, in the Model Order box, enter the order for the scenario.
6. In the Model Length box, enter the model length for the scenario.
7. In the Prediction Interval box, enter the prediction interval for the scenario.
8. Click Transform.
To transform linear modeling scenarios into transfer function models:
1. On the Linear Modeling tab, on the tab for the associated linear model, in the scenarios
grid, select the check boxes for the scenarios that you want to transform.

2. Click Transform ARX/FIR to TF . The Confirm Action dialog box appears.

116 Version 2016 SP2

Modeling SimSci APC User Guide

3. Click First Order to restrict the transfer functions for the variable relationships to the First
Order and Integrating types.Click Any Order to allow the SimSci APC software to set
the transfer functions for the variable relationships to any type of transfer function.
4. View the new transfer function models on the Transfer Function tab.

Identification Algorithms
The SimSci APC software uses three different algorithms to identify scenarios:
Ordinary Least Squares (OLS) algorithm
Partial Least Squares (PLS) algorithm
Recursive Least Squares (RLS) algorithm
These algorithms are used to calculate the model coefficients, residuals, process gains, and
steady-state responses for each scenario. They can be derived from a generalized equation
for models. See Generalized Equation on page 117 for more information.

Generalized Equation
You can use the following generalized equation for models:
Y1 k p Y2 k 1W1,k r Wr ,k
where
Y1 is a vector that contains the prediction values for the process outputs, that is, the CVs.
is the transition matrix containing the autoregressive terms. A model uses the
autoregressive terms only if its order of dynamics is greater than zero.
Y2 is a vector that contains the current and past values for the CVs.
1 is the driving matrix for the input variable corresponding to W1

W1 is the vector of increments for the first input variable that is added to the model. It
contains the current and past increments for the first input variable.
r is the driving matrix for the input variable corresponding to Wr

Wr is the vector of increments for the last input variable that is added to the model. It
contains the current and past increments for the last input variable.
r is the total number of input variables that are added to the model
k is the "current" time instant
p is the prediction interval ratio

The generalized equation represents an autoregressive model with exogenous inputs (ARX
model), where is the matrix of autoregressive terms. If we set to zero, the algorithm does
not consider any past outputs and the equation represents a Finite Impulse Response (FIR)
model.

Version 2016 SP2 117

SimSci APC User Guide Modeling

In the generalized equation, the vectors for the input variables, W1 through Wr , are given in
incremental terms, that is, they contain differenced values. The vectors for the output
variables, Y1 and Y2 , are given in non-incremental terms. When you are using the model to
subsequently evaluate controllers, this form of the equation is most suitable. However, when
you are performing the identification procedures for the model, a completely incremental
version of the equation is more efficient. See Incremental Equation on page 123 for more
information.

Order of Dynamics
The order of the dynamicsthat is, the Order value for a scenariois defined as n . When
you set n to zero, the generalized equation for a model has no autoregressive matrix ( = 0)
and therefore represents an FIR model. See Generalized Equation on page 117 and Order
on page 111 for more information.

Prediction Interval Ratio

The prediction interval ratio, p , is calculated as the ratio of the Prediction Interval value for
a scenario and the scan interval:
PredictionInterval
p
ScanInterval
See Prediction Interval on page 112 for more information.

Discrete Time Instants

We denote discrete time instants at which process data is sampled as k 1 , k , k 1 ,

k 2 , and so on. The time instant k represents the current time instant being considered
by the algorithm. The discrete time instants are separated from each other by the amount of
time specified for the scan interval.

Vectors for the Process Outputs

In the generalized equation for a model, the Y1 and Y2 vectors contain values that relate to
the process outputs (CVs) at certain time instants. To derive these vectors, we can define a
y
vector of CV values for every discrete time instant, k , as follows:
y1
y
2
y k
.
.

.
yc k
where
yi
is an output of the model, that is, a CV
c is the number of CVs in the model

118 Version 2016 SP2

Modeling SimSci APC User Guide

Y1 k p
The vector containing the prediction values for the process outputs, , has the same
yk Y1 k p yk
dimensions as ; that is, the vectors and both contain a c number of individual
scalar sampled values.
Y2 k
The vector , which contains the current and past CV values that are considered by the
algorithm, is defined as follows:
yk
y
k p
y k 2 p
Y2k .

.

.
y k N 1 p ,

N n 1
where
N is an integer that determines the number of previous prediction intervals over which the
Y2 k
vector extends.
n is the order of dynamics

Vectors for the Process Inputs

In the generalized equation for a model, the W1 through Wr vectors contain values that
relate to the process inputs (MVs and FVs) at certain time instants. To derive these vectors,
w
we can define a vector of the MV and FV values for every discrete time instant, k , as
follows:
u1
.

. w1
. .
u k um
wk .
v k v1 .
. w
. r k

.
v
q k ,
r mq
where

Version 2016 SP2 119

SimSci APC User Guide Modeling

uk
is a vector of MV values for every discrete time instant
vk
is a vector of FV values for every discrete time instant
ui
is a manipulated input variable for the model, that is, an MV
vi
is a feedforward input variable for the model, that is, an FV
wj
is an input variable for the model
m is the total number of MVs added to the model
q is the total number of FVs added to the model
r is the total number of input variables added to the model

We assume that each input variable, j ( j 1,, r ), remains constant in value across each
w
prediction interval and changes only at the prediction instant.
The model length for the input variables in the modelthat is, the Model Length value for a
scenariois defined as L . The value of this parameter determines the dimensions of the
vectors of increments for the input variables. It effectively determines the amount of past
variation for the input variables that the algorithm considers. You must set the value of L to
be a positive integer. See Model Length on page 112 for more information.
w Wj
For each input variable, j , we can define a vector that contains the increments of the
input variable value across the succeeding prediction intervals:

w j ,k w j ,k p
w
j ,k p w j ,k 2 p
.
W j ,k
.

.
w j ,k R 1 p w j ,k R p
j j

where
Rj
is the number of incremental terms in the vector.
y k N 1 p Y2 k
If the term in
(see Vectors for the Process Outputs on page 118) involves only
sample values at instant k N 1 p after a truncation to n c elements, then j is
R
calculated as follows:
R j N L 1

Rj
Otherwise, is calculated as follows:
R j N L

120 Version 2016 SP2

Modeling SimSci APC User Guide

In relation to the generalized equation for a model, we can define the following summation for
the contributions of the input variables to the predicted values for the output variables:
r

W
j 1
j j ,k 1W1,k r Wr ,k

See Generalized Equation on page 117 for more information.

Steady-State Properties
You can use the following generalized equation for models:
Y1 k p Y2 k 1W1,k r Wr ,k
where
Y1 is a vector that contains the prediction values for the process outputs, that is, the CVs.
is the transition matrix containing the autoregressive terms. A model uses the
autoregressive terms only if its order of dynamics is greater than zero.
Y2 is a vector that contains the current and past values for the CVs.
1 is the driving matrix for the input variable corresponding to W1

W1 is the vector of increments for the first input variable that is added to the model. It
contains the current and past increments for the first input variable.
r is the driving matrix for the input variable corresponding to Wr

Wr is the vector of increments for the last input variable that is added to the model. It
contains the current and past increments for the last input variable.
r is the total number of input variables that are added to the model
k is the "current" time instant
p is the prediction interval ratio

In the steady-state condition, all of the input variables are static. Therefore, all of the vectors
of increments for the input variables, W1 through Wr , are zero:
Y1 k p Y2 k

Furthermore, the values of the output variables in vectors Y1 and Y2 are static.
For a single-output system, the equation for the model output is:

yk p 1 2 3 yk yk p
yk 2 p
T

Since the output variables are static, all of the y values are equal to a constant, y :
y y1 2 3

Version 2016 SP2 121

SimSci APC User Guide Modeling

To fulfill this condition, the coefficients in the associated non-incremental transition matrix
must sum to unity:

i 1

Examples of Typical Model Structures

You must specify the following parameters, which determine the structure of the model:
The number of input variables (MVs and FVs), r
The number of output variables (CVs), c
The order of process dynamics, n
The prediction interval ratio, p
The model length, L
The order of dynamics, the prediction interval ratio, and the model length are specific to each
scenario for a model. The number of input variables and the number of output variables are
the same for each scenario for a model.
Example 1

Consider a single-input, single-output (SISO) system. The controlled variable (CV) is y and
the manipulated variable (MV) is u . When we add the CV and the MV to the model, the
following is true:

The number of CVs is 1, c 1 .

The number of MVs is 1, m 1.

The number of FVs is zero, q 0 .

We make the following specifications for a scenario in the model:

The Order value is set to 1. Thus, the order of dynamics is 1, n 1

The Prediction Interval value is equal to the scan interval. Thus, the prediction interval
ratio must be 1, p 1 .
The Model Length value is set to zero, L 0 .
Based on the above specifications, the model has the following form:
yk
yk p 1 2
y

1 uk uk p
k p
Example 2

Consider a two-input, two-output system. The two output variables (CVs) are y1 and y 2 . The
two input variables (MVs) are u1 and u 2 . When we add the CVs and the MVs to the model,
the following is true:

The number of CVs is 2, c 2 .

122 Version 2016 SP2

Modeling SimSci APC User Guide

The number of MVs is 2, m 2 .

The number of FVs is zero, L 0 .

We make the following specifications for a scenario in the model:

The Order value is set to 2. Thus, the order of dynamics is 2, n 2

The Prediction Interval value is equal to four times the scan interval. Thus, the prediction
interval ratio is 4, p 4
The Model Length value is set to three, L 3 .
Based on the above specifications, the model has the following form:
y1,k
y u1,k u1,k 4 u2,k u2,k 4
2,k u1,k 4 u1,k 8 u
2 , k 4 u 2 , k 8
y1,k 4 y1,k 4
y y 1 u1,k 8 u1,k 12 2 u2,k 8 u 2,k 12
2,k 4 u u
1, k 12 u1, k 16 2 , k 12 u 2 , k 16
2,k 4

1,k 8
y
y
u 1, k 16 u1, k 20

u 2 , k 16 u 2 , k 20
2 , k 8
where
is a 26 matrix
1 is a 25 matrix

2 is a 25 matrix

Incremental Equation
In the generalized equation, the vectors for the input variables, W1 through Wr , are given in
incremental terms, that is, they include differenced values. The vectors for the output
variables, Y1 and Y2 , are given in non-incremental terms. When you are using the model to
subsequently evaluate controllers, this form of the equation is most suitable. However, when
you are performing the identification procedures for the model, a completely incremental
version of the equation is more efficient. An incremental identification is more immune to
influences of drift and unmeasured disturbances in the data. Therefore, the incremental
model is used for all identification algorithms. The incremental model is converted to an
absolute form when you subsequently use the model in the control system design.
You can use the following incremental equation for models:
Y3 k p Y4 k 1W1,k r Wr ,k
where
Y3 is the vector of increments that contains the prediction increments for the process
outputs, that is, the CVs, with respect to the current values.
Y4 is the vector of increments that contains the current and past increments for the CVs

Version 2016 SP2 123

SimSci APC User Guide Modeling

All other variables in the incremental equation are the same as those in the generalized
W
equation. See Generalized Equation on page 117 for more information. The vectors 1, k
Wr ,k
through are unaffected by the transition from an absolute form of the equation to an
incremental form of the equation.
All of the CV increments are referenced to the same data point, that is, the same values in
the vector y at instant k .

The Y3 vector can be defined as follows:

Y3 k p y k p y k
The Y4 vector can be defined as follows:
y k p y k Y3k p
y
k 2 p y k Y3k 2 p
. .
Y4k
. .

. .
y k N 1 p y k Y3k N 1 p

Recursive Least Squares (RLS) Algorithm

The Recursive Least Squares (RLS) algorithm is the set of equations and procedures in the
SimSci APC software that are performed when a recursive least squares analysis is applied
to a data set.
Given a model structure and a set of suitable plant data, you can apply recursive least
squares analysis to the data to establish model coefficients in the transition matrix and the
driving matrices 1 through r (see Generalized Equation on page 117). These model
coefficients characterize the incremental form of the model. The model is then validated by
using the calculated residuals, process gains, and steady-state responses to assess the
properties of the model. In particular, it is important that the model effectively describes the
systems input and output relationships.
The identification algorithm for the recursive least squares (RLS) analysis uses the
incremental equation for models (see Incremental Equation on page 123). The incremental
equation can be condensed further to give a new form of the equation:
Y3 k p Wk
where
1 r
See Generalized Equation on page 117 and Incremental Equation on page 123 for further
descriptions of the variables in these equations.
The identification problem is to determine the coefficients for that give rise to the most
Y3 k p W
accurate predictions of for the given k , taking into account the complete span of the
data that is available for analysis.

124 Version 2016 SP2

Modeling SimSci APC User Guide

The RLS algorithm uses the following equations:

T
k p k Y3k p k Wk Wk Pk
W
Pk W TP
Pk p Pk k k k
dk
TP W
d k 1 W
k k k

3
Y W
k p k p k

Y3
k p

Y3 k 2 p
.
4
Y
k
.
.

Y3 k N 1 p

Y 4
k p

W1,k p
.

W
k p
.
.

Wr ,k p
where
k 0, p, 2 p, , Rt p Rt
, in which is the total number of prediction intervals in the analysis
range

k is the estimate of available at instant k

Pk
is the symmetric covariance matrix evaluated at instant k

at instant k p and is based on

Y3k p Y3 k p
is the estimate of the process output vector

the estimate of the model parameters,

k , and the contribution vector, Wk , at instant k

Wk is the same as the vector Wk except that is incorporates Y4k , which is the estimate of
Y4 k Y4 k
the process output vector , instead of
These equations provide a method for obtaining the next estimate of the model parameters
based on the current estimate. This estimate is updated by using a contribution that is
dependent on the prediction error over the corresponding prediction interval.

Version 2016 SP2 125

SimSci APC User Guide Modeling

These equations are evaluated for each prediction instant in the current analysis range. The
actual implementation of this algorithm is done in a robust manner, utilizing the upper
diagonalization concept. This, in effect, takes into account the symmetry of the covariance
P
matrix, k , and ensures that the covariance matrix is always positive definite.

Unbiased Least Squares

Because the process data is usually contaminated with measurement noise, the
straightforward application of the identification procedure without applying a filtering
mechanism gives rise to a biased estimation of the model transition matrix, , and driving

matrices, 1 through r . The filter is implemented in the equation for

Y3k p
that is seen in
W
Recursive Least Squares (RLS) Algorithm on page 124, which leads to k p in the following
equation. A biased estimate of the model parameters ultimately leads to inferior control
Y3
k p
performance. To avoid this pitfall, the "unbiasing" filter in the equation for is applied to
the identification, leading to the correct estimates of the model. A biased procedure simply
P d
uses W rather than W throughout the equations for k p , k p , and k .

Implementation
To consider the manner in which the RLS algorithm functions, we assume that the analysis

commences with k 0 . The calculation for k p is performed first:

T
k p k Y3k p k Wk Wk Pk

We must therefore have values available for the following variables:

Y3 p

0

W0
P0

Y3 p
The vector is calculated by using the actual plant values:

Y3 p y p y 0
The algorithm also initializes
W0 with actual plant values rather than with estimated values:

Y40
W
1, p
W .
W0 0 .

.
Wr , p

126 Version 2016 SP2
Modeling SimSci APC User Guide

Initially, we have no knowledge of the model coefficients. Therefore, the algorithm initially

sets them to zero in

0 :

0
0

P0
The initial covariance matrix, , is set to a diagonal matrix with very large coefficients:
P0 bI
where
b is the Covariance Weight value (see Covariance Weight). It is a large value, 104 by
default.
I is the identity matrix.
P0
In effect, the values in indicate a total lack of confidence in the current estimate of the

model coefficients, that is,

0 .
Using these established values, the algorithm calculates the first estimate of the model
p
coefficients, :

p 0 p
Y3
W
0
W
0
TP
0 0
Pp
The algorithm then computes the covariance matrix that is used at the next stage, , by
using the following equations:
TP W
d0 1 W
0 0 0

W
P0 W TP
Pp P0 0 0 0
d0

Y3 p Y3 p
The algorithm then computes the estimate of the process output vector , that is, :
Y W
3
p p 0

Wp
Finally, the algorithm computes the contribution vector that is used at the next stage, , by
using the following equation:

Y 4

p

W1, p

.
W p
.
.

Wr , p

Version 2016 SP2 127

SimSci APC User Guide Modeling

After all of the values have been calculated, the data pointer advances by an increment of p
in the data set (that is, k p ) and the algorithm repeats the complete procedure. This cycle
k 0, p, 2 p, , Rt p
repeats until the entire analysis range of the data is processed (that is, ,
Rt
in which is the total number of prediction intervals in the analysis range).
NOTE: You can find the general forms of these equations in Recursive Least Squares (RLS)
Algorithm on page 124.
As the analysis proceeds through the data, there is normally a general convergence, with the
estimated values of the model parameters tending to stabilize. Such convergence can be
slow, particularly if the data is noisy. In these circumstances, to improve the convergence,
you can cycle the analysis through the data a number of times. At the beginning of each new
W
cycle, the 0 reinitializes with the actual data, but
0 and P0 remain equal to the estimates
from the end of the previous cycle.
You can specify the number of times the algorithm cycles through the data by using the
Number of Cycles parameter for a scenario for a model. See Number of Cycles on page 113
for more information. By default, the RLS algorithm executes three such cycles.
You can use the Covariance Trace parameter as a measure of the convergence. The
Covariance Trace parameter is calculated at the end of the RLS algorithm. See Covariance
Trace on page 140 for more information.

Normalization
Data monitored from industrial processes is usually unsuitable for the direct application of
RLS identification. In general, the magnitudes of the various input and ouptut data are
diverse. For example, a model may include an MV which varies between 40 and 60 tons/h
and a CV between 0.5 and 0.6%; the ratio between the two ranges is 200:1. If you apply RLS
analysis to this data, there is a high probability that a numerical instability will occur. This
results in a failure to correctly identify the process dynamics.
Furthermore, industrial data rarely has a constant, steady-state mean that underlies the
transients; there is usually some superimposed drift that is caused by unmeasured, long-term
disturbances. This condition causes difficulty in directly applying RLS analysis to the absolute
input and output values. For this reason, the SimSci APC software uses an incremental
equation for the model identification. See Incremental Equation on page 123 for more
information.
To avoid these potential difficulties, the SimSci APC software normalizes the identification
data prior to the application of the RLS algorithm. This normalization takes place internally
and the processed information is never shown to you. The SimSci APC software calculates
the normalized value by subtracting the mean for the analysis range from each incremental
value, and then dividing by the standard deviation for the same range. It applies the
procedure to all of the data involved in the identification and creates data which is normalized
to vary across a consistent span. The normalization expression for the incremental data is:
yi Meany
Normyi
stdevy
where

128 Version 2016 SP2

Modeling SimSci APC User Guide

yi is the incremental value of the input or output variable with respect to some constant
y yi yi yk
reference value, k , that is,
y is the entire set of increments for the data analysis range.
The RLS algorithm uses this normalized data for the identification. The result is a
"normalized" model, which relies on the input and output variables expressed in their
normalized range.
This form of the model data leads to numerical stability during the identification. It is also used
to directly compare the magnitudes of the model coefficients associated with the different
input and output variables; thus, you can quickly assess the relative significance of the
different input and output variables to the model.

Ordinary Least Squares (OLS) and Partial Least Squares (PLS)

Algorithms
The Ordinary Least Squares (OLS) algorithm is the set of equations and procedures in the
SimSci APC software that are performed when an ordinary least squares analysis is applied
to a data set. The Partial Least Squares (PLS) algorithm is the set of equations and
procedures in the SimSci APC software that are performed when a partial least squares
analysis is applied to a data set.
The OLS algorithm is similar to the Recursive Least Squares (RLS) algorithm except that it
simultaneously considers all of the plant data in the analysis range. The OLS algorithm
obtains a set of model coefficients that minimize the square of the error for the model
prediction.
The PLS algorithm is the same as the OLS algorithm except that it uses more sophisticated
mathematics to deal with insufficient or bad data that would otherwise not reveal any
significant relationship.

Results of a Linear Model

When you identify a linear model, the SimSci APC software produces the following results:
Model Coefficients on page 130
Gains on page 138
Residuals on page 138
Steady-state Responses on page 140
You can use these results to judge the accuracy of the linear model as a representation of the
process dynamics.

Version 2016 SP2 129

SimSci APC User Guide Modeling

Model Coefficients
The model coefficients are the coefficients for a linear model that determine the contribution
of the current and past values for the variables (MVs, FVs, and CVs) to the calculated future

values for the set of controlled variables. They represent the elements in the and the j
matrices in the identification algorithm. See Identification Algorithms on page 117 for more
information. You can view the model coefficients for a linear model on the Coefficients tab
on the Result tab, on the model tab associated with the linear model of interest.
The model coefficients are displayed as sparklines for each CV and MV relationship and for
each CV and FV relationship. You can expand the sparklines to view the exact values of the
model coefficients. See Expanding Sparklines on page 141 for more information. Each
individual sparkline represents the model coefficients that are calculated for the associated
variable relationship. See Example of Model Coefficients on the Result Tab on page 131 for
more information.
You can view two types of model coefficients:
Sum: The model coefficients that are calculated by using the normalized data for the
input and output variables. See Viewing the Sum Model Coefficients on page 130 for
more information.
Impulse: The model coefficients that are calculated by using the raw data for the input
and output variables. See Viewing the Impulse Model Coefficients on page 130 for more
information.

Viewing the Sum Model Coefficients

You can view the sum model coefficients by using the Coefficients tab on the Result tab.
The identification algorithm calculates the sum model coefficients by using the normalized
data for the input and output variables. See Normalization on page 128 for more information.
To view the sum model coefficients:
1. In the upper left corner of the Coefficients tab, click Sum to display the sparklines for the
sum model coefficients.
2. Expand the sparklines as desired. See Expanding Sparklines on page 141 for more
information.

Viewing the Impulse Model Coefficients

You can view the impulse model coefficients by using the Coefficients tab on the Result tab.
The identification algorithm calculates the impulse model coefficients by using the raw data
for the input and output variables. See Normalization on page 128 for more information.
To view the impulse model coefficients:
1. In the upper left corner of the Coefficients tab, click Impulse to display the sparklines for
the impulse model coefficients.
2. Expand the sparklines as desired. See Expanding Sparklines on page 141 for more
information.

130 Version 2016 SP2

Modeling SimSci APC User Guide

Copying and Pasting Model Coefficients

You can copy the model coefficients for a variable relationship and then paste those model
coefficients into another variable relationship. The variable relationships can be in the same
linear model or in different linear models. The copied model coefficients are specific to a
scenario. When you paste the copied model coefficients, you override all the model
coefficients for a specified scenario.
Note that the scenario from which you copy the model coefficients and the scenario into
which you paste the model coefficients must have the same order of dynamics. The scenario
from which you copy the model coefficients must have a model length that is less than or
equal to the model length of the scenario into which you paste the model coefficients. The
SimSci APC software automatically sets any missing model coefficients to zero.
To copy model coefficients:
1. On the Result tab, on the desired linear model tab, on the Coefficients tab, right-click the
desired variable relationship.
2. On the Copy from Scenario submenu, click the desired scenario. Scenarios are listed by
their ID values.
To paste model coefficients:
1. Copy model coefficients.
2. On the Result tab, on the desired linear model tab, on the Coefficients tab, right-click the
desired variable relationship.
3. On the Paste to Scenario submenu, click the desired scenario. Scenarios are listed by
their ID values.

Viewing Names or Descriptions of Tags

You can change the default view of the tag names to view tag descriptions instead in the
matrix view on the Results tab.
To toggle the tag names with tag descriptions:

1. On the Results tab, click on the TagID button . The tag names are replaced by tag
descriptions.
If a tag does not have a tag description, you will see "No Description" when you click on
the TagID button.

2. On the Results tab, click on the TagID button again. The tag descriptions are
replaced by tag names.
Note: You are able to toggle the tag names with tag descriptions anywhere you see the
TagID button in the APC software.

Example of Model Coefficients on the Result Tab

For this example, we consider a linear model that contains:
Two CVs, TIC355.MEAS and TIC412.MEAS,
Two MVs, FIC034.SUP_IN and FIC376.SUP_IN

Version 2016 SP2 131

SimSci APC User Guide Modeling

One FV, TIC187.MEAS

When we identify a scenario for this linear model, the results for the model coefficients are
displayed in the following format on the Result tab, on the Coefficients tab:

Based on the figure above, we can write an equation for each of the CVs, TIC355.MEAS and
TIC412.MEAS:
y1,k p 11u1 12u 2 13 v1 1Y4 1,k

y2,k p 21u1 22u 2 23 v1 2 Y4 2,k

where
y1,k p
is the predicted value, in incremental terms, for the TIC355.MEAS tag
y2,k p
is the predicted value, in incremental terms, for the TIC412.MEAS tag
Y4 1,k
is a single-column matrix that contains the current and past values, in incremental
terms, of the TIC355.MEAS tag that are used to calculate the predicted value for the
TIC355.MEAS tag
Y4 2 , k
is a single-column matrix that contains the current and past values, in incremental
terms, of the TIC412.MEAS tag that are used to calculate the predicted value for the
TIC412.MEAS tag
u 1 is a single-column matrix that contains the current and past values of the FIC034.SUP_IN
tag, in incremental terms, that are used to calculate the predicted value for the TIC355.MEAS
tag and the TIC412.MEAS tag
u 2 is a single-column matrix that contains the current and past values of the FIC376.SUP_IN
tag, in incremental terms, that are used to calculate the predicted values for the
TIC355.MEAS tag and the TIC412.MEAS tag
v 1 is a single-column matrix that contains the current and past values of the TIC283.MEAS
tag, in incremental terms, that are used to calculate the predicted value for the TIC355.MEAS
tag and the TIC412.MEAS tag

132 Version 2016 SP2

Modeling SimSci APC User Guide

11 is a single-row matrix that contains the model coefficients for the contribution of the
current and past values of the FIC034.SUP_IN tag to the predicted value for the
TIC355.MEAS tag, all in incremental terms
12 is a single-row matrix that contains the model coefficients for the contribution of the
current and past values of the FIC376.SUP_IN tag to the predicted value for the
TIC355.MEAS tag, all in incremental terms
13
is a single-row matrix that contains the model coefficients for the contribution of the
current and past values of the TIC283.MEAS tag to the predicted value for the TIC355.MEAS
tag, all in incremental terms
21 is a single-row matrix that contains the model coefficients for the contribution of the
current and past values of the FIC034.SUP_IN tag to the predicted value for the
TIC412.MEAS tag, all in incremental terms
22 is a single-row matrix that contains the model coefficients for the contribution of the
current and past values of the FIC376.SUP_IN tag to the predicted value for the
TIC412.MEAS tag, all in incremental terms
23
is a single-row matrix that contains the model coefficients for the contribution of the
current and past values of the TIC283.MEAS tag to the predicted value for the TIC412.MEAS
tag, all in incremental terms
1 is a single-row matrix that contains the model coefficients for the contribution of the current
and past values of the TIC355.MEAS tag to the predicted value for the TIC355.MEAS tag, all
in incremental terms
2 is a single-row matrix that contains the model coefficients for the contribution of the
current and past values of the TIC412.MEAS tag to the predicted value for the TIC412.MEAS
tag, all in incremental terms

Modifying the Model Coefficients

You can modify the model coefficients for a variable relationship by using the expanded
sparkline for that relationship. This includes all CV and MV relationships, all CV and FV
relationships, and all autoregressive relationships (that is, the relationship found in the ARX
column). You can change only the impulse model coefficients.
You can use the following curve operations to modify the impulse model coefficients and the
autoregressive coefficients:
Modify: Use this operation to directly set the value of the model coefficients.
Gain Multiply: Use this operation to apply a gain factor to the model coefficients. The
new value for a model coefficient is equal to the original value of the model coefficient
multiplied by the specified gain factor.
Shift Left: Use this operation to shift the model coefficients to the left by a specified
number of coefficients. Due to the shift, a number of model coefficients at the end of the
model coefficient set do not have corresponding calculated values. The SimSci APC
software sets these model coefficients to zero.

Version 2016 SP2 133

SimSci APC User Guide Modeling

Shift Right: Use this operation to shift the model coefficients to the right by a specified
number of coefficients. Due to the shift, a number of model coefficients at the beginning of
the model coefficient set do not have corresponding calculated values. The SimSci APC
software sets these model coefficients to zero.
Smooth: Use this operation to smooth the model coefficients by using a smoothing spline
method. This operation can eliminate noise from the calculated values. You can specify
how many coefficients the operation uses to calculate the new values for the model
coefficients. You can also set a threshold value that filters out negligible model
coefficients at the beginning or end of the model coefficient set. See Curve Smoothing for
the Model Coefficients on page 136 for more information.
Flush All: Use this operation to set all the model coefficients to zero. This operation
effectively deletes all the model coefficients.
Multiply Selected: Use this operation to apply a multiplying factor to a single model
coefficient. The new value for a model coefficient is equal to the original value of the
model coefficient multiplied by the multiplying factor. Note that you must click a model
coefficient on the expanded sparkline to apply the multiplying factor to that model
coefficient.
You can use more than one curve operation to change the values of the model coefficients.
However, you can use only one curve operation at a time. When you switch to a different
curve operation, you must save your changes to keep the changes from the previous curve
operation.
To directly set the value of a model coefficient:
1. On the Result tab, on the Coefficients tab, click Impulse, and then expand the sparkline
for the relationship containing the model coefficients you want to modify. See Expanding
Sparklines on page 141 for more information.
2. Expand the Change Coefficients area at the top of the graph.
3. In the Operation list, click Modify.
4. On the graph, click the model coefficient you want to modify. The current value, index,
and associated scenario of the selected model coefficient appear in the Change
Coefficients area.
5. In the Selected Scenario list, click the scenario containing the model coefficient you want
to modify.
6. In the Selected Value box, type a new value for the model coefficient.

7. Click Preview to view the new value of the model coefficient on the graph.

8. Click Save to set the change or click Restore to revert to the previously saved
value for the model coefficient.
To apply a gain factor to a set of model coefficients:
1. On the Result tab, on the Coefficients tab, click Impulse, and then expand the sparkline
for the relationship containing the model coefficients you want to modify. See Expanding
Sparklines on page 141 for more information.
2. Expand the Change Coefficients area at the top of the graph.
3. In the Operation list, click Gain Multiply.

134 Version 2016 SP2

Modeling SimSci APC User Guide

4. In the Gain Factor box, type the factor you want to apply to all the model coefficients for
the relationship.

5. Click Preview to view the new values of the model coefficients on the graph.

6. Click Save to set the changes or click Restore to revert to the previously saved
values for the model coefficients.
To shift the model coefficients to the left or to the right:
1. On the Result tab, on the Coefficients tab, click Impulse, and then expand the sparkline
for the relationship containing the model coefficients you want to modify. See Expanding
Sparklines on page 141 for more information.
2. Expand the Change Coefficients area at the top of the graph.
3. In the Operation list, click Shift Left to shift the model coefficients to the left, or click
Shift Right to shift the model coefficients to the right.
4. In the Shift By box, type the number of coefficient by which you want to shift all the model
coefficients for the relationship.

5. Click Preview to view the new values of the model coefficients on the graph.

6. Click Save to set the changes or click Restore to revert to the previously saved
values for the model coefficients.
To smooth the model coefficients:
1. On the Result tab, on the Coefficients tab, click Impulse, and then expand the sparkline
for the relationship containing the model coefficients you want to modify. See Expanding
Sparklines on page 141 for more information.
2. Expand the Change Coefficients area at the top of the graph.
3. In the Operation list, click Smooth.
4. In the No of Coefficient(s) box, type the number of coefficients you want to use to
calculate each model coefficient in the relationship.
5. In the Threshold Value box, type the threshold you want to use to filter negligible values
of the model coefficients.
NOTE: See Curve Smoothing for the Model Coefficients on page 136 for more
information on the limitations for the values of the No of Coefficient(s) and Threshold
Value parameters.

6. Click Preview to view the new values of the model coefficients on the graph.

7. Click Save to set the changes or click Restore to revert to the previously saved
values for the model coefficients.
To delete all the model coefficients:
1. On the Result tab, on the Coefficients tab, click Impulse, and then expand the sparkline
for the relationship containing the model coefficients you want to modify. See Expanding
Sparklines on page 141 for more information.

Version 2016 SP2 135

SimSci APC User Guide Modeling

2. Expand the Change Coefficients area at the top of the graph.

3. In the Operation list, click Flush All.

4. Click Preview to view the new values of the model coefficients on the graph.

5. Click Save to set the changes or click Restore to revert to the previously saved
values for the model coefficients.
To apply a multiplying factor to a single model coefficient:
1. On the Result tab, on the Coefficients tab, click Impulse, and then expand the sparkline
for the relationship containing the model coefficients you want to modify. See Expanding
Sparklines on page 141 for more information.
2. Expand the Change Coefficients area at the top of the graph.
3. In the Operation list, click Multiply Selected.
4. On the graph, click the model coefficient that you want to change.
5. In the Multiply Factor box, type the factor you want to apply to the model coefficient.

6. Click Preview to view the new value of the model coefficient on the graph.

7. Click Save to set the change or click Restore to revert to the previously saved
value for the model coefficient.

Curve Smoothing for the Model Coefficients

The Smooth curve operation uses a smoothing spline method to smooth a set of model
coefficients and reduce noise in the calculated model coefficients. The Smooth curve
operation requires two user-specified parameters: No of Coefficient(s) and Threshold
Value.
The No of Coefficient(s) parameter determines the number of model coefficients that
participate in the smoothing calculations. The smoothing calculations use the values of the
following model coefficients to calculate the current model coefficient:
A number of model coefficients before the current model coefficient
The current model coefficient
A number of model coefficients after the current model coefficient
The number of model coefficients before the current model coefficient is always equal to the
number of model coefficients after the current model coefficient. Therefore, the No of
Coefficient(s) parameter must always equal an odd number.

136 Version 2016 SP2

Modeling SimSci APC User Guide

The Threshold Value parameter is a threshold factor that determines the minimum value for
a model coefficient in the variable relationship. The SimSci APC software multiplies the
threshold factor by the largest absolute value across all the model coefficients to give the
threshold. If the absolute value of a model coefficient is not greater than the threshold, the
SimSci APC software considers the model coefficient as negligible to the smoothing
operation and sets its value to zero. The SimSci APC software also removes the model
coefficient from the smoothing calculations. The threshold applies only to negligible model
coefficients at the beginning and the end of the set of model coefficients for the variable
relationship. If a negligible model coefficient falls between significant model coefficients (that
is, model coefficients with an absolute value greater than the threshold), the SimSci APC
software does not remove the model coefficient from the smoothing calculations and does not
set its value to zero.
The Smooth curve operation uses the following equations to calculate the current model
coefficient:

k 0 t t 1 i
k f k i 0 k i 0 2

t 1 i 1 t 1 for t k p t

t
n 1
2
where
n is the number of model coefficients participating in the smoothing calculations, that is, the
value of the No of Coefficient(s) parameter.
is a vector of the model coefficients for the variable relationship.

k is the index in vector of the current model coefficient (that is, the model coefficient that
the SimSci APC software is currently calculating)
k 0
is the original value of the current model coefficient
k f
is the calculated value of the current model coefficient
p is the total number of calculated model coefficients. This number may be less than the
total number of model coefficients in the variable relationship due to the application of the
threshold. That is, this number does not include the model coefficients that the SimSci APC
software removes from the calculations when the curve operation applies the threshold.

Note that the preceding equation applies only to the range of t k p t . For model
coefficients that fall in the range of 1 k t , there is an insufficient number of model
coefficients before the current model coefficient to properly calculate the current model
coefficient. For model coefficients that fall in the range of k p t , there is an insufficient
number of model coefficients after the current model coefficient to properly calculate the
current model coefficient. In these cases, the curve operation substitutes the value of the
current model coefficient for all the missing model coefficients.

Version 2016 SP2 137

SimSci APC User Guide Modeling

For example, if k 1 and n 5 , the curve operation uses

10 , 20 , and 30 to calculate
1 f
, as shown in the following equation:
10 210 20 10 30
1 f
3 9 9
p 30 p 2 0 p 10
Similarly, if k p 1 and n 5 , the curve operation uses , , , and
p 0 p 1 f
to calculate , as shown in the following equation:
p 10 2 p 20 p 0 p 30 p 10
p 1 f
3 9 9

Gains
The gains are the steady-state parameters that are calculated when the identification
algorithm determines the steady-state responses for each variable relationship in the linear
model. See Steady-state Responses on page 140 for more information.
There are two types of gains:
Absolute: The absolute gains represent the process gains relating the CVs to the MVs
and FVs. The absolute gains are expressed in engineering units.
Incremental: The incremental gains represent the rate of change over the final iteration.
The incremental units are expressed in engineering units per second.
The absolute gains are equal to the value of the steady-state response at the final iteration.
The incremental gains are calculated as follows:
SteadyStateResponse f SteadyStateResponse f 1
Incrementa lGain
PredictionInterval
where
SteadyStateResponse f
is the value of the steady-state response at the final iteration
SteadyStateResponse f 1
is the value of the steady-state response at the second to last
iteration
The incremental gains should always be zero. If the incremental gains are not zero, you
should assume that the responses are integrating in nature.

Residuals
The residuals for a model scenario are the statistical measurements of the accuracy of the
linear model as a representation of the process dynamics. You can view the residuals by
using the Residuals tab on the Result tab. Each output variable has its own set of the
following residuals for each model scenario:
Absolute Mean Error
Absolute RMS Error
Incremental Mean Error

138 Version 2016 SP2

Modeling SimSci APC User Guide

Incremental RMS Error

RSquared
When you identify a model scenario, the identification algorithm uses the following equations
to calculate the residuals for each output variable in the linear model:
Rt p

y y k
1
AbsMeanError
Rt k 0
k
k 0, p,2 p,, Rt p
,
Rt p

y y k
1
AbsRMSError
2

Rt k 0
k
k 0, p,2 p,, Rt p
,

y yk p y k y k p
Rt p
1
IncMeanError
Rt k p
k
k p,2 p,, Rt p
,

y yk p y k y k p
Rt p
1
IncRMSError
2

k p,2 p,, Rt p
k
Rt k p
,
2

a 1 a 1 a 1

a
y y
i i y i yi
RSquared i 0 i 0 i 0

a 1
a 1
2 a 1
a 1
2

a y i y i r yi yi
2 2

i 0 i 0 i 0 i 0
where
y i
is the value of the output variable that is predicted by the linear model
yi
is the value of the output variable that is obtained from the plant
a is the number of data points in the analysis range
Rt
is the total number of prediction intervals in the analysis range
The SimSci APC software calculates the residuals based on the normalized data values (see
Normalization on page 128). The residuals are therefore meaningful in the absolute sense as
a basis for interpreting the accuracy of the linear model. The values for the residuals are
restricted to the range of 0 to 1.
Except for the RSquared value, smaller values for the residuals indicate a more accurate
model. The Absolute Mean Error value and the Absolute RMS Error value measure the
extent to which the data and the linear model drift apart. The Incremental Mean Error value
should always be close to zero. The Incremental RMS Error value measures the extent to
which the shape of the actual data from the plant and the predicted data from the linear
model compare for the process outputs.
A larger value for the RSquared value indicates a more accurate model. A value of 1
indicates that the model scenario for the linear model perfectly predicts the associated output
variable.

Version 2016 SP2 139

SimSci APC User Guide Modeling

Covariance Trace
P
The covariance trace is the sum of the diagonal elements in the covariance matrix, k (see
Recursive Least Squares (RLS) Algorithm on page 124). It serves as a measure of the
convergence of the linear model. You can view the covariance trace by using the Residuals
tab on the Result tab. The Covariance Trace values are listed in the Covariance Trace area
on the left of the Residuals tab.
The covariance trace is calculated only for the model scenarios that have Recursive
selected for the Least Square Type parameter, that is, the model scenarios that use the RLS
algorithm for calculations. When you identify a model scenario, the identification algorithm
converges on a set of model coefficients. As the convergence progresses, the elements
within the covariance matrix should become very small. Thus, the covariance trace should
also become very small.
When the identification for a model scenario completes, the covariance trace should have a
value of less than 1. If the covariance trace remains large (that is, greater than 10) after the
identification completes, you may need to adjust the scenario parameters for the model
scenario. The following situations can lead to a large covariance trace:

Situation Solution

There are too many model coefficients You can do the following to reduce the number of model
in the model. coefficients in the model:
Increase the Prediction Interval paramet er for the
scenario.
Decrease the Model Length parameter for the
scenario
Reduce the dimension of the linear model.

There is not enough data. Obtain more plant data by starting scanning or by
importing more data.

One or more of the input variables for Remove the variables that have inactive data from the
the linear model has inactive data. linear model.

Steady-state Responses
The steady-state responses for a model provide additional insight into the model predictions.
For the steady-state responses, the model is subjected to separate step changes for each of
the model inputs (both MVs and FVs), and the CV responses are calculated by iterating the
model over a sufficient number of prediction intervals to achieve CV stability. The magnitude
of the step change applied to a model input is always one engineering unit. In all cases, the
step change is an increase in the MV or FV value.
You can view the steady-state responses on the Result tab, on the Steady State tab. The
steady-state responses are displayed as sparklines. You can expand the sparkline for a
variable relationship to view the steady-state responses in more detail. See Expanding
Sparklines on page 141 for more information.

140 Version 2016 SP2

Modeling SimSci APC User Guide

Each point on the trend for the steady-state responses represents one iteration. Each
iteration represents one prediction interval. Therefore, the time between iterations is equal to
the amount of time specified for the prediction interval, in seconds. At the end of the
iterations:
The final CV value is used to calculate the absolute gains for each CV and MV pair and
for each CV and FV pair.
The final rate of change in the response is used to calculate the incremental gains.
See Gains on page 138 for more information.

Expanding Sparklines
Results for the model coefficients and steady-state behavior for a model are displayed in the
form of sparklines on the Result tab. You can expand these sparklines to better analyze the
relationship.
Expanded sparklines are displayed in a Relation window, which is titled according to the CV
and the MV or FV relationship for the displayed sparklines. You can cycle through the
different relationships by using the Previous Relation and Next Relation buttons at the top
of the Relation window.

Acti vity Directions

Expanding a sparkline for the On the associated model tab, on the Coefficients tab, right-click
model coefficients the sparkline for the relationship that you want to analyze, and
then click Zoom Re sults.

Expanding a sparkline for the On the associated model tab, on the Steady State tab, right-click
steady-state behavior the sparkline for the relationship that you want to analyze, and
then click Zoom Re sults.

Model Data Comparisons

You can compare the predicted values of the linear model to the actual plant data by using
model data comparison trends. You can create a model data comparison trend by creating a
custom trend and then adding tag data to the trend by using the model-specific tag
collections. See Creating a Trend on page 32 for more information.
A model data comparison trend contains tag data for CVs only. You can add the Current
Value parameter as well as the following model data parameters to a Strip in the model data
comparison trend:
Prediction: This parameter is the model-predicted data for the tag.
IncrementError: This parameter is the error between the incremental plant data and the
incremental model-predicted data for the tag, in engineering units.
ResidualError: This parameter is the error between the plant data and the model-
predicted data for the tag, in engineering units.
Note that the model data parameters are specific to a linear modeling scenario within a
specified linear model. They include data only for the time interval in which you identify the
associated linear modeling scenario.

Version 2016 SP2 141

SimSci APC User Guide Modeling

You can use the model data comparison trends to determine how well the linear modeling
scenario models the responses and behavior of the plant process. If the linear modeling
scenario provides a well-fitting model for the tag behavior within the plant process, the
following is true:
The model-predicted data is the same as or nearly the same as the actual tag data.
The incremental error is always zero or nearly zero.
The residual error is always zero or nearly zero.

Composite Linear Model

The composite linear model is a combination of different linear models and scenarios from
those linear models, which the SimSci APC software uses to create a complete
representation of the dynamics of the process plant. You can create multiple composite linear
models for each APC Project.
A composite linear model contains the steady-state responses and impulse model
coefficients for each variable relationship that you want to include in the final model-predictive
controller.

Creating a Composite Linear Model

You can create a composite linear model by using the sparklines for the model coefficients on
the Result tab. You can use the sparklines to add variable relationships to the composite
linear model. A variable relationship is a set of results from a model scenario for a single CV
and MV pair or for a single CV and FV pair. You can use the results of only one scenario for
each available variable relationship.
In the composite linear model, the SimSci APC software groups the variable relationships
according to their associated CV. For each group of variable relationships, the first variable
relationship that you add to the composite linear model determines which scenarios that you
can use for the remaining variable relationships within that group.
The variable relationships that include the same CV must all have the same order of
dynamics. If you add a zero-order variable relationship to the composite linear model, you
can use multiple zero-order scenarios for the other variable relationships that include the
same CV. If you add a nonzero-order variable relationship to the composite linear model, you
must use the same scenario for all of the variable relationships that include the same CV.
The prediction intervals of the variable relationships that you add to the composite linear
model determine which scenarios that you can use for the variable relationships that you
subsequently add to the composite linear model. See Effect of Prediction Intervals on page
143 for more information.
To create a composite linear model:
1. On the Result tab, on the associated linear model tab, on the Coefficients tab, click Add

Composite Model .
2. In the Composite Models list, select the check box for the composite linear model.
3. Select the check boxes for the variable relationships that you want to add to the
composite linear model. Note that you can select the Select All check box to select the
check boxes for all the variable relationships.
4. Right-click one of the selected variable relationships.
142 Version 2016 SP2
Modeling SimSci APC User Guide

5. In the Add relationship to Composite model from Scenario submenu, click the
scenario that you want to add to the composite linear model. The submenu lists the
scenarios by their ID values.

Effect of Prediction Intervals

The SimSci APC software uses a least common multiple of the prediction intervals to
determine which variable relationships that you can add to the composite linear model. It sets
the least common multiple of the prediction intervals to the prediction interval of the first
variable relationship that you add to the composite linear model. This least common multiple
applies to all the variable relationships in the model.
If you want to add a variable relationship that has a prediction interval that is greater than the
least common multiple of the prediction intervals, the prediction interval of the desired
variable relationship must be a multiple of the least common multiple of the prediction
intervals.
If you want to add a variable relationship that has a prediction interval that is less than the
least common multiple of the prediction intervals, the least common multiple of the prediction
intervals must be a multiple of the prediction interval of the desired variable relationship. After
you add the desired variable relationship to the composite linear model, the prediction interval
of the desired variable relationship becomes the new least common multiple of the prediction
intervals.
Note that the order of dynamics also determines which variable relationships that you can
add to the composite linear model. If you add a variable relationship that has an order of
dynamics that is greater than zero, you can subsequently add only variable relationships that
have a prediction interval that equals the least common multiple of the prediction intervals.
See Creating a Composite Linear Model on page 142 for more information.
We recommend that you select a single prediction interval for all of the model scenarios at
the beginning of the model identification to avoid problems when you develop the composite
linear model and the final linear model.

Viewing the Composite Linear Model

You can view a composite linear model by using the Composite Linear Model tab. The
Composite Linear Model tab automatically opens on the Page View when you attempt to
add a variable relationship to the composite linear model. You can also open the Composite
Linear Model tab by clicking Composite Linear Model on the APC Explorer, under Control
Design.
The Composite Linear Model tab displays the steady-state responses and impulse model
coefficients for each variable relationship that you add to the composite linear model. It also
contains a Relationship(s) area that displays the following for each variable relationship that
you added to the composite linear model:
The model from which you added the variable relationship. This information appears in
the Model Name column.
The scenario parameters for the scenario that you used to add the variable relationship to
the composite linear model. This information appears in the following columns:
o Scenario ID

Version 2016 SP2 143

SimSci APC User Guide Modeling

o Order
o Prediction Interval
o Model Length
o Least Square Type
o ScaleMode
o Start Time
o End Time
o Number of Cycles
o Covariance Weight
See Scenario Parameters on page 110 for more information.
To view the composite linear model:
1. On the Composite Linear Model tab, click Steady State to view the steady-state
responses for each variable relationship that you added to the composite linear model.
2. Click Impulse to view the impulse model coefficients for each variable relationship that
you added to the composite linear model.
3. If desired, view a specific variable relationship in a zoomed state by doing one of the
following:
o Right-click on a variable relationship and select Zoom Results.
o Double-click on a variable relationship.
4. Expand the Relationship(s) area to view information regarding each variable
relationship.

Configuring the Composite Linear Model

You can configure the composite linear model by adding, replacing, and deleting variable
relationships from the composite linear model.
The prediction intervals of the variable relationships that you add to the composite linear
model determine which scenarios that you can use for the variable relationships that you
subsequently add to the composite linear model. See Effect of Prediction Intervals on page
143 for more information.
If you change and then identify any linear models after you add variable relationships to the
composite linear model, you must replace the variable relationships to update the composite
linear model with the most recent information. If you try to create a controller from a
composite linear model that contains outdated information, an error message appears in the
Diagnostics Window.
You can replace a variable relationship with a different scenario or with the same scenario
that you used previously for the variable relationship. The prediction intervals of the variable
relationships that you add to the composite linear model determine which scenarios that you
can use to replace the variable relationships in the composite linear model. See Effect of
Prediction Intervals on page 143 for more information.

144 Version 2016 SP2

Modeling SimSci APC User Guide

Acti vity Directions

Adding a variable relationship On the Re sul t tab, on the associated model tab, on the
to the composite linear model Coefficients tab, right-click the desired variable relationship. In
the Add relationship to Composite model from scenario
submenu, click the scenario that you want to use for the variable
relationship. The submenu lists the scenarios by their ID values.

Replacing a variable On the Re sul t tab, on the associated model tab, on the
relationship to the composite Coefficients tab, right-click the desired variable relationship. In
linear model the Add relationship to Composite model from Scenario
submenu, click the scenario that you want to use for the variable
relationship. The submenu lists the scenarios by their ID values.

Deleting variable relationships On the Composite Linear Model tab, in the Relationship(s)
from the composite linear area, select the Select check box for the variable relationships
model that you want to delete, and then click Delete the Selected

Relationships . You can click Select All Relationships

to select all the variable relationships in the composite linear
model.
Alternatively, on the Composite Linear Model tab, right-click the
sparkline for the variable relationship that you want to delete, and
then click Delete.

NOTE: When you select a variable relationship in the Relationship(s) area, the SimSci APC
software highlights the sparkline for the variable relationship. The reverse is also true.
Example: Replacing a Variable Relationship in the Composite Linear Model
1. On the Result tab, on the desired linear model tab, on the Coefficients tab, right-click the
desired variable relationship.
2. In the Add relationship to Composite model from Scenario submenu, click the desired
scenario. The submenu lists the scenarios by their ID values.
3. A confirmation message appears. Click Yes.

Version 2016 SP2 145

 SimSci APC User Guide

C HAPTER 6
Controller Design

In This Chapter
Overview.................................................................................................................................... 147
Final Linear Model ...................................................................................................................... 147
Cont roller Assembly .................................................................................................................... 155
Simulation Mode ......................................................................................................................... 194
Cost Function ............................................................................................................................. 195
Profit Function ............................................................................................................................ 201

Overview
When you design a model-predictive controller in the SimSci APC software, you create a
multi-variable dynamic model that accurately predicts the variation in the control variables for
the process plant in question. You can use the results from the linear model identifications to
build this multi-variable dynamic model. The first step is to build a composite linear model.
See the following for more information:
Linear Models on page 101
Scenarios for a Linear Model on page 106
Results of a Linear Model on page 129
Composite Linear Model on page 142
From the composite linear model, you can build a final linear model. You can then use the
final linear model to create the model-predictive controller.

Final Linear Model

The final linear model is the linear model that the SimSci APC software uses to build the
model-predictive controller. The SimSci APC software derives the final linear model from the
composite linear model. See Composite Linear Model on page 142 for more information.
The final linear model is not an exact copy of the composite linear model. The final linear
model adjusts the configuration and impulse coefficients of the composite linear model to
obtain a single, cohesive linear model that represents the dynamics of the process plant. The
final linear model therefore has:
A single order of dynamics
A single prediction interval
A single model length

Version 2016 SP2 147

SimSci APC User Guide Controller Design

A single steady-state response for each variable relationship within the linear model
A single set of impulse coefficients for each variable relationship within the linear model
A single absolute gain for each variable relationship within the linear model

Creating a Final Linear Model

You can create a final linear model by using the Composite Linear Model tab. You can
create multiple final linear models for each APC Project.
A final linear model is linked to the composite linear model that you use to create the final
linear model. The final linear model has the same name as the composite linear model.
To create a final linear model:
1. On the Composite Linear Model tab, on the tab for the associated composite linear
model, verify that you have added the required variable relationships to the composite
linear model.

2. Click Create Final Model .

Updating and Viewing the Final Linear Model

When you make a change to the composite linear model, such as changing the selected
scenario for a variable relationship, the SimSci APC software does not automatically update
the final linear model with the latest data. Instead of creating a new final linear model to use
the latest data, you can update the existing final linear model.
When you update the final linear model, the SimSci APC software does not update the values
for any user-modified model coefficients.
To update the final linear model:

1. On the Composite Linear Model tab, click Create Final Model .

2. On the Final Linear Model tab, view the new data for the final linear model.
3. If desired, view a specific variable relationship in a zoomed state by doing one of the
following:
o Right-click on a variable relationship and select Zoom Results.
o Double-click on a variable relationship.

Configuration for the Final Linear Model

You can configure the final linear model by using the Final Linear Model tab, the Composite
Linear Model tab, and the Result tab.
On the Final Linear Model tab, you can view the following for the final linear model:
The prediction interval
The order of dynamics
The model length
The steady-state response for all of the variable relationships
148 Version 2016 SP2
Controller Design SimSci APC User Guide

The impulse coefficients for all of the variable relationships

The absolute gain for all of the variable relationships
These values represent the configuration for the final linear model. You can change this
configuration by changing the configuration of the composite linear model. You can change
the configuration of the composite linear by using the Composite Linear Model tab, and the
Result tab. See Configuring the Composite Linear Model on page 144 for more information.

Prediction Interval for the Final Linear Model

The prediction interval for the final linear model is the greatest common factor of the
prediction intervals for all of the variable relationships that you added to the composite linear
model.
You can change the prediction interval for the final linear model by adding or deleting variable
relationships from the composite linear model, and then recreating the final linear model. See
Configuring the Composite Linear Model on page 144 for more information.

Order of Dynamics for the Final Linear Model

The order of dynamics for the final linear model depends on the prediction intervals for all of
the variable relationships that are added to the composite linear model.
If all of the variable relationships have the same prediction interval, the order of dynamics for
the final linear model is equal to the maximum order of dynamics for all of the variable
relationships that are added to the composite linear model.
If the variable relationships have different prediction intervals, the order of dynamics for the
final linear model is calculated as follows:

PredictionIntervali
OrderFinal max Orderi
PredictionIntervalLCM
where
OrderFinal
is the order of dynamics for the final linear model
PredictionIntervali
is the prediction interval for a variable relationship that is added to the
composite linear model
PredictionIntervalLCM
is the prediction interval for the final linear model
Orderi
is the order of dynamics for a variable relationship that is added to the composite
linear model
You can change the order of dynamics for the final linear model by adding or deleting variable
relationships from the composite linear model, and then recreating the final linear model. See
Configuring the Composite Linear Model on page 144 for more information.

Model Length of the Final Linear Model

The model length for the final linear model depends on the prediction intervals for all of the
variable relationships that are added to the composite linear model.

Version 2016 SP2 149

SimSci APC User Guide Controller Design

If all of the variable relationships have the same prediction interval, the model length for the
final linear model is equal to the minimum model length for all of the variable relationships
that are added to the composite linear model.
If the variable relationships have different prediction intervals, the model length is calculated
as follows:

PredictionIntervali
max R j ,i OrderFinal , if OrderFinal 0
PredictionIntervalLCM
ModelLengthFinal

max Prediction Interval
PredictionInterval
i
R 1, if OrderFinal 0
LCM
j , i

where
ModelLengthFinal
is the model length for the final linear model
PredictionIntervali
is the prediction interval of a variable relationship that is added to the
composite linear model
PredictionIntervalLCM
is the prediction interval for the final linear model
R j ,i
is the number of model coefficients present for the variable relationship that is added to
the composite linear model
OrderFinal
is the order of dynamics for the final linear model

Impulse Model Coefficients for the Final Linear Model

You can view the impulse model coefficients for the final linear model on the Final Linear
Model tab, on the Coefficients tab. The impulse model coefficients for the final linear model
depend on the impulse coefficients for the composite linear model.
The number of impulse model coefficients depends on the order of dynamics and the model
length for the final linear model. If the order of dynamics is zero, the number of impulse model
R
coefficients, j , is calculated as follows:
Rj n L 1
where
n is the order of dynamics for the final linear model
L is the model length for the final linear model
If the order of dynamics is greater than zero, the number of impulse model coefficients is
calculated as follows:
Rj n L

150 Version 2016 SP2

Controller Design SimSci APC User Guide

If the prediction interval for the variable relationship in the composite linear model is not equal
to the prediction interval for the final linear model, the values of the impulse model
coefficients in the composite linear model are not available for all of the values of the impulse
model coefficients in the final linear model. In this case, the values of the impulse model
coefficients in the composite linear model are distributed across the values of the impulse
model coefficients in the final linear model for the duration of the prediction interval for the
variable relationship in the composite linear model. That is, the values of the impulse model
coefficients in the composite linear model are divided by the number of prediction intervals for
the final linear model that pass before the value occurs. See Example of Impulse Model
Coefficients for the Final Linear Model on page 151 for more information.
If the value of an impulse model coefficient in the final linear model occurs after the last value
of the impulse model coefficients in the composite linear model occurs, the value of that
impulse model coefficient in the final linear model is set equal to zero. See Example of
Impulse Model Coefficients for the Final Linear Model on page 151 for more information.
You can change the impulse model coefficients for the final linear model. See Modifying the
Model Coefficients on page 133 for more information.

Example of Impulse Model Coefficients for the Final Linear Model

Consider a final linear model that has the following configuration:

Parameter Value

Order 4

Prediction Interval 2

Model Length 4

Each variable relationship in the final linear model must have 8 impulse model coefficients.
A variable relationship in the composite linear has the following configuration:

Parameter Value

Order 1

Prediction Interval 6

Model Length 1

The variable relationship has two impulse model coefficients in the composite linear model:

Coefficient Value

0 -0.03684

1 -0.02068

There are 3 prediction intervals for the final linear model for every 1 prediction interval for the
variable relationship in the composite linear model. Therefore, for this variable relationship:
Version 2016 SP2 151
SimSci APC User Guide Controller Design

The first three impulse model coefficients in the final linear model are equal to a third of
the first impulse model coefficient in the composite linear model
The second three impulse model coefficients in the final linear model are equal to a third
of the second impulse model coefficient in the composite linear model.
All remaining impulse model coefficients in the final linear model are equal to zero.

Coefficient Value

0 -0.01228

1 -0.01228

2 -0.01228

3 -0.00689

4 -0.00689

5 -0.00689

6 0.0

7 0.0

Autoregressive Coefficients for the Final Linear Model

You can view the autoregressive model coefficients for the final linear model on the Final
Linear Model tab, on the Coefficients tab, in the ARX column. The autoregressive
coefficients in the final linear model depend on the autoregressive coefficients in the
composite linear model.
The number of autoregressive coefficients is set equal to the order of dynamics of the final
linear model. The values of the autoregressive coefficients in the final linear model are set
equal to the autoregressive coefficients in the composite linear model.
If the prediction interval for the variable relationship in the composite linear model is not equal
to the prediction interval for the final linear model, the values of the autoregressive
coefficients in the composite linear model are not available for all of the values of the
autoregressive coefficients in the final linear model. In this case, any missing values are set
equal to zero.
You can change the autoregressive coefficients for the final linear model. See Modifying the
Model Coefficients for the Final Linear Model for more information.

Example of Autoregressive Coefficients for the Final Linear Mo del

Consider a final linear model that has the following configuration:

Parameter Value

Order 4

Prediction Interval 2

152 Version 2016 SP2

Controller Design SimSci APC User Guide

The final linear model requires 4 autoregressive coefficients for each CV.
A variable relationship in the composite linear has the following configuration:

Parameter Value

Order 2

Prediction Interval 4

The CV related to this variable relationship has two autoregressive coefficients in the
composite linear model:

Coefficient Value

0 -0.66271

1 -0.05408

There are 2 prediction intervals for the final linear model for every 1 prediction interval for the
variable relationship in the composite linear model. Therefore, for this CV, the values of the
autoregressive coefficients in the final linear model are:

Coefficient Value

0 0

1 -0.66271

2 0

3 -0.05408

Steady-state Responses for the Final Linear Model

The steady-state responses for the final linear model provide additional insight into the model
predictions. For the steady-state responses, the final linear model is subjected to separate
step changes for each of the model inputs (both MVs and FVs), and the CV responses are
calculated by iterating the final linear model over a sufficient number of prediction intervals to
achieve CV stability, that is, a steady state for the CV. The magnitude of the step change
applied to a model input is always one engineering unit. In all cases, the step change is an
increase in the MV or FV value.
You can view the steady-state responses for the final linear model on the Final Linear Model
tab, on the Steady State tab. The steady-state responses are displayed as sparklines. You
can expand the sparkline for a variable relationship to view the steady-state responses in
more detail. See Expanding Sparklines on page 141 for more information.

Version 2016 SP2 153

SimSci APC User Guide Controller Design

Each point on the trend for the steady-state responses represents one iteration. Each
iteration represents one prediction interval. Therefore, the time between iterations is equal to
the amount of time specified for the prediction interval, in seconds. At the end of the
iterations:
The final CV value is used to calculate the absolute gains for each CV and MV pair and
for each CV and FV pair.
The final rate of change in the response is used to calculate the incremental gains.
See Gains for the Final Linear Model on page 154 for more information.
The steady-state responses are calculated based on the impulse model coefficients for the
final linear model. If you modify the impulse model coefficients for the final linear model, the
final linear model automatically recalculates the steady-state responses.

Gains for the Final Linear Model

You can view the gains for the final linear model on the Final Linear Model tab, on the Gains
tab. The gains are the steady-state parameters that are calculated when the identification
algorithm determines the steady-state responses for each variable relationship in the final
linear model. See Steady-state Responses for the Final Linear Model on page 153 for more
information.
There are two types of gains:
Absolute: The absolute gains represent the process gains relating the CVs to the MVs
and FVs. The absolute gains are expressed in engineering units.
Incremental: The incremental gains represent the rate of change over the final iteration.
The incremental units are expressed in engineering units per second.
The absolute gains are equal to the value of the steady-state response at the final iteration.
The incremental gains are calculated as follows:
SteadyStateResponse f SteadyStateResponse f 1
Incrementa lGain
PredictionInterval
where
SteadyStateResponse f
is the value of the steady-state response at the final iteration
SteadyStateResponse f 1
is the value of the steady-state response at the second to last
iteration
The incremental gains should always be zero. If the incremental gains are not zero, you
should assume that the responses are integrating in nature.
The gains are calculated based on the impulse model coefficients for the final linear model. If
you modify the impulse model coefficients for the final linear model, the final linear model
automatically recalculates the gains.

Deleting the Final Linear Model

You can delete the final linear model by using the Final Linear Model tab.

154 Version 2016 SP2

Controller Design SimSci APC User Guide

To delete the final linear model:

1. On the Final Linear Model tab, in the Configuration area, click the area directly to the
left of the Model Name column to select the final linear model.
2. Press Delete.
3. A confirmation message appears. Click Yes.

Controller Assembly
In the SimSci APC software, there are two types of controllers:
MVC controller
Optimizer
You can create multiple MVC controllers and Optimizers in a single APC Project. Each MVC
controller is associated with a single Optimizer. Together, an MVC controller and Optimizer
are known as a controller pair. The name of the controller pair is the same as the name of the
MVC controller.
A controller pair can also contain subcontrollers. The subcontrollers allow you to organize the
MVs, CVs, and FVs in your controllers so that you can choose which MVs, CVs, and FVs in
the controller are online.

Creating the MVC Controller

The MVC controller is a controller that calculates the appropriate MV movements, that is, the
MV increments, that minimize the quadratic cost function. See Cost Function on page 195 for
more information.
You can create the MVC controller by using the Final Linear Model tab. The SimSci APC
software builds the MVC controller from the final linear model. The MVC controller contains
the same CVs, MVs, and FVs as the final linear model. You cannot add or delete variables
from the MVC controller after you create it.
To create the MVC controller:
1. On the Final Linear Model tab, ensure that the final linear model contains the desired
variable relationships, order of dynamics, prediction interval, and model length.

2. Click Create Controller .

Configuring the MVC Controller

You can configure the MVC controller by using the Controller Details tab. The MVC
controller contains the following configuration parameters:
Controller Name on page 156
Control Interval on page 156
Coupling on page 156
Controller Model Order on page 157
Long Range Prediction on page 157

Version 2016 SP2 155

SimSci APC User Guide Controller Design

Blocking Width on page 157

Blocking Horizon on page 157
Control Horizon on page 157
Predictions on page 158
Control Modelset on page 158
Watchdog From DCS on page 158
Watchdog To DCS on page 159
Fail Time on page 159
Recover Time on page 159
MVC/LP Modes on page 159
Simulation Modelset on page 159
You can change any of the preceding parameters as desired.
To configure the MVC controller:
1. On the Controller Details tab, on the MVC tab, expand the Configuration area.
2. Change the configuration parameters as desired.

Controller Name
The Controller Name parameter is the name of the MVC controller. It is a string type of
parameter that can contain any alphanumeric character as well as underscores. When you
first create the MVC controller from the final linear model, the SimSci APC software sets the
value of this parameter to MVC1. You can change this value as desired.

Control Interval
The Control Interval parameter is the interval at which the MVC controller calculates the
appropriate MV increments that the SimSci APC software applies to the DCS. When you first
create the MVC controller, the SimSci APC software sets the value of this parameter to the
prediction interval for the final linear model.
The value of this parameter must be a multiple of the scan interval.

Coupling
The Coupling parameter defines the amount of coupling between the MVC controller and
ithe Optimizer. It can have one of three values:
Tightly Coupled: The MVC controller takes the Optimizer values into account and cannot
exceed the bounds set by the Optimizer.
Loosely Coupled: The MVC controller takes the Optimizer values into account, but can
exceed the bounds set by the Optimizer.
None: The MVC controller does not take the Optimizer values into account.
The default value for this parameter is Tightly Coupled.

156 Version 2016 SP2

Controller Design SimSci APC User Guide

Controller Model Order

The Controller Model Order parameter is the order of dynamics for the MVC controller.
When you first create the MVC controller, the SimSci APC software sets the value of this
parameter to the order of dynamics for the final linear model.
You can modify the Controller Model Order, which will update the modelset.

Long Range Prediction

The Long Range Prediction parameter is the number of future data points that the MVC
controller predicts for each CV and MV. You can set this parameter to any integer value that
is greater than zero and less than or equal to the Control Horizon.
When you first create the MVC controller, the SimSci APC software sets the value of this
parameter to the value of the Control Horizon parameter.

Blocking Width
The Blocking Width parameter is the number of control intervals that the SimSci APC
software groups together to reduce the computational demand of the MVC controller when it
minimizes the quadratic cost function.
Within a group, or "block," of control intervals, the SimSci APC software forces the MVs in the
MVC controller to change by an equal amount at each control interval in the group; the MVC
controller calculates the MV increments for each group of control intervals instead of for each
individual control interval. As a result, the time required to design the MVC controller is
significantly reduced.
Typically, increasing the Blocking Width parameter reduces the aggression of controller
performance and may increase the time the MVC controller takes to drive the MVs and CVs
to their target. If the value of the Blocking Width parameter changes significantly, you may
need to adjust the controller tuning by changing the design weights. See Parameters for the
Variables in the MVC Controller for more information.
The default value for this parameter is 10.

Blocking Horizon
The Blocking Horizon parameter is an initial portion of the Control Horizon parameter
within which the cost function is minimized at intervals equal to one control interval. After the
specified number of control intervals pass, the cost function is minimized at intervals equal to
the number of control intervals specified in the Blocking Width parameter.
The default value for this parameter is 5.

Control Horizon
The Control Horizon parameter is the number of control intervals in which the MVC
controller minimizes the cost function. See Cost Function on page 195 for more information.
This value must be sufficiently large to obtain convergence of the design procedure. It should
equal or exceed the settling time of the system under consideration.
When you first create the MVC controller, the SimSci APC software sets the value of this
parameter to the model length for the final linear model.

Version 2016 SP2 157

SimSci APC User Guide Controller Design

Predictions
The Predictions parameter determines how the recursive least squares (RLS) algorithm
calculates the future predictions of the MVs when the SimSci APC software identifies the
linear models (that is, the modelsets) in the MVC controller. It can have one of two values:
Unbiased Predictions: The recursive least squares (RLS) algorithm uses the calculated
model predictions rather than the actual process data to calculate the future prediction of
the MVs.
Biased Predictions: The recursive least squares (RLS) algorithm uses the actual
process data rather than the calculated model predictions to calculate the future
prediction of the MVs.
See Unbiased Least Squares on page 126 for more information.
The default value for this parameter is Unbiased Predictions.

Control Modelset
The Control Modelset parameter is the modelset that contains the linear model coefficients
and the model-predictive gainsets that the MVC controller currently uses as the basis for
control. When the MVC controller is online, the SimSci APC software recalculates the
associated model coefficients and gainsets by using an iterative procedure that involves
minimizing the quadratic cost function over the number of control intervals specified by the
Control Horizon parameter. See Cost Function on page 195 for more information.
The SimSci APC software also recalculates the model coefficients and gainsets when the
value of the Scales parameter for any CV or MV changes or when the value of the Control
Modelset parameter changes.
By default, the SimSci APC software sets this parameter to Model Set 1, which is the
modelset that the SimSci APC software automatically generates when you create the MVC
controller from the final linear model. You can change this parameter to any modelset in the
MVC controller.

Watchdog From DCS

The Watchdog From DCS parameter is the name of the tag that contains the watchdog
signal for communication to the SimSci APC software from the DCS. A watchdog is a tag that
the SimSci APC software can use to monitor the communications between the SimSci APC
software and the DCS. It generates a continuously changing value that is typically a saw-
toothed pattern that ranges from zero to 100. The SimSci APC software continuously tests
the value of the watchdog to ensure that it is changing.
If the value of the watchdog that originates from the DCS does not change, the SimSci APC
software assumes that there is a failure in the communications between the DCS and the
SimSci APC software. If a failure in the communications occurs for the duration of the
specified Fail Time parameter, the MVC controller fails to the Waiting state. See Fail Time on
page 159 for more information.
When you restore communications between the DCS and the SimSci APC software and the
watchdog continues to change for the duration of the specified Recover Time parameter, the
MVC controller automatically initializes to the Ready state. See Recover Time on page 159
for more information.

158 Version 2016 SP2

Controller Design SimSci APC User Guide

You must manually activate the MVC controller to write values from the MVC controller to the
DCS.
By default, this parameter is blank.

Watchdog To DCS
The Watchdog To DCS parameter is the name of the tag that contains the watchdog signal
for communication to the DCS from the SimSci APC software. A watchdog is a tag that can
be used to monitor the communications between the SimSci APC software and the DCS. It
generates a continuously changing value, typically a saw-toothed pattern that ranges from
zero to 100. The value of the watchdog is continuously tested to ensure that it is changing.
If the value of the watchdog that originates from the SimSci APC software does not change,
the DCS assumes that there is a failure in the communications between the SimSci APC
software and the DCS. The DCS then takes action accordingly. See the documentation for
the DCS for more information.
By default, this parameter is blank.

Fail Time
The Fail Time parameter is the time, in seconds, in which a watchdog value can remain static
before the MVC controller fails to the Waiting state.
When you first create the MVC controller, the SimSci APC software sets the value of this
parameter to 2.5 times the prediction interval for the final linear model.

Recover Time
The Recover Time parameter is the time, in seconds, in which a watchdog tag value must
continue to change to indicate that the MVC controller has recovered from a failure due to the
loss of signal communication between the DCS and the SimSci APC software. After the MVC
controller recovers, it automatically initializes to the Ready state. You must manually activate
the MVC controller to write values from the MVC controller to the DCS.
When you first create the MVC controller, the SimSci APC software sets the value of this
parameter to the prediction interval for the final linear model.

MVC/LP Modes
The MVC/LP Modes parameter determines the mode in which the MVC controller and the
Optimizer operate when integrated with DCSs that utilize the Foxboro I/A Series software.
By default, the SimSci APC software sets this parameter to Operator Choice, indicating that
the operator can activate the MVC controller and the Optimizer independently.
You can edit this parameter only when you activate the APC Bridge software. Please see the
SimSci APC Bridge User Guide for more information.

Simulation Modelset
The Simulation Modelset parameter is the modelset that contains the linear model
coefficients and the model-predictive gainsets that the MVC controller uses as the basis for
simulation.

Version 2016 SP2 159

SimSci APC User Guide Controller Design

Parameters for the Variables in an MVC Controller

You can specify the parameters for both input and output variables in an MVC controller by
using the MVC tab on the Controller Details tab, on the tab for the MVC controller. The MVC
tab contains a tab for each type of variable in the MVC controller.
On the MVC tab, on the Control variables tab, you can access the following parameters for
the CVs:
Initial Value on page 162
Vital on page 162
Switch on page 162
Constraint Mode on page 163
Track Setpoint on page 163
Low Constraint on page 164
High Constraint on page 164
Low Limit on page 165
High Limit on page 165
Low Validity on page 165
High Validity on page 166
Scales on page 166
Error Factor on page 166
Deadzone on page 167
Cushion on page 167
Priority on page 167
Low Importance on page 168
High Importance on page 168
Lr Horizon on page 168
Linearizing Function on page 171
Define on page 171
Rate of Change on page 172
Frozen Time on page 172
Infer Time on page 172
Data Valid Time on page 172
Display on page 173
Page Number on page 173
On the MVC tab, on the Manipulated variables tab, you can access the following
parameters for the MVs:

160 Version 2016 SP2

Controller Design SimSci APC User Guide

Initial Value on page 162

Vital on page 162
Switch on page 162
Increment on page 164
Low Constraint on page 164
High Constraint on page 164
Low Limit on page 165
High Limit on page 165
Low Validity on page 165
High Validity on page 166
Scales on page 166
MV Priority on page 167
Readback on page 169
DCS Readback on page 169
DCS Status on page 169
Linearizing Function on page 171
Define on page 171
Display on page 173
Page Number on page 173
On the MVC tab, on the Feed Forward variables tab, you can access the following
parameters for the FVs:
Initial Value on page 162
Vital on page 162
Switch on page 162
Linearizing Function on page 171
Define on page 171
Low Validity on page 165
High Validity on page 166
Rate of Change on page 172
Frozen Time on page 172
Data Valid Time on page 172
Display on page 173
Page Number on page 173

Version 2016 SP2 161

SimSci APC User Guide Controller Design

Initial Value
The Initial Value parameter displays the initial value that the associated variable takes during
a simulation run, and is calculated based on available process data. The initial value should
be in engineering units. See Engineering Units and Normalized Units on page 44 for more
information.
This parameter applies to CVs, MVs, and FVs.
You can edit this parameter only when you activate Simulation mode. See Simulation Mode
on page 194 for more information.

Vital
The Vital parameter is a check box that determines whether or not a variable is vital to the
MVC controller. If you select this check box, the MVC controller considers the associated
variable as vital to the operations of the MVC controller. If you clear this check box, the MVC
controller considers the associated variable as nonessential to the operations of the MVC
controller.
If the MVC controller considers a variable as vital to the operations of the MVC controller, the
MVC controller fails to the Waiting state when the variable is turned off or has a bad quality.
You cannot activate the MVC controller until the vital variable is turned on and has a good
quality.
If the MVC controller considers a variable as nonessential to the operations of the MVC
controller, the MVC controller continues its normal operations and does not use the variable
in any calculations or procedures when the variable is turned off or has a bad quality.
Nonessential CVs are an exception to this rules. If a nonessential CV is turned off or has a
bad quality, the MVC controller uses prediction values for the CV instead of the actual values
for the CV for the amount of time specified by the Infer Time parameter. After the amount of
time specified by the Infer Time parameter elapses, the MVC controller does not use the CV
in any calculations or procedures. See Infer Time on page 172 for more information.
If the MVC controller has a subcontroller and a vital CV or MV is turned off on the
subcontroller, the MVC controller will remain active. See Managing Subcontrollers on page
173 for more information.
You can specify this parameter for CVs, MVs, or FVs.

Switch
The Switch parameter is a list that determines whether a variable is on or off with regards to
MVC controller operations. You can set this parameter to one of the following values:
On: The MVC controller uses the associated variable in all calculation and procedures.
Off: The MVC controller does not use the variable in any calculations or procedures.
Infer: The MVC controller uses the inferred value of the variable in all calculations and
procedures. You can use this option for CVs only. If you want to use this value for a CV,
the CV cannot be vital to the operations of the MVC controller. Typically, the MVC
controller uses the inferred value of a CV when a communication error occurs with the CV
or the CV has a bad quality. The MVC controller can infer the value of a CV only for the
amount of time specified by the Infer Time parameter. See Vital on page 184 and Infer
Time on page 189 for more information.

162 Version 2016 SP2

Controller Design SimSci APC User Guide

FFWD: The MVC controller uses the values of the readback signal for the variable in all
calculations and procedures instead of the value of the variable. It also treats the variable
as a feedforward variable (FV).You can use this option for MVs only. This option is also
known as feedforward mode. If you want to use feedforward mode for an MV, the MV
cannot be vital to the operations of the MVC controller. Typically, the MVC controller uses
feedforward mode for an MV when a communication error occurs with the MV or the MV
has a bad quality. See Vital on page 162 for more information.
You can specify this parameter for CVs, MVs, and FVs.

Constraint Mode
The Constraint Mode parameter determines which method is used for constraint calculations
for each variable. You can set this parameter to one of five values:
Setpoint: The MVC controller controls the value of the variable to be as close as possible
to its associated setpoint.
Controlled Constraints: The MVC controller controls the value of the variable to be
within its high and low constraints.
Setpoint+QP: The MVC controller controls the value of the variable to be as close as
possible to its associated setpoint. If the MVC controller cannot control the value of the
variable tightly to the setpoint due to either an insufficient number of MVs or conflicts with
other CVs, the MVC controller controls the value of the variable to be within its high and
low constraints by using strong quadratic programming (QP) action. We recommend that
you use this configuration only when the CV is very critical to the process.
Controlled+QP: The MVC controller controls the value of the variable to be within its high
and low constraints, if possible. If the variable violates either its high or low constraint, the
MVC controller adjusts the MVs by using strong QP action to bring the value of the
variable back within its constraints. We recommend that you use this configuration only
when the CV is very critical to the process.
QP: The MVC controller uses strong QP action to control the value of the variable to be
within its high and low constraints.
Almost all MVC controllers require only the first two of the preceding configurations, because
the MV action under the QP constraints is very aggressive and typically provides sufficient
control.
The default value for this parameter is Controlled Constraints.
You can specify this parameter for CVs only.

Track Setpoint
The Track Setpoint option determines whether a controlled variable tracks the setpoint to the
actual process value. If you have a CV with Constraint Mode set to Setpoint or Setpoint +
QP, you can enable setpoint tracking.
If you enable setpoint tracking, the controller automatically moves the setpoint value to equal
the parameter's process value when the controller updates to Ready state. SimSci APC
displays the updated setpoint value on the Controller Overview tab. This parameter
provides a bumpless transfer when you activate the controller if the process value is initially
different than the setpoint value.

Version 2016 SP2 163

SimSci APC User Guide Controller Design

By default, the Track Setpoint parameter is not enabled. You can specify this parameter for
CVs only.

Increment
The Increment parameter is the maximum amount by which an MV value can change from
one control action to the next. The value of the Increment parameter represents a constraint
on the rate of change for an MV. The MVC controller respects this constraint when it
calculates the MV increments that minimize the quadratic cost function.
The MVC controller reports the value of the Increment parameter in engineering units.
You can specify this parameter for MVs only.

Low Constraint
The Low Constraint parameter is the lower constraint that you place on the value of a
variable. That is, the value of the variable should not be less than the value of the Low
Constraint parameter.
The MVC controller can move the value of the variable to be less than the lower constraint
during calculations. However, the final solution of the MVC controller cannot contain a
variable with a value that is less than its associated lower constraint.
The MVC controller uses the lower constraint as a soft bound or as a hard bound based on
the associated priority of the variable. For CVs,the value of the Priority parameter
determines the priority (see Priority on page 167). For MVs, the value of the MV Priority
parameter determines the priority (see MV Priority on page 167). If the priority of the variable
is low, the MVC controller uses the lower constraint as a hard bound and does not relax the
bound during calculations. If the priority of the variable is high, the MVC controller uses the
lower constraint as a soft bound and relaxes the bound during calculations, if required.
The default value for this parameter is -1,000.00. You can set the value of this parameter to
any real value that is:
Greater than or equal to the value of the Low Limit parameter.
Less than the value of the High Constraint parameter.
You can specify this parameter for CVs and MVs only.

High Constraint
The High Constraint parameter is the upper constraint that you place on the value of a
variable. That is, the value of the variable should not be greater than the value of the High
Constraint parameter.
The MVC controller can move the value of the variable to be greater than the upper constraint
during calculations. However, the final solution of the MVC controller cannot contain a
variable with a value that is greater than its associated upper constraint.

164 Version 2016 SP2

Controller Design SimSci APC User Guide

The MVC controller uses the upper constraint as a soft bound or as a hard bound based on
the associated priority of the variable. For CVs, the value of the Priority parameter
determines the priority (see Priority on page 167). For MVs, the value of the MV Priority
parameter determines the priority (see MV Priority on page 167). If the priority of the variable
is low, the MVC controller uses the upper constraint as a hard bound and does not relax the
bound during calculations. If the priority of the variable is high, the MVC controller uses the
upper constraint as a soft bound and relaxes the bound during calculations, if required.
The default value for this parameter is 1,000.00. You can set the value of this parameter to
any real value that is:
Greater than or equal to the Low Constraint parameter
Less than or equal to the value of the High Limit parameter
You can specify this parameter for CVs and MVs only.

Low Limit
The Low Limit parameter is the lower limit that you place on the value of a variable. For CVs,
the setpoint value of the variable cannot be less than the lower limit. For MVs, the actuation
value of the variable that the MVC controller calculates cannot be less than the lower limit.
The default value for this parameter is -1,000.00. You can set the value of this parameter to
any real value that is:
Greater than -1,000,000,000.00
Less than the value of the High Limit parameter
You can specify this parameter for CVs and MVs only.

High Limit
The High Limit parameter is the upper limit that you place on the value of a variable. For
CVs, the setpoint value of the variable cannot be greater than the lower limit. For MVs, the
actuation value of the variable that the MVC controller calculates cannot be greater than the
upper limit.
The default value for this parameter is 1,000.00. You can set the value of this parameter to
any real value that is:
Greater than 0.00 or the value of the Low Limit parameter, whichever value is greater
Less than 10,000,000,000.00.
You can specify this parameter for CVs and MVs only.

Low Validity
The Low Validity parameter is the least value of a variable in the validation range. If the
value of the variable is less than the value of the Low Validity parameter, the MVC controller
sets the quality of the variable to bad.
The default value for this parameter is -1,000.00. You can set the value of this parameter to
any real value that is:
Greater than -1,000,000,000.00
Less than the value of the High Validity parameter

Version 2016 SP2 165

SimSci APC User Guide Controller Design

You can specify this parameter for CVs, MVs, and FVs.

High Validity
The High Validity parameter is the greatest value of a variable in the validation range. If the
value of the variable is greater than the value of the High Validity parameter, the MVC
controller sets the quality of the variable to bad.
The default value for this parameter is 1,000.00. You can set the value of this parameter to
any real value that is:
Greater than 0.00 or the value of the Low Validity parameter, whichever is greater
Less than 10,000,000,000.00
You can specify this parameter for CVs, MVs, and FVs.

Scales
The Scales parameter is a normalization factor that the MVC controller uses to normalize the
weighting matrices in the cost function. See Cost Function on page 195 for more information.
The default value for this parameter is 1.00.
You can specify this parameter for CVs, MVs, and FVs.

Error Factor
The Error Factor parameter is a factor that modifies the CV setpoints that the Optimizer
calculates. The MVC controller uses the modified CV setpoints to calculate the various error
terms in the cost function. See Cost Function on page 195 for more information.
The MVC controller calculates the modified CV setpoints by using the following equation:
yi , ss yi ,k yi ,k Si ErrorFacto r
where
y i , ss
is the modified setpoint of the ith CV.
yi , k
is the value of the ith CV at instant k .
Si
is the actual setpoint of the ith CV, which the Opitmizer calculates.
When you modify the CV setpoints, you modify the sensitivity of the MVC controller by
altering the amount of setpoint error that the MVC calculates for each CV. If you set the value
of this parameter to 1.00, the modified CV setpoint is equal to the actual CV setpoint and the
sensitivity of the MVC controller does not change. If you set the value of this parameter to a
value less than 1.00, the modified CV setpoint is less than the actual CV setpoint and the
sensitivity of the MVC controller decreases. That is, the MVC controller responds less
vigorously to actual setpoint errors.
The default value for this parameter is 1.00. You can set the value of this parameter to any
real value that is:
Greater than or equal to zero
Less than or equal to 1.00

166 Version 2016 SP2

Controller Design SimSci APC User Guide

You can specify this parameter for CVs only.

Deadzone
The Deadzone parameter is a bound that you can use to desensitize the MVC controller to
certain levels of error in the CVs. If the absolute value of the error between the CV and its
setpoint is equal to or less than the value of the Deadzone parameter, the error is set to zero.
The default value for this parameter is 0.00.
You can specify this parameter for CVs only.

Cushion
The Cushion parameter is a modification to the effective constraints on a variable. That is,
the MVC controller effectively raises the low constraint and lowers the high constraint for the
associated variable by the amount you specify for the Cushion parameter.
You can use the Cushion parameter to set a safety margin for the constraints. The safety
margin ensures that the value of a CV never rises above the high constraint or falls below the
low constraint due to uncontrollable noise in the process.
You can specify this parameter for CVs only.

Priority
The Priority parameter is a ranking of the relative importance of a CV with regards to
maintaining the degree of freedom in the MVC controller. The control system must have a
sufficient degree of freedom to solve the control problem without using more MVs than
necessary and without attempting to control more CVs than are possible.
You can set the Priority parameter to any integer equal to or greater than 1. A value of 1
represents the highest priority. You can assign the same priority to more than one CV.
If the number of CVs is greater than the number of MVs, the MVC controller controls only an
m number of CVs, where m is the number of MVs. The MVC controller excludes the CVs
with the lowest priority first.
The default value for this parameter is 1.
You can specify this parameter for CVs only.

MV Priority
The MV Priority parameter is a ranking of the relative importance of an MV with regards to
maintaining the degree of freedom in the MVC controller. The control system must have a
sufficient degree of freedom to solve the control problem without using more MVs than
necessary and without attempting to control more CVs than are possible.
You can set the MV Priority parameter to any integer equal to or greater than 1. A value of 1
represents the highest priority. You can assign the same priority to more than one MV.
If the number of MVs is greater than the number of CVs, the MVC controller uses only a c
number of MVs to control the CVs, where c is the number of CVs. The MVC controller
excludes the MVs with the lowest priority first.
The default value for this parameter is 1.
You can specify this parameter for MVs only.

Version 2016 SP2 167

SimSci APC User Guide Controller Design

Low Importance
The Low Importance parameter is a measure of the relative freedom of the MVC controller
to relax the lower constraint on the associated variable. For situations in which it is impossible
to obtain a solution that respects all of the constraints for the variables in the MVC controller,
the MVC controller must relax some of the soft constraints to achieve feasibility and obtain a
solution. You can use the Low Importance parameter to specify the lower constraints that
the MVC controller can relax first.
If you specify a large value (for example, 10,000.00) for this parameter, the MVC controller
relaxes the lower constraint on the associated variable by a very small amount, if at all. If you
specify a small value (for example, 1.00) for this parameter, the MVC controller relaxes the
lower constraint on the associated variable as needed to achieve feasibility.
The default value for this parameter is 1,000.00. You can set the value of this parameter to
any real value that is:
Greater than or equal to zero
Less than 10,000,000,000.00
You can specify this parameter for CVs only.

High Importance
The High Importance parameter is a measure of the relative freedom of the MVC controller
to relax the upper constraint on the associated variable. For situations in which it is
impossible to obtain a solution that respects all of the constraints for the variables in the MVC
controller, the MVC controller must relax some of the soft constraints to achieve feasibility
and obtain a solution. You can use the High Importance parameter to specify the upper
constraints that the MVC controller can relax first.
If you specify a large value (for example, 10,000.00) for this parameter, the MVC controller
relaxes the upper constraint on the associated variable by a very small amount, if at all. If you
specify a small value (for example, 1.00) for this parameter, the MVC controller relaxes the
upper constraint on the associated variable as needed to achieve feasibility.
The default value for this parameter is 1,000.00. You can set the value of this parameter to
any real value that is:
Greater than or equal to zero
Less than 10,000,000,000.00
You can specify this parameter for CVs only.

Lr Horizon
The Lr Horizon parameter is the number of control intervals for which the MVC controller
predicts the future trajectory of a variable. The MVC controller uses the predictions to
anticipate any future violations of the constraints for the variable. You should set the value of
the Lr Horizon parameter to a value that is equal to or greater than the settling time of the
associated variable.
The default value for this parameter equals the model length for the final linear model that
you use to create the MVC controller.
You can specify this parameter for CVs only.

168 Version 2016 SP2

Controller Design SimSci APC User Guide

Readback
The Readback parameter is a check box that determines whether the MVC controller
archives data for the previous increments of the associated variable or for the previous
increments of the readback tag for the associated variable. The MVC controller maintains an
internal archive of previous control moves that it uses to correctly predict and compute the
appropriate control moves.
If you select this check box, the MVC controller uses the previous increments of the readback
tag for the associated MV to compute the appropriate control moves. The MVC controller
uses the tag specified by the DCS Readback parameter as the readback tag for the
associated MV. See DCS Readback on page 169 for more information.
If you clear this check box, the MVC controller uses the previous increments of the
associated MV to compute the appropriate control moves.
When you archive data for the previous increments of the readback tag, the MVC controller
computes the control moves based on actual data from the process instead of its own
predictions. Therefore, we recommend that you select this check box whenever possible.
You can specify this parameter for MVs only.

DCS Readback
The DCS Readback parameter is the readback tag that contains the process data for the
variable that the MVC controller manipulates by changing the associated MV. That is, the
MVC controller makes changes to the MV to control the readback tag. For example, if the
MVC controller uses the associated MV to change the setpoint of a reflux flow PID controller,
the readback tag for the associated MV typically is the input tag that monitors the actual reflux
flowrate in the process plant.
You can use the readback tag to:
Compute the control moves based on actual data from the process. See Readback on
page 169 for more information.
Implement anti-windup protection. See DCS Status on page 169 for more information.
Validate the value of the associated MV with regards to its upper and lower validity limits.
See Low Validity on page 165 and High Validity on page 166 for more information.
You can specify this parameter for MVs only.

DCS Status
The DCS Status parameter is an input tag that monitors the wind-up status of the underlying
regulatory control loop in the plant process.

Version 2016 SP2 169

SimSci APC User Guide Controller Design

The MVs in an APC Project are usually output tags that provide suggested actuation values
to a physical controller in the plant process. Typically, the actuation values take the form of
setpoints that are inputs to the physical controllers. The physical controllers then calculate
actuation values based on their internal regulatory loops. A regulatory loop is "wound up"
when it reaches the maximum or minimum actuation value and can no longer make
adjustments to the process value of a variable by means of the actuator. For example, a
valve that is fully open can no longer increase flow to meet the requested setpoint:

Typically, the DCS Status tag has one of four values:

0: The PID loop is not wound up.
1: The PID loop is wound up high.
2: The PID loop is wound down low.
3: The PID loop is not in supervisory mode, that is, the PID loop is not monitoring its wind-
up status.
Foxboro I/A Series platforms use the preceding values. Other DCS platforms provide similar
wind-up information, but may assign different values to particular conditions. You may need
to translate the values of the DCS Status tag to match the preceding values. See The
Translation Table on page 77 for more information.
If the MVC controller detects a high PID wind-up:
The MVC controller temporarily overrides the specified value of the High Constraint
parameter for the associated MV with the current value of the readback tag. See High
Constraint on page 164 and DCS Readback on page 169 for more information.

170 Version 2016 SP2

Controller Design SimSci APC User Guide

The MVC controller uses the previous increments of the readback tag for the associated
MV to compute the appropriate control moves. That is, it selects the Readback check
box. See Readback on page 169 for more information.
If the MVC controller detects a low PID wind-up:
The MVC controller temporarily overrides the specified value of the Low Constraint
parameter for the associated MV with the current value of the readback tag. See Low
Constraint on page 164 and DCS Readback on page 169 for more information.
The MVC controller uses the previous increments of the readback tag for the associated
MV to compute the appropriate control moves. That is, it selects the Readback check
box. See Readback on page 169 for more information.
When the wind-up condition clears, the MCV controller incrementally returns the constraint
limit for the associated MV to its previous value over 10 successive scan intervals.
You can specify this parameter for MVs only.

Linearizing Function
The Linearizing Function parameter is a function that the MVC controller applies to the
values of the associated variable to provide a linear relationship for the variable. The MVC
controller transforms the process data for the associated variable based on the function that
you select, and then uses the newly calculated values for all calculations and procedures.
You can select one of six functions for this parameter:
NONE: The MVC controller uses the process values of the associated variables for all
calculations and procedures.
LN(): The MVC controller uses the natural log of the process values of the associated
variable for all calculations and procedures.
EXP(): The MVC controller uses the natural exponential function for the value of the
associated variable for all calculations and procedures.
SQUARE: The MVC controller uses the square of the value of the associated variable for
all calculations and procedures.
SQUARE ROOT(): The MVC controller uses the square root of the value of the
associated variable for all calculations and procedures.
XY Lookup(): The MVC controller uses a piecewise nonlinear transform function for all
calculations and procedures. You should define a lookup table for this function by using
the Define button. See Define on page 171 for more information.
You can specify this parameter for CVs, MVs, and FVs.

Define
The Define parameter contains a button that opens the Piecewise Nonlinear Transform
dialog box. You can use the Piecewise Nonlinear Transform dialog box to define a lookup
table for piecewise nonlinear transforms. See Piecewise Nonlinear Transforms on page 193
for more information.

You cannot click the Define button unless you set the Linearizing Function parameter
to XY Lookup().

Version 2016 SP2 171

SimSci APC User Guide Controller Design

Rate of Change
The Rate of Change parameter is the maximum amount by which the value of a variable can
change from one data point to the next. The value of the Rate of Change parameter
represents a validation constraint on the rate of change for a CV or an FV. If the value of the
CV or FV changes by an amount that is greater than the value of the Rate of Change
parameter, the MVC controller sets the quality of the CV or FV to bad.
If you set the value of the Rate of Change parameter to zero, the MVC controller does not
consider this validation constraint for the associated CV or FV.
The default value for this parameter is 0.00.
You can specify this parameter for CVs and FVs only.

Frozen Time
The Frozen Time parameter is the maximum amount of time in which the value of a variable
can remain unchanged. The value of the Frozen Time parameter represents a validation
constraint on the dynamic quality of a CV or an FV. If the value of the CV or FV is static for a
duration that is greater than the value of the Frozen Time parameter, the MVC controller sets
the quality of the CV or FV to bad.
If you set the value of the Frozen Time parameter to zero, the MVC controller does not
consider this validation constraint for the associated CV or FV.
The default value for this parameter is 0.00.
You can specify this parameter for CVs and FVs only.

Infer Time
The Infer Time parameter is the amount of time in which the MVC controller infers values for
a variable. The Infer Time parameter applies to nonessential variables only, that is, variables
that are not specified as vital to the operations of the MVC controller. See Vital on page 162
for more information.
If a nonessential CV is turned off or has a bad quality, the MVC controller uses prediction
values for the CV instead of the actual values for the CV for the amount of time specified by
the Infer Time parameter. After the amount of time specified by the Infer Time parameter
elapses, the MVC controller does not use the CV in any calculations or procedures.
If you set the value of the Infer Time parameter to zero, the MVC controller does not infer
values for the CV and immediately stops using the CV in any calculations or procedures if the
CV is turned off or has a bad quality.
The default value for this parameter is 0.00.
You can specify this parameter for CVs only.

Data Valid Time

The Data Valid Time parameter is the minimum amount of time in which the value of a
variable must remain valid to set the quality of the variable to good. The Data Valid Time
parameter applies to nonessential variables only, that is, variables that are not specified as
vital to the operations of the MVC controller. See Vital on page 162 for more information.

172 Version 2016 SP2

Controller Design SimSci APC User Guide

If the MVC controller excludes the associated variable from any calculations or procedures
due to the bad quality of the variable, the variable must have a good quality for the amount of
time specified by the Data Valid Time parameter to once again include the associated
variable in all calculations and procedures in the MVC controller.
You can specify this parameter for CVs and FVs only.

Display
The Display parameter determines which variables are included in the standard I/A Series
displays created by the APC Bridge software. This parameter is specific to integration with
DCSs that utilize the Foxboro I/A Series software.
You can set this parameter to one of two values:
Yes: The variable is included in the I/A Series display.
No: The variable is not included in the I/A Series display.
The default value for this parameter is Yes.
This parameter applies to CVs, MVs, and FVs.
You can edit this parameter only when you activate the APC Bridge software. Please see the
SimSci APC Bridge User Guide for more information.

Page Number
The Page Number parameter determines the location of a variable in the standard I/A Series
displays created by the APC Bridge software. This parameter is specific to integration with
DCSs that utilize the Foxboro I/A Series software.
The default value for this parameter is 0.
This parameter applies to CVs, MVs, and FVs.
You can edit this parameter only when you activate the APC Bridge software. Please see the
SimSci APC Bridge User Guide for more information.

Subcontrollers
A subcontroller is a collection of CVs and MVs within an MVC controller. The subcontroller
allows you to selectively put online or take offline a set of variables in a controller.
The subcontroller contains the same parameters for the CVs and MVs as does the main MVC
controller. Any changes you make to these parameters in the subcontroller are automatically
carried over to the main MVC controller.

Managing Subcontrollers
You can create a subcontroller within an MVC controller. After you create the subcontroller,
you can add CVs and MVs to it, delete it, or rename it.
You can view an overview of the subcontrollers within an MVC controller on the Sub
Controller Summary tab. Note that you can open the Sub Controller Summary tab by
clicking Sub Controller Summary on the APC Explorer, under Promote.

Version 2016 SP2 173

SimSci APC User Guide Controller Design

Acti vity Directions

Creating a subcontroller On the Controller Details tab, on the tab for the controller pair,
right-click the Sub Controller tab, and then click Add Sub
Controller.

Renaming a subcontroller On the Controller Details tab, on the tab for the controller pair,
on the Sub Controller tab, double-click the title of the associated
subcontroller tab. Type a new name for the subcontroller, and
then press Enter.

Adding controller variables On the Controller Details tab, on the tab for the controller pair,
(CVs) to a subcontroller on the MVC tab, on the Control variables tab, in the
SubController list for each CV you want to add to a
subcontroller, expand the SubController list, and then select the
check boxes for the subcont roller to which you want to add the
CV. Note that you can add a single CV to multiple subcontrollers.

Adding manipulat ed variables On the Controller Details tab, on the tab for the controller pair,
(MVs) to a subcontroller on the the MV C tab, on the Manipulated variables tab, in the
SubController list for each MV you want to add to a
subcontroller, select the subcontroller to which you want to add
the MV. Note that you can add a single MV to only one
subcontroller.

Turning off a subcontroller On the Controller Details tab, on the tab for the controller pair,
on the the Sub Controller tab, on the tab for the associated
subcontroller, expand the Configuration area. In the Switch list,
select Off.

Turning on a subcontroller On the Controller Details tab, on the tab for the controller pair,
on the the Sub Controller tab, on the tab for the associated
subcontroller, expand the Configuration area. In the Switch list,
select On.

Deleting a subcontroller On the Controller Details tab, on the tab for the controller pair,
on the the Sub Controller tab, right-click the title of the
associated subcont roller tab, and then click Delete
SubController.

Viewing the steady-state On the Sub Controller Summary tab, on the tab for the
responses for variable associated MVC controller, in the Sub-Controller Gridview,
relationships in a subcontroller select the check box for a subc ontroller to highlight the s teady-
state respons es for the variable relationships in that
subcontroller.

Configuring Subcontrollers
You can configure a subcontroller by using the Controller Details tab. The subcontroller
contains the following configuration parameters:
Vital on page 175
Switch (Subctrlr) on page 175

174 Version 2016 SP2

Controller Design SimSci APC User Guide

You can change any of the preceding parameters as desired.

To configure a subcontroller:
1. On the Controller Details tab, on the tab for the controller pair, on the Sub Controller
tab, on the tab for the associated subcontroller, expand the Configuration area.
2. Change the configuration parameters as desired.

Vital
The Vital parameter is a check box that determines whether or not the subcontroller is vital to
the MVC controller. If you select this check box, the MVC controller considers the associated
subcontroller as vital to the operations of the MVC controller. If you clear this check box, the
MVC controller considers the associated subcontroller as nonessential to the operations of
the MVC controller.
If the MVC controller considers a subcontroller as vital to the operations of the MVC
controller, the MVC controller fails to the Waiting state when the subcontroller is turned off.
You cannot activate the MVC controller until the vital subcontroller is turned on.
If the MVC controller considers a subcontroller as nonessential to the operations of the MVC
controller, the MVC controller continues its normal operations and does not use the variables
in the subcontroller in any calculations or procedures when the subcontroller is turned off.
When you turn off a subcontroller, the SimSci APC software automatically turns off the CVs
and MVs in the subcontroller.
Note that you can set the value of this parameter on the Controller Details tab, the
Controller Overview tab, or the Sub Controller Summary tab.

Switch (Subctrlr)
The Switch parameter is a list that determines whether a subcontroller is on or off with
regards to MVC controller operations. You can set this parameter to one of the following
values:
On: The MVC controller uses the associated subcontroller and its associated CVs and
MVs in all calculation and procedures.
Off: The MVC controller does not use the subcontroller or its associated CVs and MVs in
any calculations or procedures.
Note that you can set the value of this parameter on the Controller Details tab, the
Controller Overview tab, or the Sub Controller Summary tab.

Modelsets
A modelset is the set of model coefficients that the MVC controller uses to predict future
values for the CVs within the MVC controller. See Cost Function on page 195 for more
information on how the MVC controller uses the predicted values for the CVs.
Modelsets are specific to a controller pair, that is, the MVC controller and the Optimizer. The
controller pair can contain multiple modelsets. The controller pair can use only one modelset
at a time. You can specify which modelset the controller pair uses by using the Control
Modelset configuration parameter for the MVC controller. See Control Modelset on page 158
for more information.

Version 2016 SP2 175

SimSci APC User Guide Controller Design

Managing Modelsets
The SimSci APC software automatically creates a modelset named Model Set 1 when you
create the MVC controller from the final linear model. You can rename, copy, export, or
delete this modelset as desired. Note that you cannot delete the modelset that the MVC
controller uses for calculations.
When you export a modelset, the SimSci APC software saves the exported modelset as a
.model file, which you can then import into another APC Project. With the .model extension
you can also import it as a modelset or as a linear model. The exported modelset contains
data for all the model coefficients for all variable relationships present in the modelset. You
cannot export data for a single variable relationship.
You can create a blank modelset into which you can import an exported modelset from
another APC Project. The blank modelset must have the same tag structure as the exported
modelset to successfully import the exported modelset.
Typically, the quickest way to create a new modelset is to create a copy of an existing
modelset (for example, Model Set 1), and then modify the model coefficients in the modelset
to suite your needs. You can modify the model coefficients in the modelset by using the same
curve operations and procedures that you use with the model coefficients for a linear model.
See Modifying the Model Coefficients on page 133 for more information.
When you import a modelset, you can choose to import the entire modelset or a selection of
the variable relationships within the modelset. The SimSci APC software always overwrites
the current model coefficients with the model coefficients that you import, even if you modified
the current coefficients by using curve operations.
If you import a modelset that has a model length that is greater than the model length of the
associated MVC controller, the SimSci APC software increases the model length of the
associated MVC controller to equal that of the imported modelset. That is, it increases the
model length of all other modelsets within the MVC controller to match the model length of
the imported modelset. To account for the increase in model length in the other modelsets, it
adds a number of model coefficients to the end of the model coefficient set for each variable
relationship and sets the value of these added model coefficients to zero.
If you import a modelset that has a model length that is less than the model length of the
associated MVC controller, the SimSci APC software sets the value of any missing model
coefficients to zero.
You cannot import a modelset that has a control interval that does not equal the control
interval of the associated MVC controller.
You can import a modelset of a lower order of dynamics than the controller. You can also
update a controller with a modelset of a lower order of dynamics.
You can copy and paste individual model coefficient curves in a modelset.

Acti vity Directions

Creating a modelset After you complete the configuration for the final linear model, on

the Final Linear Model tab, click Create Controller .

Creating a blank modelset On the Controller Details tab, right-click the ModelSets tab, and
then click Add Model Set.

176 Version 2016 SP2

Controller Design SimSci APC User Guide

Renaming a modelset On the Controller Details tab, on the ModelSets tab, double-
click the title of the associated modelset tab. Type a new name
for the modelset, and then press Enter.

Exporting a modelset On the Controller Details tab, on the ModelSets tab, right-click
the associated modelset tab, and then click Export Model Set.
The Export Model Set data dialog box appears. Browse to the
folder in which you want to save the .model file. In the File name
box, type the name of the .model file, and then click Save.

Importing an entire exported On the Controller Details tab, on the ModelSets tab, right-click
modelset the associated modelset tab, and then click Import Model Set.
The Import Model Set data dialog box appears. Browse to and
select the .model file you want to import, and then click Open.
The existing modelset must have the same tag structure as the
modelset you want to import.

Importing a selection of On the Controller Details tab, on the ModelSets tab, right-click
variable relationships within an the associated modelset tab, and then click Import Model Set
exported modelset With Selection. The Import Model Set data dialog box appears.
Browse to and select the .xml file you want to import, and then
click Open. The Applicable Curves dialog box appears. Select
the associated check box for each variable relationship you want
to import, and then click Import Data. The existing modelset
must contain the variable relationships you want to import.

Deleting a modelset On the Controller Details tab, on the ModelSets tab, right-click
the associated modelset tab, and then click Delete Model Set.

Copying a modelset On the Controller Details tab, on the ModelSets tab, right-click
the associated modelset tab, and then click Copy Model Set.

Viewing the steady-state On the Controller Details tab, on the ModelSets tab, on the
responses for a modelset associated modelset tab, click Steady State.

Viewing the impulse model On the Controller Details tab, on the ModelSets tab, on the
coefficients for a modelset associated modelset tab, click Impulse.

Copying a curve from a On the Controller Details tab, on the ModelSets tab, on the
modelset. associated modelset tab, right-click the model coefficient curve
that you want to copy, and then click Copy Curve.

Pasting a curve into a After you copy a curve from a modelset, on the Controller
modelset Details tab, on the ModelSets tab, on the associated modelset
tab, right-click the model coefficient curve into which you want to
paste the copied curve, and then click Paste Curve.

Version 2016 SP2 177

SimSci APC User Guide Controller Design

Creating the Optimizer

The Optimizer is the controller that calculates the optimum values of the CVs and the MVs
that maximize the profitability of the plant process. It contains the set of steady-state values
that the Optimizer requires to maximize the profit function (see Profit Function on page 201).
The steady-state values act as the setpoints for the CVs and the applicable steady-state
targets for the MVs, which the MVC controller requires to minimize the cost function (see Cost
Function on page 195).
You can create the Optimizer by using the Final Linear Model tab. The SimSci APC software
builds the Optimizer from the final linear model. The Optimizer contains the same CVs, MVs,
and FVs as the final linear model.
To create the Optimizer:
1. On the Final Linear Model tab, ensure that the final linear model contains the desired
variable relationships, order of dynamics, prediction interval, and model length.

2. Click Create Controller .

Configuring the Optimizer

You can configure the Optimizer by using the Controller Details tab. The Optimizer contains
the following configuration parameters:
Optimizer Name on page 178
Control Interval on page 179
Optimize Interval on page 179
Origin Source on page 179
Iteration on page 179
Control GainSet on page 180
You can change any of the above parameters as desired.
To configure the Optimizer:
1. On the Controller Details tab, on the Optimizer tab, expand the Configuration area.
2. Change the configuration parameters as desired.

Optimizer Name
The Optimizer Name parameter is the name of the Optimizer. It is a string type of parameter
that can contain any character. It is automatically updated each time that you create the
Optimizer from the final linear model.

178 Version 2016 SP2

Controller Design SimSci APC User Guide

Control Interval
The Control Interval parameter is the interval at which the Optimizer implements the
optimum steady-state values for the MVs and the CVs with regards to the MVC controller.
The optimum steady-state values represent the MV steady-state targets and the CV setpoints
that the MVC controller uses in the calculations for the cost function. See Cost Function on
page 195 for more information.
The Optimizer calculates the optimum steady-state values each time it calculates a solution
to the optimization problem. The Optimize Interval parameter determines the frequency at
which the the Optimizer calculates the optimum steady-state values. See Optimize Interval on
page 179 for more information.
The value of this parameter must be a multiple of the scan interval.

Optimize Interval
The Optimize Interval parameter is the interval at which the Optimizer calculates a solution
to the optimization problem, that is, maximizes the profit function while taking into account all
control constraints. The results of the solution are the optimum steady-state values for the
MVs and the CVs. The Control Interval parameter determines the frequency at which the
optimum steady-state values are implemented with regards to the MVC controller. See
Control Interval on page 179 for more information.
The value of this parameter must be a multiple of the Control Interval parameter.

Origin Source
The Origin Source parameter is the source that the Optimizer uses to calculate the set of
MV, CV, and FV values that are the starting point, that is, the origin, for the maximization of
the profit function. See Profit Function on page 201 for more information.
The Origin Source parameter can have one of three values:
MVC model for Steady State: Uses the steady-state linear model from the MVC
controller to calculate the origin of the maximization of the profit function.
Origin: Uses a fixed set of specified values for the CVs, MVs, and FVs as the origin of the
maximization of the profit function. You can update the fixed set of values for the CVs,
MVs and FVs by using a Director calculation.
LP ARX Model: Uses the steady-state linear model from the LP process as the origin of
the maximization of the profit function.
By default, the SimSci APC software sets this parameter to MVC model for Steady State. If
the MVC controller and the Optimizer are tightly coupled, you cannot change this parameter
from the default value.

Iteration
The Iteration parameter is the number of iterations that are performed to calculate the values
of the CVs and MVs that are used for the origin of the maximization of the profit function. See
Profit Function on page 201 for more information.
The value of this parameter must be large enough to obtain convergence of the steady-state
model that is used to determine the origin of the maximization of the profit function.
The default value for this parameter is 0.
Version 2016 SP2 179
SimSci APC User Guide Controller Design

Control GainSet
The Control Gainset parameter is the gainset that contains the gains that the Optimizer uses
for the calculations. See Profit Function on page 201 for more information.
By default, the SimSci APC software sets this parameter to Gain Set 1, which is the gainset
that the SimSci APC software automatically generates when you create the Optimizer from
the final linear model. You can change this parameter to any gainset in the Optimizer.

Parameters for the Variables in the Optimizer

You can specify the parameters for both input and output variables in the Optimizer by using
the Optimizer tab on the Controller Details tab. The Optimizer tab contains a tab for each
type of variable in the Optimizer.
On the Optimizer tab, on the Control variables tab, you can access the following
parameters for the CVs:
Cost on page 181
Low Importance on page 182
High Importance on page 182
Integrating on page 182
Integrating Scale on page 183
Vital on page 184
Switch on page 184
Low Constraint on page 185
High Constraint on page 185
Low Limit on page 186
High Limit on page 186
Low Validity on page 187
High Validity on page 187
Cushion on page 187
Linearizing Function on page 188
Rate of Change on page 188
Frozen Time on page 189
Infer Time on page 189
Data Valid Time on page 189
On the Optimizer tab, on the Manipulated variables tab, you can access the following
parameters for the MVs:
Cost on page 181
Increment on page 181
Vital on page 184
180 Version 2016 SP2
Controller Design SimSci APC User Guide

Switch on page 184

Low Constraint on page 185
High Constraint on page 185
Low Limit on page 186
High Limit on page 186
Low Validity on page 187
High Validity on page 187
Linearizing Function on page 188
On the Optimizer tab, on the Feed Forward variables tab, you can access the following
parameters for the FVs:
Switch on page 184
Vital on page 184
Linearizing Function on page 188
Low Validity on page 187
High Validity on page 187
Rate of Change on page 188
Frozen Time on page 189
Data Valid Time on page 189

Cost
The Cost parameter is the cost factor associated with the variable. The cost factor
determines the relative contribution of its associated variable to the profit function. That is, the
cost factor influences the relative significance of the variable to the LP solution. See Profit
Function on page 201 for more information.
You can specify this parameter for CVs and MVs only.

Increment
The Increment parameter is the maximum amount by which an MV value can change from
one iteration to the next. The value of the Increment parameter represents a constraint on
the rate of change for an MV. The Optimizer respects this constraint when it calculates the
optimum steady-state values that maximize the profit function. The value of the Increment
parameter therefore determines the rate at which the LP solution moves toward the true
optimum. If you set this parameter too low, the Optimizer may not reach the optimum LP
solution by the end of the iterations for the maximization of the profit function. You should set
this parameter to a large value so that it does not constrain the LP solution. See Profit
Function on page 201 for more information.
The Optimizer reports the value of the Increment parameter in engineering units.
You can specify this parameter for MVs only.

Version 2016 SP2 181

SimSci APC User Guide Controller Design

Low Importance
The Low Importance parameter is a measure of the relative freedom of the Optimizer to
relax the lower constraint on the associated variable. For situations in which it is impossible to
obtain a solution that respects all of the constraints for the variables in the Optimizer, the
Optimizer must relax some of the soft constraints to achieve feasibility and obtain a solution.
You can use the Low Importance parameter to specify the lower constraints that the
Optimizer can relax first.
If you specify a large value (for example, 10,000.00) for this parameter, the Optimizer relaxes
the lower constraint on the associated variable by a very small amount, if at all. If you specify
a small value (for example, 1.00) for this parameter, the Optimizer preferentially relaxes the
lower constraint on the associated variable as needed to achieve feasibility.
The default value for this parameter is 1,000.00. You can set the value of this parameter to
any real value that is:
Greater than or equal to zero
Less than 10,000,000,000.00
You can specify this parameter for CVs only.

High Importance
The High Importance parameter is a measure of the relative freedom of the Optimizer to
relax the upper constraint on the associated variable. For situations in which it is impossible
to obtain a solution that respects all of the constraints for the variables in the Optimizer, the
Optimizer must relax some of the soft constraints to achieve feasibility and obtain a solution.
You can use the High Importance parameter to specify the upper constraints that the
Optimizer can relax first.
If you specify a large value (for example, 10,000.00) for this parameter, the Optimizer relaxes
the upper constraint on the associated variable by a very small amount, if at all. If you specify
a small value (for example, 1.00) for this parameter, the Optimizer preferentially relaxes the
upper constraint on the associated variable as needed to achieve feasibility.
The default value for this parameter is 1,000.00. You can set the value of this parameter to
any real value that is:
Greater than or equal to zero
Less than 10,000,000,000.00
You can specify this parameter for CVs only.

Integrating
The Integrating parameter is a check box that indicates whether or not a variable exhibits
integrating behavior in response to process disturbances. You can specify this parameter for
CVs only.
Through linear programming (LP), the Optimizer considers only the steady-state behavior of a
system. Many process systems contain integrating variables. Integrating variables do not
settle at a new steady state when disturbed and thus, do not have an overall steady-state.
Instead, these variables continue to increase or decrease at a constant rate. For example,
vessel levels, which are commonly required to participate in a constrained optimization
scheme, typically exhibit integrating behavior.
182 Version 2016 SP2
Controller Design SimSci APC User Guide

An integrating variables does not have a set of steady-state gains linking it to the incremental
changes in the process inputs. Therefore, you cannot directly include an integrating variable
in the Optimizer. Otherwise, the LP fails. Instead, you must use the rate of change of an
integrating variable to include it in the Optimizer. The rate of change of a integrating variable
has a conventional steady-state relationship with the system inputs and does not cause an
LP failure.
If you select the Integrating check box, the Optimizer uses the rate of change of the variable
when it performs calculations and procedures and returns the steady-state gains and steady-
state reference value for the rate of change of the CV instead of for the instantaneous value
of the CV.When you select this check box for a CV:
The Optimizer assumes that the CV is vital to the operations of the Optimizer. You cannot
select or clear the Vital check box for the CV.
The Optimizer assumes that the CV is always on. You cannot select or clear the Switch
check box for the CV.
The High Constraint and Low Constraint parameters for the CV apply to the rate of
change of the CV instead of the instantaneous value of the CV. Typically, the constraints
are set close to zero so that the Optimizer balances the integrating variable, making the
integrating variable static at some arbitrary instantaneous value. See High Constraint on
page 185 and Low Constraint on page 185 for more information.
The High Limit and Low Limit parameters for the CV apply to the rate of change of the
CV instead of the instantaneous value of the CV. See High Limit on page 186 and Low
Limit on page 186 for more information.
The High Validity and Low Validity parameters for the CV apply to the rate of change of
the CV instead of the instantaneous value of the CV. The values of the validity limits also
determine the normalization range for the rate of change of the CV instead of the
instantaneous value of the CV. See High Validity on page 187 and Low Validity on page
187 for more information.
If you clear the Integrating check box, the Optimizer uses the instantaneous value of the CV
when it performs calculations and procedures.
By default, the Integrating check box is cleared.

Integrating Scale
The Integrating Scale parameter is the scaling that the Optimizer uses to normalize the rate
of change for the associated integrating variable. The value of this parameter is given in
engineering units per second. It should reasonably encompass the maximum rate of change
that the variable can experience. The Optimizer applies the value of this parameter in an
additive and subtractive sense () around zero to normalize the rate of change and account
for negative rates of change. The Optimizer calculates the normalized value of an integrating
variable according to the following equation:
ValueScaled ScaleInt
ValueNorm 1
2ScaleInt
where
ValueNorm
is the normalized value. Note that this value must be between 1.0 and 2.0.

Version 2016 SP2 183

SimSci APC User Guide Controller Design

ValueScaled
is the value to be scaled, that is, the rate of change of the variable.
ScaleInt
is the value of the Integrating Scale parameter.
This parameter applies to integrating variables only. That is, you can set this parameter only
when you select the Integrating check box for the associated variable. See Integrating on
page 182 for more information.
You can specify this parameter for CVs only.

Vital
The Vital parameter is a check box that determines whether or not a variable is vital to the
Optimizer. If you select this check box, the Optimizer considers the associated variable as
vital to the operations of the Optimizer. If you clear this check box, the Optimizer considers
the associated variable as nonessential to the operations of the Optimizer.
If the Optimizer considers a variable as vital to the operations of the Optimizer, the Optimizer
fails to the Waiting state when the variable is turned off or has a bad quality. You cannot
activate the Optimizer until the vital variable is turned on and has a good quality.
If the Optimizer considers a variable as nonessential to the operations of the Optimizer, the
Optimizer continues its normal operations and does not use the variable in any calculations or
procedures when the variable is turned off or has a bad quality. Nonessential CVs are an
exception to this rule. If a nonessential CV is turned off or has a bad quality, the Optimizer
uses prediction values for the CV instead of the actual values for the CV for the amount of
time specified by the Infer Time parameter. After the amount of time specified by the Infer
Time parameter elapses, the Optimizer does not use the CV in any calculations or
procedures. See Infer Time on page 172 for more information.
If you specify a variable as integrating, the Optimizer automatically considers the variable as
vital to the operations of the Optimizer. See Integrating on page 182 for more information.
If the MVC controller is tightly coupled to the Optimizer, the Optimizer uses the value of the
Vital parameter from the MVC controller instead of the value of the Vital parameter from the
Optimizer.
You can specify this parameter for CVs, MVs, or FVs.

Switch
The Switch parameter is a list that determines whether a variable is on or off with regards to
Optimizer operations. You can set this parameter to one of the following values:
On: The Optimizer uses the associated variable in all calculation and procedures.
Off: The Optimizer does not use the variable in any calculations or procedures.
Infer: The Optimizer uses the inferred value of the variable in all calculations and
procedures. You can use this value for CVs only. If you want to use this value for a CV,
the CV cannot be vital to the operations of the Optimizer. Typically, the Optimizer uses
the inferred value of a CV when a communication error occurs with the CV or the CV has
a bad quality. The Optimizer can infer the value of a CV only for the amount of time
specified by the Infer Time parameter. See Vital on page 184 and Infer Time on page 189
for more information.

184 Version 2016 SP2

Controller Design SimSci APC User Guide

If you specify a variable as integrating, the Optimizer assumes that the CV is always on. See
Integrating on page 182 for more information.
If the MVC controller is tightly coupled to the Optimizer, the Optimizer uses the value of the
Switch parameter from the MVC controller instead of the value of the Switch parameter from
the Optimizer.
You can specify this parameter for CVs, MVs, and FVs.

Low Constraint
The Low Constraint parameter is the lower constraint that you place on the value of a
variable. That is, the value of the variable should not be less than the value of the Low
Constraint parameter.
The Optimizer can move the value of the variable to be less than the lower constraint during
calculations. However, the final solution of the Optimizer cannot contain a variable with a
value that is less than its associated lower constraint.
Typically, the lower constraints in the Optimizer should be equal to or greater than the
corresponding lower constraints in the MVC controller. This forces the Optimizer to locate an
optimum operating point that the MVC controller can achieve. The SimSci APC software does
not have an internal check to ensure that the lower constraints in the Optimizer are
compatible with the lower constraints in ithe MVC controller. You should set the lower
constraints in the Optimizer with care.
If you specify a variable as integrating (see Integrating on page 182 for more information), the
lower constraint for the variable applies to the rate of change of the variable instead of the
instantaneous value of the variable. Typically, the upper and lower constraints for an
integrating variable are set close to zero so that the Optimizer balances the integrating
variable, making the integrating variable static at some arbitrary instantaneous value. If you
set the lower constraint for an integrating variable to a nonzero value, the Optimizer
calculates the optimum rate of change of the variable, which causes the value of the variable
to change constantly at the optimum rate.
If the MVC controller is tightly coupled to the Optimizer, the Optimizer uses the value of the
Low Constraint parameter from the MVC controller instead of the value of the Low
Constraint parameter from the Optimizer.
The default value for this parameter is -1,000.00. You can set the value of this parameter to
any real value that is:
Greater than or equal to the value of the Low Limit parameter.
Less than the value of the High Constraint parameter.
You can specify this parameter for CVs and MVs only.

High Constraint
The High Constraint parameter is the upper constraint that you place on the value of a
variable. That is, the value of the variable should not be greater than the value of the High
Constraint parameter.
The Optimizer can move the value of the variable to be greater than the upper constraint
during calculations. However, the final solution of the Optimizer cannot contain a variable with
a value that is greater than its associated upper constraint.

Version 2016 SP2 185

SimSci APC User Guide Controller Design

Typically, the upper constraints in the Optimizer should be less than or equal to the
corresponding upper constraints in the MVC controller. This forces the Optimizer to locate an
optimum operating point that the MVC controller can achieve. The SimSci APC software does
not have an internal check to ensure that the upper constraints in the Optimizer are
compatible with the upper constraints in the MVC controller. You should set the upper
constraints in the Optimizer with care.
If you specify a variable as integrating (see Integrating on page 182 for more information), the
upper constraint for the variable applies to the rate of change of the variable instead of the
instantaneous value of the variable. Typically, the upper and lower constraints for an
integrating variable are set close to zero so that the Optimizer balances the integrating
variable, making the integrating variable static at some arbitrary instantaneous value. If you
set the upper constraint for an integrating variable to a nonzero value, the Optimizer
calculates the optimum rate of change of the variable, which causes the value of the variable
to change constantly at the optimum rate.
If the MVC controller is tightly coupled to the Optimizer, the Optimizer uses the value of the
High Constraint parameter from the MVC controller instead of the value of the High
Constraint parameter from the Optimizer.
The default value for this parameter is 1,000.00. You can set the value of this parameter to
any real value that is:
Greater than or equal to the value of the Low Constraint parameter
Less than or equal to the value of the High Limit parameter
You can specify this parameter for CVs and MVs only.

Low Limit
The Low Limit parameter is the lower limit that you place on the value of a variable. The
optimum steady-state value of the variable that is calculated by the Optimizer cannot be less
than the lower limit.
If the MVC controller is tightly coupled to the Optimizer, the Optimizer uses the value of the
Low Limit parameter from the MVC controller instead of the value of the Low Limit
parameter from the Optimizer.
The default value for this parameter is -1,000.00. You can set the value of this parameter to
any real value that is:
Greater than -1,000,000,000.00
Less than the value of the High Limit parameter
You can specify this parameter for CVs and MVs only.

High Limit
The High Limit parameter is the upper limit that you place on the value of a variable. The
optimum steady-state value of the variable that is calculated by the Optimizer cannot be
greater than the upper limit.
If the MVC controller is tightly coupled to the Optimizer, the Optimizer uses the value of the
High Limit parameter from the MVC controller instead of the value of the High Limit
parameter from the Optimizer.

186 Version 2016 SP2

Controller Design SimSci APC User Guide

The default value for this parameter is 1,000.00. You can set the value of this parameter to
any real value that is:
Greater than 0.00 or the value of the Low Limit parameter, whichever value is greater
Less than 10,000,000,000.00.
You can specify this parameter for CVs and MVs only.

Low Validity
The Low Validity parameter is the least value of a variable in the validation range. If the
value of the variable is less than the value of the Low Validity parameter, the Optimizer sets
the quality of the variable to bad.
The Optimizer uses the value of the Low Validity parameter to calculate the normalized
values of a variable. If the MVC controller is tightly coupled to the Optimizer, the Optimizer
uses the value of the Low Validity parameter from the MVC controller instead of the value of
the Low Validity parameter from the Optimizer to calculate the normalized values of a
variable. See Normalization for the Profit Function on page 202 for more information.
The default value for this parameter is -1,000.00. You can set the value of this parameter to
any real value that is:
Greater than -1,000,000,000.00
Less than the value of the High Validity parameter
You can specify this parameter for CVs, MVs, and FVs.

High Validity
The High Validity parameter is the greatest value of a variable in the validation range. If the
value of the variable is greater than the value of the High Validity parameter, the Optimizer
sets the quality of the variable to bad.
The Optimizer uses the value of the High Validity parameter to calculate the normalized
values of a variable and its associated cost factor. If the MVC controller is tightly coupled to
the Optimizer, the Optimizer uses the value of the High Validity parameter from the MVC
controller instead of the value of the High Validity parameter from the Optimizer to calculate
the normalized values of a variable and its associated cost factor. See Normalization for the
Profit Function on page 202 for more information.
The default value for this parameter is 1,000.00. You can set the value of this parameter to
any real value that is:
Greater than 0.00 or the value of the Low Validity parameter, whichever value is greater
Less than 10,000,000,000.00
You can specify this parameter for CVs, MVs, and FVs.

Cushion
The Cushion parameter is a modification to the effective constraints on a variable. That is,
the Optimizer effectively raises the low constraint and lowers the high constraint for the
associated variable by the amount you specify for the Cushion parameter.

Version 2016 SP2 187

SimSci APC User Guide Controller Design

You can use the Cushion parameter to set a safety margin for the constraints. The safety
margin ensures that the process value of a CV never rises above the high constraint or falls
below the low constraint due to uncontrollable noise in the process.
If the MVC controller is tightly coupled to the Optimizer, the Optimizer uses both the value of
the Cushion parameter from the MVC controller and the value of the Cushion parameter
from the Optimizer to determine the safety margin. Otherwise, the Optimizer uses only the
value of the Cushion parameter from the Optimizer to determine the safety margin.
You can specify this parameter for CVs only.

Linearizing Function
The Linearizing Function parameter is a function that the Optimizer applies to the values of
the associated variable to provide a linear relationship for the variable. The Optimizer
transforms the process data for the associated variable based on the function that you select,
and then uses the newly calculated values for all calculations and procedures.
You can select one of five functions for this parameter:
NONE: The Optimizer uses the process values of the associated variables for all
calculations and procedures.
LN(): The Optimizer uses the natural log of the process values of the associated variable
for all calculations and procedures.
EXP(): The Optimizer uses the natural exponential function for the value of the associated
variable for all calculations and procedures.
SQUARE: The Optimizer uses the square of the value of the associated variable for all
calculations and procedures.
SQUARE ROOT(): The Optimizer uses the square root of the value of the associated
variable for all calculations and procedures.
You can specify this parameter for CVs, MVs, and FVs.

Rate of Change
The Rate of Change parameter is the maximum amount by which the value of a variable can
change from one data point to the next. The value of the Rate of Change parameter
represents a validation constraint on the rate of change for a CV or an FV. If the value of the
CV or FV changes by an amount that is greater than the value of the Rate of Change
parameter, the Optimizer sets the quality of the CV or FV to bad.
If you set the value of the Rate of Change parameter to zero, the Optimizer does not
consider this validation constraint for the associated CV or FV.
If the MVC controller is tightly coupled to the Optimizer, the Optimizer uses the value of the
Rate of Change parameter from the MVC controller instead of the value of the Rate of
Change parameter from the Optimizer.
The default value for this parameter is 0.00.
You can specify this parameter for CVs and FVs only.

188 Version 2016 SP2

Controller Design SimSci APC User Guide

Frozen Time
The Frozen Time parameter is the maximum amount of time in which the value of a variable
can remain unchanged. The value of the Frozen Time parameter represents a validation
constraint on the dynamic quality of a CV or an FV. If the value of the CV or FV is static for a
duration that is greater than the value of the Frozen Time parameter, the Optimizer sets the
quality of the CV or FV to bad.
If you set the value of the Frozen Time parameter to zero, the Optimizer does not consider
this validation constraint for the associated CV or FV.
If the MVC controller is tightly coupled to the Optimizer, the Optimizer uses the value of the
Frozen Time parameter from the MVC controller instead of the value of the Frozen Time
parameter from the Optimizer.
The default value for this parameter is 0.00.
You can specify this parameter for CVs and FVs only.

Infer Time
The Infer Time parameter is the amount of time in which the Optimizer infers values for a
variable. The Infer Time parameter applies to nonessential variables only, that is, variables
that are not specified as vital to the operations of the Optimizer. See Vital on page 184 for
more information.
If a nonessential CV is turned off or has a bad quality, the Optimizer uses prediction values
for the CV instead of the actual values for the CV for the amount of time specified by the Infer
Time parameter. After the amount of time specified by the Infer Time parameter elapses, the
Optimizer does not use the CV in any calculations or procedures.
If you set the value of the Infer Time parameter to zero, the Optimizer does not infer values
for the CV and immediately stops using the CV in any calculations or procedures if the CV is
turned off or has a bad quality.
If the MVC controller is tightly coupled to the Optimizer, the Optimizer uses the value of the
Infer Time parameter from the MVC controller instead of the value of the Infer Time
parameter from the Optimizer.
The default value for this parameter is 0.00.
You can specify this parameter for CVs only.

Data Valid Time

The Data Valid Time parameter is the minimum amount of time in which the value of a
variable must remain valid to set the quality of the variable to good. The Data Valid Time
parameter applies to nonessential variables only, that is, variables that are not specified as
vital to the operations of the Optimizer. See Vital on page 184 for more information.
If the Optimizer excludes the associated variable from any calculations or procedures due to
the bad quality of the variable, the variable must have a good quality for the amount of time
specified by the Data Valid Time parameter to once again include the associated variable in
all calculations and procedures in the Optimizer.
If the MVC controller is tightly coupled to the Optimizer, the Optimizer uses the value of the
Data Valid Time parameter from the MVC controller instead of the value of the Data Valid
Time parameter from the Optimizer.

Version 2016 SP2 189

SimSci APC User Guide Controller Design

You can specify this parameter for CVs and FVs only.

Gainsets
A gainset is the set o steady-state gains that the MVC controller and the Optimizer use to
perform calculations. See Cost Function on page 195 and Profit Function on page 201 and
for more information on how the MVC controller and the Optimizer use the steady-state gains.
Gainsets are specific to the controller pair, that is, the MVC controller and the Optimizer. The
controller pair can contain multiple gainsets, but can use only one gainset at a time.

Managing Gainsets
The SimSci APC software automatically creates a gainset named Gain Set 1 when you
create the Optimizer from the final linear model. You can rename, copy, or delete this gainset
as desired. Note that you cannot delete the gainset that the Optimizer uses for calculations.
You can create a gainset from a modelset. After you create a gainset, it is no longer
associated with the modelset from which you created it. Any changes that you make to the
model coefficients in a modelset do not affect any existing gainsets. If you want to apply your
modelset changes to a gainset, you must create a new gainset from the modified modelset.
NOTE: The SimSci APC software automatically creates Gain Set 1 from the automatically
created Model Set 1. See Managing Modelsets on page 176 for more information.
You can modify the gains in a gainset to suite your needs. You cannot modify the gains in the
gainset that the Optimizer uses for calculations if the Optimizer is in the Initialize, Ready, or
Active state.

Acti vity Directions

Creating the default gainset After you complete the configuration for the final linear model, on

the Final Linear Model tab, click Create Controller .

Creating a gains et from a On the Controller Details tab, on the ModelSets tab, right-click
modelset the tab for the modelset from which you want to create the
gainset, and then click Create Gain Set.

Renaming a gainset On the Controller Details tab, on the GainSets tab, double-click
the title of the associated gains et tab. Type a new name for the
gainset, and then press Enter.

Copying a gainset On the Controller Details tab, on the GainSet tab, right-click the
associated gainset tab, and then click Copy Gain Set.

Modifying the gains for a On the Controller Details tab, on the GainSet tab, on the
gainset associated gainset tab, double-click the gain you want to modify.
An editable box that contains the full gain value appears. Click
the box, type a new value for the gain, and then press Enter.

Deleting a gainset On the Controller Details tab, on the GainSet tab, right-click the
associated gainset tab, and then click Delete Gain Set.

190 Version 2016 SP2

Controller Design SimSci APC User Guide

Updating the MVC Controller and Optimizer

When you update an MVC controller and its associated Optimizer, you typically start by
updating the linear modeling scenarios that you used to build the MVC controller and its
associated Optimizer. After you make changes to the linear modeling scenarios, you must
recreate the composite linear model, recreate the final linear model, and then update the
MVC controller and Optimizer from the final linear model. When you update an MVC
controller and its associated Optimizer from the final linear model, you can choose to create a
new modelset in the MVC controller or to overwrite an existing modelset in the MVC
controller.
The SimSci APC software does not update the gainset for the Optimizer when you update the
MVC controller and Optimizer from the final linear model.
When you make a change to the final linear model, such as changing the tag structure for the
final linear model, the SimSci APC software does not automatically update the MVC controller
and the Optimizer with the latest data. Instead of creating a new final linear model to use the
latest data, you can update the existing MVC controller and Optimizer.
The results of updating the MVC controller and the Optimizer vary based on the configuration
of the final linear model in comparison to the configuration of the MVC controller and
Optimizer. See Results of Updating the MVC Controller and Optimizer on page 191 for more
information.
To update the final linear model:

1. On the Final Linear Model tab, click Update Controller .

2. On the Controller Details tab, view the new data for the MVC controller, the Optimizer,
and the associated modelsets and gainsets.

Results of Updating the MVC Controller and Optimizer

The results of updating the MVC controller and the Optimizer vary based on the configuration
of the final linear model in comparison to the configuration of the MVC controller and
Optimizer.
Updating the MVC controller and Optimizer does not update the values for any existing model
coefficients or gains if all the following conditions are true:
The prediction interval for the final linear model equals the control interval for the MVC
controller and Optimizer.
The order of dynamics for the final linear model equals the order of dynamics for the MVC
controller and Optimizer.
The model length for the final linear model equals the model length for the MVC controller
and Optimizer.
In this case:
If you add or remove a variable relationship from the final linear model, the software does
not add or remove that variable relationship from the MVC controller and Optimizer.
If you remove all the variable relationships for a single CV, MV, or FV from the final linear
model, the software removes that CV, MV, or FV from the MVC controller and Optimizer.

Version 2016 SP2 191

SimSci APC User Guide Controller Design

If you add a variable relationship for a new CV, MV, or FV to the final linear model, the
software creates a single new modelset and adds the variable relationships for the new
CV, MV, or FV to the existing modelsets. Note that you cannot add more variable
relationships for the new CV, MV, or FV by once again updating the MVC controller and
Optimizer from the final linear model.
Updating the MVC controller and Optimizer removes any existing modelsets and gainsets
from the controllers and replaces them with the single default modelset, Model Set 1, and the
single default gainset, Gain Set 1, if any of the following conditions are true:
The prediction interval for the final linear model does not equal the control interval for the
MVC controller and Optimizer.
The order of dynamics for the final linear model does not equal the order of dynamics for
the MVC controller and Optimizer.
The model length for the final linear model is less than the model length of the controllers.
If the model length for the final linear model is greater than the model length of the
controllers, the software creates a single new modelset. It also appends the existing
modelsets with the model coefficients from the new modelset to ensure that all the modelsets
have the same model length.
The following table summarizes the result of updating the MVC controller and the Optimzer
after you make changes to the final linear model:

Change to the Final Linear Model Result after Updating Controllers

Add or remove variable relationships for No change

1
existing CVs, MVs, or FVs
1
Remove CVs, MVs, or FVs Removes the associated CVs, MVs, and FVs from
the MVC controller and Optimizer
1
Add CVs, MVs, or FVs Creates a new modelset and adds variable
relationships for the new CVs, MVs, and FVs to
existing modelsets

Change the prediction interval Removes any existing modelsets and gainsets from
the controllers and replaces them with the single
default modelset, Model Set 1, and the single default
gainset, Gain Set 1

Change the order of dynamics Removes any existing modelsets and gainsets from
the controllers and replaces them with the single
default modelset, Model Set 1, and the single default
gainset, Gain Set 1

Decrease the model length Removes any existing modelsets and gainsets from
the controllers and replaces them with the single
default modelset, Model Set 1, and the single default
gainset, Gain Set 1

2
Increase the model lengt h Creates a single new modelset and appends the
existing modelsets with the model coefficients from
the new modelset

192 Version 2016 SP2

Controller Design SimSci APC User Guide

1 Applies only if you do not change the predic tion interval, order of dynamics, or model length for the final linear model.
2 Applies only if you do not change the predic tion interval or order of dynamics for the final linear model.

Piecewise Nonlinear Transforms

A piecewise nonlinear transform is a linearizing function that uses a set of X and Y values
defined in a lookup table to transform various parameter values of a variable in a controller.
The controller uses the transformed values to perform all calculations and procedures. Note
that the controller stores the transformed values only for calculation purposes and you cannot
access the transformed values from the UI. That is, in the controller, you can see only the
untransformed parameter values for the variable.
The following table list the parameter values that the controller transforms for each type of
variable:

Parameter Controlled Variables Manipulated Variables Feedforward Variables

Current Value Yes Yes Yes

Low Constraint Yes Yes

High Constraint Yes Yes

Low Limit Yes Yes

High Limit Yes Yes

Low Validity Yes Yes Yes

High Validity Yes Yes Yes

Setpoint Yes

Target Yes

Increment Yes
1
Readback value Yes

1 This value is internal to the controller. You cannot access this value from the UI.

A lookup table is specific to a variable in a controller. You can define the series of X and Y
values in a lookup table by using the Define parameter for the variable. See Define on page
171 for more information.
The SimSci APC software uses the lookup table to determine how it should change the
parameter values for a variable. You must specify at least two sets of X and Y values to
transform the parameter values for the variable. That is, if you specify only one X and Y
value, the controller does not transform the parameter values for the variable.
If you specify two sets of X and Y values, the SimSci APC software fits the parameter values
to a single line.

Version 2016 SP2 193

SimSci APC User Guide Controller Design

If you specify more than two sets of X and Y values, the SimSci APC software fits the
parameter values to a set of lines. There is one line for each pair of X values The SimSci
APC software compares the parameter value to the specified X values to determine which
line it should use for the transformation.
If the parameter value is less than the first X value in the lookup table, the SimSci APC
software replaces the parameter value with the Y value associated with the first X value.
If the parameter value is greater than first X value in the lookup table, the SimSci APC
software compares the parameter value to each X value in the lookup table in the order in
which the X values appear on the table.
If the parameter value is equal to any X value, the SimSci APC software replaces the
parameter value with the Y value associate with that X value.
If the parameter value falls between two X values, the SimSci APC software uses the linear
interpolation method shown in the following equation to transform the parameter value:

Value X i 1
Result Yi 1 Yi Yi 1
X i X i 1
where
Result is the transformed value
Value is the value you want to transform
X i 1
is the preceding X value in the lookup table
Xi
is the current X value in the lookup table
Yi 1 X i 1
is the Y value associated with in the lookup table
Yi Xi
is the Y value associated with in the lookup table
X i 1 Xi
Note that in this case, the parameter value is greater than and less than .
If the parameter value is greater than all the X values in the lookup table, the SimSci APC
software replaces the parameter value with the Y value associated with the last X value in the
lookup table.
To avoid undesirable results, you should specify the X values in ascending order. The X
values should cover beyond the full range of possible operating data for the specific tag.

Simulation Mode
You can perform process simulations by using Simulation mode for the controllers.
Simulations allow you to tune and test the performance of a model-based controller before
you apply the controller to a real process. Typically, you use simulations to fine tune the
configured settings for a controller until you obtain an appropriate set of weights and costs for
the CVs, MVs, and FVs.

194 Version 2016 SP2

Controller Design SimSci APC User Guide

In Simulation mode, the MVC controller does not receive process values from the DCS.
Instead, the MVC controller uses the predictions calculated from its associated modelset and
the SimSci APC algorithms as the current process values. You can use the responses of the
variables to this simulated process data and the overall performance of the MVC controller
during the simulation to determine if you need to modify the controllers before you apply them
to the real process.
To make a simulation as real as possible, you should provide accurate initial values from the
process plant for all the CVs, MVs, and FVs before you start the simulation. Initial values are
calculated based on available process data. The values should be in engineering units. You
can enter these values in the Initial Value parameters for the variables in the MVC controller.
See Initial Value on page 162 for more information.
You can activate Simulation mode by clicking On on the Home tab, in the Simulation group.
You can specify the MVC controller should operate in Simulation mode by using the
Simulation Mode check box for the MVC controller. See Simulation Mode on page 209 for
more information.

Cost Function
The cost function is a mathematical equation that expresses the cost of attaining the control
target. It is used within the MVC controller to identify the optimum control moves. The general
form of the cost function is:

M
J C eTk 1Pe k 1 uTk Qu k f kT Rf k
k 1 ,
e k 1 y k 1 y ss
,
u k u k uk
,
f k u k u ss
where
JC
is the cost associated with the control moves
M is the Control Horizon parameter (see Control Horizon on page 157)
k is the "current" time instant
e k 1
is a vector that is composed of the errors between the values of the CVs and the
associated setpoint at instant k 1

is a vector that is composed of the values of the CVs at instant k 1

y k 1

y ss
is a vector of setpoints that defines the values to which the elements of the output vector,
y , are to be set by the MVC controller

P is a diagonal weighting matrix that has all non-diagonal elements set to zero. It is a square
matrix that has the dimensions c c , where c is the number of CVs in the MVC controller.

Version 2016 SP2 195

SimSci APC User Guide Controller Design

u k is a vector that is composed of the MV increments made in each actuation at instant k .

uk
is a vector that is composed of the values of the MVs at instant k
uk
is a vector that is composed of the values of the MVs at the instant immediately before the
u
instant k . In most cases, this vector is equal to k 1
Q is a diagonal weighting matrix that has all non-diagonal elements set to zero. It is a square
matrix that has the dimensions m m , where m is the number of MVs in the MVC
controller.
fk
is a vector that is composed of the deviations between the values of the MVs instant k
and the associated steady-state target.
u ss
is a vector that is composed of the steady-state target values for the MVs in the MVC
controller.
R is a diagonal weighting matrix that has all non-diagonal elements set to zero. It is a square
matrix that has the dimension m m , where m is the number of MVs in the MVC controller.

NOTE: The interval of k to k 1 equals one control interval, in seconds.

Each diagonal element of P relates to a particular CV and gives rise to an error-squared term
associated with that CV. Thus, minimizing the cost, J , leads to a minimization of each error-
squared term in P . As the minimization of the cost progresses, the controller outputs, that is,
the MVs, approach their respective setpoints.

Each diagonal element of Q relates to a particular MV and gives rise to a term that involves
the square of the changes that are to be applied to the MV setting. The minimization of the
cost invokes a constraint upon the degree to which the MVC controller can adjust an MV
during the course of control implementation.
Each diagonal element of R relates to a particular MV and generates a term that involves the
square of the deviation of the MVs from their respective steady-state targets. Thus, when the
SimSci APC software applies the MVC controller to the plant process, the MVs are driven
towards their steady-state targets.

Infer Time
All control systems must operate subject to the practical constraint imposed by the process
hardware and operational procedures. For example, a flow control valve has an operating
range of 10% to 90%, with no signal move of greater than 10% permitted. Process outputs
are permitted to vary within declared limits, perhaps corresponding to product specification
limits. Such constraints place a fundamental restriction on the performance of the control
system. If constraints are not considered at the design state, the performance of the control
system may be comprised when they are encountered in the process.
There are two possible approaches for dealing with constraints in the context of control
system design:

196 Version 2016 SP2

Controller Design SimSci APC User Guide

Design a control system that avoids constraint violation under all anticipates
circumstances by using a judicious choice of weightings. This may be referred to as an
unconstrained design since the constraints are considered informally rather than explicitly.
Design a control system that is subject to an explicit declaration of t he constraints. The
resulting control system should systematically cope with constraints when they are
encountered. There is no unique solution to the problem (that is, no algorithmic solution)
and the design mst be re-evaluated at each instant that control is executed.
The discussions on the minimization of the quadratic cost function and the control algorithm
focuses on an unconstrained design for the sake of simplicity.

Design Weights for the Cost Function

The design weights for the cost function are given in the P , Q , and R matrices of the cost
function. See Cost Function on page 195 for more information.
An increase in the value of an element of P implies that the associated error is reduced more
urgently, that is, that the associated output is to be driven more quickly to setpoint.

An increase in the value of an element of Q implies that the associated actuator is more constrained,
that is, that successive adjustments are smaller.

An increase in the value of an element of R implies that the associated MV is driven more
urgently to its target. Note that the scope for steady-state MV targeting depends on the
availability of spare MVs. There is no scope if the problem is square, that is, the number of
MVs equals the number of CVs, in which case the actuator positions are dictated by the
control requirement. You can choose to retain the default values of zero in the matrix R to
exclude steady-state actuator considerations from the control system.
The minimization of the cost function uses the normalized linear model. Using this model
simplifies the initial selection of weights. If the weights in P and Q are initially set to 1, the
minimization procedure generally yields a realistic control system, one that gives rise to
responses of a reasonable rate given the time constants of the process to be controlled.
Thus, an initial control system can be established without the need for any decisions. Given
an initial design, you can then change the design weights from the initial value of 1 to adjust
the control system responses to be closer to the desired responses.
Elements of P should be set to zero if the corresponding CVs are not to be driven to setpoint
at all. If the number of CVs exceeds the number of MVs (that is, c m ), only an m number of
CVs can be driven to setpoint.
You can specify the design weights for the cost function by using the following online
parameters for the variables in the MVC controller:
CV Weight on page 214
MV Weight on page 215
Target Weight on page 215
See Online Parameters for the Variables in the MVC Controller on page 210 for more
information.

Version 2016 SP2 197

SimSci APC User Guide Controller Design

Control Algorithm
Minimization of the cost function is carried out over a control horizon of M control intervals.
At each stage, that is, for each value of k 1 to M , the MVC controller calculates a set of
control systems gains. These system gains minimize the cost function up to that stage. These
gains normally converge to a constant set of values as the stages progress. The converged
set of gains represents the optimal solution to the control problem, subject to the specified
design weights and control horizon.
The solution of the controller design is represented by the control algorithm. The control
algorithm is:
u k uk G1 Y2 k Y2 ss G2 Vk G3 u k u ss ,
wk wk p
w w
k p k 2 p

.
Vk
.

.
w k S 1 p w k Sp

where
u k uk is a vector of increments to be made to each of the MVs in the MVC controller at
instant k relative to the current values at instant k
Y2 k Y2 ss is a vector of measurement errors from the setpoint, adhering to the structured
of the vector Y2 , which is a vector of the process outputs, that is, the CVs. See Vectors for
the Process Outputs on page 118 for more information.
Vk
is vector that is composed of the previous manipulated and feedforward increments.
wk
is a vector of the process inputs, that is, the MVs and FVs, at instant k .
u k u ss is a vector that is composed of the deviations of each MV in the MVC controller, at
instant k , from the associated steady-state target.
G1 is a matrix that holds the error gains of the control system, that is, the gain coefficients for
the current and past CV deviations from the setpoint. This matrix has m rows and n c
columns.
G2 is a matrix that holds the input gains of the control system, that is, the gain coefficients
for the current and past actuation and feedforward increments. This matrix has m rows and
r S q columns. The value of S depends on R j . See Vectors for the Process Inputs on
page 119 for more information.

198 Version 2016 SP2

Controller Design SimSci APC User Guide

G3 is a matrix that holds the steady-state actuator gains of the control system, that is, the
gain coefficients for the deviations of each actuator value from the associated steady-state
target. This matrix has m rows and n columns.
c is the total number of CVs in the MVC controller
m is the total number of MVs in the MVC controller
q is the total number of FVs in the MVC controller
r is the total number of input variables in the MVC controller
n is the order of the linear model that the MVC controller uses during calculations

V R
The dimension of k and G2 depend on j . Consider the input gain matrix G2 . For any
process input j ( j 1 to r ), the number of samples required in k for the specific process
w V
S
input is given by j :
S j Rj

Vk
Given the above, the total number of samples required in , that is, S , is equal to the
maximum of j for j 1 to r :
S

S max S j

S V
For any particular process input j , if S is greater than j , there are more elements in k
than required. In this case, the associated gains in G2 are set to zero so that the unwanted
elements do not contribute to the MVC controller.
V
The k and G2 matrices include terms that are associated with the FVs. These terms give
rise to feedforward control. The MVC controller anticipates a disturbance and corrects for it
before the effect of the disturbance is apparent in the output.
The feedforward terms and the other terms in the matrices are based on a common selection
of design weights. However, the cost function does not include an independent basis for
defining the urgency of feedforward control relative to feedback control. You can use a set of
alternate design weights for the design of feedforward terms. A complete set of controller
gains is established by using this alternative, but only those gains that relate to the
disturbance terms are updated. The remaining gains are held at their previously designed
settings.

Example of a Single-Input, Single-Output (SISO) Controller Structure

Consider a single-input, single-output (SISO) system. The controlled variable (CV) is y and
the manipulated variable (MV) is u . The following is true:

The number of CVs is 1, c 1 .

The number of MVs is 1, m 1.

The number of FVs is zero, q 0 .

Version 2016 SP2 199

SimSci APC User Guide Controller Design

We make the following specifications for the MVC controller:

The Controller Model Order value is set to 1. Thus, the order of dynamics is 1, n 1 .

The Control Interval value is equal to the scan interval, p 1 .

The Model Length value for the model that is used to create the MVC controller is set to
zero, L 0 .
We make the following assumption:

No steady-state MV gain has been derived ( G3 0 ) by setting the associated weight in

the matrix R to zero.
This is the only sensible option for a SISO problem; the MV must change value to whatever is
needed to achieve control and therefore, cannot also follow a steady-state target.
Based on the above specifications and assumptions, the control system has the following
form:
yk yss
uk uk g11 g12
g13 uk uk p
yk p yss

Example of a Multiple-Input, Multiple-Output (MIMO) Controller Structure

Consider a two-input, two-output system. The two output variables (CVs) are y1 and y 2 . The
two input variables (MVs) are u1 and u 2 . When we add the CVs and the MVs to the model,
the following is true:

The number of CVs is 2, c 2 .

The number of MVs is 2, m 2 .

The number of FVs is zero, q 0 .

We make the following specifications for a scenario in the model:

The Controller Model Order value is set to 2. Thus, the order of dynamics is 2, n 2 .

The Control Interval value is equal to four times the scan interval, p 4 .
The Model Length value for the model that is used to create the MVC controller is set to
two, L 2 .
We make the following assumption:

No steady-state MV gain has been derived ( G3 0 ) by setting the associated weight in

the matrix R to zero.

200 Version 2016 SP2

Controller Design SimSci APC User Guide

Based on the above specifications and assumptions, the model has the following form:

uk u k 4
y k y ss
u1,k u1,k G11 G12 G13 G21 G22 G23 G24 u k 4 u k 8
u u y k 4 y ss
2, k 2, k G14 G15 G16 y y G25 G26 G27 G28 u k 8 u k 12
k 8 ss
u k 12 u k 16
where

y
yk 1
y2 k
u
uk 1
u2 k
Gxy gxy1 gxy2

Profit Function
The profit function is a mathematical equation that expresses the optimal overall profitability
of the process. It is used within the Optimizer to identify the optimum set of steady-state
values for the MVs and CVs of the system that maximizes the profit while respecting all of the
specified process constraints. The steady-state values that are calculated from the profit
function are used as the setpoints in the cost function. See Cost Function on page 195 for
more information.
The general form of the profit function is a linear combination of the values of the CVs and
MVs, each multiplied by an appropriate cost factor:
c m
J P aiYi b jU j
i 1 j 1

where
J P is the profit of the process
c is the number of CVs in the Optimizer
ai
is the cost assigned to the ith CV
Yi
is the ith CV
m is the number of MVs in the Optimizer
bj
is the cost assigned to the jth MV
Uj
is the jth MV
Yi Uj
The maximization of the profit function generates a set of values for and . These sets
y u
are known as ss and ss , respectively, in the cost function.

Version 2016 SP2 201

SimSci APC User Guide Controller Design

The Optimizer carries out the maximization of the profit function in association with a steady-
state model for the process so that the generated values for the input and output variables
are consistent. The steady-state model can include feedforward inputs, but the FVs are not
considered in the profit function. See Steady-state Model on page 205 for more information.
The Optimizer carries out the maximization of the profit function subject to all of the
constraints specified for the process variables. See Low Constraint on page 185 and High
Constraint on page 185 for more information.
The maximization of the profit function is an iterative process. The Optimizer calculates the
profit of the process at the origin that is determined by the initial steady-state values. See
Initial Steady-state Values on page 204 for more information. For each iteration, the
Optimizer adjusts the values of the CVs and MVs and recalculates the profit by using the
adjusted values. This process continues until the profit converges. If the profit does not
converge within 500 iterations, the Optimizer stops the maximization of the profit function and
an error message appears in the Diagnostics Window.

Normalization for the Profit Function

All of the tag-related values in the profit function are normalized by using the low validity limit
and the high validity limit, that is, the values of the Low Validity and High Validity
parameters for each variable, respectively. The following values are normalized for each CV,
MV, and FV tag:
Value of the tag
Cost coefficient
Constraint values
Steady-state model values
The normalized values are not reported and are used for calculation purposes only. That is,
all reported values are given in engineering units.

Normalization of Variables for the Profit Function

Normalized values of the CVs, MVs, and FVs have a range of 1.0 to 2.0. The value of the CV,
MV, or FV is normalized by using the following equation:
ValueEng ,i LowValidityi
ValueNorm ,i 1
HighValidityi LowValidityi
where
ValueNorm ,i
is the normalized value of the ith CV, MV, or FV
ValueEng ,i
is the engineering value of the ith CV, MV, or FV
LowValidityi
is the low validity limit for the ith CV, MV, or FV
HighValidityi
is the high validity limit for the ith CV, MV, or FV

202 Version 2016 SP2

Controller Design SimSci APC User Guide

Normalization of the Cost Coefficients for the Profit Function

Cost coefficients are stored internally in their normalized form and are reported in their
engineering form. If you change the validity limits for a tag, you must also change the
engineering cost coefficient, that is, the value of the Cost parameter for the tag, to achieve
the same internal cost coefficient that was stored for the previous validity limits.
The Optimizer uses the normalized cost coefficient for all calculations regarding the profit
function. It uses the following equation to normalize the cost coefficients:
Cost Norm ,i Cost Eng ,i HighValidityi LowValidityi
where
Cost Norm ,i
is the normalized cost coefficient for the ith CV or MV
Cost Eng ,i
is the engineering cost coefficient for the ith CV or MV
HighValidityi
is the high validity limit for the ith CV, MV, or FV
LowValidityi
is the low validity limit for the ith CV, MV, or FV

Incremental Form
The principal source of error in the solution of the Optimizer is inaccuracy in the linear steady-
state model, which is a localized approximation of the actual steady-state properties of the
system. To minimize this error, the Optimizer formulates the problem values as incremental
quantities that are referred to the origin for the maximization of the profit function. That is, the
following values are calculated with reference to the origin:
High constraints
Low constraints
Values derived from the steady-state model (see Steady-state Model on page 205)
Values of the CVs, MVs, and FVs
Incremental values are not reported and are used for calculation purposes only. That is, all
reported values are given in non-incremental terms.
The incremental form of the profit function is:
c m
J P ai Yi b j U j
i 1 j 1 ,
where
J P is the difference between the current profit of the process and the profit that is calculated
at the origin for the maximization of the profit function
c is the number of CVs in the Optimizer
ai
is the cost assigned to the ith CV

Version 2016 SP2 203

SimSci APC User Guide Controller Design

Yi is difference between the current value of the ith CV and its value at the origin for the
maximization of the profit function
m is the number of MVs in the Optimizer
bj
is the cost assigned to the jth MV
U j
is the difference between the current value of the jth CV and its value at the origin for the
maximization of the profit function

Initial Steady-state Values

The first step in the maximization of the profit function is the calculation of the initial steady-
state values for the CVs, MVs, and FVs.
If you set the Origin Source parameter to Origin, the initial steady-state values are set equal
to a set of user-specified values. You can specify the initial steady-state values for the
variables by using the Origin parameters for the variables. See Origin on page 226 for more
information.
If you set the Origin Source parameter to MVC model for Steady State or LP ARX Model,
the Optimizer uses a linear model to calculate the initial steady-state values for the CVs, MVs,
and FVs. If you set the Origin Source parameter to MVC model for Steady State, the
Optimizer uses the linear model for the MVC controller for the calculations. If you set the
Origin Source parameter to LP ARX Model, the Optimizer uses its own linear model for the
calculations.
The Optimizer uses the incremental equation for a linear model to calculate the predicted
values for the CVs. See Incremental Equation on page 123 for more information. The
following equations define the incremental equation for a linear model:
Y3 k p Y4 k 1W1,k r Wr ,k
,

Y3 k p y k p y k ,
y k p y k Y3k p
y
k 2 p y k Y3k 2 p
. .
Y4k
. .

. .
y k N 1 p y k Y3k N 1 p
,

N n 1
where
Y3 is the vector of increments that contains the prediction increments for the process
outputs, that is, the CVs, with respect to the current values.
is the transition matrix containing the autoregressive terms. A model uses the
autoregressive terms only if its order of dynamics is greater than zero.

204 Version 2016 SP2

Controller Design SimSci APC User Guide

Y4 is the vector of increments that contains the current and past increments for the CVs.
1 is the driving matrix for the input variable corresponding to W1 .

W1 is the vector of increments for the first input variable that is added to the model. It
contains the current and past increments for the first input variable.
r is the driving matrix for the input variable corresponding to Wr .

Wr is the vector of increments for the last input variable that is added to the model. It
contains the current and past increments for the last input variable.
r is the total number of input variables that are added to the model.
k is the "current" time instant.
p is the prediction interval ratio.

N is an integer that determines the number of previous prediction intervals over which the
Y4 k
vector extends.
n is the order of dynamics.
All of the CV increments are referenced to the same data point, that is, the same values in
the vector y at instant k . Based on the preceding equations, the Optimizer uses the
yk p
following equation to calculate the predicted CV values, :
y k p y k Y4 k 1W1,k r Wr ,k
The Optimizer calculates the initial steady-state values of the CVs, MVs, and FVs by using an
iterative process. It calculates the predicted CV values for each iteration, and then uses the
y
predicted CV values as the reference values, k , for the next iteration. The iterations
continue until the CV values converge at a tolerance of 10-5 or until the maximum number of
iterations has been reached. The maximum number of iterations is the value of the Iteration
parameter. If the CV values do not converge at the end of the iterations, an error message
appears in the Diagnostics Window.
If the CV values converge, the final values of the CVs, MVs, and FVs represent the initial
steady-state values of the CVs, MVs, and FVs, which are used as the origin for the
y 1 FilterFactor y k p FilterFactor
maximization of the profit function. f

Steady-state Model
The steady-state model is expressed as a set of steady-state gains relating incremental
changes in the MVs and FVs to incremental responses in the CVs. The steady-state gains
are automatically calculated from the linear model that is associated with the Optimizer.
Based on the steady-state model, we can develop equations for the CV responses with
respect to changes in the MVs and FVs. The equation for each CV has the following form:
m q
Yi gij U j gik Vk
j 1 k 1

Version 2016 SP2 205

SimSci APC User Guide Controller Design

where
Yi is difference between the current value of the ith CV and its value at the origin for the
maximization of the profit function
m is the number of MVs in the Optimizer
g ij
is the steady-state gain that relates the ith CV to the jth MV
U j
is the difference between the current value of the jth CV and its value at the origin for the
maximization of the profit function
q is the number of FVs in the Optimizer

g ik
is the steady-state gain that relates the ith CV to the kth FV
Vk
is difference between the current value of the kth FV and its value at the origin for the
maximization of the profit function
The Optimizer must maximize the profit of the process in association with the steady-state
model so that the generated values for the input and output variables are consistent. The
equations from the steady-state model represent equality constraints on the CVs.

206 Version 2016 SP2

 SimSci APC User Guide

C HAPTER 7
Enabling Controllers

In This Chapter
Overview.................................................................................................................................... 207
Enabling the MV C Controller ....................................................................................................... 207
State of the MV C Controller ......................................................................................................... 208
Online Configuration Parameters for the MVC Controller ............................................................... 209
Online Parameters for the Variables in the MV C Controller ............................................................ 210
Changing the Parameters for the Enabled MV C Controller ............................................................ 216
Managing Trends for the MV C Controller...................................................................................... 218
Creating Radial Plots for the MVC Cont roller ................................................................................ 219
Enabling the Optimizer ................................................................................................................ 221
State of the Optimizer ................................................................................................................. 222
Online Configuration Parameters for the Optimizer ....................................................................... 223
Online Parameters for the Variables in the Optimizer .................................................................... 223
Changing the Parameters for an Enabled Optimizer ...................................................................... 226

Overview
When you enable an MVC controller and its associated Optimizer, the SimSci APC software
implements the MVC controller and the Optimizer onto the target process. The extent to
which the SimSci APC software implements the controllers depends on the current state of
the controller. See State of the MVC Controller on page 208 and State of the Optimizer on
page 222 for more information.
You can implement the MVC controller and the Optimizer separately. See Enabling the MVC
Controller on page 207 and Enabling the Optimizer on page 221 for more information.
You can enable controllers only when the APC Project is scanning. You can enable more
than one MVC controller and more than one Optimizer at a time, provided that the MVC
controllers are not in Simulation mode. For Simulation mode, you can enable only one MVC
controller at a time.

Enabling the MVC Controller

You can enable the MVC controller by initializing it and then activating it. You can initialize
and activate the MVC controller by using the Controller Overview tab, which you can open
by clicking Controller Overview in the APC Explorer, under Promote. You can initialize and
activate the MVC controller only when the APC Project is scanning.

Version 2016 SP2 207

SimSci APC User Guide Enabling Controllers

The MVC controller begins in the Manual state. When you initialize the MVC controller, it
transitions to the Waiting state, then to the Initialize state, and finally to the Ready state.
When you activate the MVC controller, it transitions from the Ready state to the Auto state,
and then to the Active state. The MVC controller must be in the Ready state before you can
activate it. That is, you must initialize the MVC controller before you activate it.
You can view the current state of the MVC controller by using the Current Status parameter
on the Controller Overview tab, in the Configuration area. When the state of the MVC
controller changes, informational messages indicating the change appear in the Diagnostics
Window.

Acti vity Directions

Starting scanning On the Home tab, in the Scan group, click On.

Initializing the MVC cont roller On the Controller Overview tab, in the Configuration
area, select Initialize for the Controller Mode parameter,

and then click Change Status . The MVC controller

transitions to the Waiting state, then to the Initialize state,
and finally to the Ready state.

Activating the MVC controller On the Controller Overview tab, in the Configuration
area, select Active for the Controller Mode parameter,

and then click Change Status . The MVC controller

transitions to the Auto state and then to the Active state.
The MVC controller must be in the Ready state before you
can activate it.

Deactivating the MVC controller On the Controller Overview tab, in the Configuration
area, select either Ready, Initialize, or Manual for the
Controller Mode parameter, and then click Change

Status . If you select Ready, the MVC controller

returns to the Ready state. If you select Initialize, the
MVC controller reinitializes and ret urns to the Ready state.
If you select Manual, the MVC controller returns to the
Manual state.

Stopping scanning On the Home tab, in the Scan group, click Off.

Example: Initializing the MVC Controller

1. On the Controller Overview tab, expand the Configuration area.

2. In the Controller Mode list, select Initialize, and then click Change Status .

State of the MVC Controller

The state of the MVC controller is the current mode of operation for the MVC controller. The
MVC controller can be in one of five states:

208 Version 2016 SP2

Enabling Controllers SimSci APC User Guide

Manual: The MVC controller does not perform any calculations and does not write values
to the output tags. When the APC Project is scanning, the MVC controller updates the
current values for the input tags and output tags with the latest process data.
Waiting: The MVC controller performs steps to transition to the Initialize state, and then
automatically transitions to the Initialize state. It typically stops at the Waiting state only if
an error occurs due to one of the following:
o A vital tag is off
o A communication error occurs, which is indicated by static watchdog tags
o The MVC controller does not contain CVs or MVs
Initialize: The MVC controller refreshes the values for all parameters in the MVC
controller, and then automatically transitions to the Ready state.
Ready: The MVC controller minimizes the cost function at the specified control interval for
the MVC controller. That is, the MVC controller calculates the optimum control moves for
the process. It does not write values to the output tags. See Cost Function on page 195
for more information.
Auto: The MVC controller performs steps to transition to the Active state.
Active: The MVC controller minimizes the cost function and writes values to the output
tags based on its configuration and calculations.
You can use the Current Status parameter on the Controller Overview tab, in the
Configuration area, to determine the current state of the MVC controller.

Online Configuration Parameters for the MVC Controller

You can set and view the online configuration parameters for the MVC controller by using the
Controller Overview tab. You can open the Controller Overview tab by clicking Controller
Overview in the APC Explorer, under Promote. The MVC controller contains the following
online configuration parameters:
Simulation Mode on page 209
Condition Number Calculation on page 210
Controller Mode on page 210
Current Status on page 210

Simulation Mode
The Simulation Mode check box determines whether or not the MVC controller operates in
Simulation mode. You cannot select the Simulation Mode check box unless the APC Project
is in Simulation mode. See Simulation Mode on page 194 for more information.
If you select the Simulation Mode check box for the MVC controller, the MVC controller
operates in Simulation mode when you activate it. That is, the MVC controller does not
receive values from the DCS. Instead, the MVC controller uses the predictions calculated
from its associated modelset and the SimSci APC algorithms as the current process values.

Version 2016 SP2 209

SimSci APC User Guide Enabling Controllers

Condition Number Calculation

The Condition Number Calculation check box is a control that determines whether or not
the condition number is calculated for the MVC controller. The condition number gives the
ratio of the strongest effect of the inputs on the outputs to the weakest effect of the inputs on
the outputs. A large condition number (>30) indicates that the controllability is low or that the
resulting controller is sensitive to model uncertainty.
If you select the Condition Number Calculation check box, the MVC controller calculates
the condition number. If you clear the Condition Number Calculation check box, the MVC
controller does not calculate the condition number.
The value of the condition number appears next to the Condition Number Calculation
check box. If the MVC controller does not the calculate the condition number, the value is
zero.
By default, the Condition Number Calculation check box is cleared.

Controller Mode
The Controller Mode parameter allows you to select and change the state of the MVC
controller.
The amount of states that you can select depends on the current state of the MVC controller.
If the MVC controller is in a Manual state, you can select only Manual or Initialize. If the MVC
controller is in a Ready or Active state, you can select Manual, Initialize, Ready, or Active.

After you select a state, you must click Change Status to change the state of the MVC
controller.See Enabling the MVC Controller on page 207 and State of the MVC Controller on
page 208 for more information.
You can edit this parameter only when the APC Project is scanning.

Current Status
The Current Status parameter indicates the current online state of the MVC controller. See
State of the MVC Controller on page 208 for more information.
You cannot edit this parameter.

Online Parameters for the Variables in the MVC Controller

You can view and specify the online parameters for both input and output variables in the
MVC Controller by using the Controller Overview tab. The Controller Overview tab
contains a tab for each type of variable in the MVC controller.
On the Controller Overview tab, on the Control variables tab, you can access the following
parameters for the CVs:
Set Point on page 211
Vital on page 212
Switch on page 212
Current Value on page 212
Trend on page 213

210 Version 2016 SP2

Enabling Controllers SimSci APC User Guide

Low Constraint on page 213

High Constraint on page 214
Status on page 216
CV Weight on page 214
Tracking on page 216
On the Controller Overview tab, on the Manipulated variables tab, you can access the
following parameters for the MVs:
Opt Target on page 211
Vital on page 212
Switch on page 212
Current Value on page 212
Trend on page 213
Low Constraint on page 213
High Constraint on page 214
Move on page 214
MV Weight on page 215
Target Weight on page 215
Status on page 216
Increment on page 216
On the Controller Overview tab, on the Feed Forward variables tab, you can access the
following parameters for the FVs:
Switch on page 212
Current Value on page 212
Trend on page 213
Status on page 216
Vital on page 212

Set Point
The Set Point parameter displays the current setpoint for the associated CV. If the MVC
controller is coupled with the Optimizer, the Optimizer calculates the setpoint and you cannot
edit this parameter.
This parameter applies to CVs only.

Opt Target
The Opt Target parameter is the current value of the optimized steady-state target for an MV.

Version 2016 SP2 211

SimSci APC User Guide Enabling Controllers

If the MVC controller is tightly coupled to the Optimizer, the MVC controller uses the
optimized steady-state target that the Optimizer calculates. In this case, you cannot edit this
parameter.
If the MVC controller is loosely coupled to the Optimizer or is not coupled to the Optimizer,
you can change this parameter to change the optimized steady-state target for the associated
MV. The value of this parameter must be between the high and low constraints for the
associated MV.
You can specify this parameter for MVs only.

Vital
The Vital parameter displays the value of the Vital parameter in the MVC controller
assembly, which is a check box that determines whether or not a variable is vital to the MVC
controller. If the check box is selected, the MVC controller considers the associated variable
as vital to the operations of the MVC controller. If the check box is cleared, the MVC
controller considers the associated variable as nonessential to the operations of the MVC
controller. See Vital on page 162 for more information.
You cannot edit this parameter.
This parameter applies to CVs, MVs, and FVs.

Switch
The Switch parameter is a list that determines whether a variable is on or off with regards to
MVC controller operations. You can set this parameter to one of the following values:
On: The MVC controller uses the associated variable in all calculation and procedures.
Off: The MVC controller does not use the variable in any calculations or procedures.
Infer: The MVC controller uses the inferred value of the variable in all calculations and
procedures. You can use this option for CVs only. If you want to use this value for a CV,
the CV cannot be vital to the operations of the MVC controller. Typically, the MVC
controller uses the inferred value of a CV when a communication error occurs with the CV
or the CV has a bad quality. The MVC controller can infer the value of a CV only for the
amount of time specified by the Infer Time parameter. See Vital on page 184 and Infer
Time on page 189 for more information.
FFWD: The MVC controller uses the values of the readback signal for the variable in all
calculations and procedures instead of the value of the variable. It also treats the variable
as a feedforward variable (FV).You can use this option for MVs only. This option is also
known as feedforward mode. If you want to use feedforward mode for an MV, the MV
cannot be vital to the operations of the MVC controller. Typically, the MVC controller uses
feedforward mode for an MV when a communication error occurs with the MV or the MV
has a bad quality. See Vital on page 162 for more information.
You can specify this parameter for CVs, MVs, and FVs.

Current Value
The Current Value parameter displays the current value of the variable.

212 Version 2016 SP2

Enabling Controllers SimSci APC User Guide

The value appears on a slider. The greatest value on the slider is set equal to the upper limit
for the variable. The least value on the slider is set equal to the lower limit for the variable.
See High Limit on page 165 and Low Limit on page 165 for more information. Two nodes
appear on the slider to indicate the upper and lower constraints for the variable. See High
Constraint on page 164 and Low Constraint on page 164 for more information. The color of
the slider indicates the proximity of the current value of the variable to the constraints.
The color of the slider is green if both of the following statements are true:
The current value of the variable is greater than the lower constraint by a margin of 5% of
the lower constraint.
The current value of the variable is less than the upper constraint by a margin of 5% of
the upper constraint.
The color of the slider is yellow if either of the following statements is true:
The current value of the variable is within a 5% margin of the lower constraint.
The current value of the variable is within a 5% margin of the upper constraint.
The color of the slider is red if either of the following statements is true:
The current value of the variable is greater than the upper constraint by a margin of 5% of
the upper constraint.
The current value of the variable is less than the lower constraint by a margin of 5% of the
lower constraint.
The color of the slider is always gray for the FVs, which do not have defined constraints.
This parameter applies to CVs, MVs, and FVs.

Trend
The Trend parameter displays a sparkline for the variable. You can use the sparkline to view
recent changes in a variable between scan intervals.
This parameter applies to CVs, MVs, and FVs.

Low Constraint
The Low Constraint parameter is the lower constraint that you place on the value of a
variable. That is, the value of the variable should not be less than the value of the Low
Constraint parameter.
The MVC controller can move the value of the variable to be less than the lower constraint
during calculations. However, the final solution of the MVC controller cannot contain a
variable with a value that is less than its associated lower constraint.
The MVC controller uses the lower constraint as a soft bound or as a hard bound based on
the associated priority of the variable. For CVs,the value of the Priority parameter
determines the priority (see Priority on page 167). For MVs, the value of the MV Priority
parameter determines the priority (see MV Priority on page 167). If the priority of the variable
is low, the MVC controller uses the lower constraint as a hard bound and does not relax the
bound during calculations. If the priority of the variable is high, the MVC controller uses the
lower constraint as a soft bound and relaxes the bound during calculations, if required.
The default value for this parameter is -1,000.00. You can set the value of this parameter to
any real value that is:

Version 2016 SP2 213

SimSci APC User Guide Enabling Controllers

Greater than or equal to the value of the Low Limit parameter.

Less than the value of the High Constraint parameter.
You can specify this parameter for CVs and MVs only.

High Constraint
The High Constraint parameter is the upper constraint that you place on the value of a
variable. That is, the value of the variable should not be greater than the value of the High
Constraint parameter.
The MVC controller can move the value of the variable to be greater than the upper constraint
during calculations. However, the final solution of the MVC controller cannot contain a
variable with a value that is greater than its associated upper constraint.
The MVC controller uses the upper constraint as a soft bound or as a hard bound based on
the associated priority of the variable. For CVs, the value of the Priority parameter
determines the priority (see Priority on page 167). For MVs, the value of the MV Priority
parameter determines the priority (see MV Priority on page 167). If the priority of the variable
is low, the MVC controller uses the upper constraint as a hard bound and does not relax the
bound during calculations. If the priority of the variable is high, the MVC controller uses the
upper constraint as a soft bound and relaxes the bound during calculations, if required.
The default value for this parameter is 1,000.00. You can set the value of this parameter to
any real value that is:
Greater than or equal to the Low Constraint parameter
Less than or equal to the value of the High Limit parameter
You can specify this parameter for CVs and MVs only.

Move
The Move parameter displays the value by which an MV changed from the previous control
interval. The MVC controller recalculates this parameter every control interval.
You cannot edit this parameter.
This parameter applies to MVs only.

CV Weight
The CV Weight parameter is a weighting factor that is applied to the associated CV during
the minimization of the cost function. The values of the CV Weight parameters are held in the
weighting factor matrix P . See Cost Function on page 195 for more information.
If you increase this parameter, the error between the current value of the CV and the
associated setpoint has a greater contribution to the cost function. Therefore, the MVC
controller more urgently drives the CV to the setpoint at the expense of other control
objectives.
If you decrease this parameter, the error between the current value of the CV and the
associated setpoint has a lesser contribution to the cost function. Therefore, the MVC
controller less urgently drives the MV to the optimization target to the advantage of other
control objectives.

214 Version 2016 SP2

Enabling Controllers SimSci APC User Guide

If you set this parameter to zero, the error between the current value of the CV and the
associated setpoint does not contribute to the cost function and the MVC controller ignores
the CV.
The default value for this parameter is 1.00. You must change this value to include the CV in
the minimization of the cost function.
You can specify this parameter for CVs only.

MV Weight
The MV Weight parameter is a weighting factor that is applied to the associated MV
increment during the minimization of the cost function. The values of the MV Weight
parameters are held in the weighting factor matrix Q . See Cost Function on page 195 for
more information.
If you increase this parameter, the current incremental change for the associated MV has a
greater contribution to the cost function. Therefore, the MVC controller moves the MV less
freely at the expense of other control objectives.
If you decrease this parameter, the current incremental change for the associated MV has a
lesser contribution to the cost function. Therefore, the MVC controller moves the MV more
freely to the advantage of other control objectives.
If you set this parameter to zero, the current incremental change for the associated MV does
not contribute to the cost function and the MVC controller ignores the increments for the MV.
The default value for this parameter is 1.00. You must change this value to include the MV
increments in the minimization of the cost function.
You can specify this parameter for MVs only.

Target Weight
The Target Weight parameter is a weighting factor that is applied to the associated MV
target during the minimization of the cost function. The values of the Target Weight
parameters are held in the weighting factor matrix R . See Cost Function on page 195 for
more information.
If you increase this parameter, the error between the current value of the MV and the
associated optimization target has a greater contribution to the cost function. Therefore, the
MVC controller more urgently drives the MV to the optimization target at the expense of other
control objectives.
If you decrease this parameter, the error between the current value of the MV and the
associated optimization target has a lesser contribution to the cost function. Therefore, the
MVC controller less urgently drives the MV to the optimization target to the advantage of
other control objectives.
If you set this parameter to zero, the error between the current value of the MV and the
associated optimization target does not contribute to the cost function and the MVC controller
ignores the MV target.
The default value for this parameter is 0.00. You must change this value to include the MV
targets in the minimization of the cost function.
You can specify this parameter for MVs only.

Version 2016 SP2 215

SimSci APC User Guide Enabling Controllers

Status
The Status parameter indicates the current status of the quality for the variable tag. It can
have one of the following values:
Blank: The tag has a good quality. There are no communication errors or validation
errors associated with the tag.
Bad: The tag has a bad quality. There are either communication errors or validation
errors associated with the tag.
Off: The tag is turned off.
High Constraint: The tag is close to or at its associated upper constraint.
Low Constraint: The tag is close to or at its associated lower constraint.
You cannot edit this parameter.
This parameter applies to CVs, MVs, and FVs.

Increment
The Increment parameter displays the value of the Increment parameter in the MVC
controller assembly. It displays the maximum amount by which an MV value can change from
one control action to the next. See Increment on page 164 for more information.
You cannot edit this parameter.
This parameter applies to MVs only.

Tracking
The Tracking parameter is the mean squared error of the model predictions for the
associated variable. The MVC controller evaluates the mean squared errors by using the 5-
step-ahead predictions and the archived data from the previous 30 control actions. The
Tracking parameter should have a value of less than 0.5 to indicate acceptable accuracy in
the predictions for the associated variable.
You cannot edit this parameter.
This parameter applies to CVs only.

Changing the Parameters for the Enabled MVC Controller

You can change the configuration parameters for the MVC controller and parameters for the
variables in the MVC controller while the MVC controller is enabled, that is, while the MVC
controller is in the Ready or Active state.
On the Controller Details tab, on the MVC tab, you can change all of the configuration
parameters for the MVC controller while the MVC controller is enabled except for the
following parameters:
Control Interval on page 156
Coupling on page 156
On the Controller Details tab, on the MVC tab, you can change all of the parameters for the
variables in the MVC controller while the MVC controller is enabled except for the Constraint
Mode parameter for CVs. See Constraint Mode on page 163 for more information.

216 Version 2016 SP2

Enabling Controllers SimSci APC User Guide

On the Controller Overview tab, you can change all of the editable online configuration
parameters for the MVC controller and all of the editable online parameters for the variables
in the MVC controller while the MVC controller is enabled.
The only online configuration parameter that you can edit for the MVC controller is the
Controller Mode parameter. See Controller Mode on page 210 for more information.
You can edit only the Set Point and the CV Weight online parameters for the CVs in the
MVC controller. See Set Point on page 211 and CV Weight on page 214 for more
information.
You can edit the following online parameters for the MVs in the MVC controller:
Opt Target on page 211
MV Weight on page 215
Target Weight on page 215

Acti vity Directions

Changing the configuration paramet ers On the Controller Details tab, on the MVC tab, in
while the MVC controller is enabled the Configuration area, change the configuration
parameters as desired.

Changing the parameters for CVs while the On the Controller Details tab, on the MVC tab, on
MVC controller is enabled the Control variables tab, change the parameters
for the CVs as desired.

Changing the parameters for MVs while the On the Controller Details tab, on the MVC tab, on
MVC controller is enabled the Manipulated variables tab, change the
parameters for the MVs as desired.

Changing the parameters for FVs while the On the Controller Details tab, on the MVC tab, on
MVC controller is enabled the Feed Forward variables tab, change the
parameters for the FVs as desired.

Changing the online configuration On the Controller Overview tab, in the

parameters while the MVC cont roller is Configuration area, change the Controller Mode
enabled parameter as desired.

Changing the online parameters for CVs On the Controller Overview tab, on the Control
while the MVC controller is enabled variables tab, change the editable online
parameters for the CVs as desired.

Changing the online parameters for MVs On the Controller Overview tab, on the
while the MVC controller is enabled Manipulated variables tab, change the editable
online parameters for the MVs as desired.

Example: Changing the Parameters for CVs While the MVC Controller is Enabled
1. On the Controller Details tab, on the MVC tab, select the Control variables tab.
2. Change the parameters for a single CV or for multiple CVs as desired.

Version 2016 SP2 217

SimSci APC User Guide Enabling Controllers

Managing Trends for the MVC Controller

You can create trends for an MVC controller by using the Controller Overview tab. You can
create trends for a subcontroller in an MVC controller by using the Sub Controller Summary
tab. You can create two kinds of controller trends:
CV trends
MV trends
NOTE: You cannot create controller trends for the FVs.
For an MVC controller, you can create a CV trend only when you select the Control
variables tab on the Controller Overview tab. For a subcontroller, you can directly create a
CV trend by using commands on the tab for the associated MVC controller, on the Sub
Controller Summary tab. The SimSci APC software displays a CV trend as a normal trend.
However, a CV trend automatically contains a superimposed Strip for each CV you select
when you create the CV trend. Each superimposed Strip displays data for a number of
parameters for the associated CV. You can view the following CV parameters on the
associated superimposed Strip:
Current Value
High Constraint
Low Constraint
Set Point
OneStep Prediction
FiveStep Prediction
LongRangeModel Prediction
You can add and remove the preceding parameters as desired by using the same procedures
you use for normal trends. See Managing Trends on page 34 for more information.
For an MVC controller, you can create an MV trend only when you select the Manipulated
variables tab on the Controller Overview tab. For a subcontroller, you can directly create an
MV trend by using commands on the tab for the associated MVC controller, on the Sub
Controller Summary tab.The SimSci APC software displays an MV trend as a normal trend.
However, an MV trend automatically contains a superimposed Strip for each MV you select
when you create the MV trend. Each superimposed Strip displays data for a number of
parameters for the associated MV. You can view the following MV parameters on the
associated superimposed Strip:
Current Value
High Constraint
Low Constraint
Opt Target
You can add and remove the preceding parameters as desired by using the same procedures
you use for normal trends. See Managing Trends on page 34 for more information.

218 Version 2016 SP2

Enabling Controllers SimSci APC User Guide

On CV and MV trends, you can view the current value, maximum value, and minimum value
for the CV or MV parameters for a Strip by using the Configure area for that Strip. You can
also adjust the scaling for the associated superimposed Strip by using the same procedures
you use for normal trends. See Adjusting the Scaling for a Strip on page 39 for more
information.

Acti vity Directions

Creating CV trends from an MV C On the Controller Overview tab, select the Control
controller variables tab, and then select the CVs for which you want to

create a CV trend. Click Controller Trends . A CV trend

that contains a superimposed Strip for each selected CV
appears.

Creating CV trends from a On the Sub Controller Summary tab, on the tab for the
subcontroller associated MVC controller, in the CV Trends column, click

ControlVariable Trends for the subcont roller. A CV

trend that contains a superimposed Strip for each CV in the
subcontroller appears.

Viewing parameter data for a CV On the CV trend, on the Strip for the desired CV, click
Configure to expand the Configure area. On the Variables
tab, view the parameter data associated with the CV.

Creating MV trends from an MVC On the Controller Overview tab, select the Manipulated
controller variables tab, and then select the MVs for which you want to

create an MV trend. Click Controller Trends . An MV

trend that contains a superimposed Strip for each selected MV
appears.

Creating MV trends from a On the Sub Controller Summary tab, on the tab for the
subcontroller associated MVC controller, in the MV Trends column, click

Manipulated Trends for the subc ontroller. An MV trend

that contains a superimposed Strip for each MV in the
subcontroller appears.

Viewing parameter data for an On the MV trend, on the Strip for the desired MV, click
MV Configure to expand the Configure area. On the Variables
tab, view the parameter data associated with the MV.

Creating Radial Plots for the MVC Controller

In the SimSci APC software, radial plots are the radii of a radar chart that displays the current
values of the CVs or MVs in relation to their associated targets values, upper constraints, and
lower constraints. You can use radial plots to identify commonalities among variables and to
quickly determine the proximity of a variable to its associated constraints and target value.

Version 2016 SP2 219

SimSci APC User Guide Enabling Controllers

Each radial plot displays the following values for its associated variable:
Current Value
High Constraint
Set Point
Low Constraint
You can create radial plots for the MVC controller by using the Controller Overview tab. You
can create two kinds of radial plots:
CV radial plots
MV radial plots
NOTE: You cannot create radial plots for the FVs.
You can create CV radial plots only when you select the Control variables tab on the
Controller Overview tab. The SimSci APC software displays CV radial plots in the CV
Targets and Constraints window. The CV Targets and Constraints window contains a
single radar chart that includes a scaling radius and a CV radius (that is, a CV radial plot) for
each CV in the MVC controller.
You can create MV radial plot only when you select the Manipulated variables tab on the
Controller Overview tab. The SimSci APC software displays MV radial plots in the MV
Targets and Constraints window. The MV Targets and Constraints window contains a
single radar chart that includes a scaling radius and an MV radius (that is, an MV radial plot)
for each MV in the MVC controller.
The scaling radius displays the axis scaling for all of the radial plots on the radar chart. The
axis scaling is determined by the maximum constraint value and the minimum constraint
value for all of the variables on the radar chart.
To simplify the appearance of the radial plots, you can add or remove tags from the radar
chart. When you add or remove tags from the radar chart, the scaling changes to
accommodate the new set of tags.

Acti vity Directions

Creating CV radial plots On the Controller Overview tab, select the Control variables tab,

and then click Radial Plots . The CV Targets and Constraints

window appears.

Removing tags from the In the CV Targets and Constraints window, in the
radar chart (CVs) Check/UnCheck Tags area, clear the associated check boxes for
the tags you want to remove from the radar chart.

Adding tags to the radar In the CV Targets and Constraints window, in the
chart (CVs) Check/UnCheck Tags area, select the associated check boxes for
the tags you want to add to the radar chart.

Creating MV radial plots On the Controller Overview tab, select the Manipulated variables

tab, and then click Radial Plots . The MV Targets and

Constraints window appears.

220 Version 2016 SP2

Enabling Controllers SimSci APC User Guide

Removing tags from the In the MV Targets and Constraints window, in the
radar chart (MVs) Check/UnCheck Tags area, clear the associated check boxes for
the tags you want to remove from the radar chart.

Adding tags to the radar In the MV Targets and Constraints window, in the
chart (MVs) Check/UnCheck Tags area, select the associated check boxes for
the tags you want to add to the radar chart.

Example: Creating CV Radial Plots

1. On the Controller Overview tab, select the Control variables tab.

2. Click Radial Plots . The CV Targets and Constraints window appears.

Enabling the Optimizer

You can enable the Optimizer by initializing it and then activating it. You can initialize and
activate the Optimizer by using the Optimizer Overview tab, which you can open by clicking
Optimizer Overview in the APC Explorer, under Promote. You can initialize and activate the
Optimizer only when the APC Project is scanning.
The Optimizer begins in the Manual state. When you initialize the Optimizer, it transitions to
the Waiting state, then to the Initialize state, and finally to the Ready state. When you active
the Optimizer, it transitions from the Ready state to the Auto state, and then to the Active
state. The Optimizer must be in the Ready state before you can activate it. That is, you must
initialize the Optimizer before you activate it.
You can view the current state of the Optimizer by using the Current Status parameter on
the Optimizer Overview tab, in the Configuration area. When the state of the Optimizer
changes, informational messages indicating the change appear in the Diagnostics window.

Acti vity Directions

Starting scanning On the Home tab, in the Scan group, click On.

Initializing the Optimizer On the Optimizer Overview tab, in the Configuration

area, select Initialize for the Controller Mode parameter,

and then click Change Status . The Optimizer

transitions to the Waiting state, then to the Initialize state,
and finally to the Ready state.

Activating the Optimizer On the Optimizer Overview tab, in the Configuration

area, select Active for the Controller Mode parameter,

and then click Change Status . The Optimizer

transitions to the Auto state and then to the Active state.
The Optimizer must be in the Ready state before you can
activate it.

Version 2016 SP2 221

SimSci APC User Guide Enabling Controllers

Deactivating the Optimizer On the Optimizer Overview tab, in the Configuration

area, select either Ready, Initialize, or Manual for the
Controller Mode parameter, and then click Change

Status . If you select Ready, the Optimizer returns to

the Ready state. If you select Initialize, the Optimizer
reinitializes and returns to the Ready state. If you select
Manual, the Optimizer returns to the Manual state.

Stopping scanning On the Home tab, in the Scan group, click Off.

Example: Initializing the Optimizer

1. On the Optimizer Overview tab, expand the Configuration area.

2. In the Controller Mode list, select Initialize, and then click Change Status .

State of the Optimizer

The state of the Optimizer is the current mode of operation for the Optimizer. The Optimizer
can be in one of five states:
Manual: The Optimizer does not perform any calculations and does not write values to
the MVC controller. When the APC Project is scanning, the Optimizer updates the current
values for the input tags and output tags with the latest process data.
Waiting: The Optimizer performs steps to transition to the Initialize state, and then
automatically transitions to the Initialize state. It typically stops at the Waiting state only if
an error occurs due to one of the following:
o A vital tag is off
o A communication error occurs, which is indicated by static watchdog tags
o The Optimizer does not contain CVs or MVs
Initialize: The Optimizer refreshes the values for all parameters in the Optimizer, and
then automatically transitions to the Ready state.
Ready: The Optimizer maximizes the profit function at the specified optimization interval.
That is, the Optimizer calculates the optimum steady-state values for the process. It does
not write values to the MVC controller. See Profit Function on page 201 for more
information.
Auto: The Optimizer performs steps to transition to the Active state.
Active: The Optimizer maximizes the profit function at the specified optimization interval.
If the Optimizer is coupled to the MVC controller, the Optimizer also writes the optimum
steady-state values to the MVC controller at the specified control interval.
You can use the Current Status parameter on the Optimizer Overview tab, in the
Configuration area, to determine the current state of the Optimizer.

222 Version 2016 SP2

Enabling Controllers SimSci APC User Guide

Online Configuration Parameters for the Optimizer

You can set and view the online configuration parameters for the Optimizer by using the
Optimizer Overview tab. You can open the Optimizer Overview tab by clicking Optimizer
Overview in the APC Explorer, under Promote. The Optimizer contains the following online
configuration parameters:
Controller Mode on page 223
Current Status on page 223

Controller Mode
The Controller Mode parameter allows you to select and change the state of the Optimizer.
The amount of states that you can select depends on the current state of the Optimizer. If the
Optimizer is in a Manual state, you can select only Manual or Initialize. If the Optimizer is in
a Ready or Active state, you can select Manual, Initialize, Ready, or Active.

After you select a state, you must click Change Status to change the state of the
Optimizer.See Enabling the Optimizer on page 221 and State of the Optimizer on page 222
for more information.
You can edit this parameter only when the APC Project is scanning.

Current Status
The Current Status parameter indicates the current online state of the Optimizer. See State
of the Optimizer on page 222 for more information.
You cannot edit this parameter.

Online Parameters for the Variables in the Optimizer

You can view and specify the online parameters for both input and output variables in the
Optimizer by using the Optimizer Overview tab. The Optimizer Overview tab contains a tab
for each type of variable in the Optimizer.
On the Optimizer Overview tab, on the Control variables tab, you can access the following
parameters for the CVs:
Current Value on page 224
Trend on page 224
Optimum on page 225
Status on page 225
Vital on page 225
Origin on page 226
On the Optimizer Overview tab, on the Manipulated variables tab, you can access the
following parameters for the MVs:
Current Value on page 224
Trend on page 224

Version 2016 SP2 223

SimSci APC User Guide Enabling Controllers

Optimum on page 225

Status on page 225
Vital on page 225
Origin on page 226
On the Optimizer Overview tab, on the Feed Forward variables tab, you can access the
following parameters for the FVs:
Current Value on page 224
Trend on page 224
Status on page 225
Vital on page 225
Origin on page 226

Current Value
The Current Value parameter displays the current value of the variable.
The value appears on a slider. The greatest value on the slider is set equal to the upper limit
for the variable. The least value on the slider is set equal to the lower limit for the variable.
See High Limit on page 186 and Low Limit on page 186 for more information. Two nodes
appear on the slider to indicate the upper and lower constraints for the variable. See High
Constraint on page 185 and Low Constraint on page 185 for more information. The color of
the slider indicates the proximity of the current value of the variable to the constraints.
The color of the slider is green if both of the following statements are true:
The current value of the variable is greater than the lower constraint by a margin of 5% of
the lower constraint.
The current value of the variable is less than the upper constraint by a margin of 5% of
the upper constraint.
The color of the slider is yellow if either of the following statements is true:
The current value of the variable is within a 5% margin of the lower constraint.
The current value of the variable is within a 5% margin of the upper constraint.
The color of the slider is red if either of the following statements is true:
The current value of the variable is greater than the upper constraint by a margin of 5% of
the upper constraint.
The current value of the variable is less than the lower constraint by a margin of 5% of the
lower constraint.
The color of the slider is always gray for the FVs, which do not have defined constraints.
This parameter applies to CVs, MVs, and FVs.

Trend
The Trend parameter displays a sparkline for the variable. You can use the sparkline to view
recent changes in a variable between scan intervals.

224 Version 2016 SP2

Enabling Controllers SimSci APC User Guide

This parameter applies to CVs, MVs, and FVs.

Optimum
The Optimum parameter is the value of the optimum steady-state setpoint or target for the
associated variable. When the Optimizer maximizes the profit function, it calculates the
optimum steady-state setpoint for all CVs in the Optimizer and the optimum steady-state
targets for all MVs in the Optimizer. See Profit Function on page 201 for more information.
The Optimizer calculates the value of the Optimum parameter at the interval specified by the
Optimize Interval parameter. See Optimize Interval on page 179 for more information.
If the Optimizer is coupled with the MVC controller, the Optimizer:
Writes the value of the Optimum parameter for all CVs to the associated Set Point
parameter in the MVC controller. See Set Point on page 211 for more information.
Writes the value of the Optimum parameter for all MVs to the associated Opt Target
parameter in the MVC controller. See Opt Target on page 211 for more information.
The Optimizer writes the values of the Optimum parameters to the MVC controller at the
interval specified by the Control Interval parameter. See Control Interval on page 179 for
more information.
You cannot edit this parameter.
This parameter applies to CVs and MVs only.

Status
The Status parameter indicates the current status of the quality for the variable tag. It can
have one of the following values:
Blank: The tag has a good quality. There are no communication errors or validation
errors associated with the tag.
Bad: The tag has a bad quality. There are either communication errors or validation
errors associated with the tag.
Off: The tag is turned off.
High Constraint: The tag is close to or at its associated upper constraint.
Low Constraint: The tag is close to or at its associated lower constraint.
You cannot edit this parameter.
This parameter applies to CVs, MVs, and FVs.

Vital
The Vital parameter displays the value of the Vital parameter in the Optimizer assembly,
which is a check box that determines whether or not a variable is vital to the Optimizer. If the
check box is selected, the Optimizer considers the associated variable as vital to the
operations of the Optimizer. If the check box is cleared, the Optimizer considers the
associated variable as nonessential to the operations of the Optimizer. See Vital on page 184
for more information.
You cannot edit this parameter.
This parameter applies to CVs, MVs, and FVs.
Version 2016 SP2 225
SimSci APC User Guide Enabling Controllers

Origin
The Origin parameter is the initial steady-state value for the associated variable. The
Optimizer uses the initial steady-state values of the variables as the origin for the
maximization of the profit function. See Profit Function on page 201 and Initial Steady-state
Values on page 204 for more information.
You can edit this parameter only if you set the Origin Source parameter for the Optimizer to
Origin. In this case, the Optimizer uses the values that you set for the Origin parameters.
Otherwise, the Optimizer calculates the Origin parameter for all the variables based on the
selection for the Origin Source parameter:
MVC model for Steady State: The Optimizer calculates the Origin parameter for the
CVs from the linear model for the MVC controller. It sets the Origin parameters for the
MVs and FVs to the current values of the associated variables.
LP ARX Model: The Optimizer calculates the Origin parameter for the CVs from the
linear model for the Optimizer. It sets the Origin parameters for the MVs and FVs to the
current values of the associated variables.
See Origin Source on page 179 for more information.
You can specify this parameter for CVs, MVs, and FVs.

Changing the Parameters for an Enabled Optimizer

You can change the configuration parameters for the Optimizer and the parameters for the
variables in the Optimizer while the Optimizer is enabled, that is, while the Optimizer is in the
Ready or Active state.
On the Controller Details tab, on the Optimizer tab, you can change all of the configuration
parameters for the Optimizer while the Optimizer is enabled except for the following
parameters:
Control Interval on page 179
Optimize Interval on page 179
Origin Source on page 179
On the Controller Details tab, on the Optimizer tab, you can change all of the parameters
for the variables in the Optimizer while the Optimizer is enabled. If the MVC controller is
coupled with the Optimizer, the MVC controller sets some of these parameters. You cannot
change the parameters that the MVC controller sets.
On the Optimizer Overview tab, you can change all the editable online configuration
parameters for the Optimizer while the Optimizer is enabled. Currently, the only online
configuration parameter that you can edit for the Optimizer is the Controller Mode
parameter. See Controller Mode on page 223 for more information.
You cannot edit any online parameters for the variables in the Optimizer.

Acti vity Directions

Changing the configuration paramet ers On the Controller Details tab, on the Optimizer
while the Optimizer is enabled tab, in the Configuration area, change the
configuration parameters as desired.

226 Version 2016 SP2

Enabling Controllers SimSci APC User Guide

Changing the parameters for CVs while the On the Controller Details tab, on the Optimizer
Optimizer is enabled tab, on the Control variables tab, change the
parameters for the CVs as desired.

Changing the parameters for MVs while the On the Controller Details tab, on the Optimizer
Optimizer is enabled tab, on the Manipulated variables tab, change
the paramet ers for the MVs as desired.

Changing the parameters for FVs while the On the Controller Details tab, on the Optimizer
Optimizer is enabled tab, on the Feed Forward variables tab, change
the paramet ers for the FVs as desired.

Changing the online configuration On the Optimizer Overview tab, in the

parameters while the Optimizer is enabled Configuration area, change the Controller Mode
parameter as desired.

Changing the online parameters for CVs On the Optimizer Overview tab, on the Control
while the Optimizer is enabled variables tab, change the editable online
parameters for the CVs as desired.

Changing the online parameters for MVs On the Optimizer Overview tab, on the
while the Optimizer is enabled Manipulated variables tab, change the editable
online parameters for the MVs as desired.

Example: Changing the Parameters for CVs While the Optimizer is Enabled
1. On the Controller Details tab, on the Optimizer tab, select the Control variables tab.
2. Change the parameters for a single CV or for multiple CVs as desired.

Version 2016 SP2 227

 SimSci APC User Guide

C HAPTER 8
Monitoring Controllers

In This Chapter
SimSci APC Reports ................................................................................................................... 229
Tag Data Reports ....................................................................................................................... 229
Managing Reports ...................................................................................................................... 230

SimSci APC Reports

You can monitor controllers by using the APC Process Monitor software distributed with the
Connoisseur software.
The APC Process Monitor software uses .fdax files created by the Connoisseur software to
create a Microsoft Excel workbook that you can use to monitor the process dynamics.
You can use the Process Monitor Configuration tab in the SimSci APC software to
generate SimSci APC report (.fdax) files. You can open the Process Monitor Configuration
tab by clicking Configuration in the APC Explorer, under Monitor.
The SimSci APC software stores the .fdax files in the folder that you specify in the Browse
box on the Configure tab, in the Data Management group. Note that you must click Browse
to specify a folder.
Please see the APC Process Monitor User Guide distributed with the APC Process Monitor
software for more information.

Tag Data Reports

You can create and schedule reports for the tag data in an APC Project.
For a tag data report, the SimSci APC software exports data for all the tags in the APC
Projects as a .tagdata file that you can later import into another APC Project.
The SimSci APC software stores the tag data reports in the folder that you specify in the
Browse box on the Configure tab, in the Data Management group. Note that you must click
Browse to specify a folder.
Tag data reports are especially useful if you want to create response testing data in one APC
Project and model the response testing data in another APC Project. The workflow for this
type of application is as follows:
1. Open an APC Project with a configured PRBS Generator and a scheduled tag data
report.
2. Run the PRBS Generator to create response testing data.

Version 2016 SP2 229

SimSci APC User Guide Monitoring Controllers

3. Run the PRBS Generator until the SimSci APC software creates a new tag data report.
Alternatively, you can manually generate the tag data report at any time.
4. Copy the tag data report to a common folder (for example, ...\APC2016\User\Data).
5. Close the APC Project.
6. Create a new APC Project.
7. Import the tag data report.
8. Create models by using the imported tag data.

Managing Reports
You can create SimSci APC reports or tag data reports. Each report has an associated file
type, report time, report length, and number of versions to keep on file.
You can create as many reports as you need. The SimSci APC software stores the reports in
the folder specified on the Configure tab, in the Data Management group.

Acti vity Directions

Creating a SimSci APC (.fdax) On the Proce ss Monitor Configuration tab, click Add
report
Configuration . A new report appears in the list of reports. In
the File Type list for the new report, select Comma delimited
(*.fdax).

Creating a tag data (.tagData) On the Proce ss Monitor Configuration tab, click Add
report
Configuration . A new report appears in the list of reports. In
the File Type list for the new report, select Tag Data (*.tagData).

Renaming a report On the Proce ss Monitor Configuration tab, double-click the File
Name parameter for the report. Ty pe the new name, and then press
Enter.

Deleting a report On the Proce ss Monitor Configuration tab, select the report, and

then click Delete Configuration .

Editing a report On the Proce ss Monitor Configuration tab, double-click the

Report Time parameter for the process monitor configuration. Type
the new value and then press Enter. Repeat for the Data Hi story
and Keep Versions parameters.

Generating reports On the Proce ss Monitor Configuration tab, click Generate File(s)

Specifying the folder in which On the Configure tab, in the Data Management group, click
to save the reports
Brow se . The Brow se For Folder dialog box appears. Browse
to and select the folder in which you want to store the reports, and
then click OK.

230 Version 2016 SP2

Monitoring Controllers SimSci APC User Guide

Acti vity Directions

Specifying a time to generate On the Proce ss Monitor Configuration tab, in the Report Time
a report column, type in the time at whic h you want a report to be generated.
Once that time has passed, a report will be generated into the folder
specified on the Configure tab.

Specifying the sourc e type for On the Proce ss Monitor Configuration tab, near the File Name
the data source
column, click the Field Chooser , and make sure the check box
next to Source Type is checked. Then, select the desired tag or
controller collection from the dropdown menu in the Source Type
column.

Version 2016 SP2 231

 SimSci APC User Guide

C HAPTER 9
Director Calculations

In This Chapter
Overview.................................................................................................................................... 233
Script Editor ............................................................................................................................... 233
Managing Scripts ........................................................................................................................ 234
Managing Sequences ................................................................................................................. 235

Overview
Director calculations are customized programs that you can use to customize your SimSci
APC application. You can create Director calculations by using the Director.
The Director is the internal programming facility for the SimSci APC software. It uses
IronPython as a language base. You can use any native IronPython function to create a
Director calculation. You can also use built-in Director functions that are specific to the SimSci
APC software. See the SimSci APC Director Functions Guide for full descriptions of the
available native IronPython functions and the built-in Director functions, including syntax
information and examples.
A Director calculation is specific to a single APC Project and the SimSci APC software saves
a Director calculation as part of the APC Project. Each Director calculation contains a set of
scripts and sequences. The scripts contain the code for the Director calculations and the
sequences determine when the code runs for the project. See Managing Scripts on page 234
and Managing Sequences on page 235 for more information.
You can manage a Director calculation by using the Script Editor. See Script Editor on page
233 for more information.

Script Editor
The Script Editor is a window that you can use to manage the scripts and sequences for a
Director calculation. You can open the Script Editor by clicking Editor on the Tool tab, in the
Director Calc group. The Script Editor has three areas:
Script and Sequences Pane: You can create, delete and rename scripts and sequences
by using this pane. (A)
Editor Pane: You can edit a script or a sequence by using this pane. (B)

Version 2016 SP2 233

SimSci APC User Guide Director Calculations

Diagnostics Pane: You can view the outputs for all scripts or sequences that are running
by using this pane. (C)

Managing Scripts
A script is an IronPython (.py) file that contains code you can use to create a Director
calculation. You can manage scripts by using the Script Editor. See Script Editor on page 233
for more information.
There are two types of scripts:
System Scripts: These are scripts that are installed with the SimSci APC software. You
cannot use these scripts in a sequence. You cannot edit these scripts, but you can copy
and paste the contents of these scripts into any user script.
User Scripts: These are scripts that you create. You can use these scripts in any
sequence and you can modify them at any time.
You can create, rename, and delete scripts by using the Scripts and Sequences Pane on the
Script Editor.
After you create a script, you can edit and run the script by using the Editor Pane for the
script. You can open the Editor Pane for the script by selecting the script in the Scripts and
Sequences Pane.
The SimSci APC software interprets the code for a Director calculation at runtime instead of
compiling the code before you perform runs. You can check the syntax of the code by running
the script. When you run a script, the SimSci APC software automatically saves the code in
the script and error messages appear on the Output tab in the Script Editor if the code
contains syntax errors.

234 Version 2016 SP2

Director Calculations SimSci APC User Guide

Acti vity Directions

Creating a script Double-click Add new script. Type the name of the new script,
and then press Enter.

Renaming a script Double-click the script. Type a new name for the script, and then
press Enter.

Deleting a script Right -click the script, and then click Delete.

Editing a script On the Editor Pane, edit the code for the script.

Running a script
On the Editor Pane, click Run Scripts .

Saving a script
On the Editor Pane, click Save Script .

Saving all scripts On the Editor Pane, click Save All.

Example: Creating a Script

1. On the Scripts and Sequences Pane, on the Scripts tab, expand the User Scripts node.
2. Double-click Add new script.
3. Type the name of the script, and then press Enter.
4. On the Editor Pane, type the code for the script.
Example: Running a Script
1. On the Scripts and Sequences Pane, on the Scripts tab, select the script that you want to
run.

2. On the Editor Pane, click Run Scripts .

Managing Sequences
A sequence is a set of scripts that you can schedule to run at a specified interval. You can
manage sequences by using the Script Editor. See Script Editor on page 233 for more
information.
You can create, rename, and delete sequences by using the Scripts and Sequences Pane on
the Script Editor.
After you create a sequence, you can specify the interval for the sequence and add scripts to
the sequence by using the Editor Pane for the sequence, which you can open by selecting
the sequence in the Scripts and Sequences Pane. The interval for the sequence determines
how often the sequence runs, and therefore, how often the scripts in the sequence run. The
Director runs all the scripts in the sequence for each sequence run. The scripts in the
sequence run one after another in the order in which they appear in the Scheduled Scripts
area.

Version 2016 SP2 235

SimSci APC User Guide Director Calculations

Each script in a sequence has a specified break time. The break time determines how long a
script in a sequence can run before the Director exits the script run. If the script run does not
complete within the amount of time specified by the break time, an error message appears in
the Diagnostics window. If you set the break time to -1, the Director waits for an infinite
amount of time for the script run to complete.The Director runs any subsequent scripts in the
sequence regardless of whether the previous scripts completed their respective runs or not.
After you specify the interval for a sequence and add scripts to the sequence, you can take
the sequence online. When a sequence is online, you cannot modify it. You must take the
sequence offline to make changes. An online sequence runs at the specified interval when
the APC Project is scanning. See Scanning on page 43 for more information.
You can associate a sequence with the MVC controller. Associating a sequence with the
MVC controller does not change how or when the sequence runs.

Acti vity Directions

Creating a sequence Double-click Add new sequence, and then type the
name of the new sequence. Then, press Enter.

Renaming a sequence Double-click the sequence, and then type the new
name. Then, press Enter.

Deleting a sequence Right -click the sequence, and then click Delete.

Specifying the interval for a sequence In the Interval (Milliseconds) box, enter the run
interval, in milliseconds, for the sequence. If you set the
run interval to zero, the sequence runs continuously.

Adding a script to a sequence In the Scheduled Scripts area, click Add Scheduled

Script . In the Name list, select the script you want

to use. Then, in the Break Time (Milliseconds) box,
enter the break time, in milliseconds for the script.

Deleting scripts from a sequence In the Scheduled Scripts area, click the area to the left
of the Name column to select the scripts that you want

to delete, and then click Delete Scheduled Script .

Note that selected scripts are highlight ed.

Taking a sequence online In the Scheduler State area, click Online.

Taking a sequence offline In the Scheduler State area, click Offline.

Associating a sequence with the MVC On the Controller Details tab, on the MVC tab, in the
controller Associated Sequence list, select the sequence you
want to associate with the MVC controller.

Example: Creating a Sequence

1. On the Scripts and Sequences Pane, on the Sequences tab, expand the Sequences
node.
2. Double-click Add new sequence.

236 Version 2016 SP2

Director Calculations SimSci APC User Guide

3. Type the name of the sequence.

4. Press Enter.
Example: Adding a Script to a Sequence

1. On the Editor Pane, in the Scheduled Scripts area, click Add Scheduled Script .
2. In the Name list, select the script that you want to add to the sequence.
3. In the Break Time (Milliseconds) column, type the break time, in milliseconds, for the
script.
4. Press Enter.

Version 2016 SP2 237

 SimSci APC User Guide

A PPENDIX A
Glossary of Terms
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Actuator
The mechanical device that enacts a change in the physical system that is under control.

Comments
A description box you can use to provide any significant or useful information about the tag.
Control Interval
The frequency at which a controller calculates and updates the values of its associated MVs.
Controller
A device used to act on process MVs so that the process transfers form its current operating
point to some desired target operating point. Both the MVC controller and the Optimizer are
controllers.
Coupling
The extent to which the MVC controller and the Optimizer are linked.
CV
A controlled variable. Typically, CVs are input tags.

DCS
Distributed control system.
Description
A description box you can use to describe or otherwise identify a tag.
Director
The internal programming facility for the SimSci APC software. It uses IronPython as a
language base.

Version 2016 SP2 239

SimSci APC User Guide Glossary of Terms

Director Calculation
A customized program you can use to customize your SimSci APC application. It contains a
set of scripts and sequences.

FV
A feedforward variable. An FV can be an input tag or an output tag.

Graph Limit
The number of trends that can appear on one page of a Trend window.

LP
Linear Programming. An optimization technique used to establish the economic optimum
operating point, based on a user-supplied cost function.

MIMO
Multiple-input, multiple-output.
MV
A manipulated variable. Typically, MVs are output tags.
MVC
Multivariable Control.

OLS
Ordinary Least Squares. A mathematical technique used to derive linear process models
from plant data. This method considers all plant data simultaneously, obtaining a set of model
coefficients by minimizing the square of the model prediction error.
OPC Group
A grouping of tags on the OPC server, which the APC Project uses to communicate with the
OPC server.

240 Version 2016 SP2

Glossary of Terms SimSci APC User Guide

PLS
Partial Least Squares. A mathematical technique used to derive linear process models from
plant data. This method considers all plant data simultaneously, obtaining a set of model
coefficients by minimizing the square of the model prediction error. It uses sophisticated
mathematics to deal with insufficient or bad data that would otherwise not reveal any
significant relationships.
PRBS
Pseudo random binary sequence. A digital test signal that simulates white noise and is used
for generating statistically rich plant data.
Prediction Interval
The frequency at which the linear model generates a prediction.
Priority
A positive integer used with the CVs and MVs in a controller to define the ranking of the
relative importance of the variables. It permits the controller to balance the number of
satisfied control objectives with the number of degrees of freedom.

QP
Quadratic programming. A constrained optimization technique used by MVC controllers.

Readback Signal
An input tag that monitors the actual value of a field point that is updated by its associated
output tag (MV).
RLS
Recursive Least Squares. A mathematical technique used to derive linear process models
from plant data. It employs a recursive approach wherein plant data is considered point by
point form beginning to end until converging on a fixed set of model coefficients, minimizing
the squares of the model prediction error.
Ruleset
A set of user-specified rules used to filter OPC server tags for the purpose of automatically
adding the tags to the APC Project.

Scan Interval
The frequency at which the APC Project communicates with the OPC servers, and thus, with
the DCS.

Version 2016 SP2 241

SimSci APC User Guide Glossary of Terms

Scanning
A state of the APC Project in which the APC Projects performs real-time APC operations at
the specified scan interval. This includes, but is not limited to, updating tag values, running
scheduled Director sequences, and applying activated controllers.
Script
An IronPython (.py) file that contains code you can use to create a Director calculation.
Script Editor
A UI component that you can use to manage the scripts and sequences for a Director
calculation.
Sequence
A set of scripts that you can schedule to run at a specified interval.
SimSci APC Button

The SimSci APC Button contains options to create, import, open, save, and close APC
Projects.
SISO
Single-input, single-output.
Smooth
A mathematical curve operation that removes noise from the identified impulse model
coefficients for the final linear model.
Sparkline
A small trend that lacks axes and coordinates. It provides a quick, simplified preview for a
trend. You can use it to quickly determine the shape of the trend and compare the trend to
other trends.

Tag
An object on an OPC server containing stored and current data related to plant
measurements.
Translation Table
A UI component that you can use to manage string translations between the OPC server and
the APC Project.
Trend
An object that graphically displays the time-dependent behavior of tag values.
Trend Manager
A UI component that you can use to manage and reopen saved Trend windows.

242 Version 2016 SP2

Glossary of Terms SimSci APC User Guide

Trend Range
The maximum amount of time displayed on a trend while the APC Project is scanning.

Unit
A description box you can use to describe the units of measure associated with a tag.
Update Interval
The frequency at which an OPC Group updates its associated input tag information in the
APC Project. When an input tag is updated, the current value in the DCS overwrites the
current value in the APC Project.

Validity Limit
The maximum and minimum value a tag can take and still be considered of good quality.
Variable Relationship
The results of a model identification that are associated with a certain pair of variables. The
variable pair can be a CV coupled with an MV or a CV coupled with an FV.

Version 2016 SP2 243