You are on page 1of 7

<<Contents>> <<Index>>

General NTPS410
Specications Exasmoc
Multivariable Optimising Control
GS 36J06D10-01E Overview


Multivariable Control Technology is gaining popularity A multivariable controller
in the process industry. It is now a unanimous Exasmoc periodically adjusts the level of several
conclusion that APC (Advanced Process Control) manipulated variables, so as to bring/keep the
delivers sustainable and measurable benets by controlled variables at or within given targets, taking
simply stabilizing the plant to produce more and into account all the steady-state and dynamic
more stable product quality. The foundation for interactions between variables.
this conclusion is that control strategies should be
developed from an understanding of the process and Set point and/or minimum/maximum objective
its nuances, a grasp of control systems over which The objective of each controlled variable can be
APC will exist, the need for integrating it with wider specied either as a set-point, or as the range between
plant objectives and a knowledge of base layer control minimum and maximum constraints. The controller
loops. takes no action if the controlled variable lies within the
Exasmoc, a product of Yokogawa-Shell alliance,
combines these demanding necessities of the industry. Selectable constraints on the control actions
It is a Multivariable Control package, which is built Limits can be set on each manipulated variable,
with the end users in mind. It is designed for use by specifying absolute minimum, absolute maximum and
process engineers working in a process plant with maximum move size over one control step.
minimal advanced control theory knowledge. The limits on control valve position are recognized to
Key Characteristic of Exasmoc: avoid the controller winding-up beyond what process
Highest Uptimes in industry (i.e. Highest in can achieve.
industry) Controller can be made to respect valve gains.
Use of unmeasured disturbance models and grey Controller can continue to run in crippled mode after
box models to include a-priori process know how some manipulated variables cease to be available for
resulting in high robustness. control. Gains are adjustable online.
Easy to use design and simulation kit Robust control
Embedding in DCS The controller has an ability to learn by distinguishing
between process noise and process movement. It uses
Easy integration with other plant information
the Kalman lter.
Upgraded Web-based Human Interface. Process optimization
It is possible to minimize/maximize either any single
PCTP (Exasmoc ofine) and Exasmoc online operate variable (manipulated or controlled), or an economic
in Windows XP/VP/2003/2008 environments. function dened as a linear combination of any process
variables. This optimization task has a lower priority
PCTP offline than process control, and is performed under the
Controller condition that more degrees of freedom are available
AIDA Simulation
Parametric Design after all control objective are met.
Model File
Main-controller and Sub-controller concept
Process Controller When Multi-Variable Control is applied to a large
PRBS or Step Responses Definition File process unit, a single large controller would be divided
Signals into several sub controllers to improve controllers
Exasmoc exibility in operation and maintenance.
Kernel The Exasmoc R3 and later enable control engineers to
I/O dene sub-controllers.
A sub-controller will perform observation and control
functions for a sub-set of manipulated and controlled
variables. It will be possible to dene different control
Process period for different sub-controllers. Between 2 sub-
controllers, there will be only intermediate variables
(Intermediate variable links a process variable to
another one.).

Yokogawa Electric Corporation GS 36J06D10-01E

2-9-32, Nakacho, Musashino-shi, Tokyo, 180-8750 Japan Copyright Aug. 2001(YK)
Tel.: 81-422-52-6323 Fax.: 81-422-52-7048 9th Edition Dec. 1, 2010(YK)
<<Contents>> <<Index>> 2
A main-controller will perform coordination between GENERAL CONFIGURATION
the sub-controllers. It will execute the feasibility check
and optimization functions for the overall controller at Features
steady state. The steady state functions will download
manipulated variable horizon limits to the sub- It handles feed-forward
controllers that will enforce consistent and optimized For process variables which can be measured but not
control strategy. adjusted, and which are known to affect the controlled
variables, anticipatory action can be taken upon a
Blending control change in the disturbance variable.
An in-line blender requires a sub-controller with the
following special features. Time delay compensation
1. All manipulated variables shall be ratios. In case of a delayed process (which has no response
2. The only valid disturbance variables are non- to an adjustment until a certain time has passed), a
manipulated blended components. predictive technique is used which recognizes the
3. The sum of the ratios should be automatically kept presence of the delay and avoids over-reaction.
as 1 (One). Speed tuner for manipulated variable adjustment
4. The optimized control is performed at steady state Manipulated variable control moves are always as
only. small as possible, and the relative use of manipulated
The Exasmoc R3 and later are capable of blending variable against each other and against the control
control. objectives can be dened.
Non-linear optimization function Dening weight of each control objective
The Exasmoc R3 and later have a unique optimization If the problem becomes over constrained, not all
function at steady state. It will be possible to dene a the control objectives (setpoints or constraints) can
non-linear (Bilinear form) optimization function. The be achieved. Then the objectives, which are rst
Exasmoc R3 and later will handle classical margin abandoned, or the relative offset on each objective, can
maximization function. An example of non-linear be dened.
function could be:
Online tuning via lters
Maximize Feed * sum of (Yield * Product Price) Tuning lters are used to specify the desired speed
Where: of responding to set-point changes, compensating for
Feed and Yields are process variables and Product disturbances in the controlled variables (or responding
Prices are parameters to inaccuracies in the controllers process model)
The control engineer will have the possibility to dene as well as anticipating the effect of the measured
multiple economic functions by ofine and to select disturbance variables.
the one that is active by online. All economic function Providing Graceful degradation of abnormal
parameters are tunable by online. events
Online model gain update In case of loss of a manipulated variable (saturation,
The online tuning function is enhanced to include actuator failure, etc) optionally control can continue
controllers model gain update. The update will not with the remaining manipulated variables (as best as
cause MVs bump. possible).
It will be possible to update the gains remotely. For In case of the loss of measurement of a controlled
instance, the robust quality estimator can download variable, control of other variables can optionally
new gains and trigger the gain update. continue. Open loop control of the variable with
missing measurement is then performed. Automatic
The online gain update is an essential feature of the checks are provided on the validity of online analyzer
Green gas oil control scheme. signals that the controller may be using.
Specied execution order of application program OPC interface / Integration tool
Special calculations for SMOC input and output Exasmoc can connect to any DCS or PLC via OPC
variables are performed by application programs, which interface to communicate with process input and
are executed in the specied order. output data. Device mode/Cache mode is selectable.
For example: Error processing according to the type of OPC error is
(Step 1) To execute SMOC input calculations prepared.
(Step 2) To execute SMOC control calculations Integration function is also provided with to exchange
data between Exasmoc controller and Exarqe
(Step 3) To execute SMOC output calculations
estimator, Exasmoc controller and any OPC server.
The Exasmoc R3 and later enable control engineers to
specify the execution order of application programs. Calculation function
Using the input data of Integration tool, the calculation
function executes the predened calculation and
outputs the results to the OPC server via Integration
tool. Basic calculation (add/subtract/multiply/divide),
arithmetic calculation (Sine/Cosine/Tangent/Log/
Absolute/...), ltering, etc. are provided. Builder function
and test function are also prepared to make it easy to

All Rights Reserved. Copyright 2001, Yokogawa Electric Corporation GS 36J06D10-01E Sep. 1, 2009-00
<<Contents>> <<Index>> 3
Human interface is fully upgraded from R4, which
consists of Web-based style views. Each view is
available not only in HIS, APC client PC but also Web
PC via Web server.
Main / Sub controller overview
Trend view
Historical message
Integration tool
Scheduler view
Tree view and multi-screen

Integration Tool View

View customization
Overview of each system can be customized. The
view can be used properly according to its usage,
such as start-up of plants or steady-state operation,
maintenance work or engineering work and so on.
Specication of users who can access Overviews.
Specication of SMOC data to be displayed on the
Sub-controller Overview

Prediction Trend View

All Rights Reserved. Copyright 2001, Yokogawa Electric Corporation GS 36J06D10-01E Dec. 1, 2010-00
<<Contents>> <<Index>> 4
AIDA - Advanced Identication and Data Analysis Exasmoc can connect to variety of DCSs and control
is the empirical dynamic process-modeling tool in systems. It reads the process data, performs the control
the PCTP suite. AIDA uses process data from which action and writes the control outputs. All information
it identies linear dynamic models. AIDA makes is possible to access from HMI stations. APC function
available the powerful science and art of process model consists of the following stations:
identication, in a process control relevant fashion. - APC server: Exasmoc/Exarqe/Exaitg control
Features - APC client : APC HMI interface
Data trending Enhanced and exible trending - Web server: Web server for APC HMI
capabilities such as stacked and shared plotting axes, - Web client : Web client function
zooming, auto-scaling, manual scaling, common APC station as well as HIS equalizes engineering data
scaling, scrolling, color customization, easy tag from the ENG engineering station of the CENTUM
rearranging, tag trend locking, easily data value access system.
from trends, segmentation and bad slicing from trends. CENTUM VP System
Correlation Analysis The linear time response APC station is connected to control stations via the
characteristics of the obtained trend data can be VI/VF Control Bus Interface Card, and it reads/writes
identied by carrying out auto-correlation analysis. tag data and receives process messages.
Cross-correlation analysis is useful for identifying linear
dynamic relation between two different data trends over Web Client Web Client

a specied range of sampling lags. This is valuable for

pre-screening input and output trend candidates for Business LAN
inclusion in the model building stage and for testing
the quality of plant test data for tting models. Cross- Firewall
correlation analysis can also be valuable for identifying
PCTP offline APC Client
problem correlation between candidate trends and for Web
detection feedback relationships in data that invalidates Server
its use. Download
Control LAN
Identication AIDA model estimation technology
offers both FIR and parametric identication methods HIS ENGS
APC Exasmoc Server
Server OPC Server
for determining process models. FIR allows for explicit HIS Std View Exarqe Server
estimation of input and output dynamic step responses. APC View Exaopc Server

A priori delay and settling times can be specied for Ethernet Card
Interface Card
each input-output pair. AIDA provides statistical 95 Vnet/IP
percent condence intervals for the estimated step
responses. Other
BCV Compact DCS
Continuous time, low order transfer functions are FCS
tted to the estimated FIR step response curves. Previous Process Data
This approach is an effective approach for response Standard
APC Data

smoothing and obtaining good initial estimates for a FCS
parametric model t to data. The low order models can
directly t to data, and statistical condence intervals CENTUM CS 3000 System
are provided for estimated parameters. AIDA uses Optional

efcient, State-of-the-art numerical computational Remote


solutions that permit solving large identication Ethernet Card


problems in minimal time. WIN

OPC Client
Models estimated by AIDA are robust with regard to OS Exaopc
characteristics often encountered in process data that Server
Interface Card
produce biased results. For example, the presence of
unmeasured disturbances and signicant noise can
invalidate results if not accounted for. To eliminate this Ethernet
problem, AIDA uses data ltering. This is handled in HIS
RQE-Offline/ Optional
SMOC-PC Ethernet
Standard (Builder)
a straightforward manner by providing a way to set a Operation/

drifting ag and a smoothing lter constant. Defaults Control


are provided to make these features transparent.

AIDA yields unbiased models from data generated V net / VL net
under closed loop conditions through the automated
application of optimal ltering. SFCS
Model validation For model validity checking, a
range of different models for prediction comparison Process Data
against actual data are available. It can generate future Exasmoc Data
predictions in an open loop deterministic fashion, or
mimic model prediction during control implementation

with feedback bias updating correction. Residual trends

and goodness-of-t tests are available for auto- and
cross-correlation analysis.

All Rights Reserved. Copyright 2001, Yokogawa Electric Corporation GS 36J06D10-01E Dec. 1, 2010-00
<<Contents>> <<Index>> 5
Optional Max. number of Exasmoc main controller: 20
Remote controllers/station
Monitoring Max. number of Exasmoc sub-controller: 20 controllers
Ethernet Card /main controller
Operator Exarqe
APC Station
OPC Client Control period: Min. 10 seconds.
Displays OS Exaopc Number of MV: 100/subcontroller.
Interface Card
Number of CV: 200/subcontroller.
Monitoring Number of DV: 50/subcontroller.
Control Number of POV: 200/subcontroller.
E net
Number of EF: 100/subcontroller.
ICS EWS Exarqe RQE-offline Optional Number of input compaction point: 100/controller.
BRC/ERC Station (Builder)
Number of output compaction point: 100/controller.
Operation DCS Download The name of Exasmoc controller: Up to 10 characters
Displays Note: Actual number of controllers depends on the
ACG size of the controllers, memory size and CPU
V Net

Process Data (APC server function)
Exasmoc Data
FCS Pentium4 2.8 GHz or better (XP or Server 2003), Core 2 Duo 2.13 GHz or better (Vista or Server 2008)
Main memory:
Other vendor DCS System 2 GByte recommended (XP or Server 2003)
The Exasmoc station is connected to other vendors 3 GByte recommended Vista or Server 2008)
DCS via OPC interface and Ethernet. Disk capacity: 40 Gbyte or more recommended
Connection test between Exasmoc with other vendors Drive: DVD-ROM drive
DCS is indispensable in advance. Monitor: SXGA (1280 x 1024 resolution), True Color
Remote Communication device:
Monitoring Ethernet-ready network card with 100 MBPS
APC Exasmoc Control Bus Interface Card are required when
Operator WIN
OPC Client
connecting to CENTUM VP, CS 3000,CENTUM CS.
Station NT
Displays OS
Other Vendor DCS
Graphics: DirectX 9 corresponding GPU and 128MB
Ethernet Card
graphic memory
(APC client function)
Ethernet Pentium4 2.8 GHz or better (XP or Server 2003),
Exasmoc SMOC-PC Optional
Station Station (Builder) Ethernet Core 2 Duo 2.13 GHz or better (Vista or Server 2008)
Download Main memory:
Control 2 GByte recommended (XP or Server 2003)
3 GByte recommended Vista or Server 2008)
Other Vendor DCS Control Bus Disk capacity: 40 Gbyte or more recommended
Drive: DVD-ROM drive
Monitor: SXGA (1280 x 1024 resolution), True Color
Process Data
Communication device:
Control Stations
Exasmoc Data Ethernet-ready network card with 100 MBPS Graphics: DirectX 9 corresponding GPU and 128MB
graphic memory

All Rights Reserved. Copyright 2001, Yokogawa Electric Corporation GS 36J06D10-01E Dec. 1, 2010-00
<<Contents>> <<Index>> 6
(APC Web server function) Software
CPU: OS: Windows XP Professional (SP2 pr SP3)
Pentium4 2.8 GHz or better (Server 2003), or Windows Vista (SP1 or SP2)
Core 2 Duo 2.13 GHz or better (Server 2008) Windows Server 2003 Std (SP2) or
Main memory: Windows Server 2003 R2 (SP2) or
2 GByte recommended (Serv r 2003) Windows Server 2008 (non-SP or SP2)
3 GByte recommended (Server 2008) 64 bit version is not supported.
Disk capacity: 40 Gbyte or more recommended The Exasmoc package and Windows must use the
Drive: DVD-ROM drive same language.
Monitor: Others:
SXGA (1280 x 1024 resolution), True Colorb Mouse: IE 7.0
Communication device: CENTUM VP: R4.01 or later versions
Ethernet-ready network card with 100 MBPS CENTUM CS 3000: R2.06.00 or later versions
Graphics: DirectX 9 corresponding GPU and CENTUM CS: R2.09.00 or later versions
128MB graphic memory Exarqe R4.02
(APC Web client function) PCTP 2009 Build 1, PCTP 2009 Build 2,
CPU: PCTP 2010 Build 1
Pentium4 2.8 GHz or better (XP or Server 2003), Adobe Reader
Core 2 Duo 2.13 GHz or better (VITSA or Server 2008) Note: Exasmoc/Exarqe R4.01 APC client function is
Main memory: available from VP HIS. In the previous HIS, R3.06
1 GByte recommended (XP or Server 2003) client and Web client are available.
1 GByte recommended Vista or Server 2008) It is necessary to install Adobe Reader 9.1 to read
Instruction Manual of Exasmoc.
Disk capacity: -
Drive: - Interface Package
Monitor: SXGA (1280 x 1024 resolution), True Color Exaopc R2.01 or later is required.
Mouse: Note: Exaopc R3.60 is required when Exaopc is
installed on the same PC as Exasmoc
Communication device:
Ethernet-ready network card with 100 MBPS OPC Interface
Graphics: DirectX 9 corresponding GPU and OPC interface connected to Exasmoc should support
128MB graphic memory the following specication. Connection test in advance
is recommended.
- OPC revision: Compliant to OPC Foundation
- OPC DA 1.0a or later in case of single OPC server
OPC DA 2.05a or later in case of multiple OPC server
- OPC function: Synchronous read/write,
Asynchronous read/write
- OPC performance: 100 data read and write per every

All Rights Reserved. Copyright 2001, Yokogawa Electric Corporation GS 36J06D10-01E Dec. 1, 2010-00
<<Contents>> <<Index>> 7
Description CENTUM, Exaopc, Exasmoc, Exarqe are registered
Exasmoc Multi-variable Model trademarks of Yokogawa Electric Corporation.
Model NTPS410 Ethernet is a registered trademark of XEROX
Predictive Control Package
Software License for Small Units (with Corporation.
-S0 Windows is a registered trademark of Microsoft
Media) (*1)
Basic Software License (with Media)
-S1 Adobe Reader is a trademark or registered trademark
of Adobe Systems Incorporated.
Site License for Small Site (with
Media) (*3) Other products and company names appearing in this
document are trademarks or registered trademarks of
Sufx Site License for Medium Site (with
their respective holders.
Codes Subject to change without notice.
Site License for Large Site (with
0 Without Exasmoc online package (*4)
1 With Exasmoc online package
1 Always 1
1 English version
Software License for Additional Unit
/-ADU (1 to 7 units) (*5)
(: 1 to 7)

*1: The size of controller for the small unit is less than
5 ins and 5 outs.
*2: Basic software license includes 1 copy of SMOC
PC and AIDA.
*3: Site license includes 3 copies of SMOC PC and
Sufx Code -S5: Small Site; Less than 100,000
Sufx Code -S6: Medium Site; 100,000 BPD to
199,999 BPD
Sufx Code -S7: Large Site; 200,000 BPD or
*4: At least one Exasmoc online package has to be
quoted per PC.
Exaopc package is required for Exasmoc to
interface with CENTUM CS 3000.
*5: The number of additional unit(s) has to be entered
in (i.e., enter 1 for the second unit).

All Rights Reserved. Copyright 2001, Yokogawa Electric Corporation GS 36J06D10-01E Dec. 1, 2010-00
Subject to change without notice.