Models Unleashed - Virtual Plant and Model Predictive Control Applications, A Pocket Guide

195 Pages·2004·2.763 MB·English

by  McMillan, Gregory K., Cameron, Robert A.

Models Unleashed:
Virtual Plant and Model
Predictive Control Applications
A pocket guide
by Gregory K. McMillan
and Robert A. Cameron
Copyright © 2004
ISA–The Instrumentation, Systems and Automation Society
67 Alexander Drive
P.O. Box 12277
Research Triangle Park, NC 27709
All rights reserved.
Printed in the United States of America.
10 9 8 7 6 5 4 3 2
ISBN 1-55617-857-3
No part of this work may be reproduced, stored in a retrieval
system, or transmitted in any form or by any means, elec-
tronic, mechanical, photocopying, recording or otherwise,
without the prior written permission of the publisher.
Library of Congress Cataloging-in-Publication Data
– Data is in Progress –
ISA wishes to acknowledge the cooperation of those manufac-
turers, suppliers, and publishers who granted permission to
reproduce material herein. The Society regrets any omission of
credit that may have occurred and will make such corrections
in future editions.
Notice
The information presented in this publication is for the
general education of the reader. Because neither the author
nor the publisher has any control over the use of the informa-
tion by the reader, both the author and the publisher disclaim
any and all liability of any kind arising out of such use. The
reader is expected to exercise sound professional judgment in
using any of the information presented in a particular applica-
tion.
Additionally, neither the author nor the publisher have
investigated or considered the effect of any patents on the abil-
ity of the reader to use any of the information in a particular
application. The reader is responsible for reviewing any possi-
ble patents that may affect any particular use of the informa-
tion presented.
Any references to commercial products in the work are
cited as examples only. Neither the author nor the publisher
endorses any referenced commercial product. Any trademarks
or tradenames referenced belong to the respective owner of
the mark or name. Neither the author nor the publisher makes
any representation regarding the availability of any referenced
commercial product at any time. The manufacturer's instruc-
tions on use of any commercial product must be followed at
all times, even if in conflict with the information in this publi-
cation.
1.0—Simulation
Overview
Process control deals with change. If process
conditions were constant there would be no
need for a control system. In a plant, the operat-
ing conditions continually fluctuate, primarily
because of changes in raw materials, production
rate, product mix, equipment performance, foul-
ing, catalyst, utilities, and ambient conditions.
Except for production rate changes, most of
these disturbances, as well as their effect on key
process variables, are not measured on line.
Control systems do not eliminate this variabil-
ity, but they can transfer it from a controlled
variable to a less important manipulated vari-
able. Control systems can also move the process
to a more optimum operating point.
Unfortunately, control loops that suffer from
stick-slip, incorrect tuning, and interaction may
actually increase the variability. The feedback
and feedforward settings of most PID controllers
are tuned by a trial-and-error method and thus
reflect more the personal preferences of the
SIMULATION 1
tuner than any knowledge of the process dynam-
ics and objectives. Plants rarely try to decouple
PID controllers, but when they do it is primarily
based on just some estimated steady state gains.
While model predictive control (MPC) cannot
eliminate the stick-slip problem, it can effec-
tively address the issues of tuning and interac-
tion while it simultaneously handles process
limits and optimizes process objectives. An
MPC system’s control actions are derived from
an experimental model that is obtained by rigor-
ous plant testing, which takes into account the
interactions and the response of constraints. If
this model is accurate, MPC requires less tuning
than PID control. Furthermore, the main MPC
tuning parameter is designed to set the amount
of variability that is transferred from the con-
trolled variables to the manipulated variables.
The embedding of this dynamic model enables
MPC to simply trade off between performance
and robustness, maintain an allowable degree of
variability in both the controlled and manipu-
lated variables, decouple interrelated loops, opti-
mize set points, and honor constraints.
2 SIMULATION
MPC uses an experimental dynamic model that
is obtained by making steps in the manipulated
and disturbance variables, identifying either
matrix coefficients directly or the parameters,
such as process gain, time delay, and time lag, so
as to predict a trajectory from previous changes
in the manipulated and disturbance variables.
The models are linear, and the effects are com-
bined by linear superposition. Thus, the knowl-
edge of the future that MPC provides excludes
the effect of nonlinearities and unmeasured
upsets. These unknowns are addressed in the
present by biasing the trajectory by a fraction of
the difference between the predicted and actual
value. The old adage that you can only control
what you know still applies.
A first-principle dynamic model can quantify
nonlinearities and unmeasured process condi-
tions. Until recently, these models required that
hundreds of differential equations be set up and
numerically integrated. However, process flow
diagram (PFD) models that are used for process
design can now be made dynamic by using tech-
nically advanced simulation software. This
guidebook will explore how to develop and
SIMULATION 3
apply dynamic PFD models to improve the
capability of MPC. Consider what opportunities
would open up if the conditions, properties, and
compositions of each stream in the PFD that was
used to design the plant were updated dynami-
cally and displayed. The previously unknown
upsets could become disturbance variables, and
users could add compositions and yields that are
real indicators of product quality and process
performance as controlled variables to the MPC.
The dynamic PFD model could be driven to
explore nonlinearities and new operating regions
as well as step-tested to identify the parameters
for the MPC experimental models. At a mini-
mum, the insight and knowledge users gain from
exploring the dynamics and pathways of vari-
ability would improve the design and justifica-
tion of MPC systems.
Historically, first-principle dynamic models have
been severely limited to a few unit operations
and mostly static estimates of a few physical
properties, such as specific heat, density, and
boiling point. These models were programmed
by setting up, sorting, and numerically
integrating the differential equations for
4 SIMULATION
accumulating energy and material within a
volume. These models typically could only be
run and maintained by the programmer, who
was one of an elite handful of specialists in the
process industry.
Just before the turn of the new century, graphi-
cally configured dynamic PFD models with
extensive physical property packages appeared
that greatly expanded the scope of the potential
applications and the models’ users. Some “state-
of-the-art” software offers users the ability to
switch a steady state PFD model into the
dynamic mode.
Steady state models with comprehensive physi-
cal-property data packages have been used to
design processes since the 1960s. As shown in
figure 1-1, these models all assume that the accu-
mulation, generation, and consumption of mass
and energy are zero, so the outputs from the vol-
ume can be calculated from the inputs by an
iterative procedure. The boundary could be the
whole or subdivided volume of a piece of pro-
cess equipment, such as a heat-exchanger pass or
column tray or a section of piping. These are
classified as “lumped parameter models” since
SIMULATION 5
they do not involve partial differential equa-
tions. Since the accumulation, generation, and
consumption are all zero, steady state models
cannot be used to simulate batch operations,
startups, shutdowns, transitions, reaction kinet-
ics, crystal growth or attrition, and cell birth,
growth, or death. With respect to process
dynamics, successive runs could be made to
show the change in the process variable within
the volume for a specific change in a disturbance
or manipulated variable. In fact, a steady state
model excels at this capability since it has the
quality and complexity of detail that’s needed to
reveal process relationships and interactions.
However, some steady state models require
hours or days to converge on a solution. In fact,
for large changes, they may never converge.
These realities can render them an ineffective
tool for identifying gains and exploring new
operating regions. Dynamic models can move to
a drastically different set of operating conditions
that in a steady state model would have caused
severe, if not fatal, convergence problems. Also,
dynamic models are needed to show process
time delays and time lags. However, until
6 SIMULATION
S Figure 1-1 — Lumped Parameter Model
IM
U
L
A
T
ION recycle
R
Subsystem Boundary
Inputs Accumulation, Generation, Outputs
and Consumption of
Material and Energy
In steady state models, the accumulation, generation, and consumption are zero. Valve size, pressure
drop, and position have no effect on flow. The outputs are calculated from inputs. For recycle streams,
the program iterates until the output and the input of the recycle block are within a tolerance spec.
7