Feedback Control of Hyperbolic PDE Systems

Panagiotis D. Christofides and Prodromos Daoutidis

Dept. of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, MN 55455

This article deals with distributed parameter systems described by first-order hyper-
bolic partial differential equations (PDEs), for which the manipulated input, the con-
trolled output, and the measured output are distributed in space. For these systems, a
general output-feedback control methodology is developed employing a combination of
theory of PDEs and concepts ffom geometric control. A concept of characteristic index
is introduced and used for the synthesis of distributed state-feedback laws that guarantee
output tracking in the closed-loop system. Analytical formulas of distributed output-
feedback controllers are derived through combination of appropriate distributed state
observers with the developed state-feedback controllers. Theoretical analogies between
our approach and available results on stabilization of linear hyperbolic PDEs are also
identified. The developed control methodology is implemented on a nonisothermal
plug-flow reactor and its performance is evaluated through simulations.

Introduction
Chemical engineering processes are inherently nonlinear approach limits the controller performance, and may Iead to
and very frequently involve state variables that change in both unacceptable control quality.
time and space. Representative examples of processes with Motivated by the preceding considerations, significant re-
significant spatial variations include plug-flow reactors (Ray, search efforts have focused on the development of control
19811, countercurrent absorbers-reactors (Rhee et al., 1986), methods for PDE systems that directly account for their spa-
fixed- and fluidized-bed reactors (Stangeland and Foss, 1970; tially distributed nature. Excellent surveys of theoretical as
Ray, 1981; Georgakis et al., 1977). The mathematical models well as application articles on this topic can be found in Balas
of these processes are typically derived from the dynamic (19821, Keulen (19931, and Ray (1978). Initially, systems of
conservation equations and consist of nonlinear partial dif- linear PDEs were considered, for which key system- and con-
ferential equations (PDEs). trol-theoretic properties (e.g., existence and uniqueness of so-
The conventional approach for the control of PDE systems lutions, stability, controllability, and observability) were well-
is based on the spatial discretization of the PDE model fol- understood (Curtain and Pritchard, 1978). The well-known
lowed by the controller design on the basis of the resulting classification of PDE systems to hyperbolic, parabolic, and
(linear or nonlinear) ordinary differential equation (ODE) elliptic (Smoller, 19831, according to the properties of the
model (Sorensen et al., 1980; Dochain et al., 1992; Patward- spatial differential operator, essentially determined the ap-
han et al., 1992). However, there are certain well-known dis- proach followed for the solution of the control problem. Thus,
advantages associated with this approach. For example, fun- for parabolic PDE systems (e.g., diffusion-reaction processes),
damental control-theoretic properties, like controllability and the fact that the system dynamics is practically determined by
observability, which should depend only on the location of a finite number of modes, motivated the use of modal de-
sensors and actuators, may also depend on the discretization composition techniques to derive ODE models that capture
method and the number and location of discretization points the dominant dynamics of the system (Curtain, 1982; Geor-
(Ray, 1981). Moreover, neglecting the infinite-dimensional gakis et al., 1977; Hanczyc and Palazoglu, 1992; Gay and Ray,
nature of the original system may lead to erroneous conclu- 1995); the controller design problem was then addressed us-
sions concerning the stability properties of the open-loop ing methods for linear ODE systems. On the other hand, for
and/or the closed-loop system. Furthermore, in processes hyperbolic PDE systems (e.g., convection-reaction processes),
where the spatially distributed nature is very strong, due to for which a modal decomposition is not possible, alternative
the underlying convection and diffusion phenomena, such an approaches were followed, using mainly optimal control
methods (Wang, 1966; Lo,1973; Balas, 1986) on the basis of
P. D. Christofides is presently at the Dept. of Chemical Engineering, University of
California, Los Angeles, CA 90024. the original infinite-dimensional system, or ODE control

AIChE Journal November 1996 Vol. 42, No. 11 3063

methods on the basis of equivalent ODE realizations ob- spatial dimension with the following state-space representa-
tained by the method of characteristics (Ray, 1981). tion:
Recently, considerable attention has focused on the under-
standing of system-theoretic properties and the dynamical dX dX
behavior of nonlinear PDEs, by treating them as evolution
-= A ( x ) -
dt dZ
+f ( x ) +g(x)u
equations in appropriate infinite dimensional spaces (Temam,
1988; Brown et al., 1991). Yet, available results on the con-
trol of systems of nonlinear PDEs are rather sparse, with the
exception of optimal control approaches [(Balas, 1991; Burns subject to the boundary condition:
and Kang, 1991; Kang and Ito, 1992) and the review paper
(Lasiecka, 1995) for an overview of recent results in systems C,x(a,t)+C,x(b,t)=R(t) (2)
literature]. Due to their practical relevance and importance,
quasi-linear PDE systems have attracted particular research and the initial condition:
interest. For quasi-linear parabolic PDE systems, an ap-
proach that utilizes a combination of eigenfunction expan-
sion techniques and nonlinear control schemes was proposed
in Chen and Chang (1992). For processes modeled by a single where x ( z , t ) = [ x , ( z , t ) ... x , ( z , t)lT denotes the vector of
first-order quasi-linear hyperbolic PDE, an approach based state variables, x ( z , t ) E X " [ ( a ,b), R n ] , with X" being the
on a combination of the method of characteristics and sliding infinite-dimensional Hilbert space of n-dimensional vector
mode techniques was proposed in Sira-Ramirez (1989). This functions defined on the interval [ a , b ] whose spatial deriva-
method was further developed in Hanczyc and Palazoglu tives up to nth order are square integrable, z E [ a ,b ] c R and
(1995) to account for possible discontinuous behavior of the t E [O,m), denote position and time, respectively, u(z, t ) de-
control action and was applied to a heat exchanger. A Lya- notes the manipulated variable, y ( z , t ) denotes the controlled
punov-based approach, inspired from concepts of thermody- variable, and q ( z , t ) denotes the measured variable. A ( x ) is a
namics, to derive conditions that guarantee stability under sufficiently smooth matrix, f ( x ) and g ( x ) are sufficiently
linear control has also been proposed in Alonso and Ydstie smooth vector functions, h ( x ) , p ( x ) are sufficiently smooth
(1995). scalar functions, R ( t ) is a column vector that is assumed to
In this article, we address the feedback control problem be a sufficiently smooth function of time, x , ( z ) E X [ ( a ,b),
for systems described by quasi-linear first-order hyperbolic
R n ] , with [ ( a , b ) , R n ] being the Hilbert space of n-dimen-
PDEs, for which the manipulated input, the controlled out-
sional vector functions defined on the interval [ a , b ] which
put, and the measured output are distributed in space. Sys-
are square integrable, and C,, C, are constant matrices of
tems of this form arise naturally in chemical engineering as
dimension n x n.
models of transport-reaction processes, whenever convective The model of Eq. 1 describes the majority of convection-
mechanisms dominate diffusive and dispersive ones [see the
reaction processes arising in chemical engineering (Rhee et
book by Rhee et al. (1986) for representative examples]. For al., 1986) and constitutes a natural generalization of linear
such systems, our objective is to synthesize nonlinear dis- PDE models (see Eq. 9 below), considered in Pel1 and Aris
tributed output-feedback controllers that enforce output
(1970) and Ray (1981) in the context of linear distributed state
tracking and guarantee stability in the closed-loop system.
estimation and control. The distributed and affine appear-
The article is structured as follows: after reviewing the nec-
ance of the manipulated variable u is typical in most practi-
essary preliminaries, a concept of characteristic index be-
cal applications (Pel1 and Ark, 1970; Gay and Ray, 1995; Ray,
tween the controlled output and the manipulated input (which
19811, where the jacket temperature is usually selected as the
can be thought of as the analogue of relative order) is intro-
manipulated variable (see below for a detailed discussion on
duced and used for the synthesis of distributed state-feed- how the jacket temperature is manipulated in practice). Fur-
back controllers that induce output tracking in the closed-loop thermore, the possibility that the system of Eq. l may admit
system. A notion of zero-output constraint dynamics for boundary conditions at two separate points (e.g., countercur-
first-order hyperbolic PDEs is introduced and used to derive rent processes) is captured by the boundary condition of Eq.
precise conditions that guarantee the stability of the closed-
2.
loop system. Then, output-feedback controllers are synthe- Depending on the eigenvalues of the matrix A ( x ) , the sys-
sized through a combination of appropriate distributed state tem of Eq. 1 can be hyperbolic, parabolic, or elliptic (Smoller,
observers with the developed state-feedback controllers. The-
1983). Assumption 1, which follows, ensures that the system
oretical analogies between the proposed approach and avail- of Eq. 1 has a well-defined solution and specifies the class of
able feedback control methods for the stabilization of linear
systems considered in this manuscript.
hyperbolic PDEs are pointed out. Controller implementation
Assumption 1. The matrix A ( x ) is real symmetric and its
issues are also discussed. Finally, the application of the de- eigenvalues satisfy:
veloped control method is illustrated through a nonisother-
ma1 plug-flow reactor example modeled by a system of three
quasi-linear hyperbolic PDEs.

First-Order Hyperbolic PDE Systems for all x E X " [ ( a ,b), Rn].

Typical examples where Assumption 1 is satisfied include
Preliminaries heat exchangers, plug-flow reactors, and countercurrent ab-
We consider systems of quasi-linear first-order PDEs in one sorbers-reactors where the matrix A ( x ) is constant and diag-

3064 November 1996 Vol. 42, No. 11 AIChE Journal

onal and its elements are the fluid velocities, as well as chro- depends on the desired performance specifications and in the
matography of two interacting solutes where A ( x ) is a full majority of practical applications [see, e.g., Ray (1981),
matrix (Rhee et al., 1986). Systems of the form of Eq. 1 for Hanczyc and Palazoglu (1992)], is of the following form:
which the eigenvalues of the matrix A ( x ) are real and dis-
tinct are said to be hyperbolic, while systems for which some
of the eigenvalues of the matrix A ( x ) are identically equal
are said to be weakly hyperbolic (Smoller, 1983).
where c'(z) is a known smooth function of z . For simplicity,
Specification of the control problem the functions b'(z), c'(z), i = 1, ..., 1 will be assumed to be
Consider the system of quasi-linear PDEs of the form of normalized in the interval [ a ,b],that is,
Eq. 1, for which the manipulated variable u ( z , t ) , the mea-
sured variable q(z, t ) , and the controlled variable y ( z , t ) are
distributed in space. Let's assume that for the control of the
variable y , there exists a finite number of control actuators, 1,
and the same number of measurement sensors; clearly, it is
not possible to control the variable y ( z , t ) at all positions. Using the relations of Eqs. 5 and 6, the system of Eq. 1 takes
Therefore, it is meaningful to formulate the control problem the form
as the one of controlling y ( z , t ) at a finite number of spatial
intervals. In particular, referring to the single spatial interval
[ z;, zi+ we suppose that the manipulated input is iZ'(t), with
ii' E R,and the measured output is tj'(t), with q iE R,while
the controlled output is ?'(t), with 7' E R, such that the fol-
lowing relations hold:

where

where b'(z) is a known smooth function of z, and ei,Q are

bounded linear operators, mapping X" into R. Figure 1
shows a pictorial representation of this formulation in the
case of a prototype example. From a practical point of view,
the function b'(z) describes how the control action ii'(t) is
distributed in the spatial interval [zi,zi+ while the opera-
tor &' determines the structure of the sensor in the same
spatial interval. Whenever the control action enters the sys-
tem at a single point zo (e.g., lateral flow injections), with
zo E [ zi,zi+ (i.e., point actuation), the function b'(z) is taken
to be nonzero in a finite spacial interval of the form [ z o -
6, zo + €1, where E is a small positive real number, and zero
elsewhere in [ zi,zi+,I. Similarly, in the case of a point sensor
acting at zo, the operator (2' is assumed to act in [ z,, - E , zo
+ €1, and considered to be zero elsewhere. The operator e' ->
with H( being the standard Heaviside function.
Referring to the system of Eq. 7, we note that by setting
A ( x ) = A , f ( x ) = Bx, g ( x ) = w , h ( x ) = la;, p ( x ) = px, where
A and B are matrices, and w ,k , p are vectors of appropriate
dimensions, it reduces to the following system of linear first-
order hyperbolic PDEs:

dx dx
-= A - + BX + w ~ ( z ) E
at az
I
I
I
I
I
I
I
I

I
I
I

I 0- I subject to the boundary condition of Eq. 2, and the initial

condition of Eq. 3.
The following example will be used throughout the article
to illustrate the various aspects of our methodology.
Example. Consider a steam-jacketed tubular heat ex-
Figure 1. Control problem specification in the case of a changer (Ray, 1981). The dynamic model of the process is of
prototype example. the form:

AIChE Journal November 1996 Vol. 42, No. 11 3065

 dT dT defined on the domain in X [ ( a ,b);R"] consisting of func-
_=-
dt
ul- - aT
dZ
+ a? tions x E Xl[(d,b);R"] that satisfy the boundary condition
of Eq. 2, generates a strongly continuous semigroup U ( t ) of
bounded linear operators on X[(a, b);R"]. This fact implies
the existence, uniqueness, and continuity of solutions for the
where T ( z , t ) denotes the temperature of the reactor, z E system of Eq. 9. In particular, the generalized solution of this
[0,1], q ( z , t ) denotes the jacket temperature, ul denotes the system is given by
fluid velocity in the exchanger, and a is a positive constant.
Considering T,. as the manipulated variable, and T as the
controlled and measured variable, the preceding model can
be put in the form of Eq. 1:

where C ( t ) is a bounded linear operator for each t mapping

X[(O, t ) ;R"] into X [ ( a ,b);R"]. The notion of semigroup can
be thought of as an analog of the notion of the state transi-
tion matrix used for linear finite-dimensional systems. As can
be easily seen from Eq. 15, U ( t ) evolves the initial condition
xo forward in time. From general semigroup theory (Fried-
man, 19761, it is known that U ( t )satisfies the following growth
Consider the case where these exists one actuator with distri- property:
bution function b ( z )= 1, the controlled output is assumed to
be j ( t ) = 10'x ( z , t ) d z , and there is a point sensor located at
z = 0.5. Utilizing these relations, the system of Eq. 11 takes
the form
where K 2 1, a is the largest real part of the eigenvalues of
the operator C ,and an estimate of K , a can be obtained
-d X-- - uI-dX - ax + aii(t) utilizing the Hiller-Yoshida theorem (Friedman, 1976).
dt dZ Whenever the parameter a is strictly negative, we will say
that the operator of Eq. 14 generates an exponentially stable
semigroup U(t).We note that although there exist many sta-
bility concepts for PDEs [e.g., weak (asymptotic) stability
(Friedman, 1976; Smoller, 198311, we will focus throughout
the article on exponential stability, because of its robustness
to bounded perturbations, which is required in most practical
Review of system-theoretic properties applications, where there is always some uncertainty associ-
The objective of this subsection is to review basic system- ated with the process model. The aforementioned concepts
theoretic properties of systems of first-order hyperbolic PDEs allow stating precisely a standard (Pel1 and Ark, 1970; Balas,
that will be used in the subsequent sections. For more details 1986) detectability requirement for the system of Eq. 9, which
on these subjects, the reader may refer to Curtain and will be exploited in the sixth section for the design of dis-
Pritchard (1978) and Russell (1978). We will start with the tributed state observers.
definitions of the inner product and the norm, with respect Assumption 2. The pair [ Qkp C 1, is detectable, that is,
to which the notion of exponential stability for the systems there exists a bounded linear operator 6,mapping R' into
under consideration will be defined. X", such that the linear operator C, = 6: - 6 Q k , gener-
Let w l , w z be two elements of X([a,b];R"). Then the ates an exponentially stable semigroup.
inner product and the norm, in X ( [ a ,b];R"), are defined as This detectability assumption does not impose any restric-
follows: tions on the form of the operator Q and thus, on the struc-
ture of the sensors (e.g., distributed, point sensors).
In closing this subsection, motivated by the lack of general
stability results for systems of quasi-linear PDEs, we review a
result that allows characterizing the local stability properties
of the quasi-linear system of Eq. 7 on the basis of its corre-
sponding linearized system. To this end, let's consider the lin-
-
where the notation ( , * Iwn denotes the standard inner prod-
earization of the quasi-linear system of Eq. 7:
uct in the Euclidean space R".
Referring to the linear system of Eq. 9, for which Assump-
tion 1 holds, it is well-established (Russell, 1978) that the op-
erator
dX
d:x = A -+ Bx, (14)
dt where

3066 November 1996 Vol. 42, No. 11 AIChE Journal

 A ( x )= A ( x , ( z ) ) , B ( 2 )= ( T ),
af(x>
x = XJZ)
will also present the development for the case of systems of
linear PDEs of the form of Eq. 9. The development for the
case of quasi-linear systems will be performed by essentially
generalizing the results developed for the linear case in a
nonlinear context.

Characteristic Index
In this section, we will introduce the concept of character-
istic index between the output and the input ii for systems
and x,(z) denote some steady-state profile. of the form of Eq. 9, which will allow us to formulate and
Proposition (Smoller, 1983, p' 12'). The system Of Eq'
solve the state-feedback control problem. To reveal the ori-
(with Z = o ) , for which Assumption subject to the gin and illustrate the role of this concept, we consider the
boundary condition of Eq. 2, is locally exponentially if
operation of differentiation of the output yi of the system of
the operator of the linearized system of Eq. 17: Eq. 9 with respect to time, which yields

(18)

generates an exponentially stable semigroup.

The following remark provides conditions, which can be
easily verified in practice, that guarantee the open-loop sta-
bility of hyperbolic PDE systems of the form of Eq. 9 (Eq. 7).
Remark I . Consider the system of linear (quasi-linear)
first-order PDEs of the form of Eq. 9 (Eq. 7) with ii = 0, and
= e'k A -
( d : 1
+ B x + e'kwb'(z)ii'. (20)

assume that the following conditions hold: Now, if the scalar e'kwb'(z) is nonzero, we will say that the
A , ( X ) 5 A2(x) 5 ... 5 A,(X) < 0 characteristic index of 7' with respect to ti', denoted by a',is
equal to one. If e'kwb'(z) = 0, the characteristic index is
for all x E X " [ ( a , b ) , RE]. greater than one, and from Eq. 20 we have
c2=0.
In this case, it can be shown (Russell, 1978) that the eigenval- dy' d
ues of the operator of Eq. 18 (Eq. 14) are of the following
form:
dt
-= (
e'k A - + B
dz ) x. (21)

...,W. Performing one more time-differentiation, we obtain:

a, = -m+ p ~ i , p = -00, (19)

Thus, first-order PDE systems that satisfy the preceding con-

ditions (physical examples include plug-flow reactors and dt
cocurrent heat exchangers) possess eigenvalues that lie on a
vertical line crossing the real axis at s = -a, which, accord- (22)
ing to Eq. 16, implies that they are exponentially stable.
In analogy with the preceding, if the scalar e'k[A(d/dz)+
Methodologicalframework B]wb'(z) is nonzero, the characteristic index is equal to two,
while if e 'k[A ( d / d z ) + B]wb'(z) = 0,the characteristic index
Motivated by the fact that control methods for quasi-linear is greater than two.
distributed parameter systems should explicitly account for Generalizing the preceding development, one can give the
their nonlinear and spatially varying nature, our methodology definition of characteristic index for systems of the form of
entails the following two steps: Eq. 9.
1. Synthesize distributed nonlinear state-feedback con- Definition 1. Referring to the system of linear first-order
trollers that enforce output tracking and derive conditions PDEs of the form of Eq. 9, we define the characteristic index
that guarantee the exponential stability of the closed-loop of the output J' with respect to the input as the smallest
system. integer u i for which
2. Synthesize distributed nonlinear output-feedback con-
trollers through a combination of the developed state-feed-
back controllers with appropriate distributed state observers.
(
e'k A -
d:
+ B )u8-'wb'(z) # 0 , (23)
Motivated by the mathematical properties of these sys-
tems, the state-feedback control problem is solved on the ba-
sis of the original PDE model, by following an approach con- or u i= ~0 if such an integer does not exist.
ceptually similar to the one used for the synthesis of inver- Remark 2. According to Definition 1, the characteristic
sion-based controllers for ODE systems. In order to motivate index is the smallest order time-derivative of the output yi
the approach followed for the quasi-linear case and identify that explicitly depends on the .manipulated input 5'. In this
theoretical analogies between our approach and available re- sense, it can be thought of as a natural generalization of the
sults on feedback stabilization of linear hyperbolic PDEs, we concept of relative order for the systems under consideration.

AIChE Journal November 1996 Vol. 42, No. 11 3067

For the case of linear ODE systems, the relative order can be where S is a linear operator mapping X" into R', s is an
interpreted as the difference in the degree of the denomina- invertible diagonal matrix of functionals, and u E R' is the
tor polynomial and numerator polynomial. Such an interpre- vector of reference inputs. The structure of the control law of
tation cannot be given for the concept of characteristic index Eq. 25 is motivated by available results on stabilization of
because the frequency-domain representation of the system linear PDEs systems via distributed state feedback (e.g.,
of Eq. 9 typically gives rise to transfer functions that involve Wang, 1966; Balas, 1986) and the requirement of output
complicated transcendental forms. tracking. Substituting the distributed state-feedback law of
In analogy with the linear case, the following concept of Eq. 25 into the system of Eq. 9, the following closed-loop
characteristic index is introduced for the quasi-linear PDE system is obtained:
system of Eq. 7.
Definition 2. Referring to the system of quasi-linear first-
order PDEs of the form of Eq. 7, we define the characteristic
index of the output f ' with respect to the input U' as the
smallest integer u i for which
J = Cloc. (26)

It is clear that feedback laws of the form of Eq. 25 preserve

the linearity with respect to the reference input vector u. We
also note that the evolution of the linear PDE system of Eq.
where a, denotes the jth column vector of the matrix A(x), 26 is governed by a strongly continuous semigroup of bounded
and L o ] ,Lf denote the standard Lie derivative notation, or linear operators, because d: generates a strongly continuous
U ' = M if such an integer does not exist. semigroup and b ( z )S x , b(z)sv are bounded, finite-dimen-
Throughout the article, it will be assumed that Eq. 24 holds sional perturbations (Friedman, 1976), ensuring that the
for all x E X", z E [ a ,bl. closed-loop system has a well-defined solution (see the sub-
From Definitions 1 and 2, one can immediately see that section titled "Review of System-Theoretic Properties").
the characteristic index u' depends on the structural proper- Proposition 2 that follows allows specifying the order of the
ties of the process (the matrices A , B and the vectors w, k input/output response in the closed-loop system (the proof is
for the linear case, or the matrix A(x) and the functions f(x), given in the Appendix).
g(x), h(x) for the quasi-linear case), as well as on the selec- Proposition 2. Consider the system of linear first-order
tion of the control system and objectives (the functions b'(z) PDEs of Eq. 9 subject to the boundary condition of Eq. 2, for
and the output operators (3'). Note that in the control prob- which Assumptions 1 and 3 hold. Then, a distributed state-
lem specification of subsection titled "Specification of the feedback control law of the form of Eq. 25 preserves the
Control Problem," we have implicitly assumed that C' and characteristic index u, in the sense that the characteristic
b'(z) are chosen to act in the same spatial interval (col- index of J with respect to u in the closed-loop system of Eq.
located); in the case where C' and b' ( z ) are chosen to act in 26 is equal to u.
different spatial intervals (noncollocated), it follows directly The fact that the characteristic index between the output J
from Eqs. 23 and 24 that the characteristic index u' =m, and the reference input u is equal to u suggests requesting
which implies that this selection leads to loss of controllabil- the following input/output response for the closed-loop sys-
ity of the output J' from the input ii'. tem:
In most practical applications, the selection of (b'(z),el) is
typically consistent for all pairs (f', U'), in a sense that is made
precise in the following assumption. d'J
Assumption 3. Referring to the system of first-order PDEs
y,dt" + ... + y1-
dt
+j j = 0, (27)
of the form of Eq. 9 (Eq. 7), u 1= g 2 = ... = u i= u.
Given this assumption, u can be also thought of as the where y l , y2, ..., yv are adjustable parameters. These pa-
characteristic index between the output vector j j and the in- rameters can be chosen to guarantee input/output stability
put vector li. and enforce desired performance specifications in the
closed-loop system. Referring to Eq. 27, note that, motivated
by physical arguments, we request, for each pair (jji,ui), i = 1,
..., I, an input/output response of order u with the same
State-Feedback Control transient characteristics [i.e., the parameters yk are chosen
to be the same for each pair ( y i , u i ) ] . This requirement can
Linear systems
be readily relaxed if necessary to impose responses with dif-
In this subsection, we focus on systems of linear first-order ferent transient characteristics for the various pairs ( J ' , u').
PDEs of the form of Eq. 9 and address the problem of syn- We are now in a position to state the main result of this
thesizing a distributed state-feedback controller that forces subsection in the form of a theorem (the proof can be found
the output of the closed-loop system to track a reference in- in the Appendix).
put in a prespecified manner. More specifically, we consider Theorem 1. Consider the system of linear first-order PDEs
distributed state-feedback laws of the form: of Eq. 9 subject to the boundary condition of Eq. 2, for which
Assumptions 1 and 3 hold. Then, the distributed state-feed-
u= S x f s u , (25) back law:

3068 November 1996 Vol. 42, No. 11 AIChE Journal

 where g(x) is a nonlinear operator mapping X" into R',
S ( x ) is an invertible diagonal matrix of functionals, and u E R'
is the vector of reference inputs. The class of control laws of
Eq. 32 is a natural generalization of the class of control laws
considered for the case of linear systems (Eq. 25). Under the
enforces the input/output response of Eq. 27 in the closed- control law of Eq. 32, the closed-loop system takes the form:
loop system.
Remark 3. Referring to the controller of Eq. 28, it is clear
that the calculation of the control action requires algebraic
manipulations as well as differentiations and integrations in
space, which is expected because of the distributed nature of
the controller.
Remark 4. The distributed state-feedback controller of It is straightforward to show that the preceding system has
Eq. 28 was derived following an approach conceptually simi- locally a well-defined solution, and the counterpart of Propo-
lar to the one employed for the synthesis of inversion-based sition 2 also holds, that is, the characteristic index of the out-
controllers for ODE systems. We note that this is possible, put J with respect to u in the closed-loop system of Eq. 33 is
because, for the system of Eq. 9, (a) the solution is well- equal to u ,which suggests seeking a linear input/output re-
defined (i.e., the evolution of the state is locally governed by sponse of the form of Eq. 27 in the closed-loop system. Theo-
a strongly continuous semigroup of bounded linear opera- rem 2 that follows states the controller synthesis result for
tors); (b) the input/output spaces are finite dimensional; and this case.
(c) the manipulated input and the controlled output are dis- Theorem 2. Consider the system of quasi-linear first-order
tributed in space. These three requirements are standard in PDEs of Eq. 7 subject to the boundary condition of Eq. 2, for
most control theories for PDE systems (e.g., Balas, 1986, 1991) which Assumptions 1 and 3 hold. Then, the distributed
and only the third one poses some practical limitations ex- state-feedback law:
cluding processes where the manipulated input appears in the
boundary.
Remark 5. The class of distributed state-feedback laws of
Eq. 25 is a generalization of control laws of the form
-
u = sx, (29)

where 5 is a bounded linear operator mapping X" into R',

which are used for the stabilization of linear PDEs. The usual
approach followed for the design of the gain operator 5 uti-
lizes optimal control methods (e.g., Lo, 1973; Ray, 1981). enforces the input/output response of Eq. 27 in the closed-
Example (Continued). In the case of the heat exchanger loop system.
example introduced earlier, it can be easily verified that the Remark 6. Theorem 2 provides an analytical formula of a
characteristic index of the system of Eq. 12 is equal to one. distributed nonlinear state-feedback controller that enforces
Therefore, a first-order input/output response is requested a linear input/output response in the closed-loop system. In
in the closed-loop system: this sense, the controller of Eq. 34 can be viewed as a coun-
terpart of input/output linearizing control laws for nonlinear
ODE systems (see Kravaris and Arkun, 1991, and the refer-
(30) ences therein), in the case of infinite-dimensional systems of
the form of Eq. 7.
Using the result of Theorem 1, the appropriate control law
that enforces this response is Closed-Loop Stability
The goal of this section is to define a concept of zero
dynamics and the associated notion of minimum phaseness
for systems of first-order hyperbolic PDEs of the form of Eq.
9 (Eq. 7), subject to the boundary conditions of Eq. 2; this
will allow us to state conditions that guarantee exponential
stability of the closed-loop system. We will initially define the
concept of zero dynamics for the case of linear systems (the
definition for the case of quasi-linear systems is completely
Quasi-linear systems similar and will be omitted for brevity). Our definition is
In this subsection, we consider systems of quasi-linear analogous with the one given in Byrnes et al. (1994) [see also
first-order PDEs of the form of Eq. 7 and control laws of the Pohjolainen (198l)l for the case of linear parabolic PDE sys-
form: tems with boundary feedback control.

AIChE Journal November 1996 Vol. 42, No. 11 3069

 Definition 3. The zero dynamics associated with the sys- poles at prespecified locations, while also guaranteeing the
tem of linear first-order PDEs of Eq. 9 is the system obtained exponentially stability of the closed-loop system, if the pair
by constraining the output to zero, that is, the system: [& wb( z)] is stabilizable (i.e., the remaining infinite uncon-
trolled poles are in the open left-half of the complex plane).
Remark 9. From the result of Proposition 3 and the dis-
ax - ax
_ d
dt
-A-
az
+ BX - w b ( . ~ ) cussion of Remark 8, it is clear that the derivation of expo-
nential-stability results for the closed-loop system under the
control law of Eq. 25 (or the control law of Eq. 29) requires
X
{ id:
C?k A - + B
1")
x (35)
that the open-loop system be minimum phase (or stabiliz-
able). The a priori verification of these properties can in
principle be performed by utilizing spectral theory for opera-
tors in infinite-dimensions (Friedman, 1976; Pohjolainen,
1981). However, these calculations are difficult to perform in
the majority of practical applications. In practice, the stabiliz-
From Definition 3, it is clear that the dynamical system that ability and minimum-phase properties can be checked
describes the zero dynamics is an infinite-dimensional one. through simulations.
The concept of zero dynamics allows us to define a notion of In closing this section, we address the issue of closed-loop
minimum-phaseness for systems of the form of Eq. 9. More stability for systems of quasi-linear PDEs. Proposition 4 that
specifically, if the zero dynamics is exponentially stable, the follows provides the counterpart of the result of Proposition
system of Eq. 9 is said to be minimum phase, while if the 3 for the case of quasi-linear systems.
zero dynamics is unstable, the system of Eq. 9 is said to be Proposition 4. Consider the system of quasi-linear first-
nonminimum phase. order PDFs of Eq. 7 for which Assumptions 1 and 3 hold,
We have now introduced the necessary elements that will under the controller of Eq. 34. Then, the closed-loop system
allow us to address the issue of closed-loop stability. Proposi- is locally exponentially stable (i.e., the differential operator
tion 3 that follows provides conditions that guarantee the ex- of the linearized closed-loop system generates an exponen-
ponential stability of the closed-loop system (the proof can be tially stable semigroup) if the following conditions are satis-
found in the Appendix). fied:
Proposition 3. Consider the system of linear first-order 1. The roots of the equation
PDEs of Eq. 9 for which Assumptions 1 and 3 hold, under
the controller of Eq. 28. Then, the closed-loop system is ex-
ponentially stable (i.e., the differential operator of the
closed-loop system generates an exponentially stable semi- lie in the open left-half of the complex plane.
group) if the following conditions are satisfied: 2. The zero dynamics of the system of Eq. 7 is locally ex-
1. The roots of the equation ponentially stable.
Remark 10. The exponential stability of the closed-loop
1+ y * s + ... + yrsU = 0 (36) system guarantees, in both the linear and the quasi-linear
case, that in the presence of small modeling errors, the states
lie in the open left-half of the complex plane. of the closed-loop system will be bounded. Furthermore, since
2. The system of Eq. 35 is exponentially stable. the input/output spaces of the closed-loop system are finite
Remark 7. Referring to the preceding proposition, we note dimensional, and the controller of Eq. 34 enforces a linear
that the first condition addresses the input/output stability of input/output dynamics between 7 and u , it is possible to im-
the closed-loop system and the second condition addresses its plement a linear error feedback controller with integral ac-
internal stability. Note also that the first condition is associ- tion around the (7 - u ) loop to ensure asymptotic offsetless
ated with the stability of a finite number of poles, while the output tracking in the closed-loop system, in the presence of
second condition concerns the stability of an infinite number constant unknown model parameters and disturbances.
of poles. This is expected since the input/output spaces are
finite dimensional, while the state of the system evolves in Output-Feedback Control
infinite dimensions. In this section, we will consider the synthesis of distributed
Remark 8. From the result of Proposition 3, it follows that output-feedback controllers for systems of the form of Eq. 9
the controller of Theorem 1 places a finite number of poles (Eq. 7). The requisite controllers will be synthesized employ-
of the open-loop infinite-dimensional system of Eq. 9 at pre- ing a combination of the developed distributed state-feed-
specified (depending on the choice of parameters yk) loca- back controllers with distributed state observers. Analysis of
tions, by essentially canceling an infinite number of poles, the resulting closed-loop system allows us to derive precise
those included in the zero dynamics. Furthermore, the conditions that guarantee that the requirements of exponen-
closed-loop system is exponentially stable if the zero dynam- tial stability and output tracking are enforced in the closed-
ics of the original system is exponentially stable (Condition 2 loop system.
of Proposition 3). This result is analogous to available results The conventional approach followed for the design of state
of stabilization of systems of linear PDEs of the form of Eq. estimators for linear PDE systems is to discretize the system
9 with feedback law of the form of Eq. 29. Specifically, it is equations and then apply results from estimation theory for
well known (Russell, 1978; Balas, 1986) that control laws of ODE systems [e.g., Sorensen et al. (1980)l. It has been shown,
the form of Eq. 29 allow placing a finite number of open-loop however, that methods for state estimation that treat the full

3070 November 1996 Vol. 42, No. 11 AIChE Journal

distributed parameter system lead to state observers that yield gence of the estimated values to the actual ones with tran-
significantly superior performance (Ray, 1981; Cooper et al., sient behavior depending on the location of the spectrum of
1986). In this direction, available results on state estimation the operator of Eq. 14.
for systems of first-order hyperbolic PDEs concern mainly the Remark 12. Available results on stabilization of systems
use of Kalman filtering theory for the design of distributed of linear hyperbolic PDEs via distributed output feedback
state observers (Pell and Aris, 1970; Yu et al., 1974). [e.g., Ray (1981), Balas (198611 concern the design of con-
trollers with the following general state-space description:
Linear systems
We consider state observers with the following general
state-space description (Pell and Aris, 1970):
ii = sg,

where (R is a bounded linear operator, mapping R' into X".

We note that the main similarity between a controller of the
where 6 is a bounded linear operator, mapping R' into X", form of Eq. 40 and the controller of Eq. 39 is that both are
that has to be designed so that the operator go= d: - 6 Q k infinite dimensional (because of the state observers utilized),
generates an exponentially stable semigroup (note that this is while their main difference lies in the fact that the controller
possible by Assumption 2). The system of Eq. 38 consists of a of Eq. 39 guarantees exponential stability of the closed-loop
replica of the process system and the term 6 ( q - Q p g ) used system, if the open-loop system is minimum-phase and de-
to enforce a fast decay of the discrepancy between the esti- tectable, while a controller of the form of Eq. 40 will expo-
mated and the actual values of the states of the system. In nentially stabilize the closed-loop system, if the open-loop
practice, the design of the operator 6 can be performed by system is jointly stabilizable/detectable (Balas, 1986).
(a) simple pole placement in the case where the output mea- Example (Continued). Referring to the linear PDE system
surements are not corrupted by noise, or (b) Kalman filtering of Eq. 12, we note that A = - uI < 0, while C, = 0, and thus
theory, in the case where the output measurements are noisy. the system is open-loop stable according to the result of Re-
Theorem 3 that follows provides a state-space realization mark 1. The open-loop stability of the system allows using a
of the output-feedback controller resulting from the combi- feedback controller, which consists of the distributed state-
nation of the state observer of Eq. 38 with the state feedback feedback controller coupled with an open-loop observer. The
controller of Eq. 28 (the proof can be found in the Appendix). appropriate controller takes the form:
Theorem 3. Consider the system of linear first-order PDEs
of Eq. 9 subject to the boundary condition of Eq. 2, for which drl - all 1
_ _- u/- - ag + a-
Assumptions 1, 2,3, and the conditions of Proposition 3 hold. at drZ Y1
Consider also the linear bounded operator 6 designed such
that the operator So= d: - S Q p generates an exponen-
tially stable semigroup. Then, the distributed output-feed-
back controller

Quasi-linear systems
In this subsection, we consider the synthesis of distributed
output-feedback controllers for systems of the form of Eq. 7.
Given the lack of available general results on state estimation
of such systems, we will proceed with the design of a nonlin-
ear state observer that guarantees local exponential conver-
gence of the state estimates of the actual state values. In par-
ticular, the following state observer will be used to estimate
the state vector of the system in space and time:
(a) Guarantees exponential stability of the closed-loop sys-
tem
(b) Enforces the input/output response of Eq. 27 in the
closed-loop system if x(z, 0) = g ( z , 0).
Remark 11. In the case of open-loop stable systems, a
more convenient way to reconstruct the state of the system is where g denotes the observer state vector and s is a linear
to consider the observer of Eq. 38 with the operator 6 set operator, mapping R' into X", designed on the basis of the
identically equal to zero. This is motivated by the fact that linearization of the system of Eq. 42 so that the eigenvalues
the open-loop stability of the system guarantees the conver- of the operator E, = E - $ Q p ( z ) lie in the left-half plane.

AIChE Journal November 1996 Vol. 42, No. 11 3071

 The state observer of Eq. 42 can be coupled with the
state-feedback controller of Eq. 34 to derive an output-feed-
back controller that guarantees output tracking and closed-
loop stability. The resulting controller is given in Theorem 4
that follows (the proof is given in the Appendix).
Theorem 4. Consider the system of quasi-linear first-order Figure 2. Nonisothermal plug-flow reactor.
PDEs of Eq. 7 subject to the boundary condition of Eq. 2, for
which Assumptions 1, 2, 3, and the conditions of Proposition
4 hold. Consider also the bounded operator @ designed such the number of discretization points increases, the closed-loop
that the operator Eo= - g Q p ( z ) generates an exponen- system resulting from the PDE model plus an approximate
tially stable semigroup. Then, the distributed output feed- finite-dimensional controller converges to the closed-loop
back controller system resulting from the PDE model plus the infinite-
dimensional controller, guaranteeing the well-posedness of
the approximate finite-dimensional controller.

Application to a Nonisothermal Plug-Flow Reactor

Process description
Consider the nonisothermal plug-flow reactor shown in
Figure 2 where two first-order reactions in series take place:

where A is the reactant species, B is the desired product,

and C is an undesired product. The inlet stream consists of
pure A of concentration C,, and temperature TAo.The reac-
tions are endothermic, and a jacket is used to heat the reac-
tor. The reaction rate expressions are assumed to be of the
following form:

r2 = - k 2 0 e - E f l T t B ,
(a) Guarantees local exponential stability of the closed-loop
where k,,, kzo, E l , E , denote the pre-exponential constants
system
and the activation energies of the reactions. Under the fol-
(b) Enforces the input/output response of Eq. 27 in the
lowing assumptions:
closed-loop system if x(z, 0) = q(z,O).
Perfect radial mixing in the reactor
In analogy with the linear case, for open-loop stable sys-
Constant volume of the liquor in the reactor
tems, the operator P can be taken to be identically equal to
Constant density and heat capacity of the reacting liquid
zero, since the local exponential stability of the open-loop
Negligible diffusive and dispersive phenomena
system guarantees the local convergence to the estimated val-
the material and energy balances that describe the dynamical
ues to the actual values.
behavior of the process take the following form:
Remark 13. Note that in the case of imperfect initializa-
Mole balance for the species A
tion of the observer states (i.e., ,(z,O) # x(z,O)), although a
slight deterioration of the performance may occur (i.e., the
input/output response of Eq. 27 will not be exactly imposed
in the closed-loop system), the output-feedback controllers of
Theorems 3 and 4 guarantee exponential stability and asymp-
totic output tracking in the closed-loop system. Mole balance for the species B
Remark 14. The nonlinear distributed output feedback
controller of Eq. 43 is an infinite-dimensional one, due to the
infinite-dimensional nature of the observer of Eq. 42. There-
fore, a finite-dimensional approximation of the controller has
to be derived for on-line implementation. This task can be Reactor energy balance
performed utilizing standard discretization techniques such
as finite differences and orthogonal collocation. It is ex-
pected that some performance deterioration will occur in this
case, depending on the discretization method used and the
number and location of discretization points (see the chemi-
cal reactor application presented in the next section). Finally,
we note that it is well established (e.g., Balas, 1986) that as

3072 November 1996 Vol. 42, No. 11 AIChE Journal

 4.1
Table 1. Process Parameters
u, = 1.0 m min- ’
L = 1.0 m
v, = 10.0 It
E , = 2.0 x 104 kcal kmol-
E , = 5.0 x 104 kcal kmol-
k , , = 5.0 x 10” min-’
k,, = 5.0 X lo6 min-
R = 1.987 kcal * kmol- ‘ . K -
AH, = 548,000.1 kcale kmol-’
4 AH, = 986,000.1
cpm = 0.231
kcale kmol -
.
kcal kg- K- ’
p, = 0.09 kg-lt- ’
c p j = 2.5 kcal. kg- K - ’
pj = 1.44 kg-lt-’
3.21
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.0
11 v,= 2,000.0 kcale min- K-
Spatial Coordinate (m) CAo= 4.0 mol -It-I
c, = 0.0 mol .It-
0.8, 1 TAo= 320.0 K
qo,
= 350.0 K
”;’ = 1.0 It
yjc = 0.001 min
Fjf = 28.87 1t.min-l
Fj: = 43.66 It min-
42 = 59.63 -
It min-
$f = 72.34 It * rnin-
I;;: = 77.23 It * min-

fluid in the reactor; V , denotes the volume of the reactor; U,

denotes the heat-transfer coefficient; I; denotes the spatially
uniform temperature in the jacket; and C,, and TAo denote
the concentration and temperature of the inlet stream in the
reactor. The values used for process parameters are given in
Spatial Coordinate (m)
Table 1, while the corresponding steady-state profiles are
shown in Figure 3. The control objective is the regulation of
the concentration of the species B throughout the reactor by
manipulating the jacket temperature T . We note that, in
practice, T is usually manipulated indirectly through manip-
ulation of the jacket inlet flow rate (this implementation is-
sue is addressed in the subsection titled “Practical Imple-
mentation Issues’?.
Setting:

the original set of equations can be put in the form of Eq. 1

Spatial Coordinate (m) with:

Figure 3. Steady-state profiles of reactor state vari-

ables.

subject to the following boundary conditions:

-k10e-E@x3~1
k , o e - E f l x 3 x-
, k 20 e - E f l x 3 ~ 2 1
where C, and C , denote the concentrations of the species A
and B in the reactor; T, denotes the temperature of the reac-
tor CAs,C,,, Trs denote the steady-state profiles for the state
variables; AH,,, AH,, denote the enthalpies of the two reac-
tions; pm, cpm denote the density and heat capacity of the

AIChE Journal November 1996 Vol. 42, No. 11 3073

 u' (t) u2(t) Ti3 (t) u4 (t) 2(t) I

(45)

It is clear that the matrix A ( x ) is a real symmetric and its Figure 4. Specification of the control problem for the
eigenvalues satisfy Eq. 4. Moreover, the three eigenvalues of plug-flow reactor.
A ( x ) are identical, which implies that the preceding system of
quasi-linear PDEs is weakly hyperbolic.
Eq. 48, its characteristic index can be calculated using Defini-
Control problem formulation -Controller synthesis tion 2. In particular, we have that:
In this subsection, we proceed with the formulation and
solution of the control problem. More specifically, it will be
assumed that there are five control actuators that are charac-
terized by a unity distribution function, that is, bi(z)Hi= H i
for all i = 1, . .., 5. The control actuators are taken to act
over equispaced intervals, that is:

El(t), [0.0,0.21 1 af2

-g3dz#0, Vi=l, ..., 5 . (49)
ii2(t), [0.2,0.41
(46)
i i 4 ( t ) , [0.6,0.81 Thus, the characteristic index of the system of quasi-linear
PDEs of Eq. 48 is equal to 2. This allows us to request the
c5(t1, [0.8,1 .Ol.
following second-order response in the closed-loop system
between 7' and ui, for all i = 1, . .., 5:
The desired performance requirement is to control the aver-
aging outputs:
(50)

[
0.2
m = / 5.0X2(2,t)dZ
0.0
Moreover, the eigenvalues of the matrix A ( x ) are negative
y2(t)= / 0.4
5.0x2(z,t ) dz and the boundary conditions are specified in a single point.
Thus, the result of Remark 1 applies directly, yielding that
the system is open-loop stable. Furthermore, it was also veri-
{
0.6
jj(t)= j j 3 ( t ) = / 5.0x2(z,t)dz (47) fied through simulations that the process is minimum-phase.
0.4
Therefore, the developed control method can be applied and
the distributed output feedback controller of Theorem 4, with
$= 0, was employed in the simulations (note that due to
I jj5(t)=/
1.0
0.8
5.0x2(z,t)dz. open-loop stability of the process the controller does not use
measured process outputs). The explicit form of the con-
troller is as follows:
Using these relations, the model that will be used for the
synthesis of the output-feedback controller is given by

where the matrix A ( x ) and the vector functions f ( x ) and

g ( x ) are specified in Eq. 45. The reactor atong with the con-
trol system is given in Figure 4. Referring to the system of

3074 November 1996 Vol. 42, No. 11 AIChE Journal

 Time (min.) Time (min.)

-.-0 5 10

Time (min.) Time (min.)

I
0.65; 5 10 15

Time (min.)
Figure 5. Output profiles for a 30% change in reference inputs.

AIChE Journal November 1996 Vol. 42, No. 11 3075

 5

Time (min.) Time (min.)

4-

3.5 -
3-

Y
OO 5 10

Time (min.) Time (rnin.)

Time (min.)
Figure 6. Manipulated input profiles for a 30% increase in reference inputs.

3076 November 1996 Vol. 42, No. 11 AIChE Journal

 Figure 7. Profile of concentration evolution of species
B throughout the reactor, for a 30% increase
in the reference inputs.

+ L f ) 2 h ( q ) ] , i = 1, ..., 5, (51) The controller was tuned to give an overdamped response

j =1 between the output jj' and the reference input ui. In particu-
lar, the parameters y1 and y 2 were chosen to be:
where the analytical expressions of the terms included in the
controller are as follows: y1= 3.0min, y 2 = 0.5min2

to achieve the following time constant and damping factor:

7 = 0.707min, 5 = 2.12

Evaluation of controller performance

Several simulation runs were performed to evaluate the
performance of the distributed output feedback controller of
Eq. 51. The method of finite differences was employed to

-
- l (-ul)
i--+--+--
d f 2 d771
all1 dz
d f 2 d172
d q 2 dz
d f 2 3773
dq3 dz
idz

0.75L
0 5 10 ,
Time (min.)
Figure 8. Profile of concentration of species B in the
outlet of the reactor, for a 30% increase in ref-
(52) erence inputs.

AIChE Journal November 1996 Vol. 42, No. 11 3077

 0.045‘ I 0.16’
0 10 20 30 40 50 60 0 10 20 30 40 50

Time (min.) Time (min.)

::I,,
0.3
0 10 20
,

30
,

40
,

50
1
60
0.5 .

Time (min.)

0.75

0.85
0
Time (min.)
Figure 9. Output profiles for a 30% increase in the reference inputs-discretization-basedcontroller.

3078 November 1996 Vol. 42, No. 11 AIChE Journal

derive a finite-dimensional approximation of the output-
feedback controller of Eq. 51, with a choice of 200 discret-
ization points. In all the simulation runs, the process was l ' 0 5 L
1-
initially assumed to be at steady state.
In the first simulation run, we addressed the reference
input tracking capabilities of the controller. Initially, a 30%
increase in the reference inputs ui, i = 1, ..., 5 was imposed
-
4
0.95-

at time t = 0.0 min. Figure 5 shows the corresponding output 2

v

profiles. It is clear that the controller enforces the requested

input/output response in the closed-loop system and regu-
-8
h
c:
v
0.9-

lates the output at the new reference input values. The cor- 0.85 -
responding input profiles for each control actuator are
depicted in Figure 6. We observe that the control action,
required by each actuator to drive the corresponding output 0.8'
to the new reference value, increases as we approach the out-
let of the reactor. This is expected, because the amount of
0.75'
heat required to maintain the reaction rate that yields the 0 10 20 30 40 50
necessary conversion increases along the length of the reac- Time (min.)
tor. Figure 7 shows the evolution of the concentration of Figure 10. Profile of concentration of species 6 in the
species B throughout the reactor, while Figure 8 shows the outlet of reactor, for a 30% increase in the
profile of the concentration of species B at the outlet of the reference inputs-discretization-based con-
reactor. We observe that by using a finite number of control troller.
actuators, we achieve satisfactory control of the output vari-
able C, at all positions and times.
For the sake of comparison, we also consider the control of
the reactor using a controller that was designed on the basis
of a model resulting from discretization of the original PDE
system in space. In particular, the method of finite differ-
ences was used to discretize the original PDE model of Eq. (TO-7)
48 into a set of five (equai to the number of control actua- + ~j UfL, i = 1, ...)5, (53)
I
tors) ordinary differential equations in time. Subsequently, an
input/output linearizing controller (Kravaris and Arkun,
1991) was designed on the basis of the resulting ODE model. where qi is the steady-state jacket inlet flow rate; ";i is the
The corresponding output profiles are shown in Figure 9, jacket volume; pi, cpj,are the density and heat capacity of the
while Figure 10 shows the profile of the concentration of fluid in the jacket; T,b is the temperature of the inlet stream
species B in the outlet of the reactor. It is clear that this to a jacket; and uil is the jacket inlet flow rate (chosen as the
controller leads to poor performance, because it does not ex- new manipulated input) in deviation variable form. Request-
plicitly take into account the spatially varying nature of the ing a first-order response of the form:
process.
dT.' . ,

y. + T,' = E', (54)

dt
Practical implementation issues
The distributed output feedback controller of Eq. 51 as- where yj, is the time constant, the necessary controller takes
sumes that the jacket temperature can be manipulated di- the form:

rectly. In practice, the jacket temperature is usually manipu- The parameter yjc should be chosen such that the response
lated indirectly through the jacket inlet flow rate. This can be of Eq. 54 is sufficiently fast compared to the response of Eq.
achieved in a straightforward way by designing a controller to 50, while estimates of T$z, t ) can be obtained from the state
ensure that the jacket temperature obtains the values re- observer of Eq. 51.
quested by the distributed controller of Eq. 51. The performance of the control scheme resulting from the
Specifically, under the assumption of perfect mixing and combination of this controller with the distributed controller
constant volume, the dynamic model of the jacket takes the of Eq. 51 was evaluated through simulations on the plug-flow
form reactor. The values used for the parameters and the steady-

AIChE Journal November 1996 Vol. 42, No. 11 3079

 0.16'
0 5 10 15 20

Time (min.) Time (min.)

"."0 5 10 15 20 25
0.48'
0 5 10 15 20

Time (min.) Time (min.)

0.65
0
I
20 5 10 15 5

Time (min.)
Figure 11. Output profiles for a 30% change in reference inputs-jacket inlet flow rates are used as the manipulated
inputs.

3080 November 1996 Vol. 42, No. 11 AIChE Journal

 Time (min.) Time (min.)

Time (min.) Time (min.)

Time (min.)
Figure 12. Profiles of jacket inlet flow rates for a 30% increase in reference inputs.

AIChE Journal November 1996 Vol. 42, No. 11 3081

state values of the jacket inlet flow rate Ti, i = 1, ..., 5, are Balas, M. J., “Trends in Large Scale Structure Control Theory:
given in Table 1. Figure 11 shows the output profiles for the Fondest Hopes, Wildest Dreams,” ZEEE Trans. Automat. Contr.,
27, 522 (1982).
same 30% increase in the value of the reference inputs as Balas, M. J., “Finite-Dimensional Control of Distributed Parameter
previously (the profiles for the jacket inlet flow rate are dis- Systems by Galerkin Approximation of Infinite Dimensional Con-
played in Figure 12). Clearly, the performance of the control trollers,” J. Math. Anal. Appl., 114, 17 (1986).
scheme is excellent, enforcing closed-loop output responses Balas, M. J., “Nonlinear .Finite-Dimensional Control of a Class of
Nonlinear Distributed Parameter Systems Using Residual Mode
that are very close to the ones obtained by neglecting the Filters: A Proof of Local Exponential Stability,” J. Math. Anal.
jacket dynamics (compare with Figure 5). Appl., 162, 63 (1991).
Brown, H. S., I. G. Kevrekidis, and M. S. Jolly, “ A Minimal Model
for Spatio-Temporal Patterns in Thin Film Flow,” Pattern and Dy-
Conclusions namics in Reactive Media, R. Aris, D. G. Aronson, and H. L. Swin-
In this work, we develop an output-feedback control ney, eds., Springer-Verlag, New York, p. 11 (1991).
Burns, J. A., and S. Kang, “A Control Problem for Burgers’ Equa-
methodology for systems described by quasi-linear first-order tions with Bounded Input/Output,” Nonlinear Dyn., 2, 235 (1991).
hyperbolic PDEs, for which the manipulated input, the con- Bymes, C. A., D. S. Gilliam, and J. He, “Root-Locus and Boundary
trolled output, and the measurable output are distributed in Feedback Design for a Class of Distributed Parameter Systems,”
space. The central idea of our approach is the combination S U M J . Contr. Optim., 32, 1364 (1994).
Chen, C. C., and H. C. Chang, “Accelerated Disturbance Damping
of the theory of PDEs and concepts from geometric control. of an Unknown Distributed System by Nonlinear Feedback,”
Initially, a concept of characteristic index was introduced and AZChE J., 38, 1461 (1992).
used for the synthesis of distributed state-feedback con- Cooper, D. J., W. F. Ramirez, and D. E. Clough, “Comparison of
trollers that guarantee output tracking in the closed-loop sys- Linear Distributed-Parameter Filters to Lumped Approximants,”
tem. Conditions that ensure exponential stability of the AZChE J., 32, 186 (1986).
Curtain, R. F., “Finite-Dimensional Compensator Design for
closed-loop system were derived. Analytical formulas of out- Parabolic Distributed Systems with Point Sensors and Boundary
put-feedback controllers were also derived through a combi- Input,” JEEE Trans. Automat. Contr., 27, 98 (1 982).
nation of suitable state observers with the developed dis- Curtain, R. F., and A. J. Pritchard, Infinite Dimensional Linear Sys-
tributed state-feedback controllers. The proposed control tems Theoly, Springer-Verlag, Berlin (1978).
methodology was implemented, through simulations, on a Dochain, D., J. P. Babary, and M. N. Tali-Manaar, “Modeling and
Adaptive Control of Nonlinear Distributed Parameter Bioreactors
nonisothermal plug-flow reactor, modeled by three quasi- via Orthogonal Collocation,” Automatica, 68, 873 (1992).
linear hyperbolic PDEs. Comparisons with a control method Friedman, A., Partial Diflerential Equations, Holt, Rinehart & Win-
that involves discretization in space of the original PDE model ston, New York (1976).
and application of standard nonlinear control methods for Gay, D. H., and W. H. Ray, “Identification and Control of Dis-
tributed Parameter Systems by Means of the Singular Value De-
ODE systems established that the new control method yields composition,” Chem. Eng. Sci., 50, 1519 (1995).
superior performance. Georgakis, G., R. A r i s , and N. R. Amundson, “Studies in the Con-
trol of Tubular Reactors: I, 11, and 111,” Chem. Eng. Sci., 32, 1359
Acknowledgment (1977).
Hanczyc, E. M., and A. Palazoglu, “Eigenvalue Inclusion for Model
Financial support from the National Science Foundation (CTS- Approximations to Distributed Parameter Systems,” Ind. Eng.
9624725) is gratefully acknowledged. Chem. Res., 31, 2538 (1992).
Hanczyc, E. M., and A. Palazoglu, “Sliding Mode Control of Nonlin-
ear Distributed Parameter Chemical Processes,” Znd. Eng. Chem.
Notation Rex, 34, 455 (1995).
A ( z ) = matrices Kang, S., and K. Ito, “ A Feedback Control Law for Systems Arising
c’(z) = performance specification function for the ith output in Fluid Dynamics,”Proc. ZEEE Conf. on Dec. and Control, Tampa,
F= outlet flow rate AZ, p. 384 (1992).
F , = inlet flow rate Keulen, B., H,-Control for Distributed Parameter Systems: A State-
T, = temperature of the wall Space Approach, Birkhauser, Boston (1993).
ii = manipulated input vector Kravaris, C., and Y. Arkun, “Geometric Nonlinear Control--An
El = ith manipulated input Overview,” Proc. Znt. Conf. on Chemical Process Control, Y . Arkun
u‘= reference input for the ith actuator and W. H. Ray, eds., Padre Island, TX, p. 477 (1991).
x = vector of state variables Lasiecka, I., “Control of Systems Governed by Partial Differential
y,., yj, = adjustable parameters Equations: A Historical Perspective,” Proc. ZEEE Conf. on Dec.
u = characteristic index of j j with respect to E and Control, New Orleans, LA, p. 2792 (1995).
Lo, C. T., “Optimal Control of Counter-Current Distributed Param-
Math symbols eter Systems,” Znt. J . Contr., 18, 273 (1973).
Patwardhan, A. A., G. T. Wright, and T. F. Edgar, “Nonlinear
L f h = Lie derivative of a scalar field h with respect to the vec- Model-Predictive Control of Distributed-Parameter Systems,”
tor field f Chem. Eng. Sci., 47, 721 (1992).
Lkfh = kth order Lie derivative Pell, T. M., and R. Ark, “Control of Linearized Distributed Systems
L, L!- Ih = mixed Lie derivative on Discrete and Corrupted Observations,” Znd. Elec. Chem. Fund.,
R = real line 9, 15 (1970).
E = belongs to Pohjolainen, S. A,, “Computation of Transmission Zeros for Dis-
T= transpose tributed Parameter Systems,” Znt. J . Contr., 33, 199 (1981).
-
11 112 = 2-norm in X Ray, W. H., “Some Recent Applications of Distributed Parameter
Systems Theory-A Survey,” Aufomatica, 14, 281 (1978).
Ray, W. H., Advanced Process Control, McGraw-Hill, New York
Literature Cited (1981).
Alonso, A. A,, and E. B. Ydstie, “Nonlinear Control, Passivity and Rhee, H. K., R. Ark, and N. R. Amundson, First-OrderPartial Difler-
the Second Law of Thermodynamics,” AIChE Meeting, Miami ential Equations: Vols. Z and 11, Prentice Hall, NJ (1986).
Beach, FL (1995). Russell, D. L., “Controllability and Stabilizability Theory for Linear

3082 November 1996 Vol. 42, No. 11 AIChE Journal

 Partial Differential Equations: Recent Progress and Open Ques-
tions,” S U M Reu., 20, 639 (1978).
Sira-Ramirez, H., “Distributed Sliding Mode Control in Systems De- dt
scribed by Quasi-linear Partial Differential Equations,” Syst. Contr.
Lett., 13, 177 (1989).
Smoller, J., Shock Waves and Reaction-Difiion Equations, Springer-
Verlag, Berlin (1983). -d”f
= e k ( A z + Bd ) L T x + e k ( A z + B
d ) LT -1
wb(z)Sx
Sorensen, J. P., S. B. Jorgensen, and K. Clement, “Fixed-Bed Reac-
tor Kalman Filtering and Optimal Control: 1,”Chem. Eng. Sci., 35, dt
1223 (1980).
Stangeland, B. E., and A. S . Foss, “Control of a Fixed-Bed Chemical
Reactor,” Znd. Eng. Chem. Res., 38, 38 (1970).
Teman, R., Infinite-DimensionalDynamical Systems in Mechanics and
Physics, Springer-Verlag, New York (1988).
Wang, P. K. C., “On the Feedback Control of Distributed Parameter
Systems,” Znt. J . Contr., 3, 255 (1966). This completes the proof of the proposition.
Yu, T. K., J. H. Seinfeld, and W. H. Ray, “Filtering in Nonlinear
Time-Delay Systems,”ZEEE Trans. Automat. Contr., 19,324 (1974).
Proof of Theorem 1
Under the controller of Eq. 28, the closed-loop system takes
Appendix the form:
The proofs of Theorem 2 and Proposition 4, concerning
the case of systems of quasi-linear PDEs, are conceptually
similar to the ones given for the linear counterparts of these
results, and will therefore be omitted for brevity.

i
u
Proof of Proposition 2 X u-ekx- C y v e k
v=l
Consider the closed-loop system of Eq. 26. Differentiating
the output of this system with respect to time, we obtain the f = ekx. (A31
following set of equations:

From the result of Proposition 2, it follows that a differentia-

tion of the output of the system of Eq. A3 yields the follow-
ing equations:

d
d2f
-=
dt2
( d
e k A-+B++b(z)S
dZ
”
-dt
= (
C?kA-+Bx
dz )
d
A-+B++b(z)S
dt * 3.2
-d”f
= ~ k ( A ~ +dB ) u x + C ! k ( A ~d + B ) L T - l w b ( z )
dt

It is sufficient to show that Cr = (T.Note that e k [A( + B

w b ( ~ ) S ] =e k [ A ( d / d z ) + B ] + e k w b ( z ) S = e k [ A ( d / d z )
+ B ] , because ekwb(z)= 0. Similarly, it can be shown, by
+
induction, that e k [ A ( d / d z ) +B w b ( z ) S l i = Ck[A(d/dz)
+ B]’, for i = 1, ..., (T - 1. Using this simplification in the
expressions for the time derivatives (Eq. Al) it follows di- Substituting these relations to Eq. 27, one can easily show
rectly that that the result of the theorem holds.

y=ekx Proof of Proposition 3

Utilizing the expressions for .the output derivatives of Eq.
dt A2 the closed-loop system of Eq. A3 takes the form:

AIChE Journal November 1996 Vol. 42, No. 11 3083

 Substituting Eq. A 7 into Eq. A 8 , we have that:

Defining the state vectors 5, = ( d i - ' j ) / d t i - ' , i = 1, . . ., o,

the system of Eq. A 5 can be equivalently written in the form Let 6 = a, - a,. If S = 0, II x 112 I K 2 II x o 112 e-02' + K 2 K ,
of the following interconnection: II W(z>112 I ioI e P 2 ' t I K , II x o 112 e-a2' + K 2 K l II W ( z )112
(1 2, I/a2 - a3)e-'3', where 0 < a3 < u2. Clearly, in this case,
the closed-loop system is exponentially stable. Furthermore,
i l = 52 if 6 > 0, II x II 2 I K , II x o I1 Ze-",' + K,Kl II W(z)112(1 zo
1/6)e-"1'(1- e-(a2-ai)t),while if S < 0, II x 112 I K 2 I1 x o 112
e-"2'+ K,K, (1 W(z)112 (I to 1/1 s I)e-a2'(1- e(az-al)f).In
either case ( 8 > 0, 6 < O), and we have that:
5,- 1= 5,

dx dx d where ii = min{u,, a2}. From Eqs. A 7 - A l l the exponential

-= A-+Bx+wb(z)
at dz stability of the closed-loop system of Eq. A5 follows directly.

Proof of Theorem 3
Part I: StabilityAnabsis. Substituting the controller of Eq.
where 39 in the system of Eq. 9, we have:

Condition 1 of the proposition guarantees that the 5-subsys-

tem of the preceding interconnection is exponentially stable,
and thus the following condition holds:

where 1.1 denotes the standard Eucledian norm; K,, a , are

positive real numbers, with K , >_ 1; and lois the value of the
variable at time t = 0 {i.e., zo
= (l/y,)[u(O) - 5(1,0] -
Z~::(yv/yu)~(,+lp}.From Condition 2, we have that the dif-
ferential operator of the system of Eq. 35 generates an expo-
nentially stable semigroup 0, that is, 1 1
u112< K2e-'2', where
Introducing the error coordinate Z = x - 7, the preceding
K,, a2 are positive real numbers, with K , 2 1. Utilizing Eq. closed-loop system can be written as
15, the following estimate can be written for the state x of
the system of Eq. A5:
dx
_
dt
-
-A-
dx
dz
[
+ BX + w b ( z ) y,ek (A - d
dz + B )'-lwb(z)]-'

i
cr
X u-C?Eoc- C y,ek
u=l
where

where

3084 November 1996 Vol. 42, No. 11 AIChE Journal

Because the operator 6 is designed such that the operator
d: - 6 Q p generates an exponentially stable semigroup, the
following estimate can be written for the evolution of the state
Z of the preceding system: In order to perform a local analysis of the stability properties
of the preceding system, we consider its linearization:

where Z, denotes the vector of initial conditions and K,, ii

are positive real numbers, with K , 2 1. Furthermore, utiliz-
ing Eq. 15 and the fact that the conditions of Proposition 3
hold, the following inequality can be written for the state x
of the system of Eq. A13:

where K,, B are positive real numbers, with K , 2 1. Substi-

tuting the inequality of Eq. A15 into Eq. A16, and perform-
ing similar calculations as in the proof of Proposition 3, it can
be shown that the closed-loop system is exponentially stable.
Part 2: Input/Output Response. Under consistent initial-
ization of the states x and 7,that is, x(z,O) = v(z,O),it fol-
lows that Z(z,O) = 0. From the dynamical system for Z, it is
clear that if Z(z,O) = 0, then Z(z, t ) = 0 for all t 2 0. Thus,
the system of Eq. A13 reduces to

Introducing the error coordinate Z = x - 7, the preceding

closed-loop system can be written as

A direct application of Theorem 1 completes the proof of the

theorem.

Proof of Theorem 4
Part 1: Stability Analysis. Substituting the controller of Eq.
43 in the system of Eq. 7, we have:

-1

where

AIChE Journal . ... 42.

November 1996 Vol. ~~,No. 11 3085
From the conditions of the theorem, we have that the x- following expressions:
subsystem (with Z = 0 ) of the preceding interconnection is ex-
ponentially stable and the &subsystem is also exponentially f = eh(x)
stable, because the operator $ is designed such that the op-
erator - g Q p ( z ) generates an exponentially stable semi-
group. Following an approach analogous to the one used in
the proof of Proposition 3, one can show that the system of
Eq. A20 is exponentially stable. Utilizing the result of Propo-
sition 1, we have that the closed-loop system of Eq. A18 is
locally exponentially stable, because the linearized system of
Eq. A20 is exponentially stable.
Part 2: Input/Output Response. Under consistent initial-
ization of the states x and 7,that is, x(z,O) = v(z,O), it fol-
lows that Z(z,0) = 0. From the dynamical system for Z, it is
clear that if Z(z, 0 ) = 0, then Z(z, t ) = 0 for all t 2 0. Thus,
the system of Eq. A20 reduces to

Substituting the preceding relation into Eq. 27, one can easily
show that the result of the theorem holds.
Since the characteristic index between f and u is u ,a differ-
entiation of the output of the system of Eq. A22 yields the Manuscript receiued July 7, 1995 and reuision received Mar. 1, 1996.

3086 November 1996 Vol. 42, No. 11 AIChE Journal