Brazilian Journal

of Chemical ISSN 0104-6632

Printed in Brazil
Engineering www.abeq.org.br/bjche

Vol. 33, No. 01, pp. 133 - 143, January - March, 2016
dx.doi.org/10.1590/0104-6632.20160331s00002780

MULTIVARIABLE OPTIMAL CONTROL

OF A HEAT EXCHANGER NETWORK
WITH BYPASSES
F. Delatore1*, L. F. Novazzi3, F. Leonardi2 and J. J. da Cruz4
1
Department of Electrical Engineering, FEI University, Av. Humberto de Alencar
Castelo Branco 3972, Zip Code: 09850-901, Sao Paulo - SP, Brazil.
Phone: +55 11 43532900
E-mail: fdelatore@fei.edu.br
2
Department of Mechanical Engineering, FEI University, Sao Paulo - SP, Brazil.
3
Department of Chemical Engineering, FEI University, Sao Paulo - SP, Brazil.
4
Department of Telecommunications and Control, University of Sao Paulo, Sao Paulo - SP, Brazil.

(Submitted: June 18, 2013 ; Revised: December 23, 2014 ; Accepted: February 2, 2015)

Abstract - Heat exchanger networks present an interesting control problem due to coupling among process
streams. In this work, the linear quadratic regulator (LQR), a feedback optimal control technique, is used to
control stream temperatures on a laboratory scale heat exchanger network, through bypass manipulation, in a
multivariable system. The LQR design was based on a mathematical model of the plant and its performance
was compared to traditional PID control and to dynamical decoupling. Experimental tests were performed to
evaluate the controllers, involving regulatory and servo problems. The performance of the different controllers
was quantitatively compared by using the integral absolute error. Although LQR is not a new control
methodology, the results obtained in this work suggest that LQR is an interesting alternative to control HEN
when compared to the PID and to the dynamic decoupler. Moreover, one of the main advantages of the LQR
is its tuning simplicity, since only one parameter is sufficient for this application.
Keywords: Heat exchanger network; Optimal control; LQR.

INTRODUCTION Design and are easily found in many Chemical Engi-

Due to the oil price rise since the seventies and to
environmental issues, efficient use of energy in
chemical processes is very important. Almost thirty
years ago the theoretical foundations of Process Inte-
gration for the efficient use of energy were estab-
lished, with Pinch Technology, an elegant approach
to set energy / cost targets for heat exchanger networks
(HEN), as well as rules to design such networks
(Linhoff et al., 1982). Nowadays these synthesis tech-
niques, including some which are based on mathe-
matical programming, are well established in Process
neering textbooks.
The design of a HEN depends on nominal stream
supply temperatures and flowrates. However, during
plant operation such nominal operating conditions
can change, influencing stream target temperatures
and propagating in the network, since the heat ex-
changers introduce coupling among different parts of
the process. Therefore, HEN control is an interesting
issue and has been addressed in the literature since
the eighties. A preliminary important contribution
can be found in Marselle et al. (1982), where the
authors proposed a HEN design technique by consid-

*To whom correspondence should be addressed

134 F. Delatore, L. F. Novazzi, F. Leonardi and J. J. da Cruz
ering process operability and controllability, which
they called process resilience. Operability is defined
as the ability of the network to remain steady-state
feasible when subjected to process disturbances,
whereas controllability is regarded as the network
capacity to go from one steady-state to a different
one, in a finite time.
Later on, Calandranis and Stephanopoulos (1988)
proposed a sequence of control actions of the loops
in a network to solve the regulatory and servo control
problems in a HEN, exploiting its structural charac-
teristics. The idea of the strategy was the identifica-
tion of routes through the HEN that could allocate
disturbances or setpoint changes to available sinks,
i.e., utility heat exchangers.
Based on a previous work on HEN control by Ma-
thisen (1994), Glemmestad et al. (1996) applied a
method for optimal operation of the network and
studied the coupling of manipulated variables, repre-
sented by bypasses positions on the exchangers, with
controlled variables, represented by stream target
temperatures. In addition to the input / output pairing
in the suggested decentralized control scheme, the
proposed approach also contemplated the optimiza-
tion of utility consumption in the HEN, since the
number of manipulated variables is greater than the
number of controlled parameters, which resulted in a
positive degree of freedom. More recently and due to
this positive degree of freedom, Sun et al. (2013)
used non-square relative gain arrays to choose which
bypasses should be selected to control a HEN.
Glemmestad et al. (1999) presented an alternative
approach to the optimal operation of HEN systems
based on on-line optimization of a steady state func-
tion and a fixed control structure. Later, Giovanini
and Marchetti (2003) showed that a low-level Dis-
tributed Control System is also capable of handling
HEN control problems when a flexible control loop
structure is provided.
In the work of Lersbamrungsuk et al. (2008), a
linear programming (LP) problem for the optimal
operation of a HEN was formulated. As a conse-
quence of the LP problem, the optimal point of oper-
ation of a HEN remained at some of its constraints.
The authors also proposed an offline strategy to
switch between active constraints, identifying possi-
ble operational regions, and combined this with de-
centralized control. Previously, Aguilera and Mar-
chetti (1998) developed a procedure for optimization
and control of a HEN, in a more complex approach
than the latter, since a nonlinear programming problem
(NLP) needs to be solved online, during operation.
From the point of view of performance, there are
a lot of different control techniques that could be
Brazilian Journal of Chemical Engineering
used in heat exchanger networks. These techniques
range from methodologies that have a complex and
highly engineered design, typically with a superior
performance, down to methodologies that are easy
and effortless to design, but normally not capable to
lead to a desirable performance. As an example, the
PID is the most common controller due to its straight-
forward design procedure and easy implementation.
Nonetheless, in multivariable control problems,
as is the case of a HEN, the PID design will demand
a supplementary engineering effort to tune well the
controllers. In this situation, model predictive control
(MPC) may be considered to be a suitable control
strategy to be used within industrial process, as it can
deal with multivariable systems, complex dynamics
and constraints on input and/or output variables. In
this approach the future moves of the manipulated
variables depend on the model and on plant meas-
ured output variables, in such a way that an on-line
constrained optimization is performed.
Gonzalez et al. (2006) presented an application of
optimization and control of heat exchanger networks,
through a two-level control structure. In the lower
level, a constrained MPC was used and the higher
level was supervised by an online optimizer. The
MPC was based on a linear approximate plant model
whereas the optimizer was based on a rigorous
model. By using a moving horizon, hard constraints
on the manipulated variables were dealt with in a
straightforward way. Although the proposed method-
ology uses a consolidated control strategy, just simu-
lated results were presented. Besides, modeling er-
rors were not explicitly included in the design.
In the range of suitable controllers for the control
of a HEN, one can also point out the Linear Quadratic
Regulator (LQR), which has a very simple design in
the time domain. During the design of a LQR, the
expected dynamic response is explicitly not taken
into account, and the closed-loop response is
checked afterwards. If the time response does not
present an acceptable performance, it is possible to
try a new controller by changing the penalty matri-
ces, resulting in an extraordinary ease to design, and
normally a good performance (Delatore et al., 2009).
In order to reduce the difficulties involved in con-
troller design, in this work a control solution for a
HEN based on optimal linear control is proposed.
The LQR controller performance was successfully
checked by experimental results obtained in a pilot
plant. It must be pointed out that LQR control ap-
plied to a HEN is not commonly found in the scien-
tific literature (Delatore et al., 2009). The objective
of this work is to illustrate the proposed design pro-
cedure, as well as to show the reasonable perfor-
 Multivariable Optimal Control of a Heat Exchannger Network witth Bypasses 135
1

mance
m foundd for the LQ QR control to perform the dT
TCby ,n i 1
cllosed loop coontrol on a HEN.
H  a6TCby ,n i  a6TCby ,n i 1 (66)
dt

whhere,
THE MATHEM
MATICAL MODEL
M

The shell and tube heat exchangerrs (HE) are the  

a1  n mH 1  f H  /   H VH  , a2  UA / 2  H VH C p, H ,
most
m common HE used in chemicall plants duee to a3  n mC 1  fC  /  CVC  , a4  UA /  2 CVC C p ,C  ,
soome particuular characteeristics, maiinly associaated
with
w their loww cost and uncomplicated repair. The T a5  n mH f H /   H VHby  , a6  n mC fC /  CVCby  .
dy ynamic moddel of a shell and tube exxchanger can n be
written
w based on the energ gy balances for
f hot and cold
c Thus a HEEN can be mathematicall
m ly representeed
sttreams, as in Equation ns 1 and 2, 2 respectivvely byy the intercon
nnection of HE,
H accordinng to the HE EN
(LLuyben, 19990). In these equations, thhe subscriptss H strructure. The discretizationn scheme is represented in
annd C stand forf the hot and
a cold sidees, respectiveely, Figgure 1 for the
t first, i-thh and n-th stages,
s respeec-
while
w T is thee stream tem
mperature, t iss the time, m is tivvely (Novazzzi, 2007). Thhe system off ordinary diif-
th
he stream floowrate, ρ itss density, Cp is the streeam ferrential equatiions, constitu
uted by 4n equations,
e was
w
sppecific heat, z is the axiaal position, U is the overall sollved numericcally.
heat transfer coefficient,
c A is the heat transfer areaa, V
iss the volumee and ν is a ratio betweeen volume and a
tu
ube length.

TH m TH UA
 H  T  T  (1)
t  H vH z  H VH C p, H H C

TC m TC U
UA
 C  T  T  (2)
t C vC z CVC C p ,C H C Figure 1: Discretized heat
h exchangger model.

Equations (1) and (2)) were discrretized throuugh

th
he finite diffference methhod, resultinng in Equatio
ons THE CONT
TROLLER CHARACT
C TERISTICS
(3
3) and (4), w where the suubscript i is connected to a
discretized teemperature anda n is the number of dis-
d Classical coontrol techniiques for sinngle input siin-
crretization cellls used, wheere i = 1,2,..,n. glee output (SIISO) system ms are based on analyticcal
toools, such as Laplace traansforms, Roouth criterioon,
dT
d H ,i etcc., as well ass on graphicaal tools, suchh as Bode an nd
  a1  a2  TH ,i 1   a1  a2  TH ,i Nichols plots. Neverthelesss, for the conntroller desig gn
dt
(3) of Multiple Inp put Multiple Output (MIIMO) system ms,
 a2TC ,n i  a2TC ,n i 1 thee use of suuch techniquues may be troublesom me.
Moodern controol design tecchniques bassed on a statte-
spaace model area generally more suitabble for MIM MO
dT
d C ,n i 1
 a4TH ,i 1  a4TH ,i   a3  a4  TC ,n i sysstems (Leviine, 2010). In this woork, the LQ QR
dt fraamework is applied to design
d the coontroller in ana
(4)
expperimental heat
h exchanger networkk. Disturbance
  a3  a4  TC ,n i 1
atttenuation corrresponding to inlet tem mperatures are
a
connsidered as part
p of the performance specificationns;
In additionn to Equationns (1) and (2)), the dynamiical furrthermore, sttream outlet temperaturees must folloow
behavior of sstream bypasses in the heat exchannger steep changes inn the setpoinnts with smaall steady-staate
was
w also moddeled and disscretized, whhich resultedd in errrors and with a smooth transient ressponse, whicch
Equations
E (5)) and (6), where
w the subbscript by inndi- does not cause actuators saaturation.
caates bypass aand f is the bypass
b fractioon. It is important to compaare the propoosed approacch
in relation to other
o typicall control alteernatives, succh
dT
d Hby ,i as the dynamicc decoupler and classicaal PID controol,
 a5THby ,i 1  a5THby
H ,i (5)
dt so that one caan verify if the LQR iss also a goo od

Braziliaan Journal of Cheemical Engineerinng Vol. 33, No. 001, pp. 133 - 143, January - Marcch, 2016
1336 F Delatore, L. F. Novazzi,
F. N F. Leonnardi and J. J. da Cruz
C

ch
hoice. In thee next three sections,
s a shhort descripttion Although the
t techniquue theoreticaally is able to
of these typical alternativees and of LQQR is presented. com mpletely eliiminate interractions amoong variablees,
in practice th his decouplinng may nott occur com m-
a) Dynamic Decoupler
D pleetely, since the techniquue is based on a proceess
moodel, which is not an exxact represenntation of reaal-
In multivaariable systeems, the inpput and outtput ityy. The decouupling systemm is usually not used onn a
variables are typically cooupled. The direct couplling staand-alone baasis. After applying
a thee decoupler, a
iss the influennce of the input ui on itss correspondding scaalar design technique
t is normally used to designna
ou utput yi. Thee cross-couppling is the influence of the PID D controller..
in
nput uj on other outputs, rather thann the output di-
reectly related to it. The diirect couplinng is the desiired
foorm of the couplings thaat appear in systems
s since it
alllows it to be
b controlledd in a simpleer form (Oguun-
naike and Rayy, 1994). Thhe difficulty is i found in sys-
s
teems with crooss-couplingg, because thhey cause intter-
acctions in conntrol loops.
The techniique of variaable decouplling aims to re-
du uce the effeects of existiing cross-coouplings amoong
variables, alloowing the tunning of the controllers
c too be
made
m in a deccentralized manner
m by scaalar controlleers.
The
T decouplinng can be acchieved by aadding a prooper
coompensatingg transfer maatrix betweenn the controller
ou utput and thhe input off the plant, as depictedd in
Figure
F 2 (Oguunnaike and Ray, 1994).
Based on F Figure 2, onee can write: Figure 2: Dynamic decoupling system.
s

Y1  s    G11  G12GI 2 U1  s  b) PID Controol

(7)
  G11G
GI1  G12 U 2  s  When subjjected to bypassb manipulation, thhe
strream outlet temperature
t in a heat exxchanger con-
Y2  s    G21  G22GI 2 U1  s  neccted directlyy to this byypass presennts a lead laag
(8) ressponse with a small tim me delay (Noovazzi, 2007 7).
   G21
2 GI1  G22 U 2  s  Rootea and Marrchetti (1997)) showed thaat integral coon-
trool is the best option to co ontrol the temmperature inn a
Let the traansfer functio
ons GI1 and GI
G 2 be: heaat exchangerr with bypassses, as indiccated in Equua-
tioon (13), wherre gC is the controller
c traansfer functio
on
andd KI is the inntegral gain.
G 
G 1    12 
GI (9)
 G11  KI
gC  s   (133)
s
G 
G 2    21 
GI (10)
 G22  These authoors proposed d a tuning foormula for thhe
inttegral controoller that deepends only on the initiial
The new transfer funnctions of thhe system, now n tem
mperature jum mp K1 of a step
s responsee in the bypaass
without
w the innteractions beetween the variables,
v are: andd on processs dead time αd, as followss:

  G G  0.739
Y1  s   G11   12 21   U1  s  (11) KI  (144)
K1  d
  G22  
c) LQR Contrrol
  G G 
Y2  s   G22   12 21   U 2  s  (12)
  G11   Let the systtem be givenn in state-spacce form as:

Brazilian Jou
urnal of Chemicaal Engineering
 Multivariable Optimal Control of a Heat Exchannger Network witth Bypasses 137
1

d In this workk, an identiccal penalty iss proposed for

f
x  t   Ax  t   Bu  t  alll states to prrevent saturaation of any internal statte.
dt
d
(15) Duue to the sym mmetry of the t problem and since thhe
y  t   Cx  t  vallues have thhe same chaaracteristics, controls weere
equually penalizzed too. By appropriately
a y choosing thhe
where
w x is thee state vectoor, u is the innput vector, y is pennalties, whicch in this caase means too find just an a
he output veector and A, B, C are coonstant matriices
th unnique constannt value as a common factorfa in the Q
with
w approprriate dimenssions. The optimalo conttrol maatrix, not only can a dessired time reesponse geneer-
laaw that minim mizes the perrformance inndex: allly be obtaineed, but also a satisfactorry tradeoff be-
b
tw
ween speed of o response and control effort. If thhe
 staate vector caannot be fullly measured,, the feedbacck
J  0
 x  t T Qx  t   u  t T Ru  t   dtt
  (16) looop using the LQR gain matrixm K can be built usin ng
a state
s observeer as shown in Figure 3, where xSP are a
where
w Q = QT ≥ 0 and R = RT > 0 arre, respectiveely, thee setpoints foor all states.
th
he state and control weigghting matricces, is givenn by
th
he linear state feedback:

u  t    Kx  t  (17)

fo
orming the so-called linear quaddratic regulaator
Levine, 20100). The state feedback maatrix K is givven
(L
by
y:

Figgure 3: Feedback system m with statee observer an

nd
K  R 1 BT P (18)
staate setpoints included in the
t loop.
where
w P = PT ≥ 0 is thee solution off the follow
wing
Riccati
R Algebbraic Equatioon: A more com mmon situation occurs w when setpoinnts
aree defined no ot for the sttates x but forf the outp put
vecctor y. In thhis way a modified
m struucture can beb
PBR 1BT P  Q  0
AT P  PA  P (19) useed, where th he error betw ween the settpoint and thhe
plaant output is integrated inn order to inccrease the gaain
Since the LLQR ensuress the minimiization of a cost c at low frequenccies, as show wn in Figure 4 (Delatore et
fuunction regaardless of thee Q and R values, the soolu- al., 2009; Delaatore et al., 2010).
tiion is alwayys optimal in i the matheematical sennse.
Thus,
T they arre the tuningg parameterss of the desiign.
These
T matrices impose penalties
p on the states anda
co ontrols and can
c be used in diagonal form,f i.e., wiith-
ou ut a penalty cross betweeen variables. If one wantss to
give importannce only to thhe states assoociated with the
ou utputs y(t) aand their inttegral z(t), itt is possiblee to
reewrite the maatrix Q as:

C T Q y C 0 
Q  (220)
 0 Qz 
Figgure 4: Feeddback system m with state obbserver, outpput
a z are expplicitly includded
uch that the variables y and
su settpoints and in
ntegrator blocck included in
i the loop.
n the cost fuunction, wherre Qy penaliizes the outp
in puts
an
nd Qz penaliizes the integgral of the ouutputs. Thus the Extra statess are thus ad
dded to the system,
s givin
ng
co
ost function can be writteen as: rise to an extennded state xa:


0  y Qy y    z     x(t ) 
T T
J Qz z  uT Ru dt (221) xa (t )   (222)

 z (t ) 

Braziliaan Journal of Cheemical Engineerinng Vol. 33, No. 001, pp. 133 - 143, January - Marcch, 2016
1338 F Delatore, L. F. Novazzi,
F. N F. Leonnardi and J. J. da Cruz
C

where:
w The experim mental HEN is constituteed by two heeat
excchangers (HE E1 and HE22) and has tw
wo hot (H1 an nd
d H22) and one cold stream m (C1). Figuure 6 shows a
z (t )  y (t )  Cx(t ) (223) piccture of the physical
p pilott HEN used in this work..
dt
d

The state feedback maatrix Ka cann thus be corrre-

sp
pondingly paartitioned as::

Ka   K x Kz  (224)

By taking into accounnt Equations (22) and (223),

th
he state modeel in Equatioon (15) can be
b rewritten as:
a

d
xa (t )  Aa xa (t )  Ba u (t ) (225)
dt
d

where:
w

 A 0
Aa    (226) Fig
gure 6: Expeerimental setuup.
C 0 
The manipuulated inputts, controlledd outputs an
nd
B dissturbance vaariables of the
t HEN too perform thhe
Ba    (227) anaalyses are lissted below:
0
Controlled variables:
v
Thus, the augmented matrices Aa and Ba are the  TCOUT1: inteermediate coold stream tem
mperature
new matrices used in the LQR
L project.  TCOUT2: cold stream outtput temperatture
Manipulated variables:
METHO
ODOLOGY  fc1: bypass valve
v positioon, heat exchhanger 1
 fc2: bypass valve
v positioon, heat exchhanger 2
Before preesenting the control desiign and its per-
p Disturbancee variables:
fo
ormance appplied to a HE EN, it is important to deffine  THIN1: hot stream
s 1 inlet temperature
th
he physical arrangementt of the excchangers, detter-  THIN2: hot stream
s 2 inlet temperature
mining
m the m
manipulated, the controlleed and the dis-
d  TCIN: cold stream
s inlet temperature
t
tu
urbance variaables of the experimentaal network con-  mC: cold strream flowrate
siidered in this work, as shown
s in Figgure 5 in a grid
g
diagram. Figure 7 prresents a sim
mplified pipinng and instrru-
meentation diag gram (P&ID)) for the piloot plant, wheere
hot streams H1 H and H2 are a indicatedd by red linnes
whhereas the cold
c stream C1 is indiccated by bluue
onnes. The basiic instrumenttation of thee plant encomm-
passes volum metric flowm meters (FT),, temperatu ure
traansmitters (T
TT) and presssure indicatoors (PI). Theere
aree also two veessels (V1 annd V2) used to heat wateer,
with electric reesistances. The
T heat exchhangers in thhe
nettwork, the connections,
c pumps (P11, P2 and P3)
andd vessels aree made of ANSI
A 304 staainless steel. It
is convenient
c to
t mention thhat the bypassses valves are
a
conntrolled by a signal rannging from 0 up to 10 V,
whhere 0 V rep presents a vaalve full shuut and 10 V a
Figu
ure 5: The HE
EN basic struucture. vallve full openn.

Brazilian Jou
urnal of Chemicaal Engineering
 Multivariable Optimal Control of a Heat Exchannger Network witth Bypasses 139
1

Some charracteristics of the shell annd tube heat ex- Table 3: Heeat exchangeer nominal conditions.
c
ch hangers HE1 and HE2 used in thhe experimenntal
HEN
H plant are presented ini Table 1. Both
B exchanggers Heat
Tin / ºC Tout / ºC U / W·m-2·K
K-1
have the sam me dimension ns and the tuubes are distrrib- eexchanger
hot cold hot cold
uted in a trianngular pitch. There is 1 pass
p in the tubbes HE1 55.0 28.5 53.4 30.0 420
an w a heat transfer areaa A
nd 1 pass inn the shell, with HE2 55.0 26.9 53.3 28.5 420
eq qual to 0.1 m2.
Table 2 prresents some nominal opeerational conndi- Three experrimental testss were perforrmed to evallu-
tiions in the HEN,
H for hoot streams H1H and H2 and a atee the control system: twoo of them invvolving a reggu-
co old stream CC1, where m stands for flowrate,
f Tin for lattory problem m, with dissturbances inn supply hot h
suupply tempeeratures and Tout for targeet temperaturres. strream H2 tem mperature, THT IN2, and inn cold streaam
Specific heat of hot and coldc streams was consideered floowrate mC, an nd one test involving
i a servo
s probleem
eq qual to 41800 J·kg-1·K-1. com mbined witth a regulaatory one, with setpoiint
Table 3 prresents nomiinal operatinng conditionss in chaanges in con ntrolled variiables TCOUTT1 and TCOUUT2
th
he heat exchhangers HE11 and HE2, including innlet andd disturbances on mC. Eiither in the reegulatory or in
an nd outlet tem
mperatures of o the hot annd cold stream ms. thee servo prob blem, changees were conssidered as steep
T overall hheat transfer coefficient U was obtainned
The funnctions. The amplitudes of o these stepp changes weere
th
hrough steaddy-state experiments and the same vaalue choosen to avoid bypass satturation and in such a waay
was
w considerred for both h exchangerss. It should be thaat the variatioons of the conntrolled variaables would be
b
pointed out thhat the two controlled
c vaariables, TCOUT1
O in the accuracy y range of thee measuremennt instrumentts.
an nd TCOUT2, aassume nominal operatinng temperatu ures
of 28.5 ºC andd 30.0 ºC, respectively.
RES
SULTS AND
D DISCUSSIION

The applicaation of the control struccture based on

o
LQQR and applied to the HEN
H is show
wn in Figure 8,
whhere a Luenbberger obserrver was useed in the loo op
forr state estim
mation and an
a anti-winddup subsysteem
waas added to the integral action. Thee weighting Q
andd R matricess were chosenn as:

Q  104. I 27 x 27 R   I 2 x 2

whhere I is the identity maatrix. It shouuld be pointeed

ouut that the controller has a single desiign parameteer,
whhich was cho osen as 10-4, through a trial
t and errror
proocedure, usinng numericall simulation. Since there is
Figure 77: P&ID of the
t experimental HEN. onnly one tuninng parameter, an experimental tunin ng
proocedure coulld also have been
b used.
Table 1:: Heat exchaanger charaacteristics. The target performancee was to achhieve a stead dy
staate decoupled system wiith a closed loop constaant
Variable Descrription Value
V Uniit tim
me in a range from 200 to 300 seconds. Althoug gh
Di tube diameter
d 0
0.006 m thee system could be tuned to achieve steady staate
NT number of tubes 7 eveen faster thhan that, thiss amount off time is sttill
LT tube leength 0
0.750 m sm
mall when compared
c too the dynam mics of moost
DS shell diameter
d 0
0.060 m
A area 0.1 m2 proocess equipmment’s. Besid des, the approoach has som
me
bennefits: by seetting a not so fast respponse, it maay
Table 2: Stream nomin
nal operating conditionss. inhherently preevent saturration of tthe actuatorrs,
proovide higherr robustness and reduce amplificatio on
Stream m / (kg·s-1) Tin / ºC
C Tout / ºC
C of sensor noisee.
H1 0.1544 55.0 53.4 The gain vaalue of the annti-windup subsystem
s was
w
H2 0.1544 55.0 53.3 takken as KT eqqual to 10, a value high enough
e to prre-
C1 0.1655 26.9 30.0 vennt numericall problems. Simulations
S u
using the HEEN

Braziliaan Journal of Cheemical Engineerinng Vol. 33, No. 001, pp. 133 - 143, January - Marcch, 2016
1440 F Delatore, L. F. Novazzi,
F. N F. Leonnardi and J. J. da Cruz
C

mathematical
m model prodduced a satissfactory perffor-
T OUT1 LQR
TC TCOUT2 LQR
L TCOUT1 PID T OUT2 PID
TC TCOUT1 De
ecoupling TCOUT2 Deco
oupling

31
mance
m for seetpoint changges and for disturbance re-
jeection with a control effoort compatible with the chhar- 30.5

accteristics of tthe bypass vaalves.

30

Temperature (ºC)
29.5

29

28.5

28
0 500 1000 1500
T
Time(s)

Figgure 9: Plotts of TCOUT responses under

u step diis-
turrbance on TH
HIN2.
LQR - fci LQR - fcci PID - fci PID
D - fci Decoupling - fci
f Decoupling - fci
1 2 1 2 1 2
10

Valve Openning (Volts)

Figure
F 8: LQR controller structure prooposed for HEN 6
ontrol.
co 5

4
As for the PID controlller and the ddynamic decou-
3
pler, the tuninng parameterrs were designed accordding
2
to
o Equation (13), with an integral gainn equal to 0..04.
1
In
n the next sections thee performancce of comm mon
co
ontrol strateegies (dynam
mic decoupleer and PID)) is 0
0 500
Time (s)
1000 1500

co
ompared to tthe LQR, thee proposed appproach usedd in
th
his work. Figgure 10: Conntrol effort of
o bypass vallves under diis-
turrbance in TH
HIN2.
a) Disturbance in Hot Sttream Inlet Temperatur
T re
TH
T IN2 It can be seeen from thee plots in Figgure 9 that thhe
conntrol systemm pushes thee controlledd temperaturres
To evaluaate the perfo ormance of the controlllers bacck to their steady-state
s nominal opeerating valuees,
with
w disturbannces in hot streams H2, a step changee of buut with differrent transien nt performannce, dependin ng
3.0 ºC was appplied to THIN2 IN at 500 s, i.e., temperatture onn which contrroller is usedd. The multi--loop PID con-
TH
T IN2 was inccreased to 58.0 ºC. Figuure 9 shows the trooller openedd both bypassses valves, in a smoo oth
reesponses of T TCOUT1 and TC
T OUT2 for thhe three contrrol- waay, as indicatted by the greeen lines in Figure
F 10. Thhe
leers, where thhe black, green and redd curves are re- LQ QR controlleer also put into
i action both
b bypassees,
laated to LQR R, PID and dynamic
d deccoupler, resppec- without saturattion, but in a more aggreessive way, as
tiively. Figuree 10 indicatees the controol effort of the shoown by the black
b lines in
i Figure 10. On the oth her
hree controlllers for bypaasses fc1 and fc
th f 2 and folloows hannd, the dynaamic decouppler actuatedd only one by-
th
he same coloor patterns of o Figure 9. As A indicatedd in pass, fc2, and fcf 1 was kept shut. This means
m that thhe
Figure
F 9, the plant was inn nominal operating conndi- dyynamic decouupler tried too control both TCOUT1 an nd
tiions until 5000 s, with TCT OUT1 and TC T OUT2 equall to TCCOUT2 by the manipulatio on of only one
o bypass. In
28.5 ºC and 330.0 ºC, respeectively. Unttil this time, the adddition to thee dynamic beehavior of TCOUT temperra-
by ypass valves were clossed, as seenn in Figure 10. turres observedd in Figure 9, it is importtant to speciify
When
W a disturrbance in TH
HIN2 is imposeed, both TCOUT1O a qquantitative performancee criterion too compare thhe
an nd TCOUT2 w will be affeected, indicaating a couppled diffferent controollers. In thiis work the Integral
I Absso-
syystem, as caan be noted in the grid diagram of the lutte Error or IA
AE (Ogunnaiike and Ray, 1994) is useed.
HEN
H (Figure 5) and in thhe responses in Figure 9, at Tabble 4 presentts the controlllers’ perform
mance based ono
500 s. IA
AE for LQR, PID P and the dynamic
d decoupler.

Brazilian Jou
urnal of Chemicaal Engineering
 Multivariable Optimal Control of a Heat Exchanger Network with Bypasses 141

Table 4: Controller performance under disturbance coupler controller, leading to a quick setpoint track-
in THIN2. ing and also disturbance rejection. This fact can be
noted in Table 5, where LQR presents the smallest
TCOUT1 TCOUT2
IAE among the three controllers.
LQR 169.6 268.1
PID 132.8 187.2 TCOUT1 LQR TCOUT2 LQR TCOUT1 PID TCOUT2 PID TCOUT1 Decoupling TCOUT2 Decoupling
Decoupling 202.5 364.9 30.5

By observing the performance of the Integral Ab- 30

solute Error for the three controllers (Table 4), one

can see that the PID has the best response for TCOUT1 29.5

Temperature (ºC)
and also for TCOUT2. However, even with a more 29
vigorous action, the LQR controller also showed a
good performance when compared to the PID. These 28.5

results suggest that the tuning parameter in Q, which

multiplies the identity matrix, and chosen as Q equal 28

to 10-4·I for the LQR design, seems to be high and

leded to a faster control action when compared to the 27.5
0 500
Time(s)
1000 1500

PID and dynamic decoupler.

It is worth noting that the PID and the decoupler Figure 11: Plots of TCOUT responses under setpoint
controllers have more tuning parameters than LQR, change and flowrate disturbance.
which has only one, as considered in this work.
LQR - fci LQR - fci PID - fci PID - fci Decoupling - fci Decoupling - fci
Since this value is the unique design parameter in the 10
1 2 1 2 1 2

controller, it is simpler to tune than the PID and the 9

dynamic decoupler. 8

7
Valve Oppening (Volts)

b) Servo and Regulatory Problem 6

5
To evaluate the servo problem in the experimental 4
HEN, the setpoints of TCOUT1 and TCOUT2 were re-
3
duced 0.5 ºC at a time t equal to 500 s, i.e., TCOUT1
2
and TCOUT2 were set to 28.0 ºC and 29.5 ºC, respec-
1
tively. Besides, to evaluate the servo problem and
0
also the regulatory problem, an increase in cold 0 500
Time (s)
1000 1500

stream flowrate mC of 0.010 kg/s was imposed at

1000 s. Figure 11 shows the responses of TCOUT1 and Figure 12: Control effort of bypass valves under set-
TCOUT2, where the black, green and red curves are point change and flowrate disturbance.
related to LQR, PID and dynamic decoupler, respec-
tively. Figure 12 indicates the control effort of the Table 5: Controller performance under setpoint
three controllers. change and flowrate disturbance.
By observing Figures 11 and 12, one can see that
TCOUT1 TCOUT2
until 500 s the plant was under nominal operating con- LQR 371.9 340.9
ditions, with TCOUT1 and TCOUT2 equal to 28.5 ºC and PID 431.8 374.3
30.0 ºC, and the bypasses valves completely closed. Decoupling 458.9 366.9
After the setpoint change at 500 s, the plots in
Figures 11 and 12 suggest a relatively smooth perfor- c) Disturbance in Cold Stream Flowrate mC
mance for setpoint tracking, as well as for disturb-
ance rejection, which was imposed at 1000 s. To evaluate a disturbance rejection in cold stream
This behavior is quantitatively indicated in Table flowrate mC, a variation from 0.165 kg/s to 0.135 kg/s
5, by using the IAE criteria. Since there is a setpoint was imposed at a time t equal to 500 s and then re-
decrease at 500 s, the bypass valves tend to open moved at 1200 s. The responses of TCOUT1 and TCOUT2
until 1000 s, when an increase in mC takes place, for the three controllers are shown in Figure 13, where
disturbing both TCOUT1 and TCOUT2, and then the the black, green and red curves are related to LQR,
bypasses begin to close. It can be observed that LQR PID and dynamic decoupler, respectively. As previ-
action is fast when compared to the PID and the de- ously mentioned in Table 3, setpoints of TCOUT1 and
Brazilian Journal of Chemical Engineering Vol. 33, No. 01, pp. 133 - 143, January - March, 2016
142 F. Delatore, L. F. Novazzi, F. Leonardi and J. J. da Cruz

TCOUT2 are equal to 28.5 ºC and 30.0 ºC. Figure 14
indicates the control effort of the three controllers,
using the same color patterns as in Figure 13.
It can be seen from the plots in Figure 13 and 14
that the LQR controller reacted fast, and opened both
bypass valves near 500 s, when the cold stream flow-
rate increased. When the flowrate in C1 returned to
its nominal operating value, at 1200 s, once more the
LQR was fast and drove the controlled variables to
28.5 ºC and 30.0 ºC. Table 6 presents the controllers
performance based on IAE for LQR, PID and the
dynamic decoupler.
TCOUT1 LQR TCOUT2 LQR
31
30.5
30
Temperature (ºC)
the LQR is slightly better than the PID. As for
TCOUT2, the decoupler has once more the best perfor-
mance and the LQR is much better than the PID.
Although the LQR was not the best option in this
test, it is still a reasonable choice, due to its tuning
simplicity.
CONCLUSIONS
In this work the use of a linear quadratic regulator
to control heat exchanger networks was studied. This
control technique is an alternative approach which
presents advantages in HEN control: the LQR con-
TCOUT1 PID TCOUT2 PID TCOUT1 Decoupling TCOUT2 Decoupling troller is easily designed and its performance can be
even better than more common control techniques,
such as PID and the dynamic decoupler. The aim of
the proposed approach is not to accomplish a better
performance than the one typically obtained with

29.5
model predictive control techniques. On the other
hand, the LQR can be designed just by using an ap-
29
proximate plant model and only one tuning parame-
28.5 ter, which can be iteratively chosen.
28
The approximate dynamic plant model used in the
design of the LQR was based on energy balance equa-
27.5
0 500 1000 1500 tions, which were linearized and their order reduced.
Time(s)
However, this dynamic model could also be obtained
Figure 13: Plots of TCOUT responses under step dis- directly from plant operation, in a more straightfor-
turbance in mC. ward way and probably with equivalent results.
LQR - fci LQR - fci PID - fci PID - fci Decoupling - fci Decoupling - fci
The LQR was applied to a lab scale heat ex-
10
1 2 1 2 1 2
changer network, constituted by two heat exchangers
9 with two hot and one cold stream. Manipulated vari-
8
ables were the bypass valve positions and the con-
7
trolled outputs were outlet temperatures. The LQR
Valve Openning (Volts)

6
tuning parameter was determined by simulation and
5
the controller performance was compared to the PID
4
and the dynamic decoupler. The controllers’ perfor-
3
mance was quantitatively assessed by the integral ab-
solute error and this index showed that the LQR per-
2
formed well both in regulatory and servo problems.
1

0
0 500 1000 1500
Time (s)
ACKNOWLEDGEMENTS
Figure 14: Control effort of bypass valves under dis-
turbance in mC. The authors wish to thank all the institutional
support of FEI University.
Table 6: Controller performance under disturbance
in mC.
REFERENCES
TCOUT1 TCOUT2
LQR 403.9 145.9
PID 431.3 374.3 Aguilera, N. and Marchetti, J. L., Optimizing and
Decoupling 160.5 132.9 controlling the operation of heat-exchanger net-
works. AIChE J., 44 (5), 1090-1104 (1998).
For the first controlled variable, TCOUT1, the de- Calandranis, J. and Stephanopoulos, G., A structural
coupling technique shows the best performance and approach to the design of control systems in heat

Brazilian Journal of Chemical Engineering

 Multivariable Optimal Control of a Heat Exchanger Network with Bypasses
exchanger networks. Comp. Chem. Eng., 12(7),
651-669 (1988).
Delatore, F., Cruz, J. J., Leonardi, F. and Novazzi, L.
F., Multivariable Control of a Heat Exchanger
with Bypasses. 11th IASTED, Cambridge, UK
(2009).
Delatore, F., Cruz, J. J., Leonardi, F. and Novazzi, L.
F., Multivariable Optimal Control of a Heat Ex-
changer Network (HEN) with Bypasses. 12th
IASTED, Banff, Canada (2010)
Giovanini, L. L. and Marchetti, J. L., Low-level
flexible structure control applied to heat ex-
changer networks. Comp. Chem. Eng., 27, 1129-
1142 (2003).
Glemmestad, B., Mathisen, K. W. and Gundersen, T.,
Optimal operation of heat exchanger networks
based on structural information. Comp. Chem.
Eng., 20 (suppl.), S823-S828 (1996).
Glemmestad, B., Skogestad, S. and Gundersen, T.,
Optimal operation of heat exchanger networks.
Comp. Chem. Eng., 23, 509-522 (1999).
Gonzalez, A. H., Odloak, D. and Marchetti, J. L.,
Predictive control applied to heat exchanger net-
works. Chem. Eng. Processing, 45, 661-671 (2006).
Lersbamrungsuk, V., Srinophakun, T., Narasimhan,
S. and Skogestad, S., Control structure design for
optimal operation of heat exchanger networks.
AIChE J., 54 (1), 150-162 (2008).
Brazilian Journal of Chemical Engineering Vol. 33, No. 01, pp. 133 - 143, January - March, 2016
143
Levine, W. S., The Control Handbook. 2nd Edition,
CRC Press, Boca Raton (2010).
Linhoff, B., Townsend, D. W., Boland, D., Hewitt,
G. F., Thomas, B. E. A., Guy, A. R. and Marsland,
R. H., A User Guide on Process Integration for
the Efficient Use of Energy. Warwick Printing
Company Ltd., Birmingham (1982).
Luyben, W. L., Process Modeling, Simulation and
Control for Chemical Engineers. McGraw-Hill,
Singapore (1990).
Marselle, D. F., Morari, M. and Rudd, D. F., Design
of resilient processing plants - II. Chem. Eng.
Sci., 37(2), 259-270 (1982).
Mathisen, K. W., Integrated Design and Control of
Heat Exchanger Network. Ph.D. Thesis, Univ. of
Trondheim, Norway (1994).
Novazzi, L. F., Dynamics and Control of Heat Ex-
changer Networks. Ph.D. Thesis, Univ. of Campi-
nas, Brazil (2007). (In Portuguese).
Ogunnaike, B. A. and Ray, W. H., Process Dynamics,
Modeling and Control. Oxford University Press,
New York (1994).
Rotea, M. A. and Marchetti, J. L., Integral control of
heat-exchanger-plus-bypass systems. IEEE Int.
Conf. on Control Applic., Hartford, USA (1997).
Sun, L., Luo, X., Hou, B. and Bai, Y., Bypass selec-
tion for control of heat exchanger network. Chin.
J. of Chem. Eng., 21 (3), 276-284 (2013).