Original Paper

Measurement and Control

2019, Vol. 52(5-6) 526–539
Improved predictive functional control Ó The Author(s) 2019
Article reuse guidelines:
for ethylene cracking furnace sagepub.com/journals-permissions
DOI: 10.1177/0020294019842602
journals.sagepub.com/home/mac

Han Song1, Cheng-li Su1, Hui-yuan Shi1,2 , Ping Li1,2 and

Jiang-tao Cao1

Abstract
The objective of this paper is to show the design and application of pass temperature balance control system using an
improved predictive functional control method in eight 800 tone/year USC ethylene cracking furnaces. The advanced
pass temperature balance controller is developed using the proposed method and implemented in proprietary APC-ISYS
software, which is connected to Yokogawa distributed control system via an OPC server. The advantage of it lies in the
fact that the dynamics of pass temperature with nonlinearity and time delay are described by Takagi–Sugeno model and
transformed into time-varying extended state space model, and thus, the proposed controller can regulate pass tem-
perature based on the extended state space formulation. In addition, the control law with a linear iterative form, easily
applied to industrial process, is derived. The robust analysis for the set point, input disturbance and output disturbance
to the output verifies the ability of tracking and disturbance rejection of the proposed method. Application results from
an industrial furnace are shown to be markedly better in terms of lower variability in the outlet temperature of both the
passes compared to the current proportional–integral–derivative control scheme.

Keywords
Pass temperature balance, predictive functional control, ethylene cracking furnace, T-S model, advanced control

Date received: 16 November 2018; accepted: 12 March 2019

Introduction designed controllers. Based on the above control idea

Ethylene cracking furnace is a vital equipment to petro-
chemical industry in which a lot of important chemical
products, such as propylene, ethylene, butadiene, ben-
zene, dimethylbenzene, and so on, can be obtained by
the high-temperature cracking reaction.1 Because the
pass temperature of the radiation room in the cracking
furnace has a direct impact on the cracking effect and
the ethylene yield, it must be precisely controlled in
order to ensure the stable operation of subsequent
equipments for high quality and yield of the expected
products.
In general, the traditional pass temperature system
of ethylene cracking furnace based on proportional–
integral–derivative (PID)2 control strategy is mainly to
adjust the pass feed-in flow so that the pass temperature
is as identical as possible. Because the total feed-in flow
is required to remain constant in industrial production,
there is strong dynamic coupling between the pass feed-
in flow and pass temperature that leads to the poor
effect on pass temperature. Wang and Zheng3 proposed
a difference control technique (DCT), that is, difference
method,4,5 applied in an actual heating furnace. The
main idea of it is that the difference of pass temperature
between two streams is adjusted to be zero using
of two-passes balance, more passes can be controlled by
reiterative use of the DCT method. However, it seems
too boring to control such system at lots of iterative
times due to many passes. To solve the existing problem
of DCT method, Li et al.,6 Wang and Zheng,7,8 Wang
et al.9 and Qian et al.10 proposed a differences control
technique (DsCT), that is, average method. Its main
idea is to balance the pass temperature by the average
value of all pass temperature, without the need for the
DCT’s reiterative application frequently. But this strat-
egy using average method would lead the controller to
be large delay and insensitive. In addition, Wang et
al.11,12 also proposed a switching difference control
strategy (SDCT) for parallel streams temperature. Its
main ideal is to control two worst passes for all passes
1
School of Information and Control Engineering, Liaoning Shihua
University, Fushun, China
2
School of Automation, Northwestern Polytechnical University, Xian,
China
Corresponding author:
Cheng-li Su, School of Information and Control Engineering, Liaoning
Shihua University, Fushun 113001, China.
Email: sclwind@sina.com

Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License
(http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/
open-access-at-sage).
Song et al.
which can avoid the feed-in flow valves quite frequently
being regulated. Luan and colleagues13–15 proposed a
distributed deviation-based uniform temperature con-
trol method on feed heater. This scheme requires the
usage of the temperature information of adjacent
passes, which is more effective for such cases with too
many passes. But this method needs many computa-
tions of matrices.
As we know, the conventional PID control is used in
the above researches. However, the pass temperature is
complex with a number of problems, which is difficult
to be solved by conventional PID control, such as time
delay, multi-input and multi-output behavior and
strong nonlinear performance. Therefore, some
advanced control methods16–20 are proposed to over-
come these problems. Zhang et al.16 designed a steady
state energy pass balance control strategy to maintain
the pass outlet temperature to be same for multi-pass
fired heater. Rivas-Perez et al.17 used a modified Smith
predictor method to balance the pass outlet tempera-
ture and applied it in a crude oil preheating furnace. It
solved the problem of the real dominant time delay and
disturbance rejection on temperature uniformity of the
furnace. Li et al.18,19 proposed a pole placement control
method combined with a humanoid intelligent multi-
modality to balance the pass temperature of an ethylene
cracking furnace. In addition, some other advanced
control methods have also been used to control the pass
temperature uniformity, such as adaptive and predic-
tive method,1,20–23 coordinate control strategy of the
pass temperature and pass liquid level,24 multivariate
and fuzzy control method25–27 and so on. Ojasvi and
Kaistha28 proposed pass balancing control scheme
applying DMC as well as simple PID control. In addi-
tion, there are some other advanced control methods
such as sliding mode control,29 adaptive control30,31
and robust model predictive control.32,33
Different from previous study, in order to balance
pass temperature of the radiation room in the cracking
furnace, an improved predictive functional control
(IPFC) approach is proposed. The advantage of the
pass temperature balance control method is that the
dynamics of pass temperature with strong nonlinear
time delay, multi-input and multi-output behavior and
external disturbance is simultaneously solved. The pass
temperature with multivariable nonlinear character is
described by Takagi–Sugeno (T-S) model. Moreover,
the linear model is acquired by weighting a set of T-S
sub-model and then transformed into the extended
time-varying state space model that can compensate
steady state error where the state variables are con-
structed by the real-life output variables and current
and past incremental input and output variables.
Finally, the control law with a simple linear iterative
form is derived. The robust analysis for the set point,
input disturbance and output disturbance to the output
verifies that the proposed approach has ability of dis-
turbance rejection and tracking.
527
Using the IPFC approach, the advanced pass tem-
perature balance controller is developed and implemen-
ted in proprietary APC-ISYS software, which is
connected to Yokogawa DCS system via an OPC ser-
ver. Application results of the pass temperature balance
in the ethylene cracking furnace using proposed method
compared with recent literature and PID control are
presented. The fluctuation of the pass temperature has
been decreased from 658C to 618C. Therefore, the
developed pass temperature balance controller can
reach the improved control performance.
The rest of this paper is arranged as follows: IPFC
algorithm and its corresponding robust analysis are
detailed in ‘‘Improved predictive functional control’’
section. ‘‘Implementation case’’ section presents the
implementation case. Conclusions are presented in
‘‘Conclusion’’ section.
IPFC
Process description
In general, an industrial process can be described by
the following nonlinear equation with p inputs,
u 2 U  Rp , and q outputs, y 2 Y  Rq
yh ðk + 1Þ = Rh ðnðkÞ , uðkÞÞ ð1Þ
where h = 1, 2, ..., q and n(k) 2 Rq consists of current
and past outputs as well as past inputs as follows
 T
uðkÞ = u1 ðkÞ, u2 ðkÞ,    , up ðkÞ ð2aÞ
 T
nðkÞ = y1 ðkÞ , . . . , yq ðkÞ , u1 ðk  1Þ , . . . , up ðk  1Þ
ð2bÞ
where yh (k) =½yh (k) ,    , yh (k  ny,h +1),yh (k  ny, h )
and ul (k  1)= ½ul (k  1) , , ul (knu,l +1),ul (knu,l ),
l=1 , ... , p, where nu, l and ny, h refer to the order of the
lth input and hth output, separately.
Here, the multivariable nonlinear process described
by Equation (1) can be approximated by many of T-S
sub-models. The fuzzy rule Rhj is expressed as follows.
Rhj: If n1 (k) is Zhj, 1 , n2 (k) is Zhj, 2 ,.,nm (k) is Zhj, m ,
and u1 (k) is Zhj, m + 1 , u2 (k) is Zhj, m + 2 ,., up (k) is
Zhj, m + p , then
yhj ðk + 1Þ = mhj nðkÞ + z hj uðkÞ + uhj , j = 1 , 2,    , Rh
ð3Þ
where Zhj refer to the antecedent fuzzy set of the jth rule
and the hth output, mhj and z hj stand for the vectors of
consequent parameters, uP
P hj is a corrected item Rh, and
m = qh = 1 (ny, h + 1) + pl= 1 nu, l stands for the num-
ber of fuzzy rules of the hth output and antecedent vari-
ables, separately. Equation (1) can be weighted by the
following linear combination of a group of sub-model
(3).
528 Measurement and Control 52(5-6)

X
Rh
  I is an identity matrix. Pj (k  1) is a portion of covar-
y h ð k + 1Þ = vhj ðn, uÞ mhj nðkÞ + z hj uðkÞ + uhj iance matrix P(k  1) = ½P1 (k  1),    , PRh (k  1)T ,
j=1
e(k) = ½yh (k)  F(k  1)T u(k  1) is the residual error,
ð4aÞ and W(k  1) is a non-negative weighted coefficient.
bhj ðn, uÞ
vhj ðn, uÞ = R ð4bÞ
Ph Multivariable predictive functional control
bhj ðn, uÞ
l=1
Equation (4a) is rewritten as follows
X
Rh
vhj ðn, uÞ = 1, 04vhj ðn, uÞ \ 1 ð4cÞ ^ h nðkÞ + ^z h uðkÞ + ^uh
yh ðk + 1Þ = m ð7Þ
j=1
Q Qp where
where bhl (n, u) = mk = 1 lZhl, k (zk ) i = 1 lZhj, i (ui ) is the
Gaussian membership function of the antecedent vari- X
Rh
ables and P stands for the fuzzy operator. ^ h ðkÞ =
m vhj ðn, uÞmhj ð8aÞ
j=1

Online identification of consequent parameters X

Rh
^ h ðkÞ =
h vhj ðn, uÞz hj ð8bÞ
To eliminate the effect of unknown disturbances and j=1
model mismatch, the consequent parameters of T-S X
Rh
models are identified online using the following recur- ^uh ðkÞ = vhj ðn, uÞuhj ð8cÞ
sive weighting least square approach.34–36 Equation (4) j=1
is expressed as follows
Taking a difference operation D = 1  z1 on both
yh ðkÞ = Fðk  1ÞT u ð5Þ sides of Equation (7), it can obtain

where ^ h DnðkÞ + ^z h DuðkÞ + D^uh

Dyh ðk + 1Þ = m ð9Þ

Fðk  1Þ = ½vh1 uðk  1Þ ,    , vhRh uðk  1ÞT To eliminate steady state error, the incremental state
^ is given as
space vector DX(k)

DX^ðkÞ =
     T
Dy1 ðkÞ ,    , Dy1 k  ny, 1 ,    , Dyq k  ny, q , Du1 ðk  1Þ ,    , Du1 ðk  nu, 1 Þ ,    , Dum ðk  nu, m Þ

Then, the fuzzy models can be transformed into the lin-

T
u = ½u1 , u2 ,    , uRh  ear time-varying augmented state space model as follows
uðk  1Þ =
2     3 DX^ðk + 1Þ = A^ðkÞDX^ðkÞ + B^ðkÞDuðkÞ + DHðkÞ
y1 ðk  1Þ , . . . , y1 k  ny, 1  1 , . . . , yq k  ny, q  1 ,
6 7 DyðkÞ = C^ðkÞDX^ðkÞ
4 u1 ðk  2Þ , . . . , u1 ðk  nu, 1  1Þ , u2 ðk  2Þ, . . . , 5
  ð10Þ
up k  nu, p  1 , u1 ðk  1Þ , . . . , up ðk  1Þ, 1
2 3T
uj = 4mhj, 1 , . . . , mhj, ny, 1 , mhj, ny, 1 + 1 , . . . , m P
q , . . . , mhj, m , zhj, 1 , . . . , zhj, p , uhj 5
hj, ðny, h + 1Þ
h=1

where F(k  1) is the regression vector and u is referred

to the vector of consequent parameters that can be where
identified as follows 2 3
m
^ 1, 1 m
^ 1, 2     m
^ 1, m1 m
^ 1, m
Wðk  1ÞPj ðk  1ÞFðk  1Þ 6 7
uj ðkÞ = uj ðk  1Þ + sðkÞ 6 1 0     0 0 7
1 + Wðk  1ÞFðk  1ÞT Pðk  1ÞFðk  1Þ 6 7
6 0 1     0 0 7
6 7
ð6aÞ 6 . .. .. .. 7
6 .. .     . . 7
6 7
Wðk  1ÞPj ðk  1ÞFðk  1ÞFðk  1ÞT Pðk  1Þ A^ = 6 7
Pj ðkÞ = Pj ðk  1Þ  6m m     m ^ q, m 7
m
1 + Wðk  1ÞFðk  1ÞT Pðk  1ÞFðk  1Þ 6 ^ q, 1 ^ q, 2 ^ q, m1 7
6 0 0  1 0  0 0 7
6 7
ð6bÞ 6 7
6 0 0  0 1  0 0 7
4 5
where the initial values of it is set as u(0)T = sI and .. .. .. ..
. .     . . m3m
P(0) = g 2 I, where s is quite small positive real number,
g is the vector of quite a big positive real number, and
Song et al. 529

2 3
^z1, 1 ^z1, 2     ^z1, p linearization strategy is adapted in the whole predictive
6 0 0   0 7 
horizon. Based on Equation (13), the hstep ahead out-
6 . .. .. 7
6 . 7 put predictive vector is presented as follows
6 . . . 7
6^ ^zq, 2 7
6 zq, 1
6     ^zq, p 7
7    
y k + hjk  = C^1 A^1 h X^ðkÞ + C^1 A^1 h1 B^1 DuðkÞ +  
B=6
^
6 0. 0   0 7
6 . .. .. 7
7    
6 . . . 7 C^1 A^1 B^1 Du k + h  2 + C^1 B^1 Du k + h  1
6 7
6 1 0   0 7 
h1
6 7 + C^1 A^1 H ^ 1 +    + C^1 A^1 H
^ 1 + C^1 A^1 H
^1
4 0 0   0 5
.. .. .. ð14Þ
. . . m3p
2 3 Using the idea of PFC, the control law can be
1 0       0
60 0  1 0          07 acquired by linearly weighing a number of basic func-
C^ = 6
4 ... .. .. 7 tions. Here, we adopt the step basic function. The con-
.  .5 trol law is
0 0    1 0  0 q3m  
 T u k + h = m = uðkÞ ð15Þ
H= u^1 0   ^
uq 0   0 m31
Equations (14) and (15) can be transformed into
The output variable y(k) is extended into the state 
 as follows    
variable DX(k)  = C^1 A^1 h X^ðkÞ + C^1 A^1 h1 +    + A^1 + I
y k + hjk
 T
X^ðkÞ = DX^ðkÞ, yðkÞ ^1
3 B^1 DuðkÞ + H

h h 1
where ^
= C1 A1 XðkÞ + C1 A^1  I
^ ^ ^ A^1  I
yðkÞ = ½ y1 ðkÞ  yq ð k Þ  T ^1
3 B^1 DuðkÞ + H
Note that ð16Þ
yðk + 1Þ  yðkÞ = C^ðkÞA^ðkÞDX^ðkÞ To handle the problem of imprecision of model and
ð11Þ
+ C^ðkÞB^ðkÞDuðkÞ + C^ðkÞDuðkÞ some unknown disturbances, the predictive error is usu-
ally employed to modify the vector of predictive output
Putting together Equations (10) and (11) leads to the as follows
extended state space model as follows
eðkÞ = yðkÞ  yðkÞ ð17Þ
" # " #" #
DX^ðk + 1Þ A^ðkÞ 0m3q DX^ðkÞ where y(k) and y(k) are the actual and predictive output
=
yð k + 1Þ C^ðkÞA^ðkÞ Iq yðkÞ vector of model, separately.
" # " # Then, the vector of modified predictive output at
B^ðkÞ DHðkÞ
+ DuðkÞ + time k + h is as follows
C^ðkÞB^ðkÞ C^ðkÞDHðkÞ    
" #  = y k + hjk
yc k + hjk  + eð k Þ ð18Þ
  DX^ðkÞ
yðkÞ = 0q3m , Iq
yðkÞ The reference trajectory, which is a predefined
path we hope to track the expected set-value, is defined
ð12Þ
as
Moreover, Equation (12) is transformed as   
yr k + h = cðkÞ  ah ðcðkÞ  yðkÞÞ ð19Þ
X^ðk + 1Þ = A^1 ðkÞX^ðkÞ + B^1 ðkÞDuðkÞ + H
^ 1 ðk Þ 2 3
yr, 1 (k + h)
2
c1 (k)
3
yðkÞ = C^1 ðkÞX^ðkÞ 6 yr, 2 (k + h)
7 6 c2 (k) 7
where yr (k + h)  =6 6 ..
7 6
7, c(k) = 6 ..
7
7,
ð13Þ 4. 5 4. 5
" # 
^ yr, q (k + h) cq (k)
A(k) 0m3q  T
where A^1 (k) = ^ ^ , B^1 (k) = yðkÞ = y1 ðkÞ y2 ðkÞ    yq ðkÞ , and

ah =
C(k)A(k) Iq
      
^
B(k) DH(k) diag½a1 h a2 h    aq h , where ch(k) is the expected set-
^ B(k)
^ , ^ 1 (k) =
H ^ , and C^1 (k) = value, yh (k) is the hth actual process output at time
C(k) C(k)DH(k)
Ts =Tr, h
½0q3m , Iq . instant k, ai = e , Ts is the sampling time, and
The extended state space model (13) used the predic- Tr, h is the time constant of the reference trajectory.
tive model to predict the output of process. X(k 
^ + hjk) Generally speaking, the following performance index

is the hstep predictive state output. A multiple-step is used to obtain the predictive control law of system
530 Measurement and Control 52(5-6)

X
H2
    2 The control law is
Jp = yr k + h  yc k + hjk

Q
h = H1
ð20Þ DuðkÞ = Kr Yr  Kx X^ðkÞ  Ke e ð25Þ
X
Hu
  2
+ Du k + h  1 R
where
h = 1
T 1 T
Kr = R + B^1 FT2 QF2 B^1 B^1 FT2 Q
where Q . 0, R . 0, H1 and H2 are the lower and
upper limit of prediction horizon separately, and Hu is T 1 T
control horizon. Kx = R + B^1 FT2 QF2 B^1 B^1 FT2 QF1
Based on the principle of the single-value optimiza- T 1 T
tion, the performance (Equation (20)) is simplified Ke = R + B^1 FT2 QF2 B^1 B^1 FT2 Q

Jp = kyr ðk + HÞ  yc ðk + HjkÞk2Q + kDuðkÞk2R ð21Þ For notational simplicity, the robustness of the sys-
tem is verified for single-input and single-output (SISO)
The incremental control input at current time is case. The process is described by Equation (1) and then
derived by substituting Equations (17)–(19) into is transformed into Equation (7) based on T-S models.
Equation (21) as follows The model (Equation (7)) can be expressed as the trans-
T T 1 fer function interpretation. The transfer function of the
DuðkÞ = R + B^1 FT2 QF2 B^1 B^1 FT2 Q actual process is Bn (z)=Ad (z) and the process model is
h  i B^n (z)=A^d (z). The model order is n and the extended
^1
I  aH ðyðkÞ + cðkÞÞ + yðkÞ  F1 X^ðkÞ  F2 H ^
state variable DX(k) is 2n. The control gain vector of
ð22Þ state feedback is
H H 1
where F1 = C^1 A^1 and F2 = C^1 (A^1  I)(A^1  I) . Kx = ½k1 , k2 ,    , kn , kn + 1 , kn + 2 ,    k2n 
The optimal control law at current time is Kr = ½ks1 , ks2 ,    , ksH  ð26Þ
uðkÞ = DuðkÞ + uðk  1Þ ð23Þ Ke = ½ks1 , ks2 ,    , ksH 

where u(k) = ½u1 (k) ,    , up (k)T . Define the two corresponding polynomial functions
The detailed algorithm flow of IPFC using T-S
model is as follows: NðzÞ = k1 + k2 z1 +    + kn z(n1)
ð27Þ
MðzÞ = 1 + kn + 1 z1 +    + k2n1 z(n1)
1. Set the initial parameters: the definite matrix
P(0), the initial consequent parameters u(0) of Then from Equation (25), the corresponding polyno-
T-S model, the initial control input u(0), the mial form in Figure 1 is viewed as follows
parameters Ts , Tr , and H of IPFC;    
2. Read the process output y(k) and the desired 1  z1 MðzÞuðzÞ = Ws yr ðzÞ  1  z1 NðzÞYðzÞ
output c(k);  k2n y^ðzÞ  Ws eðzÞ ð28Þ
3. According to Equation (6), estimate the conse-
quence parameters of T-S model: mhj, 1 ,    , The block diagram with respect to Equation (28) is
mhj, m , zhj, 1 ,    , zhj, p , uhj , j = 1 ,    , Rl , depicted in Figure 1, in which yr (z) is the set-value,
h = 1,    , q. Di (z) is the input disturbance, and Do (z) is the output
4. According to Equation (8), calculate disturbance. The tracking of the set point, the output
^ h, m , ^zh, 1 ,    , ^zh, p , and ^u1 ,    , ^uq ,
^ h, 1 ,    , m
m disturbance, and the input disturbance to the output is
then calculate the matrix A, ^ B, ^ C, ^ and H;
^ discussed to verify the robustness. First, the transfer
5. If k410, then loop to step 6, or loop to step 7; function is presented from the set point to the output
6. Set u(k) = u(0) then loop 8;
yðzÞ Ws B^n ðzÞAd ðzÞ
7. Calculate the optimal control input u(k) using TðzÞ = = ð29Þ
Equation (23), loop to step 8; yr ð z Þ Gð z Þ
8. Calculate yc (k + 1) and implement u(k) to the
process;
9. Let k = k + 1, return to step 2.

Robust analysis
^ 1 = 0, Equation (22) is rewritten as
Setting the noise H
T 1 T  
DuðkÞ = R + B^1 FT2 QF2 B^1 B^1 FT2 Q  yr  e  F1 X^ðkÞ

ð24Þ Figure 1. The block diagram of the polynomial form.

Song et al. 531

where Implementation case

   Process description
GðzÞ = 1  z1 MðzÞAd ðzÞA^d ðzÞ + A^d ðzÞBn NðzÞ
+ A^d ðzÞBn k2n  A^d ðzÞBn Ws + Ws B^n ðzÞAd ðzÞ The process of the USC ethylene cracking furnace is
presented in Figure 2. There are eight 800 tone/year
USC ethylene cracking furnace (F1110-F1180) in prac-
From Equation (24) together with Equation (12), we
tice. The structure of radiation room for USC cracking
can easily acquire that Ws = k2n . Therefore, the result is
furnace is designed as two chambers labeled A and B,
lim TðzÞ = 1 ð30Þ respectively. TI112018 and TI112019 are referred to
z!1
temperature measurements of chambers A and B. Each
From Equation (30), the system can track a constant chamber consists of two groups of pipes. Each pipe is
set value without steady error. composed of 44 furnace tubes. For chamber A of the
Second, the transfer function is presented from the cracking furnace (F1120), the feedstock (NAP) is
output disturbance to the output divided into two passes (FIC112102 and FIC112302)
  and sent into the convection room of F1120 to be pre-
yðzÞ 1  z1 MðzÞAd ðzÞA^d ðzÞ heated. The first preheated temperature (HC
Soutput ðzÞ = = PREHEAT1) is about 183°C and the second (HC
D o ð zÞ Gð z Þ
PREHEAT2) about 396°C. The preheated dilution
ð31Þ
steam (DS) is injected into each pass and then the
It is mixed substance is sent to radiation section of F1120
where a number of important chemical products, such
lim Soutput ðzÞ = 0 ð32Þ as propylene, ethylene, butadiene, benzene, dimethyl-
z!1
benzene, and so on, can be obtained by the high-
Thus, the constant output disturbance can be temperature cracking reaction. For the pass flow sys-
rejected. tem, DS and pass flow are mixed using a double
Third, the transfer function from the input distur- closed-loop ratio controller where controlled variable
bance to the output is (CV) is the set-point of feed flow. They adopt PID con-
  trol method. Likewise, there is the same process flow
yðzÞ 1  z1 B^n ðzÞAd ðzÞMðzÞ for other two branches (FIC112202 and FIC112402) of
Sinput ðzÞ = = ð33Þ
D i ð zÞ Gð z Þ chamber B. The final objective is to adjust the pass feed
flow such that the pass temperature (TIC112126A and
It can be obtained as follows
TIC112326A in the south, and TIC112226A and
lim Sinput = 0 ð34Þ TIC112426 in the north) is as identical as possible.
z!1

Therefore, the constant input disturbance can be

rejected.
Through the above roust analysis, we can conclude
that the proposed approach has a better capability of
tracking and disturbance rejection.
Pass temperature balance control
To reduce the influence of the ethylene cracking furnace
to the consequence units, the total feed flow has to
remain unchanged. Also, there are two groups of

Figure 2. Process of the USC ethylene cracking furnace.

532
Figure 3. Pass temperature balance control.
Figure 4. Principle diagram of IPFC-PID transparent control.
tubular pipes (two passes), and the pass temperature is
balanced by the change of the pass feed flow. The pass
temperature balance control between the two passes
becomes a multi-input multi-output control problem.
Figure 3 presents a method where two outlet tempera-
tures serve as the CVs, and the set-values of two pass
flows are considered as the manipulate variables (MVs).
FICi , i = 1, 2 are two feed flow controllers (PI control)
of two inlet flows, and TIi , i = 1, 2 are two thermocou-
ples. The expression of the pass temperature balance is
presented as follows
X
2 can rapidly restrain the unknown disturbances for pro-
min J = jTi  Tj ð35Þ
DFi
i=1
P
2 performance. Therefore, a transparent architecture,
s:t: DFi = 0 ð36Þ
i=1
Fi, min 4Fi 4Fi, max ð37Þ
DFi, min 4DFi 4DFi, max ð38Þ
where Ti (i = 1,2) is the pass temperature; T is referred
to the average temperature; and the constraints of the
two pass flows, that is, the lower and upper limits of the
adjustments (DFi, min , DFi, max ), are applied in the pass
temperature balance control. Also the fluctuation of the
Measurement and Control 52(5-6)
total feed flow should remain constant. The adjust-
ments of two pass flows (DFi , i = 1, 2) are the outputs
of the pass temperature balance controller (IPFC).
Control strategy
The pass temperature of the ethylene cracking furnace
is a complex process dynamic with multi-input and
multi-output, nonlinearity and large delay as well as
many of unknown disturbances. The traditional PID
controllers have a good control performance, which
cess, but PID controller has been proven incapable and
insufficient because of the aforementioned complex
that is, IPFC-PID, is used to achieve the temperature
balance control as shown in Figure 4. The advanced
control level is the IPFC controller which guarantees
the capability of tracking and robustness of process,
which the conventional control level in the traditional
PID controller that rapidly eliminates the unknown dis-
turbances. The advanced strategy implements better
performance than the traditional PID control.
The control block diagram is shown in Figure 4, in
which yr1 and yr2 are the outputs of the reference
Song et al.
Figure 5. Hardware system architecture.
trajectory and y1 and y2 are actual outputs for pass
temperature, respectively. The actual process outputs
are measured by the field instruments. These output
signals are sent to the IPFC controller through trans-
mitters. Then IPFC gives the set-value of feed flow
PID (PID1andPID2). The outputs of IPFC are com-
puted by Equation (23) and PID controller adjusts the
control parameters to track the expected value. The
output of PID control is sent to each actuator
(Actuator1 and Actuator2) to control the controlled
process of each loop (Process 1 and Process 2). The
IPFC controller proposed in this paper adopts a gener-
alized process, including PID controller, actuators and
processes.
Hardware system architecture of advanced control
system
The control system for the ethylene cracking furnace in
the PetroChina Sichuan Petrochemical Company is
CENTUM VP distributed control system (DCS) made
by Yokogawa Company.1 A control loop is constructed
by field control station (FCS). The basic level uses DCS
to achieve traditional PID control because of its capa-
bility of unknown disturbance rejection. The hardware
architecture of system is presented in Figure 5. The
upper computer adopts proprietary APC-ISYS soft-
ware, while the lower computer is CENTUM VP DCS
system. The ethylene cracking furnace is connected with
the DCS system through a cable. The communication
between the upper computer and the lower computer is
achieved by OPC server. The proposed IPFC method is
written in VB-Script of the upper computer. Here, the
PID control is used in conventional control mode. In
other words, the advanced control loop is open. The
advanced controller is joined when the switching logic
between the upper computer and the lower computer is
triggered.
Industrial implementation results
The pass temperature balance control system using the
proposed control algorithm IPFC has been in
533
operation since 2014 in eight 800 k tone/year ethylene
cracking furnaces in PetroChina Sichuan
Petrochemical Company.
To reach a good control performance, T-S fuzzy
model of pass temperature is structured through
Equation (3) using the actual process data. G-K clus-
tering approach37 is used to acquire the premise fuzzy
sets and its corresponding parameters. The conse-
quence parameters are online identified by Equation
(6). The identified T-S model of pass temperature is
R1: If y1(k) is Z11, y2(k) is Z12, u1(k) is Z13, and u2(k) is
Z14, then
y1 ðk + 1Þ = a111 y1 ðkÞ + a112 y2 ðkÞ + b111 u1 ðkÞ + b112 u2 ðkÞ
y2 ðk + 1Þ = a121 y1 ðkÞ + a122 y2 ðkÞ + b121 u1 ðkÞ + b122 u2 ðkÞ
R2: If y1(k) is Z21, y2(k) is Z22, u1(k) is Z23, and u2(k) is
Z24, then
y1 ðk + 1Þ = a211 y1 ðkÞ + a212 y2 ðkÞ + b211 u1 ðkÞ + b212 u2 ðkÞ
y2 ðk + 1Þ = a221 y1 ðkÞ + a222 y2 ðkÞ + b221 u1 ðkÞ + b222 u2 ðkÞ
R3: If y1(k) is Z31, y2(k) is Z32, u1(k) is Z33, u2(k) is Z34,
then
y1 ðk + 1Þ = a311 y1 ðkÞ + a312 y2 ðkÞ + b311 u1 ðkÞ + b312 u2 ðkÞ
y2 ðk + 1Þ = a321 y1 ðkÞ + a322 y2 ðkÞ + b321 u1 ðkÞ + b322 u2 ðkÞ
ð39Þ
where the Gaussian membership functions of the fuzzy
set Z1j , Z2j , and Z3j , j = 1, 2, 3, for the pass temperature
is shown in Figure 6.
The fuzzy model (Equation (39)) is employed in the
advanced pass temperature controller (IPFC). The
parameters of it are H = 15, Tr = 0:5s, Ts = 1s,
Q = diag(0:9, 0:9, 0:9, 0:9, 1, 1), and R = diag(1, 1) that
are acquired by repeated offline test.
The parameters of the corresponding Su and Li38 are
H = 15, Tr = 0:5s, Ts = 1s, Q = diag(0:9, 0:9, 0:9, 0:9),
and R = diag(1, 1).
Before the advanced controller is operated, tradi-
tional PID controller of each loop had to be adjusted
separately. The corresponding parameters are acquired
by Cohen-Coon tuning approach.39 Based on these
tuning parameters, the engineer adjusts the parameters
again by his abundant operational experience. The
parameters of feed flow controllers of two passes are
Kp1 = 0:2, Ki1 = 0:013, Kd1 = 0; Kp2 = 0:2, Ki2 = 0:013,
Kd2 = 0.
The results of the pass temperature before and after
the introduction of the proposed IPFC control are
shown in Figures 7–14. The control cycle is set to 15 s.
The method has been applied in eight 800 k tone/year
USC ethylene cracking furnace of PetroChina Sichuan
Petrochemical Company. The results of the pass tem-
perature (TIC112126A and TIC112326A in the south)
in chamber A are presented in Figures 7, 9, 11 and 13.
534 Measurement and Control 52(5-6)

Figure 6. Degree of membership of T-S model for the pass temperature: (a) pass temperature for the first branch, (b) pass
temperature for the second branch, (c) feed flow for the first branch and (d) feed flow for the second branch.

Figure 7. Pass temperature for chamber A under PID control.

From these figures, a better control performance can be Hence, the IPFC method surpasses the PID, Su and
easily seen with the IPFC compared to the PID, Su and Li38 and standard PFC. The corresponding changes of
Li38 and standard PFC. The fluctuation of the pass set point of pass feed flow (FIC112102 and FIC112302)
temperature has been decreased from 658C to 618C. are shown in Figures 8, 10, 12 and 14. The similar
Song et al. 535

Figure 8. Set point of pass flow for chamber A under PID control.

Figure 9. Pass temperature for chamber A under Su and Li.38

Figure 10. Set point of pass flow for chamber A under Su and Li.38
536 Measurement and Control 52(5-6)

Figure 11. Pass temperature for chamber A under standard PFC control.

Figure 12. Set point of pass flow for chamber A under standard PFC control.

results on pass temperature (TIC112226A and
TIC112426 in the north) for chamber B are obtained
and omitted here for the sake of simplicity. Due to the
decrease of the fluctuation of the pass temperature, the
coking rate in the furnace will be decreased, leading to
the interval cycle of decoking prolonged. Therefore, the
goals of extending life of the furnace tube have been
achieved.
Tables 1 and 2 show the results of pass temperature
(TIC112126A and TIC112326A in the south) for cham-
ber A, in which the standard deviations of them
decrease by 42.73% and 41.79% compared to Su and
Li,38 by 80.45% and 79.25% compared to traditional
PID control and by 32.56% and 40.09% compared to
standard PFC control. Therefore, compared to Su and
Li,38 PID controller and standard PFC, the developed
pass temperature controller can achieve the improved
control performance with regard to the set point track-
ing and disturbance rejection.
Conclusion
The design and application of pass temperature balance
control using proposed IPFC algorithm on the ethylene
cracking furnace are presented in this paper. The pass
temperature system is structured by T-S fuzzy models
and then transformed into the extended time-varying
state space model. The linear control law is computed
Song et al. 537

Figure 13. Pass temperature for chamber A under IPFC control.

Figure 14. Set point of pass flow for chamber A under IPFC control.

Table 1. Comparison results of pass temperature (TIC112126A).

Method Average Max Min Extreme deviation Standard deviation

IPFC 848.0047 849.8350 846.2189 3.6161 0.6656

Wu et al.32 848.0617 851.7710 842.5104 9.2605 1.1623
Standard PFC 848.0208 852.0275 844.5004 7.5272 0.9869
PID 849.9919 860.2997 843.3713 16.9284 3.4043

Table 2. Comparison results of pass temperature (TIC112326A).

Method Average Max Min Extreme deviation Standard deviation

IPFC 847.9827 850.0922 845.7072 4.3850 0.7455

Wu et al.32 848.3626 851.8709 842.7717 9.0992 1.2807
Standard PFC 848.0097 852.7457 844.1626 8.5831 1.2444
PID 849.5174 860.1621 843.0208 17.1413 3.5934
538
by the predictive functional control method. The pass
temperature balance controller is implemented in pro-
prietary APC-ISYS software, which is connected to
Yokogawa DCS system via an OPC server. The indus-
trial application shows the developed pass temperature
balance control system has a better control
performance.
Because the uncertainties and unknown disturbance
problems are inevitable in many industrial processes, it
may lead to the instability of system and poor control
performance. How to solve such problems is a great
challenge for the control of pass temperature balance.
Recent works have demonstrated that the disturbance
observer (DOB) is good at addressing the above prob-
lems leading to the motivation of integrating to achieve
a new model predictive control (MPC) scheme for pass
temperature balance, which will be some interesting
issues as our future works.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with
respect to the research, authorship, and/or publication of this
article.
Funding
Project (61673199) was supported by the National Natural
Science Foundation of China and Project (20180550905) was
supported by Natural Science Foundation of Liaoning
Province. Project (L2016024) was supported by the Scientific
Research Fund of Education Department of Liaoning
Province.
ORCID iD
Hui-yuan Shi https://orcid.org/0000-0002-3097-0732
References
1. Shi HY, Su CL, Cao JT, et al. Nonlinear adaptive pre-
dictive functional control based on the Takagi–Sugeno
model for average cracking outlet temperature of the
ethylene cracking furnace. Indus Eng Chem Res 2015;
54(6): 1849–1860.
2. Ang KH, Chong G and Li Y. PID control system analy-
sis, design, and technology. IEEE T Contr Sys Technol
2005; 13(4): 559–576.
3. Wang XX and Zheng DZ. Difference control of parallel
streams temperatures. J Process Contr 2005; 15(5): 531–
536.
4. Li SF, Wu FM, Chen H, et al. Refinery furnace feeding-
branches outlet temperature equilibrium control. Inform
Contr 1994; 23(4): 243–246.
5. Wang X, Gui BW and Sun DM. A new control method
and its application in industrial furnace. Autom Panor
2001; 18(2): 16–18.
6. Li P, Liu Q, Qian XH, et al. Advanced control of CBL-
II ethylene cracking furnace. Auto Petrochem Indus 1999;
3: 25–27.
7. Wang X and Zheng DZ. Generalized difference control
of parallel streams temperatures. J Process Contr 2006;
16(5): 535–543.
Measurement and Control 52(5-6)
8. Wang X and Zheng DZ. Load balancing control of fur-
nace with multiple parallel passes. Contr Eng Prac 2007;
15(5): 521–531.
9. Wang ZN, Qi-An LI, Yue LI, et al. The application of
an improved difference control method in temperatures
control of ethylene cracking furnace. J Petrochem Univ
2012; 25(3): 79–82.
10. Mu JS, Wang X, Wang ZL and Qian F. Design and
application of cot control system of ethylene cracking
furnace. Comp Appl Chem 2012; 29(1): 90–94.
11. Wang XX, Zheng DZ, Zang JQ, et al. Switching differ-
ence control of parallel streams temperatures. IFAC P
Vol 2008; 41(2): 15421–15426.
12. Wang XX. Pass balancing switching control of a four-
passes furnace system. IFAC P Vol 2011; 44(1): 12078–
12083.
13. Yang M, Luan XL and Liu F. Uniform temperature con-
trol on feed heaters by distributed deviations. CIESC J
2016; 67(12): 5148–5154.
14. Luan XL, Yang M and Liu F. Distributed pass balancing
and tracking control of feed heater. Acta Auto Sin 2017;
6(6): 1056–1064.
15. Luan X, Min Y, Albertos P, et al. Feed furnace tempera-
ture control based on the distributed deviations. Indus
Eng Chem Res 2017; 56(20): 6035–6042.
16. Zhang WY, Lv WX and Huang DX. A pass temperature
balance control for fired heater based on steady state
energy balance. Comp Appl Chem 2008; 7: 777–781.
17. Rivas-Perez R, Feliu-Batlle V, Castillo-Garcia FJ, et al.
Temperature control of a crude oil preheating furnace
using a modified Smith predictor improved with a distur-
bance rejection term. IFAC P Vol 2014; 47: 5760–5765.
18. Li P, Li QA, Lei RX, et al. Development and application
of advanced process control system for ethylene cracking
heaters. CIESC J 2011; 62(8): 2216–2220.
19. Li P, Li T and Cao JT. Advanced process control of an
ethylene cracking furnace. Meas Contr 2015; 48(2): 50–
53.
20. Leva A. Autotuning process controller with improved
load disturbance rejection. J Process Contr 2005; 15(2):
223–234.
21. Zeybek Z. Role of adaptive heuristic criticism in cascade
temperature control of an industrial tubular furnace. Appl
Therm Eng 2006; 26(2): 152–160.
22. Zhu Q, Zhao D, Zhang S, et al. U-model enhanced
dynamic control of a heavy oil pyrolysis/cracking fur-
nace. IEEE/CAA J Automat Sin 2018; 5(2): 577–586.
23. Zhang WY, Huang DX, Wang YD, et al. Adaptive state
feedback predictive control and expert control for a
delayed coking furnace. Chinese J Chem Eng 2008; 16(4):
590–598.
24. Qi LG, Lu WX, Gao XY, et al. Coordinated control of
multiple liquid levels and furnace composite system.
CIESC J 2016; 67(3): 690–694.
25. Abilov AG, Zeybek Z, Tuzunalp O, et al. Fuzzy tempera-
ture control of industrial refineries furnaces through
combined feedforward/feedback multivariable cascade
systems. Chem Eng Process Process Intensif 2002; 41(1):
87–98.
26. Wang X. Temperature uniformity control of a furnace
system with multiple parallel passes using fuzzy difference
control technique. In: World congress on intelligent con-
trol and automation, Chongqing, China, 25–27 June 2008,
pp. 79–799. New York: IEEE.
Song et al.
27. Wang GJ, Luo ZM, Zhu LN, et al. Fuzzy estimation for
temperature distribution of furnace inner surface. Int J
Therm Sci 2012; 51(1): 84–90.
28. Ojasvi AK and Kaistha N. Control system design for fur-
nace with multiple parallel passes. Indus Eng Chem Res
2016; 55: 5702–5713.
29. Li HY, Wang YY, Yao DY, et al. A sliding mode
approach to stabilization of nonlinear Markovian jump
singularly perturbed systems. Automatica 2018; 97: 404–
413.
30. Li HY, Zhao SY, He W, et al. Adaptive finite-time track-
ing control of full states constrained nonlinear systems
with dead-zone. Automatica 2019; 100: 99–107.
31. Zhou Q, Zhao SY, Li HY, et al. Adaptive neural network
tracking control for robotic manipulators with dead-zone.
IEEE T Neur Net Lear. Epub ahead of print 19 October
2018. DOI: 10.1109/TNNLS.2018.2869375.
32. Wu S, Jin QB, Zhang RD, et al. Improved design of con-
strained model predictive tracking control for batch pro-
cesses against unknown uncertainties. ISA T 2017; 69:
273–280.
539
33. Shi HY, Li P, Su CL, et al. Robust constrained model
predictive fault-tolerant control for industrial processes
with partial actuator failures and interval time-varying
delays. J Process Contr 2019; 75: 187–203.
34. Aström K and Wittenmark B. Adaptive control. 2nd ed.
Boston, MA: Addison-Wesley Longman Publishing
Company, Incorporated, 1994.
35. Ljung L. System identification: theory for the user. Upper
Saddle River, NJ: Prentic Hall, 1999.
36. Eliasi H, Davilu H and Menhaj M. Adaptive fuzzy model
based predictive control of nuclear steam generators. Nuc
Eng Des 2007; 237(6): 668–676.
37. Gustafson D and Kessel W. Fuzzy clustering with a fuzzy
covariance matrix. In: Proceeding of IEEE conference on
decision and control, San Diego, CA, 10–12 January 1979,
pp. 761–766. New York: IEEE.
38. Su CL and Li P. Adaptive predictive functional control
based on Takagi-Sugeno model and its application to pH
process. J Centr South Univ Technol 2010; 17(2): 363–371.
39. Cohen G and Coon G. Theoretical consideration of
retarded control. Trans ASME 1953; 75: 827–834.