International Journal of Vol.1 (No.1). 2015. pp.

19-26
Innovation in Mechanical Engineering & Advanced Materials (IJIMEAM) Published online: December 1, 2015
www.umb-intl-journal.com ISSN: 2477-541X

A NONPARAMETRIC SYSTEM IDENTIFICATION BASED ON TRANSIENT ANALYSIS

WITH PLANT PROCESS OF HEAT EXCHANGER AS STUDY CASE

T. Mulyana

Study Program of Industrial Engineering, Faculty of Engineering Industry, Telkom University (Tel-U), Bandung,
Indonesia

Abstract
In this article, a nonparametric identification system based on transient analysis has been reviewed, by taking
the case in some of the data plant process of heat exchanger. Results of the study found that the first-order
transfer function without time-delay the proposed model to the data with a temperature constant value is 35.20
ºC and the time constant is 7200 seconds. This model has been fit to meet the existing data proving that the
results of the calculation error do not exceed 2%.

Keywords: nonparametric identification systems, transient analysis, process plant heat exchanger, temperature
constant, time constant, calculated error

E-mail address: [email protected]

1. Introduction methods, i.e. parametric and nonparametric methods.

Introducing a mathematical model to represent The parametric method is built through a model
the actual system is very important especially in structure determination which is described by a set
simulation and prediction purposes, particularly for of parameters while the nonparametric method is
designation of digital control and dynamic systems. determined based on an output system response in a
Basically, there are two ways of constructing function or a graph forms. In general, a parametric
mathematical models i.e. mathematical modeling method can be characterized as a mapping from the
and system identification [1]. A mathematical recorded data to the estimated parameter vector.
modeling is an analytical approach. For example, A typical example of the nonparametric method
basic laws from physics (such as Newton’s laws and is transient analysis where the input can be a
balance equations) are used to describe the dynamic recorded based on step signal and step response.
behavior of a phenomenon or a process. On the other This response will by itself give certain
hand, system identification is an experimental characteristics (dominating time constant, damping
approach where a model is then fitted to the recorded factor, static gain, etc.) of a process. Nonparametric
data by assigning suitable numerical values to its techniques are often sensitive to noise and do not
parameters. In many cases such as plant processes give very accurate results. However, as they are easy
are so complex that it is not possible to obtain to apply they often become useful means of deriving
reasonable models using only analytical approach preliminary or crude models. Therefore, in order to
(physical insight, e.g. balance equations). solve the difficulties to get the dynamics model of
In such cases one is forced to use system process system, one can use system identification.
identification techniques. It often happens that a
model that is based on physical insight contains a 2. Nonparametric Identification
number of known parameters even if the structure is This section describes a nonparametric system
derived from physical laws. On the other hand identification based on a transient analysis. A
system identification methods can be applied to transient is defined if an input is taken as a step or
estimate the unknown parameters. To a dynamic impulse, and the recorded output constitutes the
system, the system identification is a technique to model [1]. Such identification methods are
estimate the mathematical models based on data characterized by the property that the resulting
observed from the system [2,3]. Many researchers models form curves or functions, which are not
such as [4], [5], and [6] are identifying the dynamic necessarily parameterized by a finite-dimensional
model of heat exchanger using system identification. parameter vector. Sometimes, it is of interest to fit a
Based on [1], techniques used in system simple low-order model to a step response. This is
identification can be divided into two distinct illustrated as the first- and the second-order systems,

19
International Journal of Vol.1 (No.1). 2015. pp. 19-26
Innovation in Mechanical Engineering & Advanced Materials (IJIMEAM) Published online: December 1, 2015
www.umb-intl-journal.com ISSN: 2477-541X

which are described using the transfer function

model in Eq. (1) where Y(s) is the Laplace transform
of the output signal y(t), U(s) is the Laplace
transform of the input signal u(t), and G(s) is the
transfer function of the system [1].

(1)

Consider a transfer function system which

represented by Eq. (2). In such a case, the system is
therefore described by a first-order differential Eq.
(3). Note that a time delay τ is included in the model.
The step response of such a system is illustrated in
Fig. 1 that demonstrates a graphical method for Fig. 2 Step response damped oscillator
determining the parameters K, T and τ from the step
response. The gained K is given by the final value. Fig. 2 obviously shows that the relative
By fitting the steepest tangent, T and τ can be damping ζ influences the character of the step
obtained. The slope of this tangent is K/T, where T responses. The remaining two parameters, K and ,
is the time constant. The tangent crosses the t axis at merely act as scale factors. The gained K scale is the
t=τ, which is the time delay [1]. amplitude axis while scale is the time axis. The
three parameters of the model in Eq. (2), namely ,
(2) ζ and could be determined by comparing the
measured step response with Fig. 2 and choosing the
curve that is most similar to the recorded data.
(3) However, one can also proceed in a number of
alternative ways. One possibility is to look at the
local extreme (maxima and minima) of the step
response. With some calculation it can be found
from Eq. (4). At times, it occurs as given by Eq. (5).
Where Eq. (6) defines time and overshoots is
given by Eq. (7) [1].

(5)

(6)

(7)
Fig. 1 Step response first order system with time delay
The relationship between the overshoot and
Physically, this equation describes a damped
the relative damping is illustrated in Fig. 3. The
oscillator. After some calculations, the step response
parameters , ζ and can be determined based on
is found to be as expressed by Eq. (4). The output
response, y(t), in Eq. (4) for various ζ is illustrated values obtained in Fig. 4. The gained is easily
by Fig. 2 [1]. obtained as the final value once convergence is
achieved. The overshoot has been determined in
several ways. One possibility is to use the first
(4) maximum. An alternative is to use several extreme
and the fact as given by Eq. (6). The amplitude of
the oscillations in Eq. (5) is reduced by a factor
for every half-period. Once is determined; ζ can

20
International Journal of Vol.1 (No.1). 2015. pp. 19-26
Innovation in Mechanical Engineering & Advanced Materials (IJIMEAM) Published online: December 1, 2015
www.umb-intl-journal.com ISSN: 2477-541X

be derived from Eq. (7), and will produce Eq. (8) 3. Nonparametric Model
[1]. A transfer function of shell and tube heat
exchanger is described as Eqs. (11) up to (17). It has
complex forms and it is difficult to apply to the
design of practical controllers. In some cases, it is
found that the outlet responses are well
approximated by using the first- or second-order
system to replace the complicated forms of the
transfer functions obtained directly from the
transformed solutions. This transfer functions are a
model form based on the Laplace transform and it is
very useful in analysis and design of a linear
dynamic shell and tube heat exchanger.
To obtain a nonparametric model, a step
response analysis is used in this case. The dynamic
model in Eq. (11) can be simplified as Eq. (12). The
obtained transfer function which assumes
Fig. 3 Overshoot M versus relative damping ζ damped is shown in Eq. (13)
oscillator
where is the transfer function, is the
output signal from the cold water temperature outlet
in shell side heat exchanger, , and is the
input signal from the hot water flow rate inlet in tube
side heat exchanger, .

(11)

(12)
Fig. 4 Determination parameters damped oscillator from
the step response
(13)
(8)

(14)
From the step response, the period of the
oscillations has also been determined. From (6), the (15)
period is given by Eq. (9) [1]. Then is given by
Eq. (10) [1].

(9)
(16)

(10) (17)

A first order transfer function without time

delay is given by Eq. (18), where is the gain and
is the time constant. A step response of such a first

21
International Journal of Vol.1 (No.1). 2015. pp. 19-26
Innovation in Mechanical Engineering & Advanced Materials (IJIMEAM) Published online: December 1, 2015
www.umb-intl-journal.com ISSN: 2477-541X

order transfer function in Eq. (18) has the following is the Padé-approximation as given in Eq. (21),
characteristics, which are demonstrated by Fig. 5 [7]. where n is the order of the approximation and this
order must be chosen [7]. The coefficients are
functions of n.
(18)
(20)
The first order transfer function with time delay
is given by Eq. (19), where is the gain, is the (21)
time constant and is time delay.

The Padé-approximations are based on a

minimization of the truncation errors in a finite
series expansion of . Table 1 shows, as an
illustration, the k-values for the orders n = 1 and n =
2 [7].

Table 1: Coefficients of Pade-approximations of order

and
Order-1 Order-2

Fig. 5 Step response first order transfer function without

time delay
Table 1 shows and which are given by
A step response of such a transfer function has
the following characteristics, which are Eqs. (22) and (23). Therefore, the Padé-
demonstrated by Fig. 6 [7]. approximation transfer functions for first and second
orders are given by Eqs. (24) and (25) [7].

(19)
(22)

(23)

(24)

(25)

Fig. 6 Step response first order transfer function with time

delay Substituting Eqs. (24) and (25) into (18) will
produce Eqs. (26) and (27).
A transfer function of a time delay which is
given by Eq. (20) is irrational transfer function (it
has no s-polynomial in the numerator and in the (26)
denominator). In some situations as in frequency
response based analysis of control systems
containing a time delay, it is necessary to substitute
with an approximation in the form of a rational (27)
transfer function. The most common approximation

22
International Journal of Vol.1 (No.1). 2015. pp. 19-26
Innovation in Mechanical Engineering & Advanced Materials (IJIMEAM) Published online: December 1, 2015
www.umb-intl-journal.com ISSN: 2477-541X

The step and phase responses of the first order Table 3: Type of step response for various value of
Padé-approximation is shown in Fig. 7. Fig. 8 shows zeta
the step and phase responses of the second order
Padé-approximation. A second order transfer
functions is given by Eq. (28). Where , and
refer to system gain, relative damping factor and
undamped resonance frequency [1,2]. Table 2 shows
type of step response for various values of .
When and the poles are real and distinct, it is
given by Eq. (29).

(28)

(29)

Pade approximation of order 1: step response comparison

1

0.5
Amplitude

-0.5 Pade approximation

Pure delay
-1
0 5 10 15
Time (s)
Phase response comparison
0
-50
4. Case Study
Phase (degree)

-100
-150 The candidate of nonparametric model has been
-200 described. When the experimental data obtained are
-250
observed, the data are apparently similar to the step
-2 -1 0
10 10
Frequency (rad/s)
10 response of the first order transfer function. This
model is candidate model and has been expressed by
Fig. 7 Step and phase’s responses order-1 Padé-
approximation
Eq. (30). Therefore, the candidate model chosen is
the first order transfer function. The error of this
Pade approximation of order 2: step response comparison
model is calculated after offset, where the time of
1.5 offset is determined according to time constant that
1 occurred. The error is given by Eq. (31), where
Amplitude

0.5 is the error, is the number of the data and is the

0 Pade approximation output signal value with for the data, and
Pure delay
-0.5
0 5 10 15 for calculated value from the model. The percentage
Time (s)
Phase response comparison
of this error is given by Eq. (32), where is
0
within the range of the .
Phase (degree)

-200

-400

-600
(30)
-1 0 1
10 10 10
Frequency (rad/s)

Fig. 8 Step and phase’s responses order-2 Padé- (31)

approximation

(32)

23
International Journal of Vol.1 (No.1). 2015. pp. 19-26
Innovation in Mechanical Engineering & Advanced Materials (IJIMEAM) Published online: December 1, 2015
www.umb-intl-journal.com ISSN: 2477-541X

The step response of the resulted nonparametric model because the error is smaller than 2%. In other
model for dataexp1 is shown in Eq. (33). words, it should be accepted.
Error calculation of nonparametric model dataexp1 measured
(33) 1

0.8

0.6

The comparison of the model output and the 0.4

0.2
measured dataexp1 is shown in Fig. 9.

Error E (C)
0

-0.2

Output signals of cold water temperature outlet nonparametric model and dataexp1 measured -0.4
55 -0.6

-0.8

50 -1
0 1000 2000 3000 4000 5000 6000 7000
Time (seconds)

45 Fig. 10 Error calculation nonparametric model dataexp1

Temperature Ts (C)

40 Similar way has been shown for dataexp1 and

dataexp1 measured
nonparametric model
dataexp2. The step response of the resulted
35
nonparametric model for dataexp2 is shown in Eq.
(34). The comparison of the model output and the
30
measured data is shown in Fig. 11. The error
25
calculation to validate this model is presented in
0 1000 2000 3000 4000 5000 6000 7000
Time (seconds) Table 4. Fig. 11 shows the obtained error. From Fig.
Fig. 9 Nonparametric model output and data measured 11, it is obvious that the trend of the model output
dataexp1 which is given by Eq. (34) is similar to the measured
dataexp2. The data starts from 29 ℃ and the model
From Fig. 9, it is obvious that the trend of the also starts from 29 ℃. From time (t) of 0s until to the
model output which is given in Eq. (33) is similar to final time (t) of 7000 s, the difference of the both
the measured dataexp1. The data and the model start temperatures reaches the smallest value, i.e. between
with the same values, i.e. ., and the difference zeros to one. These facts are presented in Table 4.
of both temperatures reaches the smallest figure.
where is fluctuating between zero to one. These facts
are presented in Table 3. (34)

Table 3. Error calculation nonparametric model and data

Output signals of cold water temperature outlet nonparametric model and dataexp2 measured
measured dataexp1 55

( ) ( ) ( ) ( ) 50

0 30.0 30.0 0.0

1000 35.0 34.5 0.5 45
Temperature Ts (C)

2000 37.0 37.0 0.0

40
3000 42.5 42.5 0.0 dataexp2 measured
4000 45.0 45.0 0.0 nonparametric model
35
5000 47.0 48.0 1.0
6000 50.0 50.0 0.0 30
7000 54.0 53.0 1.0
25
0 1000 2000 3000 4000 5000 6000 7000
Time (seconds)

Fig. 10 shows these facts based on the error Fig. 11 Nonparametric model output and data measured
values in Table 3. The error has been calculated of dataexp2
using Eqs. (31) and (32) which takes 8 samples time
between 0 and 7000 s. From Table 3, it is obvious
that the error is and the percentage of the
error is . According to these results, therefore,
the model which is shown in Eq. (33) is an excellent

24
International Journal of Vol.1 (No.1). 2015. pp. 19-26
Innovation in Mechanical Engineering & Advanced Materials (IJIMEAM) Published online: December 1, 2015
www.umb-intl-journal.com ISSN: 2477-541X

Table 4. Error calculation nonparametric model and data 60

Output signals of cold water temperature outlet nonparametric model and dataexp3 measured

measured dataexp2
50
( ) ( ) ( ) ( )
0 29.0 29.0 0.0 40

Temperature Ts (C)
1000 34.0 34.0 0.0
2000 37.5 37.5 0.0 30
dataexp3 measured
3000 41.5 41.0 0.5 nonparametric model
20
4000 44.0 44.0 0.0
5000 47.0 46.5 0.5
10
6000 49.5 48.5 1.0
7000 52.0 52.0 0.0 0
0 1000 2000 3000 4000 5000 6000 7000
Time (seconds)

Fig. 13 Nonparametric model output and data measured

Fig. 12 shows the facts based on the values of of dataexp3
the error in Table 4. The error has been calculated
using Eqs. (31) and (32) which takes 8 sample times Table 5. Error calculation nonparametric model and
between 0 and 7000 s. data measured dataexp3

Error calculation of nonparametric model and dataexp2 measured ( ) ( ) ( ) ( )

1
0 26.0 26.0 0.0
0.8
1000 33.0 31.0 2.0
0.6
2000 38.0 35.0 3.0
0.4
3000 40.0 38.0 2.0
0.2 4000 42.0 41.0 1.0
Error E(C)

0 5000 45.0 44.0 1.0

-0.2 6000 47.0 46.0 1.0
-0.4 7000 48.0 48.0 0.0
-0.6

-0.8

-1
From Fig. 13, it is apparent that the trend of the
0 1000 2000 3000 4000
Time (seconds)
5000 6000 7000
model output which is given by Eq. (35) is similar to
Fig. 12 Error calculation nonparametric model dataexp2 the measured dataexp3. The data and the model start
with the same values, i.e. . From the start time
From Table 4, it is apparent that the error is of 0s until to the final time of 7000s, the difference
and the percentage of the error is . of both temperatures reaches the smallest value that
Based on these results, therefore, the model which it is presented in Table 5. Fig. 14 shows these facts
has been given by Eq. (34) is an excellent model based on the error values in Table 5. The error has
because of the error is smaller than . Therefore, it been calculated using Eqs. (31) and (32) with 8
samples of time between 0 and 7000 s. From Table
should be accepted.
5, it is obvious that the error is and the
The step response of the nonparametric model
resulted from dataexp3 is shown in Eq. (35). The percentage of the error is . Referring to these
comparison of the model output and the measured results, therefore, the model which is given by Eq.
data is given in Fig. 13. The error calculation to (35) is a sufficiently good model because the error is
validate this model is presented in Table 5. Fig. 14 less than . Therefore, it may be accepted.
shows the obtained error.

(35)

25
International Journal of Vol.1 (No.1). 2015. pp. 19-26
Innovation in Mechanical Engineering & Advanced Materials (IJIMEAM) Published online: December 1, 2015
www.umb-intl-journal.com ISSN: 2477-541X

3
Error calculation of nonparametric model and dataexp3 measured temperature each the data is different. Based on
these results, it can be concluded that the first order
2 without time delay could be taken as nonparametric
model with the gain is 35.2 ºC, and the time constant
1 of 7200 s.
Error E (C)

0
References
-1
[1] Soderstrom, T. & Stoica, P. System
Identification. New York: Prentice Hall; 2001.
-2 [2] Knudsen, M. Experimental Modelling of
Dynamics Systems. Department of Control
-3
0 1000 2000 3000 4000 5000 6000 7000 Engineering, Aalborg University, 2004.
Time (seconds)
[3] Ljung, L. Approaches to Identification of
Fig. 14 Error calculation nonparametric model dataexp3 Nonlinear System. Sweden, Division of
Automatic Control, Linkoping University,
The comparison results of the nonparametric 2011.
models are presented in Table 6. It is shown that all [4] Nithya, S. et al. Model Based Controller Design
time constant or offset and the gain of the all for Shell and Tube Heat Exchanger. Sensors &
experimental data is the same, i.e., 7200 s and 35.2 Transducers Journal 2007; 84(10): p. 1677-
while the initial temperature for each 1686.
experimental data is different. They are 30 for [5] Sharma, C. Gupta, S. & Kumar, V. Modeling
dataexp1, 29 for dataexp2 and 26 for dataexp3. and Simulation of Heat Exchanger used in Soda
The percentage of the error for each experimental Recovery. Proceeding of the World Congress
data is different too. The dataexp1 has the error on Engineering (WCE) 2011; 2.
percentage of 1.3%, dataexp2 is 1.1% and dataexp3 [6] Sivakumar, P. Prabhakaran, D. & Kannadasan,
is 6.8%. Based on these results, it concludes that the T. Comaparative Analysis of Advanced
first order without time delay could be taken as Controllers in a Heat Exchanger. American
nonparametric model with the gain and time constant Journal of Engineering Research (AJER) 2013;
have same values, i.e. 35.2 and 7200 s, while the 02(01): p. 01-06.
[7] Halvorsen, H. P. Control Theory with
initial temperature is different depend to state of the MathScript Examples. Telemark University
experimental began. College, Norway, 2012.
[8] Mulyana, T. Parametric and Nonparametric
Table 7. The comparison results of nonparametric
models
Identification of Shell and Tube Heat
dataexp Gain Gain Time Exchanger Mathematical Model. Thesis
( ) Constant ( ) UTHM, 2014.
( ) ( )
dataexp1 30 35.2 7200 0.31 1.3
dataexp2 29 35.2 7200 0.25 1.1
dataexp3 26 35.2 7200 1.5 6.8

5. Conclusion
The nonparametric model equation is generated
from the dynamic model equation through step
response analysis being used. This equation is
represented in a transfer function form. The output
signal is the cold water temperature outlet in shell
side heat exchanger, whilst the input signal is the hot
water flow rate inlet in tube side heat exchanger. A
first order without time delay transfer function has
been selected as candidate model chosen since it has
a response which is similar to the three set of the
experimental data. It can be observed that the all-
time constant and the gain of the all experimental
data is the same although the starting of the

26