fluids
Article
Study of Different Alternatives for Dynamic Simulation of a
Steam Generator Using MATLAB
Cristhian Álvarez *, Edwin Espinel *






Citation: Álvarez, C.; Espinel, E.;
Noriega, C.J. Study of Different
Alternatives for Dynamic Simulation
of a Steam Generator Using MATLAB.
Fluids 2021, 6, 175. https://doi.org/
10.3390/fluids6050175
Academic Editor: Stéphane Vincent
Received: 5 April 2021
Accepted: 26 April 2021
Published: 29 April 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Fluids 2021, 6, 175. https://doi.org/10.3390/fluids6050175
and Carlos J. Noriega
Mechanical Engineering Department, Universidad Francisco de Paula Santander, Vía Acolsure, Sede el
Algodonal Ocaña, Ocaña 546552, Colombia; cjnoriegas@ufpso.edu.co
* Correspondence: cealvarezp@ufpso.edu.co (C.Á.); eeespinelb@ufpso.edu.co (E.E.)
Abstract: This work presents the simulation of a steam generator or water-tube boiler through the
implementation in MATLAB® for a proposed mathematical model. Mass and energy balances for the
three main components of the boiler—the drum, the riser and down-comer tubes—are presented.
Three alternative solutions to the ordinary differential equation (ODE) were studied, based on Runge–
Kutta 4th order method, Heun’s method, and MATLAB function Ode45. The best results were
obtained using MATLAB® function Ode45 based on the Runge–Kutta 4th Order Method. The error
was less than 5% for the simulation of the steam pressure in the drum, the total volume of water in
the boiler, and the mixture quality in relation to what was reported.
Keywords: steam generator; mathematical model; drum; riser; down-comer
1. Introduction
The world should prepare for the demand for energy to skyrocket in the next 20 years.
By 2040, it will rise by 30% with an annual growth rate of 1.4% [1]. This growth will be
like adding another China and India to global demand, warns the annual report of the
International Energy Agency (IEA). The IEA indicates that the global economy is growing
at an average rate of 3.4% per year. Additionally, the IEA estimates that the population
will expand from 7.4 billion to 9 billion people by 2040, and there will be an urbanization
process that will add the equivalent of a city the size of Shanghai to the world’s urban
population every four months. The energy sector will experience profound changes, with
new production powers and a shift in energy sources that will supply energy to humanity.
For these reasons, the energy supply on the planet can be considered a major concern
for all countries. In addition, most energy production is carried out with fossil fuels, along
with the environmental problems associated with their use, such as global warming, air
pollution, and acid rain [2].
One of the most feasible strategies to mitigate the impact generated by the use of
fossil fuels in power generation is to improve the efficiency of power plants. The efficient
operation of steam generators or boilers is indispensable for the correct functioning of this
type of plant known as thermal power plants. The boiler is responsible for transforming
the chemical energy of a fuel such as coal, oil, gas, or nuclear energy into thermal energy,
using this heat to convert water into steam [3].
The steam generator is used in the oil [4], pharmaceutical [5], and thermoelectric
industries, among others. It is necessary to know and understand the functioning of these
systems. This has been possible through the formulation of mathematical models that
represent their behavior with excellent results [6–12].
Recent studies have used the mathematical model defined by ordinary differential
equations (ODEs) [6]. They used this model to simulate the behavior of thermal power
plants with heat recovery systems. This is the case of the works that reported the dynamic
behavior of key parameters involved in the steam generation process, such as the appro-
priate rate of water and fuel consumption according to the rate of heat required in the
https://www.mdpi.com/journal/fluids
 Fluids 2021, 6, x FOR PEER REVIEW 2 of 15

Fluids 2021, 6, 175 2 of 14

namic behavior of key parameters involved in the steam generation process, such as the
appropriate rate of water and fuel consumption according to the rate of heat required in
the boiler risers [13]. Other studies reported the response of water consumption and
boiler risers [13]. Other studies reported the response of water consumption and steam
steam generation in the boiler, based on the variables of water level, pressure, and mix-
generation in the boiler, based on the variables of water level, pressure, and mixture quality
ture quality in the boiler drum [14].
in the boiler drum [14].
In relation to the works shown in the literature where numerical models generally
In relation to the works shown in the literature where numerical models generally
include functions that have already been implemented, this work aims to show a com-
include functions that have already been implemented, this work aims to show a com-
parison between conventional numerical models (Runge–Kutta of order 4 and Heun’s
parison between conventional numerical models (Runge–Kutta of order 4 and Heun’s
method) and the Ode45 function implemented in MATLAB® (MathWorks Inc., Natick,
method) and the Ode45 function implemented in MATLAB® (MathWorks Inc., Natick, MA,
Massachusetts, USA),implementing the mathematical model and validating the results
USA),implementing the mathematical model and validating the results shown in [6] and,
shown
in in [6]
this way, and, in this
simulating theway, simulating
operation the operation
of a steam generator,of since
a steam
thisgenerator, since this
model adequately
model adequately
represents represents
the dynamics of thethe dynamics
system of the asystem
and allows andanalysis
practical allows aofpractical
moderate analysis
com-
plexity for its development [14]. A methodology is proposed for the implementation for
of moderate complexity for its development [14]. A methodology is proposed the
of the
implementation of the mathematical
® model in MATLAB ® , which can be used in real
mathematical model in MATLAB , which can be used in real practical applications. The
practical applications.
algorithm The of
for the execution algorithm
the model foristhe execution of
programmed andthe model
three is programmed
options are studied and
for
three options are studied for the solution of the
the solution of the ordinary differential equation (ODE). ordinary differential equation (ODE).
Themodel
The modelisisprogrammed
programmed in in MATLAB
MATLAB
® and validated with the data reported by
® and validated with the data reported by [6].
[6]. Three
Three numerical
numerical methods methods were analyzed
were analyzed to solveto thesolve the ODE
implicit implicit ODE
for the for the imple-
implementation
mentation of the model. The results obtained were analyzed using the
of the model. The results obtained were analyzed using the Runge–Kutta 4th order method Runge–Kutta 4th
order method used by [13],
® MATLAB ® function Ode45 implemented by [14], and Heun’s
used by [13], MATLAB function Ode45 implemented by [14], and Heun’s method as a
method alternative.
solution as a solution alternative.
The The error for
error is determined is determined for the studied
the three methods three methods studied
with reference
with
to the reference to the results
results reported in [6]. reported in [6].

2.2.Methodology
Methodology
This
Thisworkworkwas developed
was developedfollowing the the
following methodology
methodologyproposed in Figure
proposed 1. The 1.
in Figure objec-
The
tive is to present
objective a systematic
is to present process for
a systematic the implementation
process of a mathematical
for the implementation model that
of a mathematical
allows
modelthe thatdynamic simulation
allows the dynamicofsimulation
the operation of aoperation
of the water-tubeof aboiler. Figureboiler.
water-tube 1 represents
Figure
the stages in which
1 represents the project
the stages was carried
in which out, with
the project was the (*) each
carried out,programmed
with the (*)numerical
each pro-
method
grammed is indicated,
numericalasmethod
well asiseach processasvariable
indicated, well ascalculated
each processwith the application
variable ofwith
calculated the
studied methods.
the application of the studied methods.

Figure1.1.Block
Figure Blockdiagram
diagramrepresenting
representingthe
theresearch
researchdevelopment
developmentprocess.
process.

2.1. Model Analysis

The study carried out by [6] has become a reference in recent years for carrying
out new research on the behavior of steam generators. This work opened the way for
other research studies [13–15]. The mathematical model proposed by [6] was studied and
 2.1. Model Analysis
The study carried out by [6] has become a reference in recent years for carrying out
new research on the behavior of steam generators. This work opened the way for other
Fluids 2021, 6, 175 research studies [13–15]. The mathematical model proposed by [6] was studied and an-
3 of 14
alyzed. The main characteristic of this model is that, despite its moderate complexity, it
can easily be used to represent any drum boiler with high accuracy.
In Figure 2, the main element of a recovery boiler is detailed, Where, for its study,
analyzed.
three mainThe main characteristic
components of this model
can be highlighted: theisdrum,
that, despite
which its moderate
is where the complexity,
water–steam it
can easily be used to represent any drum boiler with high accuracy.
mixture is located and, in turn, has the input of the feed water (m  ) and the output of
In Figure 2, the main element of a recovery boiler is detailed, fWhere, for its study,
generated
three mainsteam  ) , can
(m
componentss andbetwo sections of
highlighted: thepipeline: one that
drum, which transports
is where
.

the water
the water–steam
mixtureasisthe
known located  has
and, in turn,
down-comer the input of the feed water m and the output of
.

(mdc ) , and the risers through which thef water rises and be-
generated steam m s , and two sections of pipeline:  ) . one that transports the water known
gins to evaporate thanks . to
a heat input rate (Q
as the down-comer m dc , and the risers through which the water rises and begins to
Equation (1) represents the overall . mass balance in the drum and Equation (2), the
evaporate
global thanks
energy to a heat input rate Q .
balance.

Figure 2. Schematic
Schematic representation
representation of the steam generator.

Equation (1) represents the overall mass balance in the drum and Equation (2), the
global energy balance.
d
dt
[ ]
ρ w∀ wt + ρs∀st = m  -m
f

s (1)
d . .
[ ρ ∀wt +ρs ∀st ]= m f −ms (1)
 dt dw
Q+m  h -m  h = [ρ ∀ h + ρ ∀ h - P∀ + m C T ] (2)
. . f f .s s dtd  s st s w wt w t m p m
Q + mf h f − ms h s = ρs ∀st hs +ρw ∀wt hw −P∀t +mm Cp Tm (2)
By solving the above balances, dtthe ordinary differential Equations (3) and (4) are
By solving
obtained, the above
which describe balances,
the the ordinary
mathematical model indifferential
a simple way Equations (3) and
for a steam (4) are
generator.
obtained, which describe the mathematical model in a simple way for a steam generator.
d∀ wt dP
e 11d∀ wt + e 12 dP dt =. m  .-m  (3)
dt f s
e11 +e12 = m f −m s (3)
dt dt
d∀ wt dP 
e 21d∀ wt + e 22 dP  h .- m
.= Q +. m  h (4)
e21 dt +e22 =dt Q + m f hff −fms hss s (4)
dt dt
where
where
e11 = ρw −ρs
e 11 = ρ w - ρ s
e12 = ∀st dρdP + ∀wt dP
s dρw

dρ s dρ w e21 = hw ρw −hs ρs
e 12 =e∀ ∀ dρs dρw
= ∀
22st dPst
+
( h dP +
s wt ρs dh
dP
s dhw
dP )+∀wt (h w dP +ρw dP )−∀t +mm Cp dP
dTm

Equation (1) can be multiplied by hw and then subtracted from Equation (2). The
result is shown in Equation (4) and this process results in Equation (5).
h
dP dhs dρs dhw dTm d∀wt
dt ∀st ρs dP + ∀st hc dP + ∀wt ρw dP − ∀t +mm Cp dP ]+ dt (− ρ s hc ) = (5)
. . .
Q − mf (h w −hf ) − ms hc
Fluids 2021, 6, 175 4 of 14

Considering that hs −hw represents the enthalpy of condensation hc , Equation (5) is

reduced to:
dP . . .
e1 = Q − mf (h w −hf ) − ms hc (6)
dt
The dominant terms in the coefficient e1 in Equation (6) are dh w dTm
dP and dP . This is
because the energy concentrated in the mass of water and metal are the physical phenomena
of the system that determine the dynamic behavior of the pressure in the drum. As a result,
e1 can be operated as shown in Equation (7).

dhw dTm
e1 = ∀wt ρw + mm C p (7)
dP dP

Riser and Down-Comer Mathematical Model

Equation (8) below allows the determination of the quality of mixture αr as a result of
the mass and energy balance in the riser and down-comer of the steam generator.

dP dαr . .
e31 +e32 = Q − αr hc mdc (8)
dt dt
where
 
dρw dTs
e31 = ρw dh w
dP α h
r c dP )( 1 − αv ∀r +mt Cp dP − ∀r +
 
(1 − αr ) hc dρ dhs dαv
dP +ρs dP )αv ∀r + ((1 − αr ) ρs − αr ρw ) hc ∀r dP
s

dαv
e32 = ((1 − αr ) ρs − αr ρw ) hc ∀r
dαr
The average steam volume fraction α v in the down-comer tube is calculated using
Equation (9) determined by [6].

(ρ − ρs ) αr
  
ρw ρs
αv = 1− Ln1+ w (9)
ρw − ρs (ρ w ρs ) αr ρs

The volume fractions are derived depending on the drum pressure, and the quality
values of the mixture are as follows:
dαv 1 dρ dρ ρ 1 ρ +ρ
= 2
(ρ w s − ρs w )(1+ w − w s Ln (1 + η))
dP ( ρ w − ρs ) dP dP ρ s 1 + η ηρs
 
dαv ρw 1 1
= Ln(1 + η) −
dαr ρs η η 1+η

where η = αρ r (ρ w − ρs .
s

2.2. Determination of Terms Using Polynomials

The terms of Equations (3)–(8) were determined from the water steam tables as a
function of saturation pressure Psat . Engineering Equation Solver (EES® ) software was
used to plot enthalpy, density, and temperature of the metal in the drum wall. The latter
was assumed to be equal to the steam saturation temperature Tm = Tsat . Then, the curve
was adjusted using linear regression to each graph, thus obtaining a polynomial of degree
n as a function of the saturation pressure Psat .

2.3. Methods for Ordinary Differential Equation Solution

To simulate the proposed model [6] and validate the results, it is necessary to solve
the ordinary differential equations through a numerical method that allows for obtaining
results adjusted to the dynamic behavior of the boiler. Considering the characteristics of
Fluids 2021, 6, 175 5 of 14

each method, such as complexity for the programming, precision in the results, steps for the
solution, and computational cost [16–18], it was decided to use the Runge–Kutta method
of order 4 and Heun’s method and compare the results with those obtained using the
MATLAB® Ode45 function that solves this type of equation based on an explicit formula
of RK 4 and RK5 [18].

2.3.1. Heun’s Method

This method is used to improve the estimation of the slope and involves the determi-
nation of two derivatives for the interval, one at the start point and one at the end point [16].
The two derivatives are averaged to obtain an improved estimate of the slope for the whole
interval. Equations (10) and (11) shown below are used to solve differential equations with
this method.
H1 = h f(xi , yi ) (10)
2 2
H2 = h f(x i + h, yi + H1 ) (11)
3 3
After having the values of H1 and H2 , each of the points in the plane (x, y) are
determined using the following Equations (12) and (13).

xi+1 = xi +h (12)

1
yi+1 = yi + (H 1 +3H2 ) (13)
4
Fluids 2021, 6, x FOR PEER REVIEW 6 of of
Figure 3 shows the flow chart representing the code programmed for the solution 15
the ODE using Heun’s method.

MATLAB®® .
Figure 3. Flow chart of Heun’s 4th order method in MATLAB

2.3.2. Runge–Kutta 4th Order Method

The solution of this method was programmed in MATLAB® , which develops mul-
tiple slope estimates to obtain an improved average slope for the interval. Each k repre-
sents a slope [16]. Equations (14)–(17) used by the Runge–Kutta fourth order method are
shown below, where h is the step width.
Fluids 2021, 6, 175 6 of 14

2.3.2. Runge–Kutta 4th Order Method

The solution of this method was programmed in MATLAB® , which develops multiple
slope estimates to obtain an improved average slope for the interval. Each k represents a
slope [16]. Equations (14)–(17) used by the Runge–Kutta fourth order method are shown
below, where h is the step width.

k1 = h f(x i , yi ) (14)

1 1
k2 = h f(x i + h, yi + k ) (15)
2 2 1
1 1
k3 = h f(x i + h, yi + k2 ) (16)
2 2
k4 = h f(x i + h, yi + k3 ) (17)
When each k is obtained, the points on the (x,y) axes of the plane are determined using
Equations (18) and (19), as follows:

xi+1 = xi +h (18)

1
yi+1 = yi + (k + 2k2 + 2k3 + k4 ) (19)
6 1
Fluids 2021, 6, x FOR PEER REVIEW 7 of of
Figure 4 shows the flow chart representing the code programmed for the solution 15
the ODE using the Runge–Kutta 4th order method.

Figure 4.
Figure 4. Flow
Flow chart
chart of
of the
the Runge–Kutta
Runge–Kutta 4th
4th order
order method
methodin MATLAB®®..
inMATLAB

2.3.3. MATLAB® Function Ode45

This method is based on an explicit formula of the Runge–Kutta (4th and 5th) order
method, which is used to solve ODEs. The syntax shown in MATLAB® is:
[t, x] = Ode45(odef un, tspan, x0)

where
1. x is a matrix where each column corresponds to the dependent variables, and t is the
time vector,
Fluids 2021, 6, 175 7 of 14

2.3.3. MATLAB® Function Ode45

This method is based on an explicit formula of the Runge–Kutta (4th and 5th) order
method, which is used to solve ODEs. The syntax shown in MATLAB® is:

[t, x] = Ode45(odefun, tspan, x0)

where
1. x is a matrix where each column corresponds to the dependent variables, and t is the
time vector,
2. odefun is the name of the function to be evaluated,
3. tspan specifies the time interval, a vector of two numbers tspan = [ti, tf], start and
end time,
4. x0 is a vector that has the initial conditions of the variables to be evaluated.

2.4. Programming and Simulation of the Model in MATLAB®

A code was developed in MATLAB® for the implementation of the mathematical
model that represents by simulation the behavior of the drum pressure P, the total water
Fluids 2021, 6, x FOR PEER REVIEW volume ∀wt , and mixture quality αr in the steam generator. Figure 5 shows the algorithm
8 of 15
that represents the code development in MATLAB . ®

Figure 5.
Figure 5. Flow
Flow chart
chart of
of the
the code
code develop
develop in MATLAB®® for
in MATLAB for the
the Ode
Ode solution.
solution.

2.5. Validation of Results

The data used for the simulation were taken from the P16–G16 unit operating in a
power plant in Malmö, Sweden, which was reported by [6]. Table 1 shows the data used.
Fluids 2021, 6, 175 8 of 14

2.5. Validation of Results

The data used for the simulation were taken from the P16–G16 unit operating in a
power plant in Malmö, Sweden, which was reported by [6]. Table 1 shows the data used.

Table 1. Data used to perform the dynamic simulation of the steam generator.
Fluids 2021, 6, x FOR PEER REVIEW 9 of 15
Operation Data
Variable Units Value
Heat flux
Total mass [kW] [kg] 86,000300,000
Mass flow of water [kg·s −1 ] 50
Gravity
Mass flow of steam [kg·s−1 ] [m • s -2 ] 50 9.81
The table was recreated according to [6]. [kJ ·(kg)·K −1 ]
Specificheat 500
Enthalpy of water −1 1080.9
[kJ·kg ]
A time interval
Raiser volumeof 200 s was considered [m 3for] the simulation of the dynamic
37 behavior
of the Down-comer
boiler. The percentage
area error was determined
[m ] 2 using Equation (20). To
0.1123 calculate the
relative error, the reported reference value V3 and the value obtained in the simula-
Total volume [m ]Ref 88
Total mass
tion VSim are taken into account. An approximation[ kg ] 300,000
error below 5% is acceptable when
Gravity [m ·s −2 ] 9.81
performing
The this type
table was recreated of study
according [19].
to [6].

A time interval of 200 s was%Error V for -the

considered VSim
simulation of the dynamic behavior of
= Ref x100% (20)
the boiler. The percentage error was determinedVRefusing Equation (20). To calculate the rela-
tive error, the reported reference value VRef and the value obtained in the simulation VSim
are taken into account. An approximation error below 5% is acceptable when performing
3. Analysis
this of Results
type of study [19].  
EES® = VRef −VSim ×100%
3.1. Polynomial Solution Using %Error (20)
 VRef 
Figure 6 shows the behavior of water and steam densities (ρw , ρs ) as a function of
saturation
3. Analysispressure Psat . It is observed that when the saturation pressure increases, the
of Results
3.1. Polynomial Solution Using ®
saturated liquid reaches the EES
saturated vapor point. For this reason, the density of the
liquid ρw 6decreases
Figure shows the and the density
behavior of the
of water andsteam increases
steamρ sdensities (ρuntil reaching the crit-
w , ρs ) as a function of
saturation
ical point. pressure Psat . It is observed that when the saturation pressure increases, the
saturated liquidthis
Knowing reaches the saturated
phenomenon, the vapor
curve point. For this reason,
fit is performed, the density
resulting of the liquid
in the polynomials
w decreases
ρshown and the
in the figure. density of the steam ρ s increases until reaching the critical point.

Figure 6.
Figure 6. Graph
Graph used
used to
to determine
determine the
the density
density polynomials
polynomials of
of saturated
saturated liquid
liquidand
andsaturated
saturatedsteam
steam
through linear
through linear regression
regression(curve
(curvefitting)
fitting)in
inthe
theEES
EES®® software.
software.

Figure 7 shows the behavior of the enthalpy of water and the enthalpy of steam as a
function of the saturation pressure. It is necessary that as the saturation pressure in-
creases, the saturated liquid reaches the saturated vapor point. For this reason, the en-
thalpy of water increases and the enthalpy of steam decreases until reaching the critical
point.
Fluids 2021, 6, 175 9 of 14

Knowing this phenomenon, the curve fit is performed, resulting in the polynomials
shown in the figure.
Figure 7 shows the behavior of the enthalpy of water and the enthalpy of steam as a
function of the saturation pressure. It is necessary that as the saturation pressure increases,
Fluids 2021,
Fluids 2021, 6,
6, xx FOR
FOR PEER
PEER REVIEW
REVIEW 10 of
10 of 15
15
the saturated liquid reaches the saturated vapor point. For this reason, the enthalpy of
water increases and the enthalpy of steam decreases until reaching the critical point.

Figure7.
Figure
Figure 7. Graph
7. Graphused
Graph usedto
used todetermine
to determinethe
determine thepolynomials
the polynomialsofof
polynomials ofenthalpy
enthalpyofof
enthalpy ofwater
waterand
water andenthalpy
and enthalpyofof
enthalpy of steam,
steam,
steam, as
asas
a
a
a function
function of
of saturation
saturation pressure
pressure through
through linear
linear regression
regression (curve
(curve fitting)
fitting) in
in the
the EES
EES
®
® software.
® software.
function of saturation pressure through linear regression (curve fitting) in the EES software.

Figure 88 shows
Figure
Understanding showswhat that the
that the temperature
wastemperature is increasing
is
done in the graph, increasing since,
since,
the curve asperformed,
as
fit is the saturation
the saturation pressure
pressure
resulting in
increases,
increases,
the the temperature
the
polynomials temperature
shown in the approaches
approaches
figure. the the critical
critical point
point where
where the the critical
critical temperature
temperature of of
the water
the water
Figureisis8 374
374
shows °C. that
°C. Knowingthis,
Knowingthis, it is
it
the temperatureis assumed
assumed that the
that
is increasing the saturation
saturation
since, temperature
temperature
as the saturation of the
of
pressurethe
liquid is
is equal
increases,
liquid equal to the
the saturation
the temperature
to saturation temperature
approaches of the
the metal
the critical
temperature of metal (Tsat ==the
point where
(T )) ..
Tmcritical temperature
◦ sat Tm
of theThus,
waterby
Thus, byis knowingthe
374 C. Knowingthis,
knowingthe temperatureit is assumed
behavior, that
an the saturation
adjustment
temperature behavior, an adjustment of the cross isof the temperature
cross of and
is made
made the
and
liquid
thus we
thus is equal
we obtain to
obtain the the saturation
the polynomial temperature
polynomial shown
shown in in the of the
the figure.
figure. metal ( T sat = T m ) .

Figure8.
Figure
Figure 8. Graph
8. Graph used
Graph used to
to determine thethe saturation temperature
saturation temperature polynomial
temperature polynomial using
polynomialusing linear
usinglinear regression
linearregression
regression
(curvefitting)
(curve
(curve fitting)in
fitting) inthe
in theEES
the EES®®® software.
EES software.
software.

The polynomials
Thus,
The polynomials
by knowingtheshown in the
the previous
temperature
shown in previous figures
behavior, are used
used of
an adjustment
figures are tothe
to solve theisdifferential
cross
solve the differential
made and
equations
thus that represent
we obtain
equations that represent the behavior
the polynomial
the behavior
shown ofof the
in steam generator.
the figure.
steam generator.
These polynomials
These polynomials apply
apply only
only to
to the
the drum,
drum, since
since this
this is
is where
where there
there are
are two-phase
two-phase
liquid–steam conditions (mixture).
liquid–steam conditions (mixture).

3.2. Analysis
3.2. Analysis of
of Simulation
Simulation Results
Results
Fluids 2021, 6, 175 10 of 14

The polynomials shown in the previous figures are used to solve the differential
equations that represent the behavior of the steam generator.
These polynomials apply only to the drum, since this is where there are two-phase
liquid–steam conditions (mixture).

Fluids 2021, 6, x FOR PEER REVIEW 3.2. Analysis of Simulation Results 11 of 15

Figure 9 shows the behavior of the pressure, which is obtained using the mentioned
numerical methods. These results are compared with the data shown in [6] and it is
served that
observed the
that Runge–Kutta
the Runge–Kuttamethod
methodandandthe
theOde45
Ode45function
functionare
are much
much closer to
to the data
the data
of
of the reported pressure
the reported pressure graph,
graph, while
while the
the Heun’s
Heun’smethod
methodresult
resultisisless
lessexact.
exact.

Figure 9.
Figure 9. Pressure
Pressure behavior
behavior with
with different
different solution
solution methods
methods (Ode45,
(Ode45, Runge–Kutta,
Runge–Kutta,Heun).
Heun).

Figure
Figure 10 10shows
showsthe behavior
the behavior of the totaltotal
of the volume of water
volume obtained
of water using using
obtained the numerical
the nu-
methods of Runge–Kutta of the fourth order, Heun, and the Ode45 function of MATLAB ®.
merical methods of Runge–Kutta of the fourth order, Heun, and the Ode45 function of
In the figure,
MATLAB it can
®® . In the be seen itthat
figure, canthe
beRunge–Kutta
seen that themethod and themethod
Runge–Kutta Ode45 function
and the are the
Ode45
ones that are
function would produce
the ones thatresults
wouldclose to the
produce results
results reported
close to the in [6]. reported in [6].
results

10. Behavior
Figure 10.
Figure Behavior of the total
of the total volume
volume of
of water
water with
with the
thedifferent
differentsolution
solutionmethod
method(Ode45,
(Ode45,Runge–
Runge–Kutta,
Kutta, Heun). Heun).

Figure
Figure 11 11 shows
shows the
the behavior
behavior of
of the
the quality
quality of
of the
the obtained
obtained mixture
mixture using
using the
the nu-
nu-
merical methods of Runge–Kutta of the fourth
merical methods of Runge–Kutta of the fourth order,
order, Heun,
Heun, and
and the
the Ode45
Ode45 function
function of
of
MATLAB ®
®® . The
The figure shows that the results of the Runge–Kutta method and Heun’s
MATLAB . figure shows that the results of the Runge–Kutta method and Heun’s
method are further from the results reported in [6], while the Ode45 function of
MATLAB®® is more accurate to the reported data.
Fluids 2021, 6, 175 11 of 14
Fluids 2021, 6, x FOR PEER REVIEW 12 of 15

Fluids 2021, 6, x FOR PEER REVIEW method ®

12 of 15
are further from the results reported in [6], while the Ode45 function of MATLAB
is more accurate to the reported data.

Figure 11. Behavior of the mixture quality with the different solution methods (Ode45,
Runge–Kutta, Heun).
Figure 11. Behavior
Figure Behaviorofofthe
themixture
mixturequality
qualitywith
withthethe
different solution
different methods
solution (Ode45,
methods (Ode45, Runge–
3.3. ValidationHeun).
Runge–Kutta, Using Approximation Error of the Results
Kutta, Heun).
To understand the dynamic operation of the boiler, it is important to select variables
3.3.
3.3. Validation
Validation
to achieve theUsing
Using Approximation
Approximation
physical interpretationError
Errorofof
of
thethe Results
theprocess,
Results which describe the storage of mass,
To
energy,To understand the dynamic operation of the boiler,
understand
and momentumthe dynamic
in the operation
system. The of the
three ititisisimportant
predominant
boiler, to
toselect
variables
important in the
select variables
process
variables
to
to achieve the physical interpretation of the process, which describe the storage of mass,
achieve
are: the the
total physical
volume ofinterpretation
water of
represents the process,
the which
accumulation describe
of water the
in storage
the of
system, the
mass,
energy,
pressureand
energy, and momentum
P inmomentum in the
the drum represents system.
in the system. The three
the energy
The three predominant
received variables
by the system,
predominant in the
variablesand, process
finally,
in the are:
the
process
the
are: total
quality
the of volume
total of water
the volume
mixture that
of represents
represents
water thethe
represents accumulation
distribution ofofwater
the accumulation ofinwater
water–steam the system,
that
in the the pressure
represents
system, the
Ppressure
in the drum
fraction in represents
ofP steamthe in the
the risers
drum energy received
[14]. Figure
represents the energy12by the
shows system, byand,
the difference
received finally, the
between
the system, quality
and, the of the
pressure
finally, the
mixture
data that
obtained, represents
using the
the distribution
different of
numerical water–steam
methods, that
and represents
the
quality of the mixture that represents the distribution of water–steam that represents the the
pressure fraction
reports of steam
shown in
in the
[6]. risers
This [14].
differenceFigure
in 12 shows
results is the difference
obtained using between
Equation the pressure
(20)
fraction of steam in the risers [14]. Figure 12 shows the difference between the pressure and data
thus obtained,
determines using
the
the
data different
approximation
obtained, numerical
error the
using methods,
with each ofand
different the the pressure
methods
numerical andreports
methods,the and shown
report the[6]. in [6]. This
pressure difference
reports shown in in
results
[6]. This is
When obtained
difference using
conducting Equation
this study,
in results (20) and
it can be
is obtained thus determines
seenEquation
using that the three the approximation
(20) andsolution error
goals are in
thus determines withan
the
each
errorofrange
the methods
approximation errorand
between withthe
2.2% report
and
each 2.5%. [6].methods and the report [6].
of the
When conducting this study, it can be seen that the three solution goals are in an
error range between 2.2% and 2.5%.

Figure 12.
Figure 12. The approximation error
error for
for the
the total
totalvolume
volumeof
ofwater
waterininthe
theboiler
boilerwith
withrespect
respecttotothe
the
referencevolume.
reference volume.
Figure 12. The approximation error for the total volume of water in the boiler with respect to the
When conducting
In Figure
reference this study,that
12, it is observed
volume. it can
thebe seenmethods
three that the three
used solution goals are
present errors in an
below error
2.5% in
range between 2.2%
the calculation of theand 2.5%. and this indicates that they can be valid to determine this
pressure,
variable, however,
In Figure 12, it iswith the that
observed RK4themethod, a lowererror
three methods is obtained
used present in the
errors below 2.5%re-
in
the calculation of the pressure, and this indicates that they can be valid to determine this
variable, however, with the RK4 method, a lowererror is obtained in the re-
Fluids 2021, 6, 175 12 of 14

Fluids 2021,
Fluids 2021, 6,
6, xx FOR
FOR PEER
PEER REVIEW
REVIEW 13 of
13 of 15
15
In Figure 12, it is observed that the three methods used present errors below 2.5%
in the calculation of the pressure, and this indicates that they can be valid to determine
this variable, however, with the RK4 method, a lower error is obtained in the results,
sults,stabilizing
stabilizing its value
its value aroundaround
2%. For 2%.
theFor the pressure
pressure calculation,
calculation, the highest
the highest error per-
error percentages
are presented
centages when using
are presented theusing
when Ode45 thefunction with a maximum
Ode45 function value ofvalue
with a maximum 2.5% offor2.5%
the
for the simulated time. Figure 13 shows the difference between the total
simulated time. Figure 13 shows the difference between the total water volume data water volume
data obtained,
obtained, using using the different
the different numerical
numerical methods,methods,
and the and the reports
volume volumeshown
reportsinshown in
[6]. This
difference in results isinobtained
[6]. This difference results using Equation
is obtained (20)Equation
using and thus determines the determines
(20) and thus approximation the
error with each error
approximation of thewith
methods
each and themethods
of the report [6].
and the report [6].

Figure 13.
Figure
Figure 13. Approximation
13. Approximation error
Approximation error for
error for calculations
for calculations of
calculations of total
of total water
total water volume
water volumein
volume inthe
in theboiler.
the boiler.
boiler.

Analyzing
Analyzing Figure
Figure 13,
13, it
it can
can be
be seen
seen that
that the
the three
three methods
methods present
present aa good
good behavior
behavior
to
to determine
determine the
the total
total volume
volume of of water
water with
with practically
practically equal
equal percentages
percentages of of error
error and
and
minimum
minimum values
values that
that are
are below
below 0.6%
0.6% during
during the thesimulated
simulatedoperating
operatingtime.
time.
Figure
Figure 14
14shows
showsthe difference
the differencebetween
between the the
quality datadata
quality of theofmixture obtained,
the mixture using
obtained,
the different
using numerical
the different methods
numerical and theand
methods quality reportsreports
the quality shownshown
in [6]. This
in [6].difference in
This differ-
results
ence inisresults
obtained using Equation
is obtained (20) and thus
using Equation (20) determines the approximation
and thus determines error with
the approximation
each
errorof the each
with methods and
of the the report
methods andin [6].
the report in [6].

Figure 14.
Figure
Figure 14. Approximation
14. Approximation error
Approximation error for
error for the
for the quality
the quality of
quality of the
of the mixture
the mixturein
mixture inthe
in theboiler.
the boiler.
boiler.

Figure
Figure 14
14 shows
shows that
that for
for Heun’smethod
Heun’smethod andand the
the RK4
RK4 function
function in
in the
the time
time range
range
between
between 75 and 100 s, the error obtained in the calculation of the quality of the mixture
75 and 100 s, the error obtained in the calculation of the quality of the mixture
tends to stabilize at a value of around 7% after an accelerated increase during the first
seconds of the simulation. Undoubtedly, the Ode45 function allows for obtaining the
lowest percentage of error in the calculation of the quality of the mixture, stabilizing its
Fluids 2021, 6, 175 13 of 14

tends to stabilize at a value of around 7% after an accelerated increase during the first
seconds of the simulation. Undoubtedly, the Ode45 function allows for obtaining the lowest
percentage of error in the calculation of the quality of the mixture, stabilizing its value
around a maximum of 4.5%, much lower than those reached by RK4 and Heun’smethod of
7.5% and 8.5%, respectively.

4. Discussion
In the present work, the Runge–Kutta numerical method of order 4, Heun’s numerical
method, and the MATLAB® Ode 45 function were used to reproduce experimental values
of the reference [6], which simulates the behavior of pressure, total volume of water, and
quality of the mix in the boiler drum. Based on the results obtained, it is valid to affirm that
the three methods can be used solve the ordinary differential equation and represent in
general terms the dynamic behavior of the three most important variables in the operation
of the boiler, however, when analyzing the percentage of error in the results, it is evidenced
that the RK4 method is the best option to represent the behavior of the pressure in the drum
and the Ode4 function to determine the quality of the mix. The three methods present very
good results to determine the total volume of water with minimum percentages of error
that do not exceed 0.6%.

Author Contributions: The author C.Á. contributed to the conceptualization. He also programmed
and validated the simulation of the applied model, supervised the execution of the project, prepared
the original draft of the work, and made the adjustments based on the editing modifications. The
author E.E. contributed to the conceptualization and definition of the methodology applied in the
development of the work, as well as in the writing of the original draft and its subsequent revision.
The author C.J.N. contributed to the programming of the code in the software, supervised the
conservation of the data and validation of the model, as well as the visualization of the results
obtained, and also contributed to the revision and editing of the document. All authors have read
and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Acknowledgments: The project was developed with the institutional support of Universidad Fran-
cisco de Paula Santander Ocaña, Colombia.
Conflicts of Interest: The authors declare no conflict of interest.

Nomenclature

Q Heat flux
.
mf Mass flow of feed water
.
ms Mass flow of steam
hs Saturated enthalpy of steam
hw Saturated enthalpy of water
hf Enthalpy of feed water
Cp Specific heat
Vr Raiser volume
Adc Down-comer area
Vt Total volume
Vm Total mass
g Gravity
k Constant
Fluids 2021, 6, 175 14 of 14

References
1. International Energy Agency IEA. The Global Authority on Energy. 2018, Volume 12, p. 8. Available online: https://www.iea.org
(accessed on 10 October 2020).
2. Nordell, B. Thermal pollution causes global warming. Glob. Planet. Chang. 2003, 38, 305–312. [CrossRef]
3. Sunil, P.; Barve, J.; Nataraj, P. Mathematical modeling, simulation and validation of a boiler drum: Some investigations. Energy
2017, 126, 312–325. [CrossRef]
4. Butler, R.M. Thermal Recovery of Oil and Bitumen; Prentice Hall: Edmonton, AB, Canada, 1991.
5. Haagen, M.; Zahler, C.; Zimmermann, E.; Al-Najami, M.M.R. Solar Process Steam for Pharmaceutical Industry in Jordan. Energy
Procedia 2015, 70, 621–625. [CrossRef]
6. Aström, K.; Bell, R. Drum-boiler dynamics. Automatica 2000, 36, 363–378. [CrossRef]
7. Äström, K.J.; Bell, R.D. Simple drum-boiler models. Power Syst. Model. Control Appl. 1988. [CrossRef]
8. Adam, E.; Marchetti, J. Dynamic simulation of large boilers with natural recirculation. Comput. Chem. Eng. 1999, 23, 1031–1040.
[CrossRef]
9. Bhambare, K.; Miltra, S. Modeling of a Coal-Fired Natural Circulation Boiler. J. Energy Resour. Technol. 2007, 129, 159–167.
[CrossRef]
10. Taimoor, A.; Alghamdi, M.; Farooqi, O.; Siddiqui, M.; Zain-ul-abdein, M.; Saleem, W.; Ijaz, H.; Ali, A. Comprehensive Dynamic
Modeling, Simulation, and Validation for an Industrial Boiler Incident Investigation. Process Saf. Prog. 2019, 38, e12040. [CrossRef]
11. Tognoli, M.; Najafi, B.; Marchesi, R.; Rinaldi, F. Dynamic modelling, experimental validation, and thermo-economic analysis of
industrial fire-tube boilers with stagnation point reverse flow combustor. Appl. Therm. Eng. 2019, 149, 1394–1407. [CrossRef]
12. González, P.; Gómez, J.; Ferruzza, D.; Haglind, F.; Santana, D. Dynamic performance and stress analysis of the steam generator of
parabolic trough solar power plants. Appl. Therm. Eng. 2019, 147, 804–818. [CrossRef]
13. Keshavar, O.; Jafarian, A.; Shekafti, M. Dynamic simulation of a heat recovery steam generator dedicated to a brine concentration
plant. J. Therm. Anal. Calorim. 2018, 135, 1763–1773. [CrossRef]
14. Ahmed, S.; Elhosseini, M.; Arafat Ali, H. Modelling and practical studying of heat recovery steam generator (HRSG) drum
dynamics and approach point effect on control valves. Ain Shams Eng. J. 2018, 9, 3187–3196. [CrossRef]
15. Huan, L.; Weide, H. A water level dynamics simulation model of AP1000’s steam generator based on Åström-Bell model. In IOP
Conference Series: Earth and Environmental Science; IOP Publishing: Bristol, UK, 2019; Volume 310. [CrossRef]
16. Griffiths, D.V.; Smith, I.M. Numerical methods for engineers. In Numerical Methods for Engineers, 2nd ed.; Mcgraw-Hill: New York,
NY, USA, 2006.
17. Caiza, W. Métodos Numéricos; Prentice Hall: Madrid, Spain, 2015.
18. MathWorks. Documentation: MATLAB. 2015. Available online: https://la.mathworks.com/ (accessed on 10 October 2020).
19. Mezo, J. Encuestas y Margen de Error: Unaguíapráctica. 2015. Available online: http://www.cuadernosdeperiodistas.com/
encuestas-y-margen-de-error-una-guia-practica/ (accessed on 10 October 2020).