INTERNATIONAL JOURNAL OF ENERGY RESEARCH

Int. J. Energy Res. 2010; 34:20–35

Published online 20 April 2009 in Wiley InterScience
(www.interscience.wiley.com). DOI: 10.1002/er.1548

A thermal nonlinear dynamic model for water tube drum boilers

M. A. Habib1,,y, H. E. Emara-Shabaik2, I. Al-Zaharnah1, and T. Ayinde1

1
Mechanical Engineering Department, King Fahd University of Petroleum & Minerals, Dhahran 31261, Saudi Arabia
2
System Engineering Department, King Fahd University of Petroleum & Minerals, Dhahran 31261, Saudi Arabia

SUMMARY

A thermal model for the prediction of possible tube overheating was developed. The model incorporates a nonlinear
state space dynamic model that captures the important physical interactions of the main variables of steam generation
in naturally circulated water tube drum boilers. This paper provides an investigation of the dynamic effects of rapid rise
in fuel flow rate (heat input) on the thermal and flow characteristics of the riser tubes in natural circulation water tube
boilers. The system under consideration includes the drum, riser and downcomer as its major components. The dynamic
response of the system’s state variables due to rapid rises in fuel flow rates was investigated.
The results show that the sudden rise in the firing rate is followed by an increase in the steam quality, which is
accompanied by a decrease in the circulation rate as a result of increase in the pressure. The riser temperature increases
slightly above the saturation temperature due to the increase in the steam temperature and due to the dynamic influence
resulting from sudden increase in the heat flux. The present calculations of the water level in the drum provide good
comparison with those in the literature. Copyright r 2009 John Wiley & Sons, Ltd.

KEY WORDS: water tube boilers; thermal models; dynamic model

1. INTRODUCTION downcomer circuits and the steam/water mixture

in the riser tubes. This flow must be adequate to
Rapid increase in the fuel flow rate causes cool the tubes and prevent overheating. Normally,
increased rates of evaporation and thus causes overheating of riser tubes are due to one of two
high increase in the quality that causes tube reasons. The first occurs under steady-state condi-
overheating. The problem impact is not only due tions when the tubes are partially blocked such
to the cost of replacing defective parts but also that the pressure losses increase. As a result the
due to the frequent need of system shutdown and mass flow rate of water decreases at the same heat
the possible imminent safety hazards. The motive flux causing less heat transfer to the water and
force driving the steam/water mixture through consequent overheating of the tube metal. The
the tubes in a natural circulation system is the second cause occurs when sudden rise in the heat
difference in density between cooler water in the flux occurs and is followed by an increase in the

*Correspondence to: M. A. Habib, Mechanical Engineering Department, King Fahd University of Petroleum & Minerals, Dhahran
31261, Saudi Arabia.
y
E-mail: mahabib@kfupm.edu.sa

Received 8 November 2008

Revised 23 February 2009
Copyright r 2009 John Wiley & Sons, Ltd. Accepted 7 March 2009
 A THERMAL NONLINEAR DYNAMIC MODEL 21

steam quality. As a result, circulation decreases as robustness in the event of faults. A digital control
a result of increase in the pressure. The present system employing optimal control has been
study is related to the second cause. developed [3] for a thermal power generator. The
Water circulation control of steam generation is system is designed to coordinate the control of the
an important problem that must be considered for excitation system and the governor system of a
plant safety and reliability. Poor control leads to thermal generator to improve the stability and to
frequent shutdown. Water circulation in natural suit practical application. Nomura and Sato [4]
circulation drum boilers is one of the critical applied adaptive optimal control to a coal-fired
problems in boiler technology. Poor water thermal power plant emulator. Drum boiler
circulation may cause tube rupture resulting in model, which runs in real time, has been
unscheduled boiler shutdown that may interrupt developed and validated [5] using dynamic data
plant operation. Such poor circulation may arise recorded on an actual plant. The model can be
from operational-type problems such as rapid rises used for dynamic simulation studies in long time
in boiler load causing rapid rises in the heat flux as frames, greater than 30 s, in particular where
a result of rapid rises in fuel flow rates. It is known assessment of deviations of internal parameters,
that each boiler tube experiences different heating such as steam pressure, drum level and steam
conditions due to the nonuniformity of heat flux temperature outside safety limits is essential.
distribution in the furnace. During the last few Motivated by model-based control, Astorm and
years, some boiler explosions were attributed to Bell [6] developed a nonlinear dynamic model for
poor water circulation [1]. As a result, calculations drum boilers. The goal was to develop a model
and measurements of water circulation and other that captures the key dynamical properties over a
operating parameters, such as steam quality and wide operating range. The model is based on
void fraction, have become more important not physical principles with a small number of
only for boiler manufacturers but also for large parameters. In their work particular attention
industrial establishments as well as insurance was given to the dynamics of the drum level as it
companies. Adequate water circulation is has been recognized that drum level control is an
necessary to cool tubes that form boiler walls. important factor in power plant operation. The
Criteria are required to determine the potential for model has four states; two account for storage of
tube overheating. These criteria can be applied total energy and total mass, one characterizes
using circulation modeling and calculations to steam distribution in the risers and another for the
identify problem areas. Modeling the steam steam distribution in the drum. The model is
generator system including drum boiler, riser and augmented by quadratic approximations of the
downcomer is one of the important problems. steam tables and physical parameters thus derived.
Many previous investigations were conducted with The model has been validated against plant data
the objective of understanding the transient with very rich excitation that covers a wide
behavior of the system. operating range. These experiments have given
Power plants are usually equipped with many insight into the behavior of the system and
local controllers that need to be coordinated to have guided the modeling effort. The model is
collectively reach a global objective under normal nonlinear and agrees well with experimental data.
and abnormal system conditions. Integrated In particular, the complex shrink and swell
autonomy is thus an important part of power phenomena associated with the drum water level
plant operations and control. Preliminary results are well captured by the model. Garduno-Ramirez
of a sliding mode controller retrofit for a thermal and Lee [7] presented a coordinated control
power plant were presented [2]. Simulation results scheme for fossil fuel power plants drive units
indicate that the performance of the sliding mode based on optimal power–pressure mappings. The
controller is better than the existing multi-loop procedure is presented through a case study, and
PID controller, with respect to minimizing its feasibility is demonstrated via simulation
deviations from set points and greater stability experiments. A modeling and control method for

Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2010; 34:20–35
DOI: 10.1002/er
22 M. A. HABIB ET AL.

thermal power plants having nonlinear dynamics conservations. The simulation results were
varying with load was developed by [8]. They compared with experimental measurements.
presented a multivariable nonlinear exponential The above examples clearly demonstrate the
model that may be applied to describe a class of value of dynamic simulation for the investigation
nonlinear systems satisfying certain smooth and development of new control strategies and
conditions. The model can closely describe the enhanced operational procedures of power plants.
nonlinear dynamic behaviors of thermal power The ability to perform transient analysis on the
plants over a wide operating range. proposed equipment with the control algorithm
Habbi et al. [9] developed a nonlinear dynamic prior to the finalization of the design phase assures
fuzzy model for natural circulation drum the supplier and customer of satisfactory
boiler–turbine is presented. The model was operation of both the hardware and the controls
derived from Astrom and Bell [6] nonlinear of power plants. The present study aims at
dynamic system and describes the complicated developing, utilizing nonlinear modeling, a
dynamics of the physical plant. It was shown that numerical procedure for the dynamic simulation
the dynamic fuzzy model gives in some appropriate of natural circulation steam generator system and
sense accurate global nonlinear prediction and at investigating the dynamic response of the pressure
the same time that its local models are close in the drum, steam quality at the exit of the riser
approximations to the local linearization of the tubes to rapid variations in fuel flow rate in order
nonlinear dynamic system. Chen and Shamma [10] to optimize the steam generator operation to
presented a gain-scheduled approach for prevent the overheating of the riser tubes.
boiler–turbine controller design. The controller is
to achieve tracking performance in the power
output and drum pressure while regulating water 2. PROBLEM FORMULATION
level deviation. The boiler–turbine dynamics is
The boiler tube walls are the medium through
characterized by nonlinear dependence on drum
which thermal energy released by the combustion
pressure. Kim and Choi [11] developed a model for
of fuel in the boiler furnace on one side are
water level dynamics in the drum–riser–downcomer
transmitted to the water on the other side during
loop of a natural circulation drum-type boiler. The
the process of steam production. The integrity of
model is based on basic conservation rules of mass,
these walls are essential to the continued and the
momentum, and energy, together with the
safe operation of the boiler and consequently to
constitutional equations. The work provides an
the continuity of service by the power plant. One
investigation of the response of water level
of the main factors is to maintain the temperature
dynamics to changes in steam demand and/or
of the tube walls within safe limits. In order to do
heating rate. The results were compared with
that it will be crucial to identify the factors leading
those of [6]. Daren and Zhiqiang [12] provided
to and predict the eminence of tube overheating.
and evaluated a nonlinear coordinated controller
The differential equations comprising the thermal
through computer simulation using a well-
nonlinear dynamic model and governing the flow
established and validated nonlinear boiler–turbine
inside the boiler tubes were formulated. The
model. The simulation results are presented to
following is a description of the mathematical
illustrate the performance of a nonlinear
models considered in the present study.
coordinated control system and compare the
proposed strategy with the conventional direct
2.1. Boiler modeling
energy balance control strategy. A computer
program, Drum Boiler Start-up Simulation The system considered includes the drum, riser and
Program for simulating the start-up behavior of downcomer. The governing equations consist of
controlled circulation and natural circulation conservation of mass and energy of the total
boilers was developed by [13]. The model is based system, the equations governing the phase change
on the principles of mass, energy and momentum in the drum including the steam and water

Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2010; 34:20–35
DOI: 10.1002/er
 A THERMAL NONLINEAR DYNAMIC MODEL 23

@a
volumes inside the drum and the rate of steam a33 ¼ ðð1  xÞrg þ xrf Þhfg Vr ð10Þ
condensation and the equations governing the flow @x
circulation in the riser–downcomer loop, which @rg 1
govern the transport of the mass, energy and a42 ¼Vsd þ
@P hfg
momentum. Thus, a set of four differential non-  
linear equations [6] representing the time depen- @hg @hf @Ts
 rg Vsd þ rf Vwd  Vsd  Vwd þ Md Cp
dence of the state variables of the pressure, steam @P @P @P
 
quality, total water volume and steam volume in @rg @r @a
þ xð1 þ bÞVr a þ ð1  aÞ f þ ðrg  rf Þ
the drum can be presented as follows: @P @P @P
dVwt dP
a11 þ a12 ¼ m_ w  m_ s ð1Þ ð11Þ
dt dt
dVwt dP @a
a21 þ a22 ¼ Q_ þ m_ w hw  m_ s hg ð2Þ a43 ¼ xð1 þ bÞðrg  rf ÞVr ð12Þ
dt dt @x

dP dx a44 ¼ rg ð13Þ
a32 þ a33 ¼ Q_  xhfg m_ dc ð3Þ
dt dt To close this set of equations, the following
dP dx dVsd relations that are needed for the calculation of
a42 þ a43 þ a44 the mass flow rates in the riser and downcomer as
dt dt dt
well as the mass flow of steam condensation in the
rg 0 hw  hf
¼ ðV  Vsd Þ þ m_ w ð4Þ drum in addition to relations governing the steam
Td sd hfg quality and void fraction are presented in the
The coefficients of these four equations are following:
given by ðhf  hw Þ 1
m_ ct ¼ m_ w þ
a11 ¼ rf  rg ¼ rfg ð5Þ hfg hfg
 
@hg @hf @Ts dP
@rf @rg  rg Vst þ rf Vwt  Vt þ Mt Cp
a12 ¼ Vwt þ Vst ð6Þ @P @P @P dt
@P @P
ð14Þ
a21 ¼ rf hf  rg hg ð7Þ
: :
  mr ¼ mdc Vr
@r @hf
a22 ¼ Vwt hf f þ rf 
@rg

@P @P @r @a
 a þ ð1  aÞ f þ ðrg  rf Þ
  @P @P @P
@rg @hg @Ts
þ Vst hg þ rg  Vt þ Mt Cp dP @a dx
@P @P @P  þ rfg Vr ð15Þ
dt @P dt
ð8Þ rg
  m_ sd ¼ ðVsd  V0sd Þ þ xm_ dc þ xbðm_ dc  m_ r Þ
td
@hf @r
a32 ¼ rf  xhfg f ð1  aÞVr ð16Þ
@P @P
  rg V0sd
@rg @hg td ¼ ð17Þ
þ ð1  xÞhfg þ rg aVr m_ sd
@P @P
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
@a @Ts 2rf Adc rfg gaVr
þ ðrg þ xrfg Þhfg Vr  Vr þ Mr Cp m_ dc ¼ ð18Þ
@P @P k
ð9Þ Vwt ¼ Vwd þ Vdc þ ð1  aÞVr ð19Þ

Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2010; 34:20–35
DOI: 10.1002/er
24 M. A. HABIB ET AL.

 ! !!
r rg rfg r r rfg
a¼ f 1 ln 1 þ x ð20Þ a ¼ f 1  s ln 1 þ x ð32Þ
rfg xrfg rs rfg xrfg rg
 
@a 1 @rg @rf Q_ ¼ m_ dc ðxhfg Þ ð33Þ
¼ r  r
@P ðrf  rg Þ2 f @P g
@P
td ðhf  hw Þ
! Vsd ¼ V0sd  m_ w ð34Þ
rf rg þ rf rg hfg
 1þ  ln ð1 þ ZÞ
rg ð1 þ ZÞ rg Z Vwt ¼ Vwd þ Vdc þ ð1  aÞ  Vr ð35Þ
ð21Þ
rg V0sd
  td ¼ ð36Þ
da rf 1 1 m_ sd
¼ ln ð1 þ ZÞ  ð22Þ
dx Zrg Z 1þZ The thermodynamic properties rf, rg, hf, hg, @r g @r
@P , @P ,
f

@hf @hg @Ts

Z ¼ xðrfg Þ=rg ð23Þ @P , @P , @P and Ts, were obtained from the
thermodynamic property tables [14].
Vwd þ Vsd Vsd
L¼ ; Ls ¼ and 2.3. Calculation of the heat transfer coefficient
Ad Ad
ð24Þ
Vwd The two-phase (liquid and vapor) flow pattern in a
Lw ¼ pipe depends on many parameters such as the flow
Ad
velocity, the quality of the mixture, the properties
In addition, the following equations for the rate of of the two phases (density, viscosity and surface
steam condensation in the drum were used in the tension) and the pipe geometry as well as its
place of Equations (16) and (17) for comparison of orientation. The flow patterns in horizontal tubes
the total water level in the drum and the are slightly different due to the effect of gravity.
contributions due to steam and water. The characteristics of the two-phase flow inside

ðhf  hw Þ 1 @hg @hf vertical and horizontal tubes are presented as
m_ cd ¼ m_ w þ r Vsd þ rf Vwd
hfg hfg g @P @P follows. The flow pattern in a given tube can be
 determined from the flow pattern maps in which
@Ts dP the region of operation of each flow pattern is
ðVsd þ Vwd Þ þ Md Cp ð25Þ
@P dt indicated [15, 16]. As shown by [16–18] and many
others, the flow patterns that are normally
rg ðV0sd  Vsd Þ
td ¼ ð26Þ encountered in vertical pipes are bubble flow, slug
m_ cd flow, churn flow, annular flow and wispy annular
flow. The flow patterns in horizontal tubes are
2.2. Steady state conditions slightly different due to the effect of gravity. These
patterns may take any of the forms of stratified
To obtain the initial conditions, the following
flow, plug flow, stratified wavy flow, dispersed
model equations at equilibrium conditions were
bubbly flow, annular flow and intermittent flow.
solved.
The flow pattern can be predicted for a given flow
m_ w ¼ m_ s ð27Þ condition using the semi-theoretical method of [19]
in which a flow pattern map is represented in terms
Q_ ¼ m_ s ðhg  hw Þ ð28Þ
of a number of dimensionless parameters.
m_ r ¼ m_ dc ð29Þ A large number of empirical and semi-empirical
correlations are available to predict heat transfer
m_ ct ¼ m_ w ðhf  hw Þ=hfg ð30Þ coefficients for the flow boiling regimes in vertical
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi and horizontal tubes. Some of these correlations
2rf Adc ðrf  rg ÞgaVr have been widely tested and are practically used.
Q_ ¼ xhfg ð31Þ
k Some lack the validation. Examples of these are

Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2010; 34:20–35
DOI: 10.1002/er
 A THERMAL NONLINEAR DYNAMIC MODEL 25

the models of [20–29]. A comparison of these 2.4. Wall temperature calculations

models with experimental data was conducted by
The following equations were used to calculate the
[30]. The model proposed by [27] was used to
inside and outside wall temperatures of the riser
calculate the heat transfer coefficient in the present
tube:
water circulation loop. The equation proposed by
[27] for the calculation of the heat transfer Ti  Tsat To  Ti
q00 pDo ¼ ; q00 pDo ¼ ð44Þ
coefficient includes a convection evaporation Rconv Rpipe
term and nucleate boiling term and is expressed as
where q00 is the heat flux at the outside surface of
h ¼ hL ½C1 ðCOÞC2 ð25Frl ÞC5 þ C3 BOC4 Fk  ð37Þ the tubes, Ti and To are the wall temperatures at
where hL is the single phase heat transfer the inner and outer walls of the tube. Tsat is the
coefficient and is calculated from the Dittus– water saturation temperature. Rconv and Rpipe are
Boelter equation [31]. the resistances of the inner water film and the
pipe thickness. Rconv and Rpipe are calculated from
Nu ¼ 0:023ðReÞ0:8 ðPrÞ0:4 ð38Þ the following relations:
where Re is the Reynolds number of the flow inside lnðDo =Di Þ 1
Rpipe ¼ ; Rconv ¼ ð45Þ
the tube and Pr is the Prandtl number for water. 2pKpipe pDi h
Gð1  xÞDi where Kpipe is the thermal conductivity of the pipe
Re ¼ ð39Þ
ml material. h is the heat transfer coefficient of the
Fk is a fluid dependent parameter (51 for water) water film inside the tube.
G2 2.5. The solution procedure
Fr is the Froude number ¼ ð40Þ
r2L gD
Equations (1)–(4) were solved simultaneously
G ¼ mass flux using an explicit method time step is 0.3 s for a
total time 150 s. The coefficients in these equations
m_ were obtained from Equations (5)–(13). These
¼ and D is the pipe diameter ð41Þ
A equations together with Equations (14)–(26) and
(37)–(45) were used to calculate the heat transfer
q00
BO is the boiling number ¼ ð42Þ coefficient and metal temperature as well as other
Ghfg
main boiler parameters. The operational data of
the simulated system are given in Table II. The
CO ¼ convection number
initial values were obtained through the solution
   
1  x 0:8 rv 0:5 of Equations (27)–(36) and are taken as given in
¼ ð43Þ Table III. Time step independence tests were
x rl
conducted and the results are shown in Figure 1.
x is the steam quality and the constants C1, C2, C3,
C4, C5 are given in Table I.
Table II. The operational data of the simulated system.
Table I. The thermal model constants.
Parameter Value Parameter Value
COp0.65 COX0.65 Comments
Drum 8.5 MPa Riser 37 m3
C1 1.136 0.6683 pressure volume
C2 0.9 0.2 Steam 49.9 kg s1 Downcomer 11 m3
C3 667.2 1058 flow rate volume
C4 0.7 0.7 Heating 85.9 MW Water surface 20 m2
C5 0.3 0.3 C5 5 0 for vertical tubes value area
and 5 0 for horizontal tubes Drum 40 m3 Feed water 238.61C
if Fr is greater than 0.04 volume temperature

Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2010; 34:20–35
DOI: 10.1002/er
26 M. A. HABIB ET AL.

The test was performed for the changes in drum 3. VALIDATION

pressure in response to step variations in firing rate
In order to validate the present nonlinear dynamic
at three different values of time steps. The figure
model, the responses of the different boiler
indicates that the model captures the changes in
parameters to step variations in firing rate were
pressure with maximum variations of 0.8% as the
compared with available experimental data. Vali-
time step is reduced from 3 to 1 s and 0.2% as the
dation of the model was performed through
time step is reduced from 1 to 0.3 s.
comparison of the present results against the
available data [6,32]. The measurements provide
Table III. The initial data (steady-state conditions) of firing rate variations with time as shown in
the simulated system. Figure 2. The responses of drum pressure to these
variations were recorded. The results of the
Parameter Units Value
present calculations are shown in Figure 3 for
Pressure, P kPa 8500 response to step variations in firing rate. The
Energy input, Q MW 85.9 results show that the model captures the variations
Steam mass flow rate, m_ s kg s1 49.9
Feedwater mass flow rate, m_ w kg s1 49.9
in pressure. A reasonably good agreement is
achieved.

10000

9500

9000
Pressure (KPa)

8500

8000

7500
Time step= 0.3 s
Time step= 1 s
7000 Time step = 2 s
Time step = 3 s
6500
0 500 1000 1500 2000 2500 3000 3500
Time (s)
Figure 1. Response of drum pressure to variations in firing rates (time step independence).

120
Firing rate (Energy

110
input), MW

100

90

80
0 500 1000 1500 2000 2500 3000 3500
Time, s

Figure 2. Experimental data for distribution of firing rate [6].

Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2010; 34:20–35
DOI: 10.1002/er
 A THERMAL NONLINEAR DYNAMIC MODEL 27

10000
Present calculations
Experimental data
9500

9000
P (KPA)
8500

8000

7500
0 500 1000 1500 2000 2500 3000 3500
Time (s)

Figure 3. Comparison of numerical results of response of drum pressure to variations in firing rate with experimental
data of [6].

9400
Step rise = 10 %
9300
Step rise = 20 %

9200
Drum pressure, kPa

9100

9000

8900

8800

8700

8600

8500
40 60 80 100 120 140 160 180 200
Time, s

Figure 4. Response of drum pressure for step rise in heat input.

4. RESULTS rise in the heat input, the rate of pressure increase

is 6.5 kPa/s. In order to explain the increase in the
The calculations were performed to investigate the drum pressure, Figure 5 is provided and presents
influence of step increase in heat input by 10% and the response of the mass flow rates in risers and
by 20% on the different boiler. Parameters such as downcomers for the two cases of step rise in heat
the pressure, the steam quality, the mass flow rates input. The riser mass flow exhibits a sudden
in the riser and downcomer, the volumes of water increase as a result of increase in the heat input.
and steam and the heat transfer coefficient as well As the heat input, Q, _ is increased, a sudden
as the metal temperature. increase from the steady-state value in the mass
Figure 4 presents the change of the pressure as a flow rate in the riser occurs. However, the
result of sudden increase in heat input, Q, _ and downcomer flow exhibits a gradual increase with
shows that the pressure increases at a constant rate time. Thus, the outlet of the riser first increases
of dP/dt 5 3.14 kPa/s as a result of a step rise in and then decreases to match the flow in the
the heat input of 10%. For the case of 20% step downcomer. Thus, the difference between the riser

Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2010; 34:20–35
DOI: 10.1002/er
28 M. A. HABIB ET AL.

Figure 5. Response of mass flow rates in riser and downcomer for step rise in heat input.

0.06
Step rise = 10 %
0.059 Step rise = 20 %

0.058

0.057
Steam quality

0.056

0.055

0.054

0.053

0.052

0.051
40 60 80 100 120 140 160 180 200
Time, s

Figure 6. Response of steam quality for step rise in heat input.

and downcomer mass flow rates, m_ r  m_ dc , higher magnitudes of flow rates in risers and
increases suddenly causing increase in the mass downcomers. As well, the difference between the
flow rate into the drum and the volume of riser and downcomers at start of the rise is much
steam–water in the drum. The difference between higher for the case of 20%. Consistent with the
both of the two flow rates diminishes at around increase in heat input, the quality at exit of the
30 s from the moment of the step rise in heat input. riser increases as a result of increased rate of
This is followed by a continuous gradual decrease evaporation. The quality increases first at a high
in the risers and downcomers mass flow rates. The rate during the first 10 s. This is followed by less
case of 20% step rise shows similar trends but with steeper increase as shown in Figure 6.

Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2010; 34:20–35
DOI: 10.1002/er
 A THERMAL NONLINEAR DYNAMIC MODEL 29

12 20.4
Step rise = 10 %
Step rise = 20 % 20.2

Drum steam volume, m3

Drum water volume, m 3

10
20

19.8
8
19.6

19.4
6
19.2

4 19
40 60 80 100 120 140 160 180 200
Time, s
Figure 7. Response of steam and water volumes in the drum for step rise in heat input.

Figure 8. Response of steam condensation for step rise in heat input.

Figure 7 presents the dynamics of steam and the riser flow rate m_ r and the quality x cause high
water volumes in the drum under the water level. increase in steam flow into the drum, which results
The difference between the mass flow rate in the in increase in Vsd. The increase in Vsd is consistent
risers and downcomers along with the increase in with the calculations of [11]. Figure 7 exhibits
the quality leads to a rise in the mass of water as continuous increase in the water volume in the
well as steam in the drum. As well, as shown in drum with high rate during the first 10 s. The rise
Figure 7, the volume of the steam within the liquid in steam inside the drum under the water level
(under water level) in the drum, Vsd, increases as a reaches a maximum at around 100 s for the case of
result of a combined increase in the riser flow rate 20% and 130 s for the case of 10%, then, it
and the quality at exit of the riser. The increase in decreases. The reduction in the steam volume in

Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2010; 34:20–35
DOI: 10.1002/er
30 M. A. HABIB ET AL.

58.2 31

30.8

Steam volume, m 3
Water volume, m 3 57.8 30.6

30.4

57.4 30.2

Step rise = 10 % 30
Step rise = 20 %
57 29.8
40 60 80 100 120 140 160 180 200
Time, s
Figure 9. Response of total water and steam volumes for step rise in heat input.

14520 15400
Heat transfer coefficient, W/m 2K

Heat transfer coefficient, W/m 2K

Step rise = 10 %
15350
Step rise = 20 %

14480 15300

15250

14440 15200

15150

14400 15100

15050
40 60 80 100 120 140 160 180 200
Time, s
Figure 10. Response of heat transfer coefficient for step rise in heat input.

the drum after 50 s is due to the condensation of Figure 9 shows a decrease in the total steam
steam inside the drum, which causes a new volume, Vst, according to the relation Vst1
addition in the water volume. Figure 8 indicates Vwt 5 V where V is the total volume of the system.
that the rate of condensation as a result of increase The response of heat transfer coefficient, h, is
in pressure increases suddenly and then continues shown in Figure 10 and the response of the liquid
to increase but at a lower rate. The main cause of heat transfer coefficient, hL is presented in Figure
condensation is due to the increase in the pressure 11. As shown in Figure 10, h decreases during the
at the same saturation temperature Ts. Figure 9 first ten seconds. This is followed by increase in h
shows that the total volume of water in the system, with time. Despite that the case of 20% exhibit
Vwt, increases. The increase in Vwt is a result of higher values of h, the rate of increase in h is
increase of Vwd and the condensation of the steam slower for the case of 20%. The temperature
under the water level in the drum. Consistent with distributions are shown in Figure 12. In order to
the increase in the total water volume, Vwt, explain the behavior of the heat transfer

Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2010; 34:20–35
DOI: 10.1002/er
 A THERMAL NONLINEAR DYNAMIC MODEL 31

8200

Liquid heat transfer coefficient, W/m 2K

8150

Step rise = 10 %
8100
Step rise = 20 %

8050

8000

7950

7900

7850
40 60 80 100 120 140 160 180 200
Time, s

Figure 11. Response of liquid heat transfer coefficient for step rise in heat input.

6.25 6.52

6.24 6.5
Temperature difference, °C

Temperature difference, °C
6.48
6.23
6.46
6.22
6.44
6.21
Step rise = 10 % 6.42
Step rise = 20 %
6.2 6.4

6.19 6.38
40 60 80 100 120 140 160 180 200
Time, s
Figure 12. Response of the difference between the metal and saturated steam temperatures for step rise in heat input.

coefficient, the response of the heat transfer the lower rate of increase in the shown above in
coefficient of the liquid, hL, Figure 11, and the Figure 10 and hL in Figure 11. Figure 14 exhibits
quality x, Figure 6, should be investigated. As the the distribution of the void fraction that indicates
pressure increases, rw increases and results in a a rapid increase in the first few seconds then
decrease in the volume flow rate, which, in turn, followed by a continuous reduction in its value. As
leads to reduction in the liquid flow velocity. The a result, the heat transfer coefficient decreases as
heat transfer coefficient of the liquid phase, hL, is shown in Figure 10. This explains the behavior of
influenced by the flow velocity, which is shown in the heat transfer coefficient, h, in Figure 10.
Figure 13. The rate of velocity decrease with time Comparison of response of water levels is
is much higher for the case of 20%. This explains made for the present results of step rise of 10%

Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2010; 34:20–35
DOI: 10.1002/er
32 M. A. HABIB ET AL.

1.17

1.16

1.15 Step rise = 10 %

Step rise = 20 %
Velocity, m/s

1.14

1.13

1.12

1.11
40 60 80 100 120 140 160 180 200
Time, s

Figure 13. Response of liquid flow velocity for step rise in heat input.

0.284

0.282 Step rise = 10 %

Step rise = 20 %
Steam volume fraction

0.28

0.278

0.276

0.274

0.272

0.27
40 60 80 100 120 140 160 180 200
Time, s
Figure 14. Response of steam volume fraction for step rise in heat input.

utilizing Equations (25) and (26) with those of [6], the steam contribution to water level. The three
which utilizes Equations (16) and (17) and also models present well the growth in bubble volume
with the results of [11]. Figure 15 provides a in the first few seconds. However, the present
comparison of the water level due to liquid and calculations and those of [11] show that the
steam contents. The present results confirm those volume of bubble continue to increase at a lower
of [11], however, with higher magnitudes. The rate rather than decrease and then increase as
present calculations and those of References given by [6]. The reason may be related to
[6, 11] present the same water contribution to Equation (16) used by [6], which is based on an
water level. However, the differences are due to empirical model that has to be validated. The

Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2010; 34:20–35
DOI: 10.1002/er
 A THERMAL NONLINEAR DYNAMIC MODEL 33

0.04

0.035

Drum water level, m 0.03

0.025

0.02
Calculations, [6]

0.015 Calculations, [11]

Present calculations
0.01

0.005

0
50 100 150 200
Time, s
Figure 15. Response of the drum level for step rise in heat input.

preliminary results of a current investigation on The results indicate that increasing the heat flux
the influence of step rise in steam flow rate show by 10% and by 20% can lead to high variations in
more close agreement of the present model and pressure, steam quality and water level in the
those of [6] and [11]. drum. As well, the changes in the heat transfer
coefficient lead to a temperature increase in the
riser metal temperature. The riser temperature
increases due to the increase in the steam
5. CONCLUSIONS
temperature and due to the dynamic influence
A thermal model for the prediction of possible tube resulting from increase in the heat flux. The
overheating was developed. The model incorporates results of the water level in the drum are in
a nonlinear state space dynamic model that reasonable comparison with those available in the
captures the important physical interactions of the literature.
main variables of steam generation in natural
circulation water tube drum boilers. The model is
used to investigate the dynamic effects of rapid NOMENCLATURE
changes in fuel flow rate on the overheating of the
riser tubes in natural circulation water tube boilers. A 5 Cross sectional area of the tube
The system under consideration includes the drum, Ad 5 The drum area at normal operat-
riser and downcomer as its major components. ing level
A numerical scheme for the solution of the governing BO 5 Boiling number
differential equations was established. The dynamic CO 5 Convection number
response of the system’s state variables due to rapid C1, C2, C3, 5 Constants used in Equation (37)
changes in fuel flow rate and steam flow rate was C 4, C 5 and their values are given in
investigated. The thermal model was used for the Table I
prediction of possible tube overheating was devised. Fr 5 Froude number

Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2010; 34:20–35
DOI: 10.1002/er
34 M. A. HABIB ET AL.

G 5 Mass flux ug 5 specific internal energy of steam

h 5 Specific enthalpy uw 5 specific internal energy of water
h 5 Heat transfer coefficient V 5 Volume
hfg 5 hfg 5 hg–hf Vd 5 Drum volume
hf 5 specific enthalpy of saturated li- Vdc 5 Downcomer volume
quid water Vr 5 Riser volume
hg 5 specific enthalpy of saturated Vsd 5 Volume of steam under the liquid
water vapor level
hw 5 specific enthalpy of feedwater V0sd 5 The volume of steam in the drum
k 5 dimensionless friction coefficient in the hypothetical situation when
used in Equation (18) there is no condensation of steam
K 5 Thermal conductivity in the drum
L 5 Drum water level Vst 5 Total volume of steam in the
Lw 5 Level variations caused by system
changes of the amount of water Vt 5 The total volume of the drum,
in the drum downcomer and risers; Vt 5 Vst1
Ls 5 Level variation caused by the Vwt
steam in the drum. Vwd 5 Volume of water under the liquid
Mf 5 Mass of saturated liquid in water level
walls and drum Vwt 5 Total volume of water in the
Mg 5 Mass of saturated vapor in water system
walls and drum vf 5 Specific volume of saturated liquid
Mr 5 Total riser mass water
Mt 5 Total mass of metal (including vg 5 Specific volume of saturated vapor
tubes and drum) water
m_ cd 5 The condensation flow rate in the x 5 Mass fraction of steam in the flow
drum xo 5 Distance from bottom at which
m_ ct 5 The total condensation flow rate boiling starts
m_ dc 5 The down comer flow rate
m_ f 5 Mass flow rate of feedwater sup- Subscripts
plied to the drum
m_ r 5 The flow rate out of the risers c 5 Condensation
m_ s 5 Mass flow rate of steam exiting the d 5 Drum
boiler to the superheater and dc 5 Downcomer
turbine m 5 Metal
m_ sd 5 Steam flow rate through the liquid s 5 Steam
surface in the drum sat 5 saturates
m_ w 5 Mass flow rate of feedwater f 5 Saturated liquid
Nu 5 Nusselt number fg 5 Difference between vapor and
P 5 Drum pressure, kPa liquid
Q_ 5 Heat flow rate to the risers r 5 Riser
00
q 5 Heat flux t 5 Total
td 5 Residence time of the steam in the w 5 Feedwater
drum
DT 5 TmetalTsat Greek Symbols
Tm 5 Metal temperature
Ts 5 Saturated steam temperature a 5 Volume fraction of steam in the
u 5 Specific internal energy flow

Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2010; 34:20–35
DOI: 10.1002/er
 A THERMAL NONLINEAR DYNAMIC MODEL 35

b 5 Parameter in empirical formula, circulation boiler start up behavior. Energy Conversion and
b 5 0.3 Management 2005; 46:533–549.
14. International Formulation Committee (IFC). A formula-
r 5 Density tion of the thermodynamic properties of ordinary water
rg 5 Density of saturated steam substance. The 1967 IFC Formulation for Industrial Use,
rw 5 Density of feedwater 1967.
15. Hewitt GF, Roberts DN. Studies of two-phase flow
patterns by simultaneous X-ray and flash photography.
Report AERE-M2159, Atomic Energy Research Establish-
ACKNOWLEDGEMENTS ment, Harwell, 1969.
16. Hewitt GF. Gas– Liquid Flow, in Handbook of Multiphase
Systems. Hemisphere Publishing: New York, 1982.
The authors wish to acknowledge the support received 17. Collier JG. Convective Boiling and Condensation. McGraw-
from King Fahd University of Petroleum & Minerals Hill Book Company (UK) Ltd.: Maidenhead, 1972.
during this study. 18. Hewitt GF, Shires GL, Bott PR. Process Heat Transfer.
CRC Press: USA, 1994.
19. Taitel Y, Dukler AE. A model for predicting flow regime
transitions in horizontal and near horizontal gas-liquid
REFERENCES
flow. AIChE Journal 1976; 22:47–55.
20. Chen JC. A correlation for boiling heat transfer to
1. Pereyra EJ, Shah BA, Edwards LL. Circulation rates and saturated fluids in vertical flow. Industrial & Engineering
tube burnout in natural circulation boilers. Tappi Journal Chemistry, Process Design and Development 1966; 5(3):
1989; 72:143–148. 322–339.
2. Sreedhar R, Fernandez B, Masada GY. Sliding control 21. Shah MM. A new correlation for heat transfer during
retrofit for a thermal power plant. Proceedings of the 4th boiling flow through pipes. ASHRAE Transactions 1976;
IEEE Conference on Control Applications, Albany, NY, 82(2):66–86.
U.S.A., 28–29 September 1995; 409–414. 22. Shah MM. Chart correlation for saturated boiling heat
3. Yokokawa S, Ueki Y, Tanaka H, Doi H, Ueda K, transfer: equations for further study. ASHRAE Transac-
Taniguchi N. Multivariable adaptive control for a thermal tions 1982; 88(1):185–196.
generator. IEEE Transactions on Energy Conversion 1988; 23. Bjorge RW, Hall GR, Rohsenow WM. Correlation of
3(3):479–486. forced convection boiling heat transfer data. Inter-
4. Nomura M, Sato Y. Adaptive optimal control of steam national Journal of Heat and Mass Transfer 1982;
temperatures for thermal power plants. IEEE Transactions 25:753–757.
on Energy Conversion 1989; 4(1):25–33. 24. Gungor KE, Winterton RHS. A general correlation for
5. Flynn ME, O’ Malley MJ. A drum–boiler model for long flow boiling in tubes and annuli. International Journal of
term power system dynamic simulation. IEEE Transactions Heat and Mass Transfer 1986; 29:351–358.
on Power Systems 1999; 14(1):209–217. 25. Klimenko VV. A general correlation for two-phase flow
6. Astrom KJ, Bell RD. Drum-boiler dynamics. Automatica heat transfer. International Journal of Heat and Mass
2000; 36:363–378. Transfer 1988; 31:541–552.
7. Garduno-Ramirez R, Lee KY. Multiobjective optimal 26. Klimenko VV. A generalized correlation for two-phase
power plant operation through coordinate control with forced flow heat transfer, second assessment. International
pressure set point scheduling. IEEE Transactions on Energy Journal of Heat and Mass Transfer 1990; 33:2073–2088.
Conversion 2001; 16(2):115–122. 27. Kandlikar SG. A general correlation for saturated
8. Peng H, Ozaki T, Haggan-Ozaki V, Toyoda Y. A nonlinear two-phase flow boiling heat transfer inside horizontal
exponential ARX model-based multivariable generalized and vertical tubes. Journal of Heat Transfer 1990;
predictive control strategy for thermal power plants. IEEE 112:219–228.
Transactions on Control Systems Technology 2002; 28. Steiner D, Taborek J. Flow boiling heat transfer in vertical
10(2):256–262. tubes correlated by an asymptotic model. Heat Transfer
9. Habbi H, Zelma M, Bouamama B. A dynamic fuzzy model Engineering 1992; B(2):43–69.
for a drum-boiler-turbine system. Automatica 2003; 29. Darabi J, Salehi M, Saeedi MH, Ohadi MM. Review of
39:1213–1219. available correlations for prediction of flow boiling heat
10. Chen P, Shamma JS. Gain–scheduled L1-optimal control transfer in smooth and augmented tubes. ASHRAE
for boiler–turbine dynamics with actuator saturation. Transactions 1995; 101(1):225–235.
Journal of Process Control 2004; 14:263–277. 30. Basu P, Cheng LM. An experimental and theoretical
11. Kim H, Choi S. A model on water level dynamics in natural investigation into the heat transfer of a finned water wall
circulation drum-type boilers. International Communica- tube in a circulating fluidized bed boiler. International
tions in Heat and Mass Transfer 2005; 32:786–796. Journal of Energy Research 2000; 24(4):291–308.
12. Daren Y, Zhiqiang X. Nonlinear coordinated control of 31. Dittus FW, Boelter LMK. Heat transfer in automobile
drum boiler power unit based on feedback linearization. radiators of the tubular type. Publ. Eng., vol. 2, University
IEEE Transactions on Energy Conversion 2005; 20(1): of California, Berkeley, 1930; 443.
204–210. 32. Astrom KJ, Eklund K. A simplified non-linear model of a
13. Li B, Chen T, Yang D. DBSSP—A computer program for drum boiler–turbine unit. International Journal Control
simulation of controlled circulation boiler and natural 1972; 16(1):145–169.

Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2010; 34:20–35
DOI: 10.1002/er