INTERNATIONAL JOURNAL OF ENERGY RESEARCH

Int. J. Energy Res. 2011; 35:897–908

Published online 21 July 2010 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/er.1747

A new method for optimum heating of steam boiler

pressure components
Jan Taler and Piotr Dzierwa

Department of Thermal Power Engineering, Faculty of Mechanical Engineering, Cracow University of Technology, Al. Jana Paw"a II
37, 31-864 Cracow, Poland

SUMMARY
A method for determining time-optimum medium temperature changes is presented. The heating and cooling of
the pressure elements will be conducted in such a way that the circumferential stresses caused by pressure and fluid
temperature variations at the edge of the opening and point of stress concentration do not exceed the allowable
values. However, the calculated optimum temperature changes are difficult to follow in practice during the initial
stage of heating. Nevertheless, it is possible to increase the fluid temperature stepwise to the minimum value and
then heat the pressure component according to the determined optimum temperature changes. Allowing stepwise
fluid temperature increase at the beginning of heating ensures that the heating time of a thick-walled component is
shorter than the heating time resulting from the calculations according to EN 12952-3 European Standard or TRD
301 (Technische Regeln für Dampfkessel) regulations. Copyright r 2010 John Wiley & Sons, Ltd.

KEY WORDS
boiler drum; thermal stresses; heating optimization; inverse heat conduction problem

Correspondence
*Jan Taler, Department of Thermal Power Engineering, Faculty of Mechanical Engineering, Cracow University of Technology, Al. Jana
Paw"a II 37, 31-864 Cracow, Poland.
y
E-mail: [email protected]

Received 12 January 2010; Revised 21 May 2010; Accepted 24 May 2010

1. INTRODUCTION The allowable rate of the temperature change of the

working medium will also be determined using the
Rapid startups and shutdowns of the steam boilers German boiler regulations—TRD 301 [4] or European
cause excessive thermal stresses in thick-walled pressure Standard 12952-3 [5], which are based on a quasi-
components [1–3]. Remnant life of the pressure steady state, one-dimensional temperature distribution
components can be significantly reduced if the heating in the pressure element. The stress concentration
or cooling rates are too high. The boiler drums of the coefficient for stresses caused by pressure and thermal
conventional power plants and feed water nozzles in load at the point P1 (Figure 1) is assumed to be in
the reactors of the nuclear power plants are among the accordance with those regulations. It has to be noted
critical pressure elements [4–10], determining the heat- that only a few types of vessel-nozzle junctions have
ing and cooling rates during the startup and the been considered in the aforementioned regulations. In
shutdown of the power unit. The lifetime of the pressure the proposed method, two points: P1 and P2 (Figure 1)
elements can be reduced significantly if the heating and are considered while determining the allowable tem-
cooling rates are too high. The optimum fluid perature changes of the fluid. The stress concentration
temperature changes can be determined with respect coefficients are determined using the three- dimen-
to the allowable rate of the temperature change [8] or sional stress analysis that is conducted with the use of
taking into account the allowable thermal stress at the the finite element method (FEM). The circumferential
component wall [6,7,9]. A method for determining the stresses, caused by the pressure, are greatest at the
allowable temperature changes of the working medium point P1. Large concentrations of circumferential
during the heating and cooling of the thick-walled thermal stresses are also observed at the point P2.
elements of the power steam boilers will be presented. Complex shapes of the analyzed junctions can easily be

Copyright r 2010 John Wiley & Sons, Ltd.

J. Taler and P. Dzierwa Optimum heating of steam boiler pressure components

Figure 1. Longitudinal section of the boiler drum—downcomer junction.

considered in the analysis of the temperature and stress A three-dimensional stress analysis using FEM was
distribution using the FEM method. Both the TRD conducted, in order to prove that the total maximum
301 regulations and the EN 12952-3 standard [5] are circumferential stresses, at the edge of the opening,
too conservative, because they exclude rapid, step during the optimum heating of the boiler drum—
changes of the medium temperature at the beginning of downcomer intersection did not exceed the allowable
the heating or cooling process. Thus, a new procedure values. The method of determining the optimum tem-
for the determination of the optimum changes of the perature changes of the medium, which is presented in
medium temperature will be presented [7]. At the be- this paper, allows the shortening of the boiler start-up
ginning of the pressure element heating process, the time from the cold state. This reduces the losses
medium temperature changes stepwise, e.g. by filling occurring during the start-up process.
the pressure element at the ambient temperature, with
boiling water. The value of the allowable step in tem-
perature rise of the medium is determined in a sim- 2. MATHEMATICAL FORMULATION
plified way, by dividing the maximum allowable values OF THE METHOD
of stress by the maximum value of the, so-called,
‘influence function’, which represents thermal circum- Maximum circumferential stress at the stress concen-
ferential stresses at the edge of the opening, caused by tration point P (Figure 1) at the edge of the opening,
a sudden unit rise of the medium temperature. Further which is given by vector rP, is the sum of pressure
temperature changes are determined while assuming stresses sp and thermal stresses SðrP ; tÞ:
the quasi-steady temperature distribution in the thick-
walled element. This method of heating the thick-
sP ¼ SðrP ; tÞ1sP ðrP ; tÞ; ð1Þ
walled elements is justified since the maximum thermal
stresses that are caused by a sudden rise in the medium
temperature occur after 60–120 s. Later, the value where thermal stress S(rP,t) caused by the change of
of the thermal stress falls, and over time it reaches zero. the medium temperature over time f(t) is determined

898 Int. J. Energy Res. 2011; 35:897–908 r 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Optimum heating of steam boiler pressure components J. Taler and P. Dzierwa

using Duhamel’s integral sequentially for fM we obtain:

Z t f1 ¼ ðyðtF11 Þ  S0 Þ=uF 11 for M ¼ 1; ð5Þ
@uðrp ; t  yÞ
Sðrp ; tÞ ¼ S0 1 f ðyÞ dy: ð2Þ
o @y P
yðtM1F Þ  S0  M1 i¼1 fi DuM1Fi
The u(rP, t) function, also called the influence func- fM ¼ ;
uF11
tion, represents the thermal stress at time t at the se- M ¼ 2; 3; . . . ð6Þ
lected point r of the pressure element, caused by a unit,
stepwise increase of the medium temperature (f(t) 5 0 where
for tp0 and f(t) 5 1 for t40). Du0 ¼ u1  u0 ¼ u1 ; Dui ¼ ui11  ui
The initial value S0 is constant and is independent of
and
the location of r. The influence function u(r, t)
allows to determine thermal stress S(rP,t) for any X
n1
un ¼ Dui : ð7Þ
continuous change of the medium temperature f(t),
i¼0
using Duhamel’s integral, defined by Equation (2).
The influence function u(rP,t) at the points P1 and P2 Equations (5) and. (6) allow determination of f1, f2,
describes the change of circumferential stresses at the f3, etc, in a sequential way.
edge of the opening, caused by a unit, stepwise increase When the component is loaded by internal pressure,
of the fluid temperature. The optimum temperature the total circumferential stresses at the points P1 and
changes for the pressure element, when it is absent P2, caused by internal pressure and thermal load, can
internal pressure, are determined from the following limit the rate of the medium temperature changes
equation: during the boiler startup and shutdown process. The
Z t appropriate procedure for determining the optimum
@uðrE ; t  yÞ medium temperature f(t) 5 Tf (t) is presented in [6].
S0 1 f ðyÞ dy ¼ yðtÞ: ð3Þ
0 @t
This is an inverse problem, since we have to solve
the Volterra integral equation (3) of the first kind. 3. COMPARISON WITH THE BOILER
The so-called future time steps, which were introduced REGULATIONS
to the analysis of the inverse problems by Beck [6–10],
will be used for the stabilization of the solution The heating rates: vT1 for p1 and vT2 for p2 can be
to the inverse problem. In order to determine f(t), determined in accordance with the German TRD 301
its real changes will be replaced by a step curve boiler regulations [4], or the European Standard EN
(Figure 2). The value of the medium temperature fM 12952-3 [5]. The rate of the allowable temperature
will be determined from the condition (3) at the tM1F change vT 5 dTf/dt for arbitrary pressure p1pppp2
time point: can be determined from the following equation:
SðrE ; tM1F Þ  yðtM1F Þ ¼ 0: ð4Þ dTf =dt ¼ vT ðpÞ: ð8Þ

It is assumed, that the determined value The rate of the temperature change vT for any p in
of fM applies only for the tM1 ptptM time the interval p1pppp2, in Equation (8), is obtained from
interval. the linear interpolation between (p1, vT1) and (p2, vT2):
Approximating the integral in Equation (2) by p2 nT1  p1 nT2 nT2  nT1
vT ¼ 1 pðTf Þ: ð9Þ
the method of rectangles and solving Equation (2) p2  p1 p2  p1

Figure 2. Approximation of the real changes of the medium temperature f(t) using the step curve.

Int. J. Energy Res. 2011; 35:897–908 r 2010 John Wiley & Sons, Ltd. 899
DOI: 10.1002/er
J. Taler and P. Dzierwa Optimum heating of steam boiler pressure components

When the fluid in the pressure component is satu- be considered as a thick plate (slab) if the inner radii of
rated steam, saturated water or steam-water mixture, vessels are large. The thermal stress at the point
then the fluid pressure depends on the saturation 0pxTpL should be equal to the allowable stress:
temperature sT ðxT ; tÞ ¼ s a ðtÞ; ð13Þ
p ¼ ½ðT1aÞ=bc ; ð10Þ
where the thermal stress sT is given by:
where the symbols p and T denote the saturation
Eb 
pressure in MPa and the temperature in 1C, respec- sT ¼ ½TðtÞ  Tðx; tÞ: ð14Þ
1v
tively. The constants are: a 5 20.81045, b 5 200.40565,
c 5 4.564173. The allowable stress s a may depend on time. The
The uniform initial condition was assumed: T 
mean temperature TðtÞ over the slab thickness is
(t 5 0) 5 T0, where T0 is the constant temperature. defined as:
The allowable rates of fluid temperature changes vT1 Z L
and vT2 are determined from the condition that the  ¼
TðtÞ Tðx; tÞ dx: ð15Þ
0
sum of circumferential stresses which are caused by
pressure and thermal loads is equal to the allowable The problem of optimum heating or cooling will be
stress sa solved under the assumption that the physical prop-
D1H Eb cr erties of the component material and the heat transfer
ap ðp  p0 Þ 1aT vT H 2 fww ¼ sa ; ð11Þ coefficient are constant.
2H 1n k
The mathematical formulation of the problem
where the symbol fww denotes is [10,11]
1 ðo2  1Þ ð3o2  1Þ  4o4 ln o @T @2 T
fww ¼ : ð12Þ cr ¼k 2; t40; ð16Þ
8 ðo2  1Þ ðo  1Þ2 @t @x
The allowable temperature rate vT1 at the beginning
Tðx; 0Þ ¼ T0 ; 0pxpL; ð17Þ
of the process is calculated from Equation (11)
for p 5 p1 and s ¼ s a . The allowable temperature @T
rate vT2 at the end of the process is calculated from jx¼0 ¼ 0; t40; ð18Þ
@x
Equation (11) for p 5 p2 and s ¼ s a .

@T 
k ¼ h½Tf ðtÞ  T jx¼L ; t40: ð19Þ
@x x¼L
4. DETERMINATION OF THE
OPTIMUM FLUID TEMPERATURE The boundary condition of the third kind is given at
CHANGES BY SOLVING THE the exposed surface. The time-varying fluid tempera-
PARAMETRIC LEAST-SQUARES ture Tf(t) in Equation (19) is unknown and will be
PROBLEM determined from the condition (13).
The solution of the direct heat conduction problem,
It is very difficult to carry out optimum changes of the which is defined by the heat conduction equation (16),
fluid temperature Tf during pressure component initial condition (17), boundary conditions (18) and
heating at the initial stage of the heating process. (19) is as follows:
However, optimum fluid temperature changes can be Tðx; tÞ ¼ Ts 1ðT0  Ts ÞFðx; tÞ1vT t  vT F1 ðx; tÞ ð20Þ
approximated by a ramp function consisting of a
step increase in fluid temperature Ts followed by the where
temperature increase with a constant rate vT (Figure 3). X
1
2sin2 mn  x  at 
As an example, the optimum fluid temperature changes Fðx; tÞ ¼ cos mn exp m2n 2
m 1 sin mn cos mn
n¼1 n
L L
Tf(t) will be determined during slab heating. It is worth
mentioning that the cylindrical or spherical vessels can ð21Þ

Figure 3. Approximation of optimum temperature changes by a ramp function: (a) slab and (b) ramp function.

900 Int. J. Energy Res. 2011; 35:897–908 r 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Optimum heating of steam boiler pressure components J. Taler and P. Dzierwa


L2 1  x 1  x 2 solution:
F1 ðx; tÞ ¼ þ 1  1 b1 a22  b2 a12
a Bi L 2 L Ts ¼ ; ð30Þ
X1
2sinmn a11 a22  a21 a12

m2 ðm þ sin mn cos mn Þ
n¼1 n n
b1 a21  b2 a11
 vT ¼ ; ð31Þ
 x  at  a12 a21  a11 a22
 cos mn exp m2n 2 ð22Þ
L L where
where a 5 k/c r is thermal diffusivity, mn are the posi-  i Þ  FðxT ; ti Þ;
FN ðxT ; ti Þ ¼ Fðt ð32Þ
tive roots of the transcendental equation mn tg mn 5 Bi,
and Bi 5 hL/k is the Biot number. F1N ðxT ; ti Þ ¼ F1 ðti Þ  F1 ðxT ; ti Þ; ð33Þ
Substituting expression (20) into Equation (15) X
nt
gives a11 ¼ FN2 ðxT ; ti Þ; ð34Þ
i¼1
TðtÞ 
 ¼ Ts 1ðT0  Ts ÞFðtÞ1v 
T t  vT F1 ðtÞ ð23Þ
 and F1 ðtÞ are given by: X
nt
where FðtÞ a12 ¼ F1N ðxT ; ti ÞFN ðxT ; ti Þ; ð35Þ
X
1
2sin2 mn i¼1
 ¼
FðtÞ expðm2n FoÞ ð24Þ
m ðm 1 sin mn cos mn Þ
n¼1 n n
X
nt X
nt
b1 ¼ T0 FN2 ðxT ; ti Þ  s a ðti ÞFN ðxT ; ti Þ; ð36Þ
i¼1 i¼1
L 2
1 1 X
1
2sin mn 2
F1 ðtÞ ¼ þ  a21 ¼ a12 ; ð37Þ
a Bi 3 n¼1 m3n ðmn þ sin mn cos mn Þ
  X
nt
2 at 2
 exp mn 2 ð25Þ a22 ¼ F1N ðxT ; ti Þ; ð38Þ
L
i¼1
Taking into account Equations (20) and (23),
X
nt X
nt
Equation (14) can be expressed in the form: b2 ¼ T0 FN ðxT ; ti ÞF1N ðxT ; ti Þ  s a ðti ÞF1N ðxT ; ti Þ:
Eb i¼1 i¼1
sT ¼ 
½Ts 1ðT0  Ts ÞFðtÞ1v 
T t  vT F1 ðtÞ  Ts
1v ð39Þ
 ðT0  Ts ÞFðtÞ  vT t1vT F1 ðx; tÞ The symbol s a ðti Þ
denotes
Eb  s a ðti Þ ð1  vÞs a ðti Þ
¼ fðT0 Ts Þ½FðtÞFðx; TÞvT ½F1 ðtÞ  F1 ðx; tÞg: s a ðti Þ
¼ ¼ : ð40Þ
1v Eb Eb
ð26Þ 1v
In the optimization problem, the unknown para- Since the time points ti are equally distributed with
meters Ts and vT are to be adjusted to approximately the time step Dt, the time points ti are given by:
satisfy the following system of equations: ti ¼ i Dt; i ¼ 1; :::; nt : ð41Þ
sT ðxT ; ti Þ ffi s a ; i ¼ 1; . . . ; nt ; ð27Þ The values of Ts and vT can also be determined by
where sT is given by Equation (26). the modified Levenberg–Marquardt method [12,13]
The least-squares method is used to estimate the using the subroutine BCLSF from the IMSL mathe-
parameters Ts and vT. The parameters Ts and vT are matical library [14]. The same results were obtained
computed by minimizing the sum of squares of the using Equations (29) and (30), as well as the
differences between the calculated stress sT and the Levenberg–Marquardt method.
allowable stress s a :
Xnt 
Eb  i Þ  FðxT ; ti Þ  vT ½F1 ðti Þ 5. A NEW PROCEDURE FOR
Se ¼ fðT0  Ts Þ½Fðt
i¼1
1 v DETERMINING ALLOWABLE
2 TEMPERATURE RATES DURING THE
F1 ðxT ; ti Þg  s a ðti Þ ¼ min: ð28Þ HEATING AND COOLING OF THICK-
WALLED BOILER COMPONENTS
It is necessary to find the values of Ts and vT, for
which the two partial derivatives are simultaneously In the case of components, loaded with pressure, the
zero: optimum medium temperature can be obtained in a
similar way. Initially, the medium temperature may
@Se @Se
¼ 0; ¼ 0: ð29Þ change stepwise and later the medium temperature
@Ts @vT changes can be conducted at a rate calculated
Finding derivatives (29) give a set of linear equations according to the TRD 301 regulations or to the EN
in the unknowns Ts and vT, which has the following 12952-3 standard.

Int. J. Energy Res. 2011; 35:897–908 r 2010 John Wiley & Sons, Ltd. 901
DOI: 10.1002/er
J. Taler and P. Dzierwa Optimum heating of steam boiler pressure components

Taking into account that the TRD 301 regulations Cooling—the calculations are only carried out for
and the EN 12952-3 standard do not permit stepwise the point P1
fluid temperature changes and the optimum fluid
temperature changes resulting from the proposed  Determine the stress concentration coefficients at
method, a new procedure for optimum heating and the edge of the opening: ap for the stresses caused
cooling of pressure components will be proposed. The by pressure and aT for the thermal stresses.
developed procedure has an advantage over the present  Determine the allowable values of stresses during
boiler regulations, since the pressure components can the cooling process s a , according to the EN
be heated or cooled much faster which results in a large 12952-3 standard.
reduction of heat loss.  Determine the allowable cooling rates according
The procedure for determining the rate of the al- to the EN 12952-3 standard: vT1 for the initial
lowable medium temperature changes is as follows: pressure p1 and vT2 for the final pressure p2.
Heating—calculations will be conducted for the  Determine the maximum value of the influence
points P1 and P2 situated at the edge of the opening function umax 5 umin, and time tmax, for which
(Figure 1). the function reaches the maximum, for the
assumed value of the heat transfer coefficient.
 Determine the stress concentration coefficients at  Calculate the initial value of the stepwise rise in
the edge of the opening: ap for the stresses caused the medium temperature, using the formula
by pressure and aT for the thermal stresses.
_ D1H
 Determine the allowable values of stresses DT00 ¼ sa  ap ðp1  p0 Þ umax ð44Þ
2H
during the start-up process s a , according to
the EN 12952-3 standard, taking into account  Assume that DT0 ¼ DT00 ; there is no need to
that the stress values s a are different for P1 reduce the temperature stepwise change, since the
and P2. reduction of stresses caused by pressure is greater
 Determine the allowable heating rates according than the rise in stresses caused by temperature
to the EN 12952-3 standard: vT1 for changes. The reason for this is the small
the initial pressure p1 and vT2 for the final temperature drop during a decrease in pressure
pressure p2. for higher values of saturation pressure,
 Determine, the minimum value of the influence  Determine the optimum medium temperature
function umin, and time tmin, for which the Tf(t) from Equation (8), solving it by the
function reaches the minimum for the assumed Runge–Kutta method and taking into considera-
value of the heat transfer coefficient. tion the initial condition stating that the initial
 Calculate the initial stepwise rise in the medium medium temperature is: Tf (t 5 0) 5 T0DT0,
temperature, using the formula where T0 is the initial temperature of the
component at time t 5 0.
D1H
DT00 ¼ s a  ap ðp1  p0 Þ umin : ð42Þ
2H
Such temperature changes can be conducted in real
life.
 Reduce the allowable value of the medium 5.1. Uncertainty analysis
temperature stepwise rise by vT1 tmin; the allow-
able value of the stepwise rise of the medium The general uncertainty analysis, which is used in
temperature is then: experimentation, is adapted to uncertainties in the
optimum fluid temperature. The values of the material
DT0 ¼ DT00  vT1 tmin : ð43Þ properties and the measured variables have uncertain-
ties. The propagation of the individual variable
 Determine the optimum medium temperature Tf uncertainty through a calculation procedure into
(t) from Equation (8) solving it by the Runge–- results will be analyzed using the error propagation
Kutta method and taking into consideration the rule of Gauss [12,13,15]. Consider a general case in
initial condition stating that the initial medium which a result r is a function of J independent variables
temperature is: Tf (t 5 0) 5 T01DT0, where T0
is the initial temperature of the component at r ¼ rðx1 ; x2 ; :::; xJ Þ: ð45Þ
time t 5 0, Uncertainty in the results is given by
 Having determined the optimum temperature
" 2  2  2 #1=2
changes and heating times with respect to @r @r @r
stresses at the points P1 and P2, the longer Ur ¼ Ux 1 Ux 1    1 Ux ;
@x1 1 @x2 2 @xJ J
lasting heating process will be assumed to be
optimum. ð46Þ

902 Int. J. Energy Res. 2011; 35:897–908 r 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Optimum heating of steam boiler pressure components J. Taler and P. Dzierwa

where Uxi are the uncertainties in the independent 6. EXAMPLES OF CALCULATION

variables xi, and the partial derivatives @r=@xi are the
sensitivity coefficients. First, the optimum medium temperature changes over
The sensitivity coefficients in Equation (46) were time, during the heating of a plate, of the thickness
calculated numerically using central difference H 5 0.1 m will be determined. The edges of the infinitely
quotients. The absolute uncertainties Uxi are expressed large plate can expand freely, but they cannot bend. The
with 95% level of confidence, i.e. Uxi ¼ 2sxi . Thus, following data were assumed for the calculations: the
the uncertainty in the result Ur is also at 95% thermal conductivity: k 5 42 W m1 K1, the specific
confidence. heat: c 5 483.1 J kg1 K1, the density: r 5 7782 kg m3,
The Young modulus is E 5 1.966  1011 N m2,
the linear thermal expansion coefficient b 5 1.32 
105 l K1 and the Poisson ratio n 5 0.29. The heat
transfer coefficient at the surface of the heated plate
equals: h 5 2000 W m2 K1. The other surface of the
plate is thermally insulated. The allowable compres-
sive stress on the heated surface of the plate is:
s a 5 109.06 MPa. At the start time, the plate tem-
perature is even and equals 01C. The time changes of the
influence function for various heat transfer coefficients
are presented in Figure 4. The time point, at which the
influence function reaches the maximum absolute value,
depends on the heat transfer coefficient. The greater
the value of the heat transfer coefficient, the earlier
the absolute maximum value of stress will occur. The
absolute maximum value of the influence function also
increases with the increase of the value of the h
coefficient. The optimum changes of the medium
temperature were determined using formulas (5) and
(6) (Figure 5). An inspection of the results presented in
Figure 5 indicates that it is impossible to achieve the
optimum medium temperature at the beginning of the
heating process, because this temperature is very high.
Thus, the optimum temperature was approximated,
using a ramp function consisting of the step temperature
rise at time t 5 0 and a linear temperature increase for
time t40. The value of the initial medium temperature
Figure 4. Plot of the influence function for the 0.1-m thick plate. jump and the rate of the linear medium temperature rise

Figure 5. Optimum medium temperature during the heating process of a thick-walled plate.

Int. J. Energy Res. 2011; 35:897–908 r 2010 John Wiley & Sons, Ltd. 903
DOI: 10.1002/er
J. Taler and P. Dzierwa Optimum heating of steam boiler pressure components

were determined using the least-squares methods, in point P1 and aT 5 2.074 at the point P2, respectively.
such a way that assures that the integral over time from Numerical model has been used for the determination
the square of the difference between the actual stress of the influence function at points P1 and P2. The FEM
and the allowable stress on the heated surface of the analysis was carried out by means of the ANSYS
plate is minimized. In the estimated optimum medium software. The maximum absolute value of the function
temperature Tf,ramp 5 54.9410.0998 t time t is expressed u(r,t) is larger at the point P2 in comparison with its
in seconds. value at the point P1. The uncertainties in properties
The 95% uncertainty 2 sDT in the calculated initial and wall dimensions are assumed as follows: 2sE 5
temperature step DT is
2 2  2  2  2 31=2
@DT @DT @DT @DT
6 2sE 1 2sb 1 2sn 1 2sk 7
6 @E @b @n @k 7
2sDT ¼ 6 6  2  2  2
7
7
4 @DT @DT @DT 5
1 2sc 1 2sr 1 2sL
@c @r @L

In a similar way the uncertainty in the heating rate 0.8  103 MPa, 2sb 5 0.1  106 K1, 2sn 5 0.001,
vT was calculated. 2sk 5 0.5 W (m1 K1), 2sc 5 15 J kg1 K1, 2sr 5
The following data have been taken for the 95% 60 kg m3, 2sH 5 0.0001 m, 2sh 5 0.00005 m. The 95%
random uncertainties associated with the independent uncertainty in the calculated minimum values of the
variables: 2sE 5 0.8  103 MPa, 2sb 5 0.1  106 K1, influence functions obtained from Equation (46) are:
2sn 5 0.001, 2sk 5 0.5 W m1 K1, 2sc 5 15 J kg1 K1, 2 su1 ;min 5 0.054 MPa1 K1, 2 su2 ;min 5 0.058 MPa K1.
2sr 5 60 kg m3, 2sL 5 0.0001 m. The minimum values of the influence functions are:
The 95% uncertainties in DT and vT are: u1;min 5 2.3470.054 MPa K1 at the point P1 and
2 sDT 5 1.29 K, 2 svT 5 0.0041 K s1 and the results u2;min 5 2.4870.058 MPa K1 at the point P2.
can be expressed as: DT 5 54.9471.29 K and vT 5 In a similar way, the uncertainties for the initial fluid
0.099870.0041 K s1. temperature jump DT0 and fluid temperature rates vT1
This optimum ramp heating can easily be conducted and vT2 were determined assuming the same un-
in power plants. The step increase of the medium certainties. The fluid temperature rates vT1 and vT2
temperature by 54.94 K can also be achieved without were calculated using Equation (11) and the initial
difficulty in real life by flooding the component with temperature jump DT0 was calculated from Equation
water at a temperature 54.94 K higher than the initial (43). Uncertainties in DT0, vT1 and vT2 were calculated
temperature of the component. for the point P1 and P2 using Equation (46) to give
In the second example, optimum water temperature
changes will be determined with respect to total stress  point P1: DT0 ¼ Ts  T0 ¼ 50:5 1:42 K, vT1 5
at the points P1 and P2 on the edge of the boiler drum- 4.1770.16 K min1, vT2 5 12.5970.49 K min1
downcomer intersection. The following boiler drum  Point P2: DT ¼ Ts  T0 ¼ 47:25 1:27 K, vT1 5
dimensions were assumed for the calculation: 3.7470.14 K min1, vT2 5 5.1970.20 K min1
D 5 1.7 m, d 5 0.09 m, H 5 0.09 m and h 5 0.006 m
(Figure 1). Also, the following properties of steel were The 95% confidence intervals for the optimum
assumed: k 5 42 W m1 K1; c 5 538.5 J kg1 K1; temperatures and pressure changes are shown in
r 5 7800 kg1 m3; E 5 1.96  1011 N1 m2; b 5 Figures 6(a) and (b) for the point P1 and in Figures 7(a)
1.32  105 l K1 and n 5 0.3. The heat transfer coef- and (b) for the point P2. Since total compressive
ficient on the inner surface of the drum and down- stresses at the point P2 reach a larger value, the opti-
comer is h 5 1000 W m2 K1). The outer surfaces of mum medium temperature change rate with regard to
the boiler drum-downcomer intersection are thermally circumferential stresses at point P2 is smaller than the
insulated. The stress distribution analysis was per- temperature change rate at point P1. Furthermore, the
formed for the elastic state. The stress concentration temperature jump at the beginning of the process is
coefficients are: ap 5 2.65 at the point P1 and ap 5 0.51 smaller at the point P2. If stresses at the point P2 affect
at the point P2, respectively. Since the diameter and the course of heating and cooling, the start-up opera-
wall thickness of the downcomer tube are much tion is longer, while shutdown lasts shorter in contrast
smaller than the diameter and wall thickness of the to durations obtained by means of the approach, when
boiler drum, the intersection resembles a plate with a P1 is the criterion point.
hole subjected to biaxial stretching stresses with 2:1 Next, the changes of total stresses, in conjunc-
ratio. For a plate with such a load, the stress tion with determined optimum temperatures, were
concentration coefficient is 2.5 at the point P1 and 0.5 determined by means of the 3D FEM analysis to check
at the point P2. The thermal stress concentration that total stresses do not exceed the allowable stresses.
coefficients in quasi-steady state are aT 5 1.86 at the From the analysis of the results, one can see that

904 Int. J. Energy Res. 2011; 35:897–908 r 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Optimum heating of steam boiler pressure components J. Taler and P. Dzierwa

Figure 6. Optimum time changes of temperature and pressure and comparison of total circumferential stress with the allowable
stress when optimum medium temperature changes are determined from the condition that the total circumferential stress at
the point P1 is equal to the allowable stress s a : (a) 1—optimum water temperature, 2—limits of the 95% confidence interval;
(b) 1—optimum pressure, 2—limits of the 95% confidence interval; and (c) circumferential stress at the points P1 and P2.

Int. J. Energy Res. 2011; 35:897–908 r 2010 John Wiley & Sons, Ltd. 905
DOI: 10.1002/er
J. Taler and P. Dzierwa Optimum heating of steam boiler pressure components

Figure 7. Optimum time changes of temperature and pressure and comparison of total circumferential stress with the allowable
stress when optimum medium temperature changes are determined from the condition that the total circumferential stress at
the point P2 is equal to the allowable stress s a : (a) 1—optimum water temperature, 2—limits of the 95% confidence interval;
(b) 1—optimum pressure, 2—limits of the 95% confidence interval; and (c) circumferential stress at the points P1 and P2.

906 Int. J. Energy Res. 2011; 35:897–908 r 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Optimum heating of steam boiler pressure components J. Taler and P. Dzierwa

allowable stresses are exceeded at the point P2 r 5 radius (m)

(Figure 6(c)), when the optimum medium temperature r 5 position vector
is determined with respect to total stresses at point P1. S 5 stress calculated using Duhamel
If the optimum heating and cooling of a boiler drum integral (MPa)
are carried out with respect to total stresses at the point Se 5 sum of the temperature differ-
P2, the maximum stresses on the edge of a hole at the ence squares (K2)
points P1 and P2 are smaller than the allowable stresses t 5 time (s)
(Figure 7(c)). It is advantageous when the boiler drum T 5 temperature (1C or K)
is heated with respect to stresses at the point P2, since T 5 mean temperature over the wall
the allowable stresses both at the point P1 and P2 are thickness (1C)
not exceeded and the lifetime of the pressure compo- u 5 influence function (temperature
nent is longer. or thermal stress for unit step-
wise increase of the fluid tem-
perature)
7. CONCLUSIONS U 5 uncertainty
vT 5 rate of temperature changes
The presented method of determining the allowable (K s1)
changes of the medium temperature during the boiler x 5 cartesian coordinate (m)
start-up and shutdown processes can be used in real life. y 5 allowable stress (MPa)
It allows a significant reduction in the boiler start-up time,
thus reducing the start-up costs. At the beginning of the Greek symbols
start-up process the pressure element can be filled with
a ¼ k=ðcrÞ 5 thermal diffusivity (m2 s1)
hot water. In this paper, the procedures for determining
ap 5 stress concentration factor for
the optimum temperature changes during the heating and
circumferential stress caused by
cooling processes of the boiler drum or other boiler
pressure
elements were presented. The proposed method can also
aT 5 stress concentration factor for
be used when the medium pressure changes independently
thermal stress
of the medium temperature. The present standards should
b 5 linear thermal expansion coeffi-
be extended to take into consideration not only the point
cient, l K1
P1 but also the P2 on the hole edge, when allowable fluid
Dt 5 time step (s)
temperature changes are determined.
mn 5 nth root of characteristic equation
n 5 Poisson’s ratio
r 5 density (kg m3)
NOMENCLATURE s 5 normal stress (MPa)
sX 5 sample standard deviation
Bi 5 Biot number, Bi 5 hL/k s a 5 allowable stress for heating
c 5 specific heat capacity (J kg1 K1) _
sa 5 allowable stress for cooling
D 5 inner diameter of the drum (m) o ¼ ðD12HÞ=D 5 ratio of the outer to inner
E 5 modulus of elasticity (MPa) radius of the cylinder
f 5 fluid temperature (1C)
F 5 number of future time steps Subscripts
F(x,t) 5 solution of the initial-boundary a 5 allowable
problem f 5 fluid
Fo 5 Fourier number, Fo 5 at/L2 0 5 initial
F, F1 5 functions defined by Equations
(21) and (22)
h 5 heat transfer coefficient
(W m2 K1) REFERENCES
H, h 5 thickness of the drum and
downcomer wall (m) 1. Sanaye S, Rezazadeh M. Transient thermal model-
k 5 thermal conductivity (W m1 K1)
ling of heat recovery steam generators in combined
L 5 slab thickness (m)
cycle power plants. International Journal of Energy
nt 5 number of time points
pn 5 gauge pressure (MPa) Research 2007; 31(11):1047–1063.
p1 5 initial pressure during start-up or 2. Habib MA, Emara-Shabaik HE, Al-Zaharnah I,
shutdown of the boiler (MPa) Ayinde T. A thermal nonlinear dynamic model for
p2 5 final pressure during start-up or water tube drum boilers. International Journal of
shutdown of the boiler (MPa) Energy Research 2010; 34:20–35.

Int. J. Energy Res. 2011; 35:897–908 r 2010 John Wiley & Sons, Ltd. 907
DOI: 10.1002/er
J. Taler and P. Dzierwa Optimum heating of steam boiler pressure components

3. Robinson GF, Holmes ML. Transient modelling of 9. Taler J, Dzierwa P, Taler D. Optimum heating
operational upset in a special-purpose boiler. Interna- of pressure components of large steam boilers.
tional Journal of Energy Research 1984; 8(4):325–349. Forschung im Ingenieurwesen 2009; 73(3):183–192.
4. TRD 301. Technische Regeln für Dampfkessel. 10. Taler J, Duda P. Solving Direct and Inverse Heat
Heymanns Beuth: Berlin, 2001, 143–185. Conduction Problems. Springer: Berlin, Heidelberg,
5. European Standard EN 12952-3. Water-tube Boilers 2006.
and Auxiliary Installations—Part 3: Design and 11. Carslaw HS, Jaeger JC. Conduction of Heat in
Calculation for Pressure Parts, European Committee Solids (2nd edn). Oxford University Press: Oxford,
for Standardization, Brussels, 2001. 2008.
6. Taler J, Dzierwa P. A new method for determining 12. Seber GAF, Wild CJ. Nonlinear Regression. Wiley:
allowable medium temperature rates during heating Hoboken, NY, 1989.
and cooling of thick-walled boiler components. 13. Press WH, Teukolsky SA, Vetterling WT, Flannery
Proceedings of the Seventh International Congress BP. Numerical Recipies in FORTRAN; the Art
on Thermal Stresses, Taipei, Taiwan, vol. 2, 2007; of Scientific Computing (3rd edn). Cambridge
437–440. University Press: Cambridge, 2007.
7. Dzierwa P. Optimization of heating and cooling of 14. BCLSF Subroutine. Optimization. IMSL MATH/
thick walled pressure components. Ph.D. Thesis, LIBRARY, FORTRAN Subroutines for Mathema-
Department of Mechanical Engineering, Cracow tical Applications, Chapter 8, vol. 2. Visual
University of Technology, Cracow, Poland, 2007. Numerics Inc.: Houston, TX, U.S.A., 1994.
8. Taler D, Taler J. Optimum heating of thick plate. 15. Coleman HW, Steele WG. Experimentation,
International Journal of Heat and Mass Transfer Validation, and Uncertainty Analysis for Engineers
2009; 52:2335–2342. (3rd edn). Wiley: Hoboken, NY, 2009.

908 Int. J. Energy Res. 2011; 35:897–908 r 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/er