66 IEEE TransactionS on Power Systems, Vo1.6, No.

1, February 1991

BOILER MODELS FOR SYSTEM DYNAMIC PERFORMANCE STUDIES

F. P. de Mello, Fellow
Power Technologies, Inc.
P.O. Box 1058
1482 Erie Boulevard
Schenectady, New York 12301-1058
ABSTRACT
The dynamic response of boiler pressure is a mass-volume-
energy balance relationship which can be derived from
changes in turbine valve position and boiler energy
input. The response characteristics so derived are
compared with more simplified models which preserve the
essential nonlinear characteristics of the process. It
is shown that the dominant influence on the pressure
response comes from the pressure controls tuning.
Errors in modeling the steady state response of the
boiler itself are neutralized by the controls.
The use of simplified boiler models in power system
dynamic performance studies is justified.
KEYWORDS

Boiler-turbine response, boiler-pressure control,

boiler-models.
INTRODUCTION
The dynamic response of boilers can be of importance
in predicting the behavior of power systems through STACK
upsets involving significant imbalances between
generation and loads beyond the 3-5 second time period
considered in typical transient stability simulations.
-
Simplified models of the steam generating process have
been suggested to account for pressure effects on the Figure 1. Boiler Process
response of steam turbines .
A boiler model based on first principles of mass, The combustion products path is also shown. The
energy and volume balance is derived to lend insight on waterwalls absorb radiant heat in the furnace. The hot
the parameters that figure in the model. The relations gases leaving the furnace transfer heat by radiation and
apply to a drum-type boiler. Typical parametes values convection to the secondary superheater, reheater primary
used are for a 270 Mw, 2400 psi coal-fired unit . superheater and economizer in succession.
Constant volume circulating pumps maintain an
The model is limited to pressure/flow effects. The
saturation characteristics of waterhteam are the essentially constant rate of circulation through the
dominant energy storage contributors. Storage in downcomers and waterwall s.
superheaters is considered from the mass, volume, and Desuperheating spray water is introduced between the
pressure effects of superheated steam assuming constant primary and secondary superheater sections for control of
temperature. main steam temperature.
The validity of the simplified models is established In this design burner tilts are used to control
with comparisons of their response characteristics reheat temperature and the firing rate is used to control
against those obtained with the more rigorous solution pressure while feedwater flow is controlled to maintain
of the physics from first principles. Methods of drum level. The modeling considers pressure/flow effects
deriving the parameters of the simplified models are only.
given. Referring to Figure 2, the relations defining the
process dynamics for this subsystem are based on the
The Boiler Process following assumption:
(1) Feedwater at enthalpy hfw and flow rate hw
Figure 1 is a schematic of the boiler process. 3 mixes with saturated water from drum and the
mixture flows down the downcomer at a
Steam generated in the waterwalls is separated in the
drum from where it flows through the primary and constant recirculation rate i .
In natural
secondary superheaters on to the high pressure (HP) circulation boilers the boyanEy head versus
turbine. It then re-enters the boiler to the reheater head loss due to friction drop
where its energy level is increased. It flows through the characteristics also tend to maintain a
intermediate and low pressure turbine to the condenser. constant recircu 1 at 1on.
(2) Heat absorbed by fluid in waterwalls is
90 SI4 305-3 PtRS A paper recommended and approved uniformly distributed i .e. proportional to
by the IEEE Power System Engineering Committee of the length or volume.
ISEE Poller Engineering Society f o r presentation a t the (3) Velocities of steam and water are assumed
IEEE/PES 1990 Summer Fleeting, Minneapolis, Minnesota, equal.
J u l y 15-19, 1990. Ilanuscript submitted February 1 ,
1990; made available f o r p r i n t i n g June 27, 1990.

Authorized licensed use limited to: Politehnica Timisoara. Downloaded on November 2, 2009 at 07:49 from IEEE Xplore. Restrictions apply.
 67

where
hf = enthalpy of saturated water
hr' = enthalpy of water entering waterwalls
h = enthalpy of saturated steam
9
PD = drum pressure

Qw = heat transferred from tubes to fluid i n

waterwalls
VATER
WALLS
Variables to be so!ved.arej M , M , MSC, PD, hr and hr'
given as inputs %, m , mr, f .

- boiling mr

DONN
M is calculate$ as proportional to the volume
requid8 to heat the incoming fluid from hr' to hf.
Since i is constant, this translates into being
proportignal to (hf- h ')/Qw. A small time constant is

-
/ -

COMERS used to represent resFdence time effects. The block

diagram defining the calculation of M is shown in

- -- I Figure 4 where the subscript "0" denote&he variables'

quiescent values. Tsc is the residence time of the

- -+a- subcooled fluid, approximately equal to MSc/ir.

Calculation o f h, and h,'
Figure 2. System Defining Steam Generation and Pressure
The enthalpy of water entering the downcomers is obtained
For the system with boundaries shown on Figure 3, from the mixing equation:
the following equations apply:
h = (ir - hW.)h + iwhfw
Mass Balance (4)
mr
where hfw .is the enthalpy of feedwater. The enthalpy of
the water entering the waterwalls h I is the same as hr
where delayed by the transport time in thzdowncomers which is
Mf = Mass of saturated liquid in waterwalls and approximated as a time constant (Figure 5).
drum
Equations (1) and (2) can be expressed as
Msc = Mass of subcooled liquid in waterwalls
= Mass of saturated vapor in waterwalls and
Mf + Mg = $ (kW - is)dt - Msc = A (5)
and
Mg drum M f v f + M 9v 9 = V - M scv f = B (6)
$ = feedwater flow rate
from which Mf = (Avg - B)/(vg - vf) (7)
1, = recirculation flow rate and M 9 = (B - Avf)/(vg - vf) (8)
ts = steam flow rate out of drum Equation (3) can be expressed as
Volume Balance 144V
(Mf+ MSc)vf + Mgvg = V (2)
Mfhf + Mghg - J PD = J (6, + irh,.'
where
vf = specific volume of saturated water
vg = specific volume of saturated steam
V = volume of waterwalls and drum
Enerav Balance

Authorized licensed use limited to: Politehnica Timisoara. Downloaded on November 2, 2009 at 07:49 from IEEE Xplore. Restrictions apply.
 , ,, , , .,

68

I
pD

i7

Figure 6. Computation of Heat Flux from Tube Metal to Fluid

-1 where
MC = heat capacitance of waterwalls

K, =
heat transfer coefficient between tubes and
f-- inner fluid
'"r Q = heat flux from furnace to waterwalls
Figure 3. Waterwall anc Drum Subsystem
Storaqe in SuDerheaterq
The effect of superheater volume is considered
h, - hl assuming temperature is constant. The distributed nature
of the process is approximated as in Figure 7 with three
lumped volumes 1 inked by orifices to represent pressure
to flow squared, the pressure flow effects are'modeled
by the following set of algebraic and differential
equations :
PD = P1 + (ms)'K1 P1 = P2 + (m12)2K2

P2 = P3 + (mZ3)'K3
Figure 4. Computation of Subcooled Resident Mass

where K1, K2, K3 are coefficients relating pressure drop

to flow rate square, Kv is proportional to turbine valve
flow area, and the symbol P denotes pressure at the
volume denoted by the suffix.
ml 2 m23 9
Figure 5. Computation of Enthalpy at Waterwall Inlet
Saturated steam properties, v v , h h are all
functions of PD. For the ran&
approximate expressions are:
OF 2306'tog2800 psi, Valve
Figure 7. Lumped Representation of Superheaters
vf = 0.02934 - 0.0078 (PD/lOOO) + 0.003 (PD/1000)2
Flow rates between volumes are calculated from the
vg = 0.52968 - 0.22085 (PD/looO) + 0.0245 (PD/looo)2 continuity equations
hf = 684.2 - 93.5 (PD/looo) + 45 (PD/looo)2 . h 3 dP3 , .&2 dP2
iz3=;hT+V - - m12 = iZ3 + v - -
hg = 1083.3 + 116.5 (PD/lOOO) - 45 (PD/1000)2 (11) ,x3dt 2bP2 dt

Calculation o f Heat Flux From Tubes to Fluid in

Waterwalls
The heat flux from tubes to fluid is assumed to
be proportional to the temperature difrerence between
metal and fluid. Temperature of the fluid is basically
a function of drum pressure. Figure 6 shows the
relations in block diagram form.

Authorized licensed use limited to: Politehnica Timisoara. Downloaded on November 2, 2009 at 07:49 from IEEE Xplore. Restrictions apply.
 69
where
VI, V2 and V3 are the volumes of the lumped
superheater sections and 4 , P 2 , p 3 are the densities o f
steam in these volumes.
Calculation o f Drum Level
1
Defining
Mfd = mass of liquid in drum
Iterate I
I
Using the last time step value of b
VD = volume of drum with last time step value of P as 2
start calculate Vf, vg. hf. hg(ll).
MF(7). MG(8) and P(10)
Volume of waterwalls occupied by the steaming mixture = 2

where.
mws
(v - vD) - Msc*Vf
The proportion of this volume occupied by liquid is

= steam generation
(16)

i - G(P - Po)
- old P,'"

No

Calculate new flows out of drum and

in superheater and to turbine
iterating on throttle pressure PR
Mass of saturated liquid in waterwalls Start with last time step
value of P3
Use equations 15 and 13 to obtain flows
and pressures up to P .
Compare Po
with P calculatedqn Block 2.
Use error Po correct next trial value
Mass of fluid in drum Mfd = Mf - Mfw (20)
value of P and iterate till
err& in pD < E.
3
Chanqe in drum level
DL =
(Mfd - Mfdo) f'
DA (21) Update values of pressure in
where DA = drum cross section area superheater nodes to be used
in next time step
4
The Comutation Scheme
The computation of the above set of nonlinear
algebraic and differential equations is done by numerical Calculate derivatives of States
integration and differentiation methods as well as ',h (Flgure 5)
iterative solution of simultaneous algebraic relations. Msc (Figure 4)
Figure 8 is a flow chart of the computation steps Msc + Mf t Msc (Figure 5)
implemented on a PC. Not shown is the initialization of
states which can be derived from the equations, setting [ (VI)
Msc + Mfhf
derivatives of states equal to zero for the given 144V
operating condition defined by a given turbine valve + M 9h 9 - s P ] (9)
opening and throttle pressure.
A, B and C are functions of states (outputs of (Figure 9)
integrators) as defined in relations 5, 6, and 9. In the 6, (Figure 6)
5
flow chart, the numbers in brackets refer to the
equations used to calculate the particular variables.
Results Take time step and update
states using integration algorithm
The above equations were solved for a boiler with 6
the following basic data defined for full load operating
conditions of 480 lbs/sec steam output at 2400 psi,
1000°F.
Calculate drum level
,V = 2850 cu ft using equations 16 to 21
7
VD = 950 cu ft

Go to Block 1

DA = 270 sq. ft, Figure 8. Solution Scheme for Process Equations

Authorized licensed use limited to: Politehnica Timisoara. Downloaded on November 2, 2009 at 07:49 from IEEE Xplore. Restrictions apply.
 70

Examination of the pressure response in Figure Y

reveals an interesting effect of pressure on steam
generation. Basically this is a positive feedback effect,
as less BTU are needed to generate steam with increases
;3p3 in pressure level.
- = 1.0, P3 = 2400 Inspection of the boiler response characteristics
3 ap,J in Figures 9 and 10 suggests a simple model shown on
Figure 11 which is essentially similar to that presented
hr = 2000 lbs/sec, PD = 2640 in Figure 6 of reference 1 except that the positive
feedback effect from pressure on steam generation has
The boiler response characteristics to step changes been included.
in turbine valve flow area and heat flux are illustrated The process of heat transfer from metal to inner
in Figures 9 and 10 with feedwater flow put on control fluid in the waterwalls is through temperature difference
from drum level with a simple proportional and integral between metal and inner fluid. Metal temperature is
control. obtained from integration of heat flux into and out of
the tube walls divided by the tube heat capacitance.
Kp = 100 lbs/in The steam generated from a given amount of heat flux
is influenced by the drum pressure through a multiplying
KI = 0.1 lbs/in effect as noted in Figure 11.
The parameters of this model are defined as follows:
Figures 9 and 10 show the response of pressure,
steam flow and drum level for step changes in turbine kW = waterwall heat transfer coefficient BTU/sec°F
valve flow area and heat flux respectively around the
full load and half load operating points. CD = main component of boiler storage PSI/lb

CSH = equivalent component o f boiler storage lumped

at turbine end. PSI/lb
K = friction drop coefficient lb/sec x /PSI 1/2
CV = turbine control valve flow area lb/sec x PSI
CM = metal capacitance BTU/OF

K,,, = steam generation per BTU lb/BTU

KT = dTsat/dPsat OF/PSI
KH = 2[hg - hfwI/aP BTU/PSI
The storage constant (C + C ) can be defined as
the change in mass of steam &orec?!n waterwalls, drum,
100% Load Porn!
and superheaters and steam leads per change in drum
pressure. The contribution to this storage constant from
Figure 9. 2% Step Change in Valve Flow Area superheaters is approximated as
b Pi

-------
where Vi is the superheater section volume and

'E $IT is the partial density with respect

to pressure at an average constant
temperature in the section.
The contribution from the waterwalls and drum (C )
containing saturated fluid is more complicated and can le
derived from considering the system as in Figure 12 where
M and M are the resident masses of saturated water and
sQeam re2pectively.
The small perturbation form of the mass balance,
volume balance and energy balance equations for this
5oX Load subsystem are:

Figure 10. Step Change in Heat Flux dMf dM

-
dt = Ai, dt (22)

Authorized licensed use limited to: Politehnica Timisoara. Downloaded on November 2, 2009 at 07:49 from IEEE Xplore. Restrictions apply.
 Turbine V a l v e

Steam Flow
>

Throttle
Pressure

m
"w
Air
Steam
Generation

Qw Heat t o Fluid

Figure 11. Simplified boiler model with pressure effect on steam generation

derived from the physical dimensions and !hara%eristics

of the boiler and use of first principles of mass
balance, energy balance etc. as illustrated in this
paper, a more practical approach is to derive the

Authorized licensed use limited to: Politehnica Timisoara. Downloaded on November 2, 2009 at 07:49 from IEEE Xplore. Restrictions apply.
 72

16 process lags, yields the responses to step changes in

---
turbine valve in Figure 15 using the detailed model and
(CO + CSH) = 2400 x 60 x .02 in Figure 16 using the simplified model.
= 180 seconds
with the variables, pressure and steam flow, expressed 'n
per unit of rated. Experience has shown that splitting
the total storage constant (CO + CsH) as 10% for C and
90% for CO yields a good approximation to the &ler
pressure response.
The parameter K is obtained from a knowledge of the
pressure drop between drum and admission to the turbine.
For instance if at full load (i = 1 pu) the drum and
throttle pressures are 2640 and 5400 psi respectively,
then

k / 2640 - 2400
2400 = 1.0 or K = 3.16

The parameter KT relates the change in saturation 100% Load 50% Load
temperature with pressure. Keeping the temperature as
degrees F and pressure in per unit Figure 13. Step Change in Valve Flow Area for Model of
Figure 1 1
2T
KT = - x 2400
hP
K typically for a 2400 psi unit (drum pressure around
Typically 7
2g40 psi) = 0.057 x 2400 = 137.
can be estimated from the assumption of
25OF drop for fu 1 load steam generation. Hence
1
$ = 25 = 0.04

The value of heat capacitance CM can be obtained from the

\ \ '
mass of water wall tubes multiplied by specific heat of 100% Load 50% Load
metal. From experience with modeling of boilers it is
found that the effective time constant of heat transfer

--
from waterwalls to inner fluid is in the order of 7 Figure 14. 2% Step Change in Heat Flux for Model of
seconds. Hence CM/KW = 7 or CM = 0.7. Figure 1 1

KH = J[hg - hfwl/dP .-_

I----4
For a 2400 psi unit with enthalpy at the economizer
outlet o f 560 BTU/lb KH = -0.12, BTU/lb psi.
In per unit
-0.12
KH= - 520 x 2400 = -0.55
where 520 = hgo - hfwo.
Using these parameters in the model of Figure 14 yields
the responses to turbine valve step and heat release step
changes as in Figures 13 and 14 which correspond very
closely to those in Figures 9 and 10 obtained from the
-.-

J,

.e0*
e--
It-----
more elaborate modeling of the process physics. 100% Load sec 50% Load

ResDonse With Pressure Control Figure 15. Detailed Model with Pressure Control
2% Step Increase in Valve Flow Area
Assuming that the fuel system dynamics are
characterized by a dead time of 20 secs in series with a The controller function is
time constant of 30 secs
AQ KI(l + TIs)(l + TRs)
c-20s
AS - 1 +30 S
a three mode control (proporbional, integral and rate)
operating on pressure error , properly tuned for the
(I+ 3
with KI = 0.04, TI .= 100 sec, TR = 30 sec.

Authorized licensed use limited to: Politehnica Timisoara. Downloaded on November 2, 2009 at 07:49 from IEEE Xplore. Restrictions apply.
 73

.. - .. r" .....- - .. W O -
sec

_.? O - 7 pu

100% Load
f
25 P S I
100% Load 50% Load 3.
Figure 16. Simplified Model With Pressure Control Figure 18. Model as in Figure 17 Without Pressure
2% Step Increase in Valve Flow Area Controls
One notes that the simplified model's response is
almost identical to that of the more detailed model.
A Sinmler Model

The pressure-temperature effect on steaming rate

(K ) in the block diagram of Figure 11 is a transient
efTect and not very significant. The pressure-enthalpy
effect on steaming rate (KH) affects the 'steady state
gains between valve flow area and pressure and between
heat flux and pressure.
Since control effects with integral action can
neutralize variations in steady state process gain, the
AQ
I
0.05 pu
control performance of boiler pressure under changes in
turbine valve flow area is basically the same where the
effects of K and K are ignored in the block diagram of
Figure 11. ?he mod& is then identical to that of Figure 100%Load 50% Load
6, reference (1) and is shown in Figure 17.
Figure 19. Model of Figure 19 With Pressure Controls
2% Step Change in Valve Flow Area
CONCLUSIONS
VALVt PVS
SCT ran In certain studies of power system dynamic
performance, steam turbine mechanical power response
including boiler pressure effects may be significant,
requiring the modeling of the steam supply system. This
paper has demonstrated the validity of simplified boiler
models that have previously been used to represent these
effects.
Boiler response characteristics derived from
modeling of the basic energy balance, mass balance and
volume balance relations using physical boiler parameters
were compared with those obtained from two other
Figure 17. Dynamics of Turbine Power Including Boiler simplified models, one which matched both the steady
Pressure Effects state and transient open loop boiler response
characteristics and a simpler model which matches the
Figure 18 shows the response of this model to a step initial open loop response. It is shown that both
change in valve flow area with no change in energy input. simplified models yield acceptable results when used to
The parameters in this model were the same as those used yield the boiler response including pressure controls.
in the model of Figure 11 except for the elimination of Suggestions are made on how to derive parameters to
the constants KT, K,,, MC and Ky and the introduction of be used with these models.
Tw = 7 sec as the water wall time constant. A partial list of references in boiler modeling and
One notes that the differences between the response control is also cited (4-10).
of this model and those of the previous two models lie in
the steady state gain. REFERENCES
The responses with pressure controls with same
settings used with the previous two models are shown in 1. F. P. de Mello and D. N. Ewart, "MW Response of
Figure 19. By comparing these with those in Figures 15 Fossil Fueled Steam Units, IEEE Working Group on
and 16 one concludes that the simpler model of Figure 19
is quite valid when used together with pressure controls.

Authorized licensed use limited to: Politehnica Timisoara. Downloaded on November 2, 2009 at 07:49 from IEEE Xplore. Restrictions apply.
 ~

74

Power Plant Response to Load Changes", IEEE

Transactions on Power Amaratus and Systems, Vol.
PAS-92, No. 2, March/April 1973, pp. 455-463.
2. F. P. de Mello, D. N. Ewart and W. M. Stephens,
"Simulation as a Design Tool for Plant Jack
McDonough Boiler Controls", ISA 7th National Power
Instrumentation Symposium Proceedings, 1964, pp. 35-
46.
3. F. P. de Mello, "Plant Dynamics and Control
Analysis", IEEE Transactions on Power Atmaratus and
Systems, Paper 63-1401, 1963, Vol. S82, pp. 664-678.
4. Durrant, 0. W. and Loeser, J. K., "Boiler-Turbine
Control System for Application to Universal Pressure
Boilers", IEEE Conference Paper CP63-1410, ASME-IEEE
National Power Conference, Cincinnati, Ohio,
September 22-25, 1963.
5. Profos, P., "Dynamics of Pressure and Combustion
Control in Steam Generators", Sulzer Technical
Review, V37, No. 4, 1955, pp. 1-15.
6. Chien, K. L., Ergin, E. I., Ling C., and Lee, A.,
"Dynamic Analysis of a Boiler", ASME V 80, December
1958, pp. 1809-1819.
7. McDonald, J. P. and Kwatny, H. G., "A Mathematical
Model for Reheat Boiler-Turbine-Generator Systems",
IEEE Paper 70 CP221-PWR, PES Winter Power Meeting,
New York, New York, January 25-30, 1970.
8. Ahner, D. J., de Mello, F. P., Dyer, C. E and
Summer, V. C., "Analysis and Design of Controls for
a Once-Through Boiler Through Digital Simulation",
Proc. 9th National Power Instrumentation Symposium,
Detroit, May 16-18, 1966.
9. de Mello, F. P. and had, F. P., "Boiler Pressure
Control Configurations", IEEE Paper CP31, IEEE
Winter Power Meeting, New York, New York, January
29-February 3, 1967.
10. Anderson, P. M., "Modeling Thermal Power Plants for
Dynamic Stability Studies":, Book Cyclone Copy
Center, Ames, Iowa 1974.
F. P. de Mello graduated with a B.Sc and M.Sc degrees in
Electrical Engineering from MIT in 1948.
In 1948 he joined Rio Light and Power and Sao Paulo
Companies in Brazil and for the next seven years held
positions of increasing technical responsibility in
system planning and design studies.
From 1955 to 1969, he worked in the General Electric
Company Electric Utility Engineering Operation where he
specialized in studies of dynamics and electrical
machines, excitation control, prime-mover systems,
overall power systems and boiler dynamics.
Mr. de Mello joined Power Technologies, Inc. at the
time of its formation in August of 1969 as a Principal
Engineer, Dynamics and Control. He was appointed Vice
President in 1973.
He is presently a Principal Consultant for Power
Technologies, Inc.

Authorized licensed use limited to: Politehnica Timisoara. Downloaded on November 2, 2009 at 07:49 from IEEE Xplore. Restrictions apply.