You are on page 1of 8

ISSN 1052-6188, Journal of Machinery Manufacture and Reliability, 2015, Vol. 44, No. 8, pp. 744–751.

© Allerton Press, Inc., 2015.


Original Russian Text © G.A. Pikina, F.F. Pashchenko, 2014, published in Problemy mashinostroeniya i avtomatizatsyi, 2014, No. 1, pp. 143–149.

Models of Steam Generator Heat Processes


G. A. Pikinaa and F. F. Pashchenkob
a
Moscow Power Engineering Institute, ul. Krasnokazarmennaya E-250, Moscow, 111250 Russia
b
Institute of Control Sciences, Russian Academy of Sciences, ul. Profsoyuznaya 65, Moscow, 117997 Russia
e-mail: pikinaga@mpei.ru, feodor@ipu.rssi.ru

Abstract—Static and dynamic models with distributed and lumped parameters have been developed
for steam generators of water-moderated nuclear reactors (WMNRs).
It has been shown that the application of static models with lumped parameters may lead to incorrect
design solutions when constructing steam generators. It is stated that dynamic models with lumped
parameters have a low accuracy, especially in the case of the primary water temperature channel used
as a control action in the thermal load control system of the nuclear reactor.

Keywords: mathematical model, distributed and lumped parameters, water-moderated nuclear reac-
tor, steam generator.
DOI: 10.3103/S105261881508004X

INTRODUCTION
A steam generator is a metal vessel [3] half-filled with secondary water (Fig. 1). The pipe still is sub-
merged in water. Primary water incoming from the power reactor flows within the pipes. From outside,
the steam–water mixture is on the saturation line. The dry saturated steam inflows to a turbine. Therefore,
four basic media can be distinguished in the object of study: primary water, pipe-still active metal, steam–
water mixture, and steam generator frame.
The isobaric thermal capacity of the external coolant (working fluid) of the WMNR steam generator
located on the saturation line is infinite, i.e., cp1 = ∞, and, hence, the Stanton number is zero, i.e., St1=
0, and the working fluid energy equation degenerates into equation θ1s(z) = const [5].
The saturation temperature depends only on pressure maintained by upstream constant pressure reg-
ulators or on the power unit thermal load. Therefore, the mathematical model of this heat exchanger is
divided into two subsystems, i.e., the systems of two differential equations of the energy of the internal
coolant (primary water) and of the thermal conductivity of pipe walls, and a separately solved equation of
frame energy. The block diagram of the model of heat processes is shown in Fig. 2.

Static Model for Steam Generator Heat Processes


To obtain the distribution of temperatures over the length of pipes (with external radius r1 and internal
radius r2), let us take a model of a plane wall with an equivalent coefficient of primary water heat transfer
2 = α 2 r2 / r1. This choice simplifies the analytic solution without introducing any addi-
to the pipe metal α eq
tional error [6]. As will be further seen, the wall model has an effect only on the structure of the formula
of heat transfer coefficient K.
Taking into account the accepted simplification, the steam generator static model is as follows:

⎧∂θ 2 = −St (θ − ϑ );
⎪⎪ dz 2 2 2

⎨ 2 (1)
⎪∂ ϑ2 = 0,
⎪⎩dy

744
MODELS OF STEAM GENERATOR HEAT PROCESSES 745

Ds

h P1s, θ1s
Ds

Ds
D2, θ2 D2, θ2
in out

Fig. 1. Simplified diagram of the WMNR steam generator.

X2 Y2
Medium 2
D2, P2, θ2 D2, P2, θ2
in in

θ2 υ

Wall

θ1s
P1s
Medium 1

θ1s

Frame

qloss

Fig. 2. Block diagram of the model of steam generator heat processes.

with boundary conditions

⎛ ∂ϑ ⎞ α ⎛ ∂ϑ ⎞ α
eq

⎜ ⎟ = 1 (ϑ1 − θ1s ); ⎜ ⎟ = 2 (θ 2 − ϑ 2 ).
⎝ ⎠ y =δ λ
∂ y ∂
⎝ ⎠ y =0
y λ

Without deriving the formula, let us write the solution to system (1) with respect to the representation
of the external coolant temperature θ2 (s):

K St 2 / α 2
eq
θ 2(s) = 1 θ in + θ (s),
eq 2 eq 1
s + K St 2 / α 2 s + K St 2 / α 2
where heat transfer coefficient K, depending on the required accuracy, is calculated by formulas for cylin-
drical or plane walls:
−1 −1
⎛ r ⎛ r ⎞⎞ ⎛ ⎞
K cyl = ⎜ 1 + 1eq + 1 ln ⎜ 1 ⎟ ⎟ ; K pl = ⎜ 1 + 1eq + δ ⎟ . (2)
⎝ α1 α 2 λ ⎝ r2 ⎠ ⎠ ⎝ α1 α 2 λ ⎠
JOURNAL OF MACHINERY MANUFACTURE AND RELIABILITY Vol. 44 No. 8 2015
746 PIKINA, PASHCHENKO

θ2(z)
330

310

DP

290
LP

270
0 0.2 0.4 0.6 0.8 1.0
z

Fig. 3. Temperature distribution of the primary loop of VVER-440.

Let us pass from the representations of variables to the original functions. For this purpose, one should
substitute the Laplace representation of saturation temperature θ1s(s), taking into account its length con-
stancy z,
θ 1s
θ 1s ( s ) =
,
s
where θ1s is the value of the working-fluid saturation temperature. Let us substitute this expression into
the solution for θ2(s):
K St 2 / α 2
eq
θ 2(s) = 1 θ in + θ1s = 1 θ 2in + b θ ,
s+b s(s + b)
eq 2 1s
s + K St 2 / α 2 s(s + K St 2 / α 2 )
eq

where b = K St 2 / α 2eq and make the inverse Laplace transformation:


θ 2(z) = (1 − e −bz )θ1s + e −bz θ 2in . (3)
A view of the primary water temperature distribution according to formula (3), which was estimated for
the steam generator of a water-moderated reactor VVER-440, is shown in Fig. 3 (DP curve), where the
static temperature distribution curve (LP curve) for the wall point model (when we assume the absence of
its thermal resistance λ = ∞) is also shown for comparison.
The difference between models with distributed parameters and the point wall model reached 12°C in
the central part of pipes, and the difference of temperatures at the output was 8°C. The relative error of
calculation of the static distribution of water temperature did not exceed 4% of the absolute temperature
values, since the pipe thickness was only 1.5 mm. However, if we assess error relative to the differential
water temperature (θ in 2 – θ 2 ), it increases up to 30%.
out

In the case of heat exchanges with thick pipes, the temperature calculation error increases significantly.
The use of the model with lumped parameters may lead to an incorrect design solution when constructing
a steam generator, since the predetermined temperature difference (θ in 2 – θ 2 ) is achieved when the pipe
out

still is half as long, according to the LP model.

LINEAR DYNAMIC MODELS AND CHARACTERISTICS


OF VVER STEAM GENERATOR HEAT PROCESSES
This article considers linear models of steam generators in which external coolant (working fluid) is on
the saturation line at constant pressure P1s = const [1, 2, 4–8].
Based on the assumption of pressure constancy, we may only consider internal coolant and wall energy
equations, with the temperature (or pressure) of the coolant saturation θ1s = f(P1s) being taken as an input
value. Below the evaporation surface in the steam generator, there is a pipe still (a system of pipes com-

JOURNAL OF MACHINERY MANUFACTURE AND RELIABILITY Vol. 44 No. 8 2015


MODELS OF STEAM GENERATOR HEAT PROCESSES 747

θ1s θin

k1 W M W2
St2W2 θ2
υ D2
kD2W2
k2 W M

Fig. 4. Signal graph of DP model.

bined by distributing and collecting headers) with a primary heating water flowing in it and secondary
boiling water flowing outside of it. Let us write a system of pipe still equations in the operator form, which
includes an energy equation for the water in pipes and the pipe metal heat conduction equation [4, 5]:
⎧θ 2( p, s) = W 2( p, s)θ 2in ( p) + St 2W 2( p, s)ϑ( p, s) + k D2W 2( p, s)D2( p, s);
⎨ (4)
⎩ϑ( p, s) = k1W M ( p)θ1s ( p, s) + k 2W M ( p)θ 2 ( p, s).
The signal graph of the system is shown in Fig. 4. Based on this, we obtain the dynamic characteristics
of the channels of interest.
With the neglect of heat loss from the primary pipelines to the environment, static water temperature
at the steam generator input θ in2 will be equal to the water temperature at the reactor output θout.
In transient regimes, these temperatures are not equal, due to the heat accumulated in the water and the
pipeline metal. If necessary, the reactor and steam generator model can be extended by pipeline models.

2 → θ2
The Dynamic Characteristics of Channel θ in
The model signal graph includes only one loop. Therefore, the two-dimensional transfer function of
the coolant temperature channel has quite a simple appearance:
W 2( p, s) 1
W θinθ ( p, s) = = . (5)
2 2
1 − k 2St 2W 2( p, s)W M ( p) s + [τ 2 p + St 2(1 − k 2W M ( p))]
Let us make the inverse Laplace transformation of the transfer function relative to argument s:
W θinθ ( p, z) = e −[τ2 p +St 2(1− k2W M )]z = e −τ2zpW *( p, z ), (6)
2 2

⎛T p + k1 ⎞
−St 2 ⎜ M ⎟z
−St 2[1− k 2W M ( p)]z ⎝ TM p +1 ⎠
where W *( p, z) = e =e .
The obtained transfer function includes transcendental function W*(p,z) with a fractional rational
exponent and a lagging element with parameter τ = τ2z. The lagging value varies from 0 to τ2, depend-
ing on the value of coordinate z. In order to construct the complex transfer function, let us study
expression (6) for frequencies ω = 0 and ω → ∞.
At ω = 0, let us determine the channel gain
W θinθ ( j 0, z ) = W *( j 0, z ) = e −St 2(1− k2 )z = e −St 2k1z = K θinθ (z ),
2 2 2 2

where K θinθ (z) is the channel transfer coefficient depending on z.


2 2

At ω → ∞, we have W *( j ∞, z) = e −St 2z , i.e., a proportional element with gain depending on length;


coordinate z is an asymptote to the fractional rational part of the CFC (complex frequency characteristic).
Taking into account the lagging element, we conclude that circle e −St 2z e − j τ2zω is an asymptote to the
complex frequency characteristic of the temperature channel. Figure 5 shows complex frequency charac-
teristics of the model with distributed parameters (DP curve for z = 1) and the point model (LP curve)
with parameters that are characteristic of the VVER-1000 steam generator: St2 = 5.75, k1 ≈ k2 = 0.5, τ2 =
3 s, TM = 0.1 s.

JOURNAL OF MACHINERY MANUFACTURE AND RELIABILITY Vol. 44 No. 8 2015


748 PIKINA, PASHCHENKO

Im(ω)
0.10

0.05

ω=0 ω=0
0

DP
−0.05

LP
−0.10

−0.15
−0.1 −0.05 0 0.05 0.10 0.15 0.20 0.25 0.30
Re(ω)

Fig. 5. CFC of channel θ in


2 → θ 2 of steam generator VVER-1000.

hθ θ (t)
2
in
2
out

0.2

LP
0.1
DP
exp(−St2) exp(−St2k1)
0 2 τ2 4 t, s

Fig. 6. Transient characteristic of channel θ in


2 → θ 2 of steam generator VVER-1000.

The respective transient characteristics of the models are given in Fig. 6.


It is not difficult to obtain the transfer function of a model with lumped parameters (LP model). For
this purpose, we should only substitute the transfer function W2 (p,s) with the transfer function of the LP
flow model W2 (p) = 1/(τ 2 p + 1 + St 2 ) in general expression (5) of the two-dimensional transfer function
of the channel. Then, we have

W 2( p)
W θinθ ( p) =
1 − k 2St 2W 2( p)W M ( p)
2 2
(7)
TM p + 1 TM p + 1
= = .
(τ 2 p + 1 + St 2 )(TM p + 1) − k 2St 2 τ 2TM p 2 + [τ 2 + (1 + St 2 )TM ]p + 1 + k1St 2

As expected, we have obtained a second-order dynamic system (two heated vessels are included in the
model: a wall and an internal coolant). Let us determine the channel gain coefficient KLP of the point

JOURNAL OF MACHINERY MANUFACTURE AND RELIABILITY Vol. 44 No. 8 2015


MODELS OF STEAM GENERATOR HEAT PROCESSES 749

model from its transfer function, assuming that p = 0, and compare it with the gain coefficient KDP of the
distributed model:

K LP = 1 ≥ K DP = e −k1St 2 = 1 .
1 + k1St 2 1
1 + k1St 2 + (k1St 2 ) + …
2

2!
Thus, as in the case of convective heat exchangers with single-phase coolants, the point model yields
an overestimated value of the gain coefficient. The level of difference is determined by the value of the
Stanton number and will decrease with the decrease of the St2 value.
By comparing the dynamic characteristics in Figs. 5 and 6, we may conclude that LP models accepted
in engineering practice are not sufficient to provide a satisfactory view concerning not only the dynamic
properties of steam generators, but also the static temperature values.

Dynamic Characteristics of Channel D2 → θ2


The gain coefficient kD2 included in the energy equation for the internal coolant depends on z:
⎛ ∂θ ⎞
k D2 = ⎜ 2 ⎟ 1 .
⎝ ∂ z in ⎠ 0 D20
Let us differentiate equation (3) of temperature distribution obtained for static conditions
⎛ ∂θ 2 ⎞ −bz
⎜ ∂ z ⎟ = be (θ1s = θ 2 )0,
in

⎝ in ⎠ 0
and take coefficient kD2 as a constant value. When choosing kD2, different variants are possible, but it seems
to be most reasonable to assume a mean integral length value equal to
⎛ ∂θ2 ⎞
⎜ ∂ z ⎟ = (θ 2 − θ 2 )0.
out in

⎝ in ⎠ 0
Then we may write the following expression from the signal graph:
k D2
W D2θ2 ( p, s) = k D2W θinθ ( p, s) = .
2
s + [τ 2 p + St 2(1 − k 2W M )]
2

Let us now make the inverse Laplace transformation, assuming that D2(p, s) =D2(p)/s:
− 1 ⎧W ( p, s)⎫ −1 ⎧ k D2 ⎫
W D2θ2 ( p, z) = L ⎨ D2θ2 ⎬=L ⎨ ⎬
⎩ s ⎭ ⎩s{s + [τ 2 p + St 2(1 − k 2W M )]}⎭ (8)
k D2 ⎡1 − W θinθ ( p, z)⎤ .
=
τ 2 p + St 2[1 − k 2W M ( p)] ⎣ 2 2 ⎦
Figure 7 provides the complex frequency characteristics of the model with distributed parameters (DP
curve for z = 1) and the point model (LP curve) of the VVER-1000 steam generator for kD2 =
0.00739°C/(kg/s).
The distribution of the input action (coolant flow) over the pipe still length significantly diminished
the influence of time constant τ2, compared with the temperature channel in Fig. 5, which leads to a sig-
nificant convergence of the frequency characteristics of the DP and LP models. Nevertheless, the error of
models with lumped parameters is still intolerably large.

Dynamic Characteristics of Channel θ1s → θ2


Let us turn again to the signal graph in Fig. 4 and write the expression of the two-dimensional transfer
function relative to the temperature θ1s of the steam–water mixture.
k1St 2W M
W θ1sθ2 ( p, s) = k1St 2W M ( p)W θinθ ( p, s) = .
s + [τ 2 p + St 2(1 − k 2W M )]
2 2

Saturation temperature θ1s is constant throughout the standardized pipe still length z and, hence, its
double image is equal to

JOURNAL OF MACHINERY MANUFACTURE AND RELIABILITY Vol. 44 No. 8 2015


750 PIKINA, PASHCHENKO

Im(ω) ω = 0 rad/s
0

−5 × 10−4

ω = 1.5 LP

−1 × 10−3
DP

−1.5 × 10−3
0 5 × 10−4 1 × 10−3 1.5 × 10−3 2 × 10−3 2.5 × 10−3
Re(ω)

Fig. 7. CFC of channel D 2 → θ 2 of steam generator VVER-1000.

Im(ω) ω = 0 rad/s
0

−0.2 LP

ω = 1.5
−0.4
ω = 1.5 DP
−0.6
−0.2 0 0.2 0.4 0.6 0.8 1.0
Re(ω)

Fig. 8. CFC of channel θ1s → θ 2 of steam generator VVER-1000.

θ1s ( p)
θ1s ( p, s) = .
s
Taking this into account, the common transfer function of the channel is

−1 ⎧W ( p, s)⎫ −1 ⎧ k1St 2W M ( p) ⎫
Wθ1sθ2 ( p, z) = L ⎨ θ1sθ2 ⎬=L ⎨ ⎬
⎩ s ⎭ ⎩s{s + [τ 2 p + St 2(1 − k2W M ( p))]}⎭
k1St 2W M ( p) ⎡1 − Wθinθ ( p, z)⎤ = k1St 2W M ( p)W D2θ2 ( p, z).
=
τ 2 p + St 2(1 − k2W M ( p)) ⎣ 2 2 ⎦ k D2
As can be seen from (9), the inertia of this channel is higher than the inertia of the flow changing chan-
nel D2, which is due to the effect on temperature θ1s that is made through the heat-transfer wall with trans-
fer function WM(p).
According to formula (9) for the above-assumed numerical values of the model parameters, the com-
plex frequency characteristic was calculated for the coolant temperature at steam generator output. Its
hodograph is shown in Fig. 8 (DP curve). For comparison, a model CFC hodograph with lumped
medium parameters (LP curve) is shown in the same figure. As expected, taking into account the distri-
bution of the input effect on temperature θ1s over the pipe still length, we managed to reduce the error of

JOURNAL OF MACHINERY MANUFACTURE AND RELIABILITY Vol. 44 No. 8 2015


MODELS OF STEAM GENERATOR HEAT PROCESSES 751

the LP model transfer coefficient to approximately 20%. However, the discrepancy between the models
grows rapidly with increases in frequency.

CONCLUSIONS
The discrepancy between the static models with distributed and lumped parameters at a pipe thickness
of 1.5 mm is 30% of the temperature difference (θ in2 – θ 2 ). The use of models with lumped parameters
out

may lead to incorrect design solutions when constructing steam generators, since the predetermined tem-
perature difference (θ in
2 – θ 2 ) is achieved when the pipe still half as long, according to the LP model.
out

Summing up the results of calculation of steam generator dynamic characteristics, one may confidently
state the low accuracy of models with lumped parameters. This is particularly noticeable in the case of the
temperature channel of the coolant (primary water), which is used as a control action in the thermal load
control system of the nuclear reactor.

ACKNOWLEDGMENTS
This work was supported by the Russian Foundation for Basic Research (project no. 12-08-01334).

REFERENCES
1. Demchenko, V.A., Avtomatizatsiya i modelirovanie tekhnologicheskikh protsessov TES i AES (The Way to Auto-
mate and Simulate Technological Processes at Thermal Power-Plants and Nuclear Power Plants), Odessa:
Astroprint, 2001.
2. Ivanov, V.A., Regulirovanie energoblokov (Power Units Control), Leningrad: Mashinostroenie, 1982.
3. Margulova, T.Kh., Atomnye elektricheskie stantsii (Nuclear Power Plants), Moscow: Vysshaya shkola, 1974.
4. Pashchenko, F.F. and Pikina, G.A., Osnovy modelirovaniya energetisheskikh ob”ektov (Simulation Foundations
for Energy Objects), Moscow: Fizmatlit, 2011.
5. Pikina, G.A., Matematicheskie modeli tekhnologicheskikh ob”ektov (Mathematical Models of the Technological
Objects), Moscow: National Research Univ. MPEI, 2007.
6. Pikina, G.A., The way to choose a wall model for calculating heat exchanges dynamics, Vestn. Mosk. Energ. Inst.,
2008, no. 1.
7. Plyutinskii, V.I., Staticheskie i dinamicheskie kharakteristiki yadernykh energeticheskikh ustanovok (Static and
Dynamic Characteristics of the Nuclear Power Plants), Moscow: National Research Univ. MPEI, 1980.
8. Plyutinskii, V.I. and Pogorelov, V.I., Avtomaticheskoe upravlenie i zashchita teploenergeticheskikh ustanovok AES:
uchebnik dlya tekhnikumov (Automatic Control and Protection of Heat-Power Engineering Plants in Nuclear
Power Plants. Student’s Book for Technical Schools), Moscow: Energoatomizdat, 1983.

Translated by D. Zabolotny

JOURNAL OF MACHINERY MANUFACTURE AND RELIABILITY Vol. 44 No. 8 2015

You might also like