Nuclear Engineering and Design 177 (1997) 147 – 153

Analysis of steam generator loss-of-feedwater experiments with

APROS and RELAP5/MOD3.1 computer codes
E. Virtanen a, T. Haapalehto a, J. Kouhia b,*
a
Lappeenranta Uni6ersity of Technology, P.O. Box 20, FIN-53851 Lappeenranta, Finland
b
VTT Energy, Nuclear Energy, P.O. Box 20, FIN-53851 Lappeenranta, Finland

Abstract

Three experiments were conducted to study the behaviour of the new horizontal steam generator construction of
the PACTEL test facility. In the experiments the secondary side coolant level was reduced stepwise. The experiments
were calculated with two computer codes RELAP5/MOD3.1 and APROS version 2.11. A similar nodalization scheme
was used for both codes so that the results may be compared. Only the steam generator was modeled and the rest
of the facility was given as a boundary condition. The results show that both codes calculate well the behaviour of
the primary side of the steam generator. On the secondary side both codes calculate lower steam temperatures in the
upper part of the heat exchange tube bundle than was measured in the experiments. © 1997 Elsevier Science S.A.

1. Introduction During normal operation all the heat exchange

Compared with the Western PWRs the Russian
design VVER-reactors have a special feature:
horizontal steam generators. This type of steam
generator (SG) has two vertical collectors and
horizontal U-shaped heat exchange tubes. The
behaviour of the horizontal SG is very different
compared with the western type vertical SG. For
example, the secondary side coolant volume rela-
tive to the reactor core power of a VVER-440 is
much larger than in a typical western PWR.
Hence, all the loss-of-feedwater transients are
slower. During a total station black-out incident a
vertical SG dries in some tens of minutes but a
horizontal SG in some hours.
* Corresponding author. Fax: +358 5 6212379; e-mail:
Jyrki.Kouhia@vtt.fi
tubes of a horizontal SG are covered with water,
but in the case of loss-of-feedwater accident the
secondary side water level starts to decrease.
When the level falls below the highest tube rows
the heat transfer from primary to secondary side
starts to deteriorate. If the heat transfer de-
creases more rapidly than the heat produced in
the core, the primary coolant temperature begins
to rise. This temperature rise in turn provokes
the expansion of primary coolant and an ascent
in the liquid level in the pressurizer. When the
primary circuit pressure rises up to the opening
pressure of the pressurizer relief valves they open
and allow primary coolant to leak out of the
system. Clearly, it is quite important that such
behaviors are completely understood, since
coolability of the core must be maintained in all
conditions.

0029-5493/97/$17.00 © 1997 Elsevier Science S.A. All rights reserved.

PII S 0 0 2 9 - 5 4 9 3 ( 9 7 ) 0 0 1 9 1 - X
148 E. Virtanen et al. / Nuclear Engineering and Design 177 (1997) 147–153

The Western thermal-hydraulic system codes, the steam generator, hence a roughly U-shaped
like RELAP5 [1] and CATHARE2 [2], are devel- pipe is needed to complete the connection. The
oped and validated to model the phenomena in cold leg loop seal is formed by the elevation
the vertical SGs. But the differences between ver- difference of the inlet and outlet of the reactor
tical and horizontal steam generators are so im- coolant pump. Fig. 1 shows how the geometry of
portant that the calculational models developed the loop seals are modelled in the PACTEL facil-
for the vertical SGs are not directly applicable for ity.
the horizontal SGs. The PACTEL loss-of-feedwa-
ter experiments have been performed to study the
2.1. Steam generator description
overall behaviour of the horizontal steam genera-
tors and especially to produce experimental re-
The number of the loops has been reduced from
sults for code assessment. The work presented
six of the reference system to three in PACTEL,
here has been done at Lappeenranta University of
thus one PACTEL steam generator corresponds
Technology.
to two in the power plant. The original steam
generator of PACTEL has full-length heat ex-
change tubes and the same tube bundle geometry
2. Facility description
as in the power plant steam generator. The tube
bundle height is less than 15% of the height of the
PACTEL (Parallel Channel Test Loop) [3] is a
reference steam generator. This fact limits the use
volumetrically scaled (1:305) out-of-pile model of
of the original PACTEL steam generator, when
the Russian design VVER-440 reactors used in
for example the effects of the decrease of the
Finland. The facility is an integral test facility and
secondary side level are studied.
it includes all the main components of the pri-
The new steam generator (Fig. 2), installed in
mary circuit of the reference reactor. The reactor
one loop, has vertical primary collectors and hori-
vessel is simulated by a U-tube construction con-
zontal heat exchange tubes. The 118 U-shaped
sisting of separate core and downcomer sections.
heat exchange tubes are arranged in 14 layers and
The core is comprised of 144 electrically heated
nine vertical columns. The average length of the
fuel rod simulators. The geometry and the pitch
tubes (2.8 m) is about one third of that in the full
of the rods are the same as in the reference
reactor. The rods are divided into three roughly
triangular shaped parallel channels, which repre-
sent the intersection of the corners of three hexag-
onal VVER rod bundles. The maximum total core
power is 1 MW, or 22% of the scaled nominal
power. The maximum primary pressure is 8.0
compared with 12.3 MPa of the reference reactor.
The component heights and the relative eleva-
tions correspond to those of the full scale reactor
to match the natural circulation pressure heads in
the reference system. The hot and cold leg eleva-
tions of the power plant have been reproduced.
This is particularly important for the loop seals.
Unlike other PWRs VVER-440 has a loop seal
also in the hot leg. This is a consequence of the
steam generator location, which is almost at the
same height as the hot leg connection to the upper
plenum. The primary collector of the steam gener-
ator is connected to the hot leg at the bottom of Fig. 1. Geometry of the PACTEL facility.
 E. Virtanen et al. / Nuclear Engineering and Design 177 (1997) 147–153
Fig. 2. New steam generator of PACTEL.
scale steam generator (9.0 m). The outer diameter
of the tubes is 16 mm, corresponding to the
reference system, and the inner diameter is 13 mm
(in the power plant 13.2 mm). In order to have a
higher tube bundle, the pitch in the vertical direc-
tion has been doubled (48 mm) compared with the
reference steam generator. The pitch in the hori-
zontal direction has been maintained. The longer
pitch in the vertical direction may cause differ-
ences especially for the heat transfer but also for
pressure losses and swelling. However, under loss-
of-feedwater conditions a relatively large error in
the estimation of boiling heat transfer coefficient
has only a small effect on the total heat transfer
coefficient, because the boiling heat transfer co-
efficient is very large compared with the convec-
tive coefficient on the primary side.
The outer diameter of the shell is 1.0 m (in the
power plant 3.34 m). Because of the higher verti-
cal pitch, the secondary side is larger than the
scaled down secondary volume. This distorts the
time scale of secondary side transients. Two com-
partments have been constructed on each side of
the steam generator to decrease the mass of water
directly involved in the primary to secondary heat
transfer process. The compartments are not to-
tally isolated from the rest of the secondary side,
but the coolant has several flow paths in and out
of the compartments.
The measurement instrumentation of the steam
generator contains mainly temperature measure-
ments. The primary and secondary side tempera-
149
tures are measured in eight tubes in three or four
different locations in each tube. The inlet and
outlet temperature of the primary coolant, the
primary mass flow rate, the mass flow rate of the
feedwater and the differential pressure on the
secondary side are measured as well. The mea-
sured collapsed level is based on the differential
pressure measurement.
3. Experiments
Three experiments, SG-2, SG-3 and SG-4, were
conducted to study the behaviour of the new
steam generator construction of PACTEL facility.
The experiments were like the loss-of-feedwater
experiments but instead of reducing the secondary
level continuously it was reduced in steps.
In these experiments only one loop with the
new steam generator was in operation. The pri-
mary coolant pump was operating during the
whole experiment. Primary and secondary side
pressures were different from one experiment to
the other (Table 1).
In the beginning of each experiment the core
power was set to about 960 kW and it was kept
constant until a steady state was reached. The
collapsed level of the secondary side was then
reduced stepwise. Feedwater injection was
stopped and when the secondary level had
dropped 50 mm the injection was started again. In
Fig. 3 the measured feedwater mass flow rate in
experiment SG-3 is shown. The system was al-
lowed to settle down for 30 min between each
step. When the secondary side coolant level had
dropped below the highest heat exchange tube
row the heat transfer from primary to secondary
side began to decrease. The temperature in the
Table 1
Primary and secondary pressures in the experiments
Test Primary pressure Secondary pressure (MPa)
(MPa)
SG-2 3.0 0.7
SG-3 1.2 0.3
SG-4 7.5 4.2
150 E. Virtanen et al. / Nuclear Engineering and Design 177 (1997) 147–153
Fig. 3. Measured feedwater mass flow rate in experiment SG-3
(measurement uncertainty 9 0.005 kg s − 1).
outlet of the cold collector as well as in the inlet
of the hot collector began to rise. In order to keep
the coolant temperature in the inlet of the hot
collector constant the core power was reduced.
Also the feedwater mass flow rate was changed to
correspond to the evaporation rate and to keep
the collapsed level constant during the settling
down period.
The results of the different loss-of-feedwater
experiments are similar. Naturally, both primary
and secondary temperatures are different because
the pressures are different. The only clear differ-
ence is that the behavior is more stable when the
secondary pressure is higher. This is because the
collapsed level is near the swell level and the
transition zone where the highest tubes start to
dry out is shorter.
4. Calculations
4.1. Computer codes used in the calculations
Two computer codes, RELAP5/MOD3.1 and
APROS version 2.11, were used in the calculation
of the experiments. RELAP5 is a thermal-hy-
draulic system code developed at the Idaho Na-
tional Engineering Laboratory (INEL) for
best-estimate transient simulation of light water
reactor coolant systems. RELAP5/MOD3.1 is the
latest version based on a nonhomogeneous and
nonequilibrium (six-equation) model for the two-
phase system. APROS (Advanced Process Simula-
tor) [4] code is developed in Finland by VTT and
Imatran Voima Oy (IVO). The APROS simula-
tion environment contains a database, graphics
and physical models for the simulation of nuclear
and conventional power plants and chemical pro-
cesses. The database includes several predefined
process components, like tanks, steam generators,
pumps and valves, which makes it easier to create
a new simulation model. The thermal-hydraulics
of APROS include single-phase and three- five-
and six- equation two-phase flow models.
4.2. Nodalization
Only the steam generator part of the facility has
been modelled, the rest of the facility has been
given as a boundary condition. The same nodal-
ization scheme has been used for both computer
codes although some code dependent alterations
had to be made.
The nodalization for RELAP5/MOD3.1 is
shown in Fig. 4. Hot and cold legs are modelled
with large time dependent volumes where the
pressure and temperature are controllable. A time
dependent junction connects the volume mod-
elling the hot leg to the hot collector. The vertical
hot and cold collectors are modelled with eight
nodes. The heat exchange tubes are lumped into
seven pipe components (numbers 705–735 in Fig.
4). The lowest component models the three lowest
tube rows, the next five components two tube
rows each and the highest component models the
highest tube row. All the pipe components are
divided horizontally into five nodes.
On the secondary side of the steam generator
the heat exchange tube area is modelled with
seven nodes and the vapor volume above the heat
exchange tubes with two nodes. The steam gener-
ator ends as well as the side compartments are
modelled with separate components. From the
side compartments (number 770 in Fig. 4) there
are only cross flow junctions to the steam genera-
tor ends (number 760 in Fig. 4), which in turn are
connected to the tube bundle area and also to the
vapour volume. This construction allows vertical
internal recirculation typically associated with
 E. Virtanen et al. / Nuclear Engineering and Design 177 (1997) 147–153
horizontal steam generators. Feedwater is injected
into the third lowest node in the heat exchange
tube area. The outlet of steam is modelled with a
time dependent volume as a mass sink and a
control valve connecting the steam line (number
780 in Fig. 4) to the sink.
All heat structures except the thin plates be-
tween the heat exchange tube area and the side
compartments are modelled. The material proper-
ties have been obtained from the manufacturers.
Heat losses to the environment are taken into
account in the simulation. A constant heat trans-
fer coefficient together with a constant environ-
ment temperature is used as a boundary
condition.
In APROS there are certain predefined process
components like tanks, pumps and valves. The
predefined steam generator model was not used in
this study, but a new model was constructed using
five-equation calculation level modules e.g. nodes
and branches. Basically it is the same model as
Fig. 4. Nodalization of the new steam generator for RELAP5/
MOD3.1.
151
was used with the RELAP5 code. Because
APROS does not have the same kind of time
dependent components as RELAP5, two control
valves had to be added, one to control the pri-
mary mass flow rate and another to control the
secondary feedwater mass flow rate. Otherwise,
the same input information was given for the
APROS code as for RELAP5.
During a steady state run the secondary side
collapsed level is controlled with a control system
in both codes. At the start of the transient this
system is excluded from the simulation. Another
control system is used to keep the secondary side
pressure constant during the whole calculation.
Primary side pressure, coolant mass flow rate
and hot leg temperature as well as secondary side
pressure and feedwater mass flow rate were given
as a boundary condition. The values were ob-
tained from the measurement data of each experi-
ment.
4.3. Results of the simulations
From the point of view of the safety analysis of
a nuclear reactor the interesting parameter is the
amount of heat transferred form the primary side
to the secondary side. The transferred power is a
function of the temperature difference between
the hot and cold collector. Because most of the
heat transfer occurs through the heat exchange
tubes, the ability of a code to calculate overall
heat transfer can be evaluated by observing the
inlet and outlet temperatures of the heat exchange
tubes.
Fig. 5 shows the measured primary tempera-
tures from one instrumented heat exchange tube
together with the corresponding temperatures cal-
culated with RELAP5/MOD3.1. The measure-
ment locations are 200 mm from the hot collector
and 500 mm from the cold collector, and the
calculation nodes are 279 mm from both collec-
tors. This causes a small difference between the
measured and calculated values. The oscillations
in the measured temperatures were not well repro-
duced because the boundary condition values for
the hot leg temperature are based on the averaged
measurement data. In Fig. 6 the same measured
temperatures and temperatures calculated with
152 E. Virtanen et al. / Nuclear Engineering and Design 177 (1997) 147–153
Fig. 5. Primary temperatures in one tube from experiment
SG-2 and from RELAP5 calculation (uncertainty in the mea-
sured value 93.0°C).
the APROS code are presented. The uncovering
of the tube can clearly be seen around 20 500 s in
the experiment and around 22 500 s in the RE-
LAP5 calculation. APROS calculates the uncover-
ing of the tube to occur a little later. The time
difference between experiment and calculation re-
sults is caused by the nodalization. In this case
two heat exchange tube rows are lumped together
and the temperature measurements are located in
the upper tube row.
In the experiment, steam superheating was ob-
served after the highest heat exchange tubes were
uncovered. In Fig. 7 the measured secondary side
temperatures in the highest tube layer together
Fig. 6. Primary temperatures in one tube from experiment
SG-2 and from APROS calculation (uncertainty in the mea-
sured value 93.0°C).
Fig. 7. Secondary temperatures in the highest tube layer from
experiment SG-4 and from RELAP5 calculation (uncertainty
in the measured value 9 3.0°C).
with the temperatures from the RELAP5/
MOD3.1 calculation are shown. In Fig. 8 the
same measured temperatures with temperatures
from two different APROS calculations are
shown. Both codes underestimate the steam su-
perheating. In APROS calculations it was ob-
served that the hydraulic diameter of the
secondary nodes has a significant influence on the
calculated steam temperature. If the hydraulic
diameter (Dh) is not given for the code in the
input deck the code calculates it based on the
given flow area of the node, Eq. (1).
Fig. 8. Secondary temperatures in the highest tube layer from
experiment SG-4 and from APROS calculations with imposed
(APROS – 1) hydraulic diameter and with Dh calculated by the
code (uncertanty in the measured value 93.0°C).
 '
E. Virtanen et al. / Nuclear Engineering and Design 177 (1997) 147–153
4Af
Dh =
p
4Af
Dh =
Pw
where Af is the flow area and Pw is the wetted
perimeter. In Fig. 8 APROS refers to the result
from the calculation where the code was allowed
to calculate the hydraulic diameter itself. In the
other calculation presented here (APROS – 1 in
Fig. 8) the hydraulic diameter was calculated us-
ing Eq. (2) and it was given to the code. The value
is the same which was used in the RELAP5
calculations. When calculated from Eq. (1) the
hydraulic diameter for a tube bundle area sec-
ondary side node is 1.002 m, and calculated from
Eq. (2) it is 0.05 m. As can be seen, the calculated
steam temperature using the hydraulic diameter of
0.05 m (APROS – 1) is nearer the measured values.
Even though the difference between measured and
calculated steam temperatures was around 15°C
in some cases the effect on the overall heat trans-
fer was negligible because the heat transfer to
single-phase steam is small compared with the
heat transfer to steam – water mixture.
In the experiment the temperature of steam in
the steam line was a little lower than in the
highest tube layer. Both codes calculated signifi-
cantly lower steam temperatures in the steam line.
The heat transfer mode, that RELAP5 uses on the
inner surface of the heat structure describing the
upper part of the steam generator shell, was
found to be condensation. In this mode the calcu-
lated heat transfer is more efficient than heat
transfer from single-phase vapor to the wall.
5. Conclusions
Three PACTEL loss-of-feedwater experiments
have been calculated using the APROS version
2.11 and RELAP5/MOD3.1 computer codes. The
main parameter to be compared was the tempera-
ture difference between hot and cold collectors,
e.g. the total power transferred from primary to
secondary side. In all the calculations the results
obtained were acceptable. Some differences were
153
(1) caused by the rather rapid changes in the experi-
mental boundary conditions, which were not com-
pletely reproduced. When the results were
(2)
compared more in detail, some differences were
observed. Especially the steam temperature above
the heat exchange tube bundle was in most of the
calculations lower than in the experiments. Even
though the difference between measured and cal-
culated steam temperatures is relatively large it
does not affect the overall heat transfer.
The assessment calculations presented here are
not sufficient to validate the codes used, but they
show that the APROS and RELAP5 codes are
able to model the overall behaviour of the hori-
zontal steam generators. Even with quite a simple
secondary side model, for example without recir-
culation, it is possible to simulate reasonably well
the heat transfer from the primary to the sec-
ondary side under loss-of-feedwater conditions.
However, some problems seem to exist in their
capabilities to model the phenomena on the sec-
ondary side. In order to validate or improve the
models new experiments are needed, because in
the experiments performed the boundary condi-
tions were too complex. Also some parameters
were not measured, like steam flow rate and heat
exchange tube surface temperatures, limiting the
usefulness of the experiments used for compari-
son.
References
[1] Carlson, K.E., Riemke, R.A., Rouhani, S.Z., Shumway,
R.W., Weawer, W.L., June 1990. RELAP5/MOD3 CODE
MANUAL, Volume 1: Code structure, System Models and
Solution Methods, E.G. and G. Idaho, Idaho Falls, ID,
USA.
[2] Barre, F., Bernard, M., 1990. The CATHARE code strat-
egy and assessment. Nucl. Eng. Des. 124, 257 – 284.
[3] Raussi, P., Munther, R., Kalli, H., Kouhia, J., Puustinen,
M., October 2 – 6, 1994. Experimental VVER Safety Stud-
ies in Finland. Trans. ENC’94, Int. Nucl. Congress, Lyon,
France.
[4] Porkholm, K., Hänninen, M., Puska, E.K., Ylijoki, J.,
November 15 – 20, 1992. APROS Code for the Analysis of
Nuclear Power Plant Thermal-Hydraulic transients. Eight
Proc. Nucl. Thermal-Hydraulics, ANS Winter Meeting,
Chicago, IL.