Hot duct break transient with two- and three-loop ALLEGRO models

Nuclear Engineering and Design 370 (2020) 110911

Contents lists available at ScienceDirect

Nuclear Engineering and Design
journal homepage: www.elsevier.com/locate/nucengdes

Hot duct break transient with two- and three-loop ALLEGRO models
´v Mayer
Guszta
Centre for Energy Research (EK), 1525 Budapest 114, P.O. Box 49, Budapest, Hungary

A R T I C L E I N F O

A B S T R A C T

Keywords:
Gas cooled fast reactor
Three-loop ALLEGRO
Hot duct break
CATHARE
Bypass transient

ALLEGRO is a concept of a small nuclear reactor with the primary aim to demonstrate the viability of the
Generation IV (GFR) Gas cooled Fast Reactor technology and to ensure the experimental and qualification
background for its new refractory fuel. In the GFR development the so-called bypass transients represent a
pivotal role during safety analysis. Since the hot duct is located inside the cold duct there is no loss of coolant
during the transient, nevertheless a huge core bypass may develop, which threatens the cladding integrity. The
current ALLEGRO design consists of two primary cooling loops but a three-loop version is also investigated. In
this paper the hot duct break transient is studied with a two- and three-loop ALLEGRO model by using the French
CATHARE thermal hydraulics code. The preliminary analysis showed that the three-loop model has a better

cooling performance.

1. Introduction
The development of ALLEGRO traces back to the beginning of the
2000′ s when the GEN IV International Forum (GIF) selected the Gas
cooled Fast Reactor (GFR) technology as a future development direction
aiming at the closing of the fuel cycle, ensuring proliferation resistance,
sustainability, reliability and high thermal efficiency (Stainsby et al.,
2011). The development started by the collaboration of European
research institutes, universities and companies.
As part of the European gas cooled fast reactor development, two
basic designs were developed parallel. Both of them have helium pri­
mary coolant at a pressure of 7.0 MPa, circulated by the primary
blowers. The first design is the GFR, which is the large-scale prototype of
the gas fast reactor technology with a thermal power of 2400 MW
(GFR2400). The second is the ALLEGRO, which is the demonstrator of
the GFR2400 and which has an envisaged thermal power of not higher
than 75 MW.
The CEA 2009 ALLEGRO design has two secondary circuits with
pressurized water coolant at a pressure of 6.5 MPa. The tertiary circuit of
ALLEGRO is an air cooler, which delivers the heat to the ambient air.
There is no electricity production envisaged with ALLEGRO. Since the
core outlet coolant temperature for these reactors is proposed to be at
about 800 ◦ C, a new refractory carbide fuel is aimed to be developed,
which withstands this high temperature and the fast neutron spectrum.
The primary aim of ALLEGRO is to qualify the new refractory carbide
fuel in fast neutron spectrum and in high temperature helium environ­
ment. This newly tested and qualified fuel will be used later in the GFR-

2400 reactors. Other important role of ALLEGRO as a demonstrator

reactor is to show the viability of the helium cooled technology for the
GFR-2400 reactors.
Significant part of the current nuclear developments focuses on SMRs
(small modular reactors) (Karol et al., 2015), which have the benefit of
being manufactured at a plant and bringing to a site to be assembled.
Due to its relatively low thermal power ALLEGRO may also serve as an
SMR. The modularity, the less on-site construction may significantly
decrease the costs, which makes the design favorable on the energy
market. Nevertheless, ALLEGRO being a fast spectrum reactor, the
closure of the fuel cycle would also be beneficiary on the long term
energy production.
The predecessor of ALLEGRO was called ETDR (Experimental
Technology Demonstration Reactor) (Morin et al., 2009). Later, the
concept was redesigned and it was renamed to ALLEGRO (Poette et al.,
2009b, 2009a). The new reactor concept, ALLEGRO featured a higher
thermal power of 75 MW and two coolant loops in comparison to the
earlier reactor concept of 50 MW and a single coolant loop. The concept
was investigated within several European and national projects.
Recently, the ALLEGRO reactor has been developed by a consortium
´, 2014), associating several
named V4G4 Centre of Excellence (Gado
research organizations and companies from Czech Republic, France,
´cha et al., 2019).
Hungary, Poland and Slovakia (Va
Since there is still no experimental reference for the new carbide fuel
in fast neutron spectrum, ALLEGRO cannot directly be started with this
fuel. For that reason, a three-step methodology is planned to be used. In
the first step, the ALLEGRO core will contain MOX or UOX fuel with

E-mail address:

/>Received 28 August 2020; Received in revised form 16 October 2020; Accepted 16 October 2020
Available online 5 November 2020
0029-5493/© 2020 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license ( />

G. Mayer

Nuclear Engineering and Design 370 (2020) 110911

stainless steel cladding for which sufficient experiences exist. In this first
core-configuration the outlet helium coolant temperature will be kept at
520 ◦ C for all the subassemblies because of the temperature limit of
stainless steel. In the second step, in some of the core positions, exper­
imental carbide subassemblies will be used at elevated outlet helium
coolant temperature of about 800 ◦ C. The remaining part of the core,
which consists of stainless steel cladding, will have lower, 520 ◦ C outlet
temperature. The different outlet temperatures will be reached by
adjusting some gagging at the inlet of each fuel sub-assembly. Finally, in
the third core-configuration the already tested carbide fuel will be
loaded into the core. Because of the good refractory properties of the
ceramic cladding the outlet temperature will be 800 ◦ C for all the
subassemblies.
The development of ALLEGRO is an iterative process. A study (Mayer
and Bentivoglio, 2015), performed for the CEA 2009 ALLEGRO MOX
core by using the CATHARE thermal hydraulics code, pointed out that
the two-loop version of ALLEGRO satisfies the defined peak cladding
temperature (PCT) values for loss of flow accident (LOFA), loss of off-site
power (LOSP), loss of heat sink (LOHS), loss of coolant accident (LOCA
up to 10 in.). A new nitrogen injection strategy was proposed in order to
handle 1 in. LOCA + blackout, 10 in. LOCA + guard vessel failure, total
hot and cold duct break. Nevertheless, another study (Mayer and Ben­

tivoglio, 2014) pointed out that the hot duct break transient aggravated
by the loss of the second loop may lead to higher PCT values than the
required limit in a conservative case. Investigating this topic further, the
idea of using three primary loops instead of two has arisen to improve
the design. The hypothesis is that in case of a three-loop design a single
failure has less effect on the PCT values, because there is one additional
loop - which is intact - for the heat removal. Accordingly, this study
focuses on the differences between the cooling capabilities of the twoand three-loop ALLEGRO models in case of hot duct break.
This study does not contain uncertainty analysis of the hot duct break
transient, because the primary goal is to show the difference between the
cooling capabilities of the two- and three-loop models. Of course, the
final safety analysis of ALLEGRO should definitely be supported by
uncertainty analysis.
The novelty of this paper is that it compares the cooling performance
of the two- and the three-loop ALLEGRO models for the hot duct break
initiating event, which is aggravated by the failure of one of the two
primary blowers. The newly developed three-loop ALLEGRO input deck
is available for future studies with wide range of scenarios like LOCA,
LOFA, etc. and it supports the designers to make decision in the two- or
three-loop question.

speed at 20% of its nominal value. The DHR loops are mostly used in
design extension conditions (DEC), when the cooling cannot be main­
tained by using the primary loops.
The secondary circuits contain pressurized water at 6.5 MPa. There is
5 bar pressure difference between the primary and the secondary cir­
cuits in order to avoid water ingress into the core at the beginning of a
primary system to secondary system leakage accident (PRISE). The
water coolant is circulated by the secondary pumps. Since the electricity
production is not envisaged in ALLEGRO, the heat is removed from the

secondary circuit by an air cooler, the ultimate heat sink is the ambient
air.
When the heat removal from the core by using the two primary loops
is not sufficient, the primary valves close, the DHR valves open and the
DHR blowers start parallel. The secondary circuit of the DHR loops
contains pressurized water at 1.0 MPa and there is no active element in
it. Since the flow is driven by natural circulation, the heat exchangers are
located highly above the core. The tertiary circuit of the DHR loop is a
pool, which ensures cooling capacity for 24 h by the vaporization of its
water content. It is sized to dissipate the decay heat by purely natural
convection, if the system is pressurized. Since the hot duct break tran­
sient belongs to the Category 4 transients (see later at fuel acceptance
criteria), the DHR loops do not play any role in this study. The heat is
removed by the use of the main heat exchangers and the main blowers
are driven by their pony (auxiliary) motors.
The three-loop model has the same structure (Fig. 2), the only dif­
ference is that it has a third primary cooling loop. The power of the core
is the same in both the two- and three-loop models. One important
advantage of the three-loop model against the two-loop model is that it
is more resistant to single failure. For instance, in case of a hot duct
break in the first loop and a blower failure in the second loop, there is
still a third cooling loop available in the three-loop model. In the twoloop model only the first blower is supposed to run, even though it is
located in the broken loop. In case of the three-loop model both the first
blower in the broken loop and the third blower in the intact loop are
supposed to operate. According to this, in case of the three-loop model
only the 1/3 of the total blower capacity is lost instead of the ½.
3. Representation of hot duct guillotine break
In the current ALLEGRO design the hot duct (hot leg) is located in­
side the cold duct (cold leg) as it can be seen in Fig. 3a. In this concentric
pipe arrangement a thermal insulation is necessary between the hot and

the cold ducts (Fig. 3b) to decrease the heat exchange. In this arrange­
ment three kinds of breaks can be imagined. In the first only the hot duct
and its insulation are broken and the cold duct remains intact. In the
second case the cold duct is broken and the hot duct remains intact. In
the third both the hot and cold ducts are broken. In the first case there is
no depressurization of the system, because the break is inside the cold
duct, which is the outer boundary of the primary circuit. Nevertheless,
there is a huge core bypass (see Fig. 3a). In the second case, when the
cold duct is broken, there is a LOCA transient in which there is a
depressurization of the primary circuit. In order to keep necessary
backup pressure for this scenario a so-called guard vessel is used, which
encompasses the whole primary circuit and can maintain maximum

2. The CEA 2009 ALLEGRO design
Fig. 1 shows the main cooling loops of CEA 2009 ALLEGRO design. It
consists of two primary and three decay heat removal (DHR) loops. The
two main loops are used parallel and the DHR loops are closed in normal
operation. The core is cooled by helium at pressure of 7 MPa. The
blowers are driven by the main motors with electrical power of 418 kW
and a smaller power pony motor is mounted to the same shaft in order to
ensure cooling in case of scram. The successful scram signal switches off
the main motors and starts the pony motors to maintain the blower

Fig. 1. The cooling loops of CEA 2009 ALLEGRO design.
2


G. Mayer

Nuclear Engineering and Design 370 (2020) 110911

Fig. 2. The cooling loops of the three-loop ALLEGRO design.

Fig. 4. Representation of the hot duct guillotine break of ALLEGRO. There is a
gap between the two hot duct parts, which causes a core bypass.
Fig. 3a. Representation of ALLEGRO hot duct break. For the sake of simplicity
only the broken loop is depicted. Dashed arrows show decreased coolant mass
flow rate through the core due to the huge core bypass. There is no LOCA, since
the outer boundary of the primary circuit remains intact.

pipe. In this study the hot duct break size was selected between 20%
(~127 mm = 5 in.) and 100% (600 mm = 23.6 in.) of the inner hot duct
diameter without supposing any other objects (debris of insulation,
liner, etc.) in the flow path.
In case of hot duct break, the pressure loss between the hot and the
cold ducts plays a major role during the transient. The lower is the
pressure loss coefficient the higher core bypass may develop. In a pre­
vious study this value was selected to be 0.549 (Mayer and Bentivoglio,
2014). In a recent paper by using the computational fluid dynamics
(CFD) technique (Farkas et al., 2019) the pressure loss coefficient was
found to be around the value of 1.0. Keeping in mind that the CFD codes
use turbulence models, which may bring some uncertainty into the
calculations, and to stay conservative, the smaller value of 0.549 was
selected for all of these calculations in the present study.
4. The CATHARE code
The French proprietary CATHARE 2 V2.5_3 code with its graphical
user interface GUITHARE is developed by CEA, EDF, FRAMATOME and
IRSN. Originally it was elaborated for thermal hydraulic simulation of
pressurized and boiling water reactors, but later its validity was
extended for gas cooled reactors. The code was validated against exist­

ing system loops with gas coolant (Bentivoglio and Tauveron, 2006,
2008; Bentivoglio and Messi´e, 2008; Polidori et al., 2013) and was used
for transient analysis of GFR2400MW (Mayer and Bentivoglio, 2015),
and code-to-code benchmarks of ALLEGRO (Bubelis et al., 2008; Kvizda
et al., 2019). Recently, the application of the code was extended for
gases where the ideal gas equation of state is not valid (Mauger et al.,
2019).

Fig. 3b. The cross section of ALLEGRO ducts. The hot duct is located inside the
cold duct.

10–13 bar backup pressure. In the third case there is a LOCA transient
and a core bypass transient at the same time. This research focuses on
the first case, in which the hot duct is broken and the cold duct remains
intact and there is no depressurization of the system and no LOCA.
Until the cold duct remains intact, the classical 200% size break –
which is usually supposed in most of the pressurized water reactor
(PWR) analyses – can geometrically be excluded because of the lack of
sufficient space in the cold duct. The two new hot duct parts (as a result
of the guillotine break) have a large diameter each and they have no
sufficient place to move in perpendicular directions generating two fully
opened pipes. On the other hand, the shrinking of the hot duct in axial
direction is possible due to the fast and high temperature change of the
pipe wall. The axial displacement of the hot duct (Fig. 4) is also possible
due to the flexible supports in the heat exchangers, which are designed
to ease the thermomechanical stresses caused by the elongation of the

5. Hot duct break model
The simulation of the ALLEGRO hot duct break with a onedimensional code like CATHARE is limited. Since the hot duct is in­
side the cold duct, in case of hot duct break the hot and the cold ducts are

connected to with a simple pipe element. There is a valve in this pipe,
which is closed during normal operation. When the break is initialized
this valve is opened. Fig. 5 shows the bypass which is formed due to the
opening of the valve.
3


G. Mayer

Nuclear Engineering and Design 370 (2020) 110911

Fig. 5. Representation of the hot duct break modeling in the two-loop ALLEGRO model. The red arrow represents the break flow. After the closure of the second
valve (Main valve 2) the second loop has no effect. The three DHR loops are closed during the transient. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)

side was also decreased by the same factor in order to get similar ve­
locity distribution in the air cooler.

6. The new three-loop ALLEGRO input deck
In the European VINCO project, a code-to-code benchmark was
performed by the members of the V4G4 consortium (Kvizda et al., 2019)
using the CATHARE2, RELAP5-3D and MELCOR codes. In that work
each participant developed an own input deck, which were based on an
earlier CEA input and on a new ALLEGRO database and benchmark
specification. In this work the VINCO input deck is used for the two-loop
modeling. On the other hand, for the three-loop calculations the
development of a new ALLEGRO input was necessary. In this study,
using the two-loop VINCO ALLEGRO CATHARE input deck, a new threeloop model was developed to simulate the hot duct break scenario. In the
followings the modifications between the two models are described.

6.4. DHR blowers
The DHR loops and blowers do not play any role in the hot duct break
transient, since the DHR valves are closed during the whole procedure.
For this reason, there were no modifications in the new input deck
regarding the DHR loops.
6.5. Main blowers
The main primary blower characteristic is described with a nondi­
mensional head and torque map in the original VINCO model. In this
study the same map is used for the two-loop calculations, but the
reference values of the blower were modified. Since the total mass flow
rate of the three-loop ALLEGRO core remained unchanged, its nominal
value in each loop was set to the 2/3 of the two-loop version. This
modification influenced the nominal torque of the blowers. The nominal
head of the blowers remained unchanged.

6.1. Core
Since this study focuses primarily on the thermal hydraulic processes
in ALLEGRO, the reactor core and its neutronic parameters were kept
identical in both the two- and three-loop models. In this way, it is easy to
compare the cooling capabilities of the two models.
In the VINCO benchmark (Kvizda et al., 2019) a new thermal hy­
draulics core model was used, in which the thermal inertias of the
reflector and shielding materials were modeled. As a result, the calcu­
lated maximum cladding temperatures were significantly lower
compared to the case when the thermal inertia of the inner elements
were neglected. Nevertheless, those calculations may contain large un­
certainties concerning the heat exchange coefficients, because the lack
of the current experimental support. In order to stay conservative, in this
paper the inner coupling between the heat structures in the core (as it
was modeled in the VINCO project) are neglected for both the two- and

three-loop models.

6.6. Piping
The length and diameter of the primary hot and cold ducts were not
modified in the three-loop ALLEGRO model. The pressure loss co­
efficients along the loops were identical in the two- and the three-loop
models.
7. Nodalization scheme
The input deck developed in the earlier projects was the starting
point of this study. Its simplified nodalization scheme is shown in Fig. 6.
On the left hand side there is the core. The upper and the lower plenums
are represented by CATHARE VOLUME elements. There are four chan­
nels that represent the inner part of the core: two bypasses, one average
and one hot channel. They are modeled by using the so-called AXIAL
elements. The reflector and shielding is also taken into account at the
bottom and at the upper part of these channels. Between the hot and the
cold ducts there is an AXIAL element which connects them. This is the
representation of the internal break used in the simulations. For the sake
of simplicity, only one secondary and one tertiary circuit is shown on the
right hand side of Fig. 6. The main heat exchanger (MHX) is represented
by using two axial elements, one on the primary and a second in the
secondary circuit. The pressure of the secondary circuit is maintained by
the pressurizer. The tertiary circuit consists of only one axial element
and two boundary conditions. The three DHR loops, the guard vessel,
the nitrogen injection lines are not depicted since they do not play any
role in the hot duct break scenario.

6.2. Main heat exchangers
The CEA 2009 ALLEGRO model contains two main heat exchangers
(MHX), which remove 37.5 MW thermal power individually. In order to

remove the same thermal power with three loops (25 MW/MHX), the
following modifications were carried out. The heating perimeter of the
heat exchangers and the flow areas were decreased to the 2/3 of the twoloop model on both the primary and secondary side of the MHXs. This
ensures similar velocity distribution in the two- and three-loop models in
the MHXs.
6.3. Air cooler
In the three-loop model, similarly to the MHXs, the air cooler was
modeled by decreasing the number of heat exchanger pipes to the 2/3 of
the original number of the two-loop model. This was achieved by
modifying the heat exchange perimeter. The flow area at the tertiary
4


G. Mayer

Nuclear Engineering and Design 370 (2020) 110911

Fig. 6. Representation of the ALLEGRO CATHARE nodalization scheme. The internal break is modeled by using an axial element, which connects the hot and cold
the ducts. Only one primary loop is depicted. The three DHR loops are not presented.

8. Acceptance criteria for fuel

“power to mass flow rate” signal activates the scram, which stops the
reactor and results in the decreasing of core power according to Fig. 7.
The sequence of events of the investigated transients can be found in
Table 1.
The scram signal lets the absorbers fall into the core and at the same
time it also switches the main blowers off. This action is needed to avoid
the fast overcooling of the fuel and its cladding in most of the scram
cases, because of the low thermal inertia of the ALLEGRO core. As the

main blower rotational speed in the first loop (Fig. 8) decreases to 20%
of its nominal value, a pony motor – which is mounted to the same shaft is activated and it maintains the 20% rotational speed of the main
blower during the whole transient.
Four hot duct break transients were investigated with a break size of
23.6 in., which is the inner diameter of the hot duct. In the first case, the
two-loop input deck of ALLEGRO with a blower inertia of 10 kg*m*m
was used. The remaining three simulations were carried out by using the
new three-loop model with blower inertias of 6.7, 10 and 20 kg*m*m.
The 6.7 kg*m*m value in the three-loop model corresponds to the same

Acceptance criteria for the ALLEGRO MOX cladding can be found in
(Kvizda et al., 2019). In case of design extension condition (DEC) the
proposed limit is 1300 ◦ C, which is the melting temperature of the
cladding. If the PCT remains below this limit the cladding integrity is
ensured. For categories 3 and 4 the 850 ◦ C temperature maximum is
proposed. On the one hand, it is the limit of cladding burst at high (100
bar or higher) pressure difference. On the other hand, it is the maximum
temperature of strong pellet cladding mechanical interaction (PCMI)
with the irradiated, brittle cladding. For categories 1 and 2 the limit of
620 ◦ C for the cladding temperature is proposed. Since the break of the
total hot duct belongs to the category 4 transients, the corresponding
temperature limit for the fuel cladding is 850 ◦ C.
9. Results
9.1. The role of blower inertia
The hot duct break transient starts with the opening of the break
between the hot and the cold ducts in the first loop. This is accomplished
by the opening of the valve between the hot and the cold ducts as it was
described previously. At the same time - as a single failure criterion - the
blower and its pony motor in the second loop are supposed to be out of
operation. Since the opening of the break causes a large core bypass, the

Table 1
Sequence of events and maximum peak cladding temperatures.

Fig. 7. Reactor power on logarithmic scale.
5

Simulation id.
number

1

2

3

4

unit

Number of primary
loops
Blower inertia
Break size
Break size
Moment of scram
Stop of 1st main
motor
Stop of 2nd main
motor

Stop of 3rd main
motor
Start of 1st pony
Start of 2nd pony
(single failure)
Start of 3rd pony
Closure of main
valve 2
Partial closure of
main valve 2
Peak cladding
temperature

3

3

3

2

6.7
600
23.6
3.4E− 04
6.8E− 04

10
600
23.6

3.4E− 04
6.8E− 04

20
600
23.6
3.4E− 04
6.8E− 04

10
600
23.6
6.9E− 04
1.4E− 03

kg*m*m
mm
inch
s
s

6.8E− 04

6.8E− 04

6.8E− 04

1.4E− 03

s

6.8E− 04

6.8E− 04

6.8E− 04

n.a.

s

11.4
no

16.9
no

33.5
no

11.4
no

s
s

12.1
41.6

18.0

62.7

35.5
126.5

n.a.
52.6

s
s

15.4

22.9

44.5

15.4

s

955.0

921.0

825.0

986.0

Celsius

G. Mayer

Nuclear Engineering and Design 370 (2020) 110911

Fig. 10. Rotational speed of the main blower in the third loop with different
blower inertias.

Fig. 8. Rotational speed of the main blower in the first loop for
different models.

to the two other blowers becomes lower than 80%. This is depicted in
Fig. 11. When the second blower rotational speed reaches the 5% of its
nominal value, the valve in the second loop closes fully (stem position is
0%). It can be seen in Fig. 11 that similarly to the previous examples the
blower inertia influences the partial and the full closure time instants of
the second valve. The larger blower inertia results in more delayed valve
closure actuation. If the total inertias of the blowers are equal, the time
instances of the actuation in the two- and three-loop models are very
close to each other. There are no primary valve actuations in the first and
in the third loop during the transient, the stem positions are kept con­
stant. Of course, in steady state conditions the stem positions are
controlled values to ensure the desired mass flow rates in the loops.
Fig. 12 depicts the core mass flow rate, which is mainly influenced by
the pressure rise of the blowers and the pressure losses throughout the
system including the hot duct break. (The core pressure loss in nominal
conditions is roughly 0.8 bar and in the heat exchanger 0.2.) It can be
seen that at the first 100 s of the transient the core mass flow rates are
very similar in the two- and three-loop models - when the total blower

inertias are the same - but after that point the mass flow rate in the twoloop model becomes higher. Other outcome for the three-loop model is
that the core mass flow rates are higher for the larger blower inertias
until the startup of ponies. Later, the mass flow rates become very
similar in the three-loop calculations.
Fig. 13 shows that the steady mass flow rate through the break be­
tween the cold and the hot ducts is one order higher compared to the
core mass flow rate (Fig. 12). During the first minute of the transient the
blower inertia has a large effect on the break mass flow rate. It is mainly
driven by the rotational speed of the blowers. Obviously, the blowers
with higher inertia generate higher break mass flow rates. After reaching
the 20% pony speed, the break mass flow rates of the different models
become very similar. In the two-loop model the final break mass flow
rate is slightly smaller than in the three-loop model.

total primary rotational inertia of the two-loop model (6.667*3 = 10*2).
It can be seen in Fig. 8 that if the blower inertias are 10 kg*m*m in both
models, then the first blower reaches the 20% rotational speed about 5 s
earlier in the two-loop model than in the three-loop model. If the 6.67
kg*m*m value for the blower inertia is set for the three-loop model then
the rotational speeds of the 1st blower are almost identical in the twoand in the three-loop model. It is not surprising because the total inertia
of the blowers in the primary circuit is the same for both models in this
case. When the 20 kg*m*m blower inertia is selected for the three-loop
model (total of 60 kg*m*m for the three blowers), the rotational speed of
the first blower reaches the 20% of its nominal value significantly later,
at about 33 s. It will soon be clear, that this ensures better cooling
performance and lower peak cladding temperature at the beginning of
the transient.
The pony in the second loop is supposed not to operate because of the
single failure criterion, and as a result of this the main blower finally
stops. Fig. 9 shows the second blower rotational speed for all of the

previously mentioned four cases. It can be seen that as the blower inertia
is higher, the time instance of the stop of the blower is also higher. It
should be noted that the value of the blower friction was the same in all
models.
In the third loop the blower rotational speed decreases up to the 20%
of its nominal value, where the pony is activated which maintains the
flow during the rest of the transient (Fig. 10). Of course, in the two-loop
model there is no third loop and there is no third blower.
Shortly after the pony motors in the first and third loops are activated
and maintain the 20% rotational speed of the nominal value of the main
blowers, the mass flow rate in the second loop becomes negative, which
may damage the second blower. In order to avoid this issue, the blower
protection signal partially closes (stem position of the valve is 5%) the
main valve in the second loop when the rotational speed ratio compared

Fig. 9. Rotational speed of the main blower in the second loop for
different models.

Fig. 11. Valve positions in the second loop.
6


G. Mayer

Nuclear Engineering and Design 370 (2020) 110911

classical 200% hot duct break is not possible because of geometrical
reasons, without the break of the cold duct (total cross duct – break of
both the hot and the cold ducts). Nevertheless, in case of a small break at
the hot duct, the steel material of the pipe in the vicinity of the break

may dramatically be overcooled in a relatively short time, which may
result in high thermal stresses and increased crack propagation. In the
roughly 10 m long DHR ducts the 100% break size is theoretically
imaginable, if conservative considerations are taken into account by
supposing the shrinking of the hot duct caused by the fast and sudden
temperature drop.
In Figs. 15 and 16 the maximum cladding temperatures can be seen
for different break sizes for the two- and three-loop models, respectively.
Considering smaller break sizes, the peak cladding temperature is
decreasing. In order to show more comparative results between the twoand the three-loop ALLEGRO versions, we suppose that the total blower
inertia is the same for both models.
According to Fig. 15 the PCTs fall below the 850 ◦ C criterion in case
of the 15 in. or smaller break sizes. When the three-loop model is used
(Fig. 16) the PCT criterion fulfils for break sizes up to 20 in., even if the
decreased 20/3 ≈ 6.7 kg*m*m inertia is used for each blower.

Fig. 12. Core mass flow rate.

9.3. Two loops versus three loops
Selecting the 23.6 in. break size and comparing the two- and threeloop models with the 10 kg*m*m and the 6.6667 kg*m*m inertia,
respectively, the whole picture may seem contradictory, because the
PCT in the three-loop model is lower by 31 ◦ C compared to the two-loop
model, despite the fact that the results show very similar core mass flow
rates for both models during the first two minutes of the transients. It can
be explained by Fig. 17, which shows the lower plenum temperatures.
Since the coolant flows from the lower plenum to the core, its temper­
ature influences the cladding temperature. It can be seen that there is a
lower inlet core temperature in case of the three-loop model than in the
two-loop model. The difference is 31 ◦ C. This explains the better per­
formance of the three-loop model. It should be noted that in the threeloop model the third primary circuit is operating during the transient,

because the single failure criterion was used only for the second blower.
The upper plenum temperature (Fig. 18) reveals more difference
between the models in case of 23.6-inch hot duct break. In the three-loop
model the upper plenum temperature is significantly lower than in the
two-loop model. Closer analysis of the data showed that in case of the
three-loop model the operating third blower changes the main flow di­
rections according to Fig. 19. A relevant portion of the third loop coolant
does not flow into the core but into the break in the first loop. It changes
the flow direction towards the upper plenum in the first loop. As a result
of that the cold helium coolant flows directly to the upper plenum,
which causes low temperature. The mass flow rates from the upper
plenum towards the first loop hot duct inlet can be seen in Fig. 20. The

Fig. 13. Break mass flow rate.

The peak cladding temperature has a limit of 850 ◦ C for the ALLE­
GRO MOX core in Category 4 transients. Fig. 14 shows the time evolu­
tion of maximum cladding temperature. The break size is 23.6 in. for
each simulation. It can be seen that the peak cladding temperature for
the two-loop model is the highest amongst all cases, even if the 6.7
kg*m*m blower inertia is used in the three-loop model. The temperature
difference is 31 ◦ C. It can also be seen that if the blower inertia in the
three-loop model is 20, then the peak cladding temperature is below the
850 ◦ C limit, even if the break size is the largest.
9.2. The effect of break size
In the previous example a guillotine break was envisaged at the hot
duct near the reactor downcomer with the conservative assumption of a
24.6 in. (100%) break size. In ALLEGRO cross duct geometry the

Fig. 15. Maximum cladding temperatures for the two-loop model for different

break sizes. The inertia of both blowers is 10 kg*m*m.

Fig. 14. Maximum cladding temperature.
7


G. Mayer

Nuclear Engineering and Design 370 (2020) 110911

figure shows that as the primary valves in the second loops are closed,
the mass flow rate in the first loop of the two-loop model becomes
positive, opposed to the three-loop model in which it stays negative. The
cold helium flows to the upper plenum directly. This explains the low
upper plenum temperature in the three-loop model in Fig. 18.

scenario. The break was initiated at the hot duct of the first loop close to
the reactor vessel. As a single failure criterion the blower was immedi­
ately stopped in the second loop and its main primary valve was sup­
posed to close after its rundown. Opposed to other reactor types this
transient does not lead to the depressurization of the system, since the
hot duct is located inside the cold duct, nevertheless a huge core bypass
develops.
The core model was the same in both the two- and three-loop ver­
sions. In the three-loop model the heat exchange surfaces and loop mass
flow rates were decreased to the 2/3 of the two-loop model. The total
blower inertia was kept constant in one of the three-loop examples for
easy comparison.
The results showed that the peak cladding temperature was lower by
31 ◦ C in case of the three-loop model than in the two-loop model, if the

total blower inertias (20 kg*m*m for all the systems, i.e. 2*10 for the
two-loop system and 3*6.667 for the three-loop system) were identical.
The main reason of this is that the lower plenum temperature is also
lower in the three-loop simulation. The calculations showed larger,
65 ◦ C difference in PCTs between the two- and three-loop models, when
the inertia of each blower was the same (10 kg*m*m for each).
The increased blower inertia led to significantly lower PCT values.
From cooling point of view, increasing the blower inertia seems a
promising tool in case of hot duct break. Nevertheless, the fast over­
cooling of the cladding should be avoided in case of other initiating
events, when the scram is actuated and the core cooling is good, for
instance in loss of flow accident (LOFA). This suggests that the blower
inertia should have a carefully selected trade-off value in ALLEGRO.
The effect of the hot duct break size was studied by varying the
diameter of the pipe, which represents the hot duct break. This pipe
connects the hot and the cold ducts in this break model. The results
showed that the break size plays a major role in the final peak cladding
temperatures. For this reason, it is important to elaborate an ALLEGRO
design in which the hot duct break with large break sizes is practically
eliminated or at least the probability of the accident is low.
To sum up the results, three main conclusions for the ALLEGRO
developers can be drawn. Firstly, the gain coming from the additional
(third) loop is not very significant compared to the two-loop version in
case of hot duct break transient. It is not necessarily worth the extra cost
which occurs due to the construction of the third primary loop. Never­
theless, it has to be emphasized that in these calculations the hot and the
cold duct diameters were the same for both the two- and three-loop
models. For this reason, further studies would be beneficial in which
the diameter of the three-loop hot and cold duct is proportionally
decreased. This decreased diameter may increase the benefit of the

three-loop model against the two-loop model in the sense of PCT. In
addition, further future studies are necessary to assess the effect of
bypass transients at different break locations of ALLEGRO. Such a study
may reveal larger gain in PCT between the two- and three-loop models.
Additionally, in this study the break was supposed not to have any debris
in the cooling path, which may significantly influence the cooling
capability of the loops. Since the debris in the cooling path increases the
risk of blower failure, the three-loop model may have more benefit.
Secondly, increasing the blower rotational inertia, instead of using
three loops, may help a lot in the sense of peak cladding temperature,
which design modification may result in a much lower building cost.
Thirdly, the relevance of the three cooling loops may be more sig­
nificant for other initiating events such as total cross duct break which
needs to be further investigated. The question of using two or three
primary loops in ALLEGRO is still under consideration and it should be
supported by probabilistic safety assessment (PSA) studies too.

10. Conclusion

CRediT authorship contribution statement

The difference between the cooling performances of the two- and
three-loop ALLEGRO models were investigated in case of hot duct break

´v Mayer: Conceptualization, Methodology, Software, Vali­
Guszta
dation, Formal analysis, Investigation, Data curation, Writing - original
draft, Visualization, Project administration, Funding acquisition.

Fig. 16. Maximum cladding temperatures for the three-loop model for different

break sizes. The inertia of each blower is 6.67 kg*m*m.

Fig. 17. Lower plenum temperatures.

Fig. 18. Upper plenum temperatures.

8


G. Mayer

Nuclear Engineering and Design 370 (2020) 110911

Fig. 19. Representation of the hot duct break modeling in the three-loop ALLEGRO model after the closure of the second loop valve. The red arrows represent the
break flow and the direction changes compared to the two-loop model. (For interpretation of the references to colour in this figure legend, the reader is referred to the
web version of this article.)
Bubelis, E., et al., 2008. A GFR benchmark: Comparison of transient analysis codes based
on the ETDR concept. Prog. Nucl. Energy. />pnucene.2007.11.090.
Farkas, I., et al., 2019. Determination of pressure loss coefficient of the hot duct break of
the CEA ALLEGRO 2009 concept using a CFD technique. May. In: Nuclear
Engineering and Design. Elsevier, pp. 234–242. />nucengdes.2019.05.016.
Gad´
o, J., 2014. The reactor ALLEGRO and the sustainable nuclear energy in Central
Europe. In: EPJ Web of Conferences. />20147808001.
Karol, M., John, T., Zhao, J., 2015. Small and Medium sized Reactors (SMR): a review of
technology. Renew. Sustain. Energy Rev. Elsevier 44, 643–656. />10.1016/j.rser.2015.01.006.
Kvizda, B., et al., 2019. ALLEGRO Gas-cooled Fast Reactor (GFR) demonstrator thermal
hydraulic benchmark. Nucl. Eng. Des. 345 />nucengdes.2019.02.006.
Mauger, G., et al., 2019. On the dynamic modeling of Brayton cycle power conversion
systems with the CATHARE-3 code. Energy. />energy.2018.11.063.

Mayer, G., Bentivoglio, F., 2014. Transient analysis of crossduct break scenarios using
the CATHARE2 Code for the 75MW ALLEGRO demonstrator. Int. Conf. Nucl. Eng.,
Proc., ICONE 3. />Mayer, G., Bentivoglio, F., 2015. Preliminary study of the decay heat removal strategy
for the gas demonstrator ALLEGRO. Nucl. Eng. Des. 286 />nucengdes.2015.02.001.
Morin, F., et al., 2009. Status of the ETDR preconceptual design studies. Conference:
SMIRT-18, Beijing, China.
Poette, C., Brun-Magaud, V., et al., 2009a. ALLEGRO: The European Gas Fast Reactor
demonstrator project. In: International Conference on Nuclear Engineering. https://
doi.org/10.1115/ICONE17-75326.
Poette, C., Malo, J., et al., 2009b. GFR demonstrator ALLEGRO design status. In:
International Congress on Advances in Nuclear Power Plants 2009, ICAPP 2009,
pp. 323–332.
Polidori, M., et al., 2013. Thermal-hydraulic codes benchmark for gas-cooled fast reactor
systems based on hefus3 experimental data. International Congress on Advances in
Nuclear Power Plants, ICAPP 2013: Nuclear Power - A Safe and Sustainable Choice
for Green Future, Held with the 28th KAIF/KNS Annual Conference.
Stainsby, R., et al., 2011. Gas cooled fast reactor research in Europe. Nucl. Eng. Des.
Elsevier B.V. 241 (9), 3481–3489. />nucengdes.2011.08.005.
V´
acha, P., et al., 2019. 000379 progress in the ALLEGRO project - neutronics and
thermal-hydraulics. ICAPP 2019 – International Congress on Advances in Nuclear
Power Plants France, Juan-les-pins – 2019, May 12-15.

Fig. 20. The evolution of mass flow rates coming from the upper plenum to the
inlet of the hot ducts in the first (broken) loop.

Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.

References
Bentivoglio, F. and Messi´
e, A., 2008. ‘Cathare simulation of a depressurization transient
for the 2400MW Gas Fast Reactor concept’, in: Societe Francaise d’Energie Nucleaire
- International Congress on Advances in Nuclear Power Plants - ICAPP 2007, ‘The
Nuclear Renaissance at Work’.
Bentivoglio, F., Tauveron, N., 2006. Validation of CATHARE code for gas-cooled
reactors: comparison with E.V.O experimental data on Oberhausen II facility.
Proceedings of the 2006 International Congress on Advances in Nuclear Power
Plants, ICAPP’06.
Bentivoglio, F., Tauveron, N., 2008. Validation of the CATHARE2 code against
Oberhausen II data. Nucl. Technol. />
9