Nuclear Engineering and Design 287 (2015) 90–94

Contents lists available at ScienceDirect

Nuclear Engineering and Design

journal homepage: www.elsevier.com/locate/nucengdes

A new method to evaluate the sealing reliability of the flanged

connections for Molten Salt Reactors
Qiming Li a,b,∗ , Jian Tian a,b , Chong Zhou a,b , Naxiu Wang a,b,∗∗
a
Shanghai Institute of Applied Physics, Chinese Academy of Sciences, Shanghai 201800, China
b
Key Laboratory of Nuclear Radiation and Nuclear Energy Technology, Chinese Academy of Sciences, Shanghai 201800, China

h i g h l i g h t s

• We novelly valuate the sealing reliability of the flanged connections for MSRs.
• We focus on the passive decrease of the leak impetus in flanged connections.
• The modified flanged connections are acquired a sealing ability of self-adjustment.
• Effects of redesigned flange configurations on molten salt leakage are discussed.

a r t i c l e i n f o a b s t r a c t

Article history: The Thorium based Molten Salt Reactor (TMSR) project is a future Generation IV nuclear reactor system
Received 26 August 2014 proposed by the Chinese Academy of Sciences with the strategic goal of meeting the growing energy
Received in revised form 19 February 2015 needs in the Chinese economic development and social progress. It is based on liquid salts served as both
Accepted 2 March 2015
fuel and primary coolant and consequently great challenges are brought into the sealing of the flanged
connections. In this study, an improved prototype flange assembly is performed on the strength of the
Freeze-Flange initially developed by Oak Ridge National Laboratory (ORNL). The calculation results of the
finite element model established to analyze the temperature profile of the Freeze-Flange agree well with
the experimental data, which indicates that the numerical simulation method is credible. For further
consideration, the ideal-gas thermodynamic model, together with the mathematical approximation, is
novelly borrowed to theoretically evaluate the sealing performance of the modified Freeze-Flange and the
traditional double gaskets bolted flange joint. This study focuses on the passive decrease of the leak driving
force due to multiple gaskets introduced in flanged connections for MSR. The effects of the redesigned
flange configuration on molten salt leakage resistance are discussed in detail.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction (TMSR) development project was started by the Shanghai Insti-

The Molten Salt Reactor (MSR) was selected by the Generation
IV International Forum (GIF) as one of the six advanced nuclear
reactors concepts for its favorable feature in inherent safety, econ-
omy, fuel cycle, and a lower pressure primary loop (GIF-IV, 2002;
MacPherson, 1985; LeBlanc, 2010). In the recent decades, there
has been a growing interest in MSRs. A Thorium based MSR
∗ Corresponding author at: Shanghai Institute of Applied Physics, Chinese
Academy of Sciences, Shanghai 201800, China. Tel.: +86 21 39194199;
fax: +86 21 39194199.
∗∗ Corresponding author at: Shanghai Institute of Applied Physics, Chinese
Academy of Sciences, Shanghai 201800, China. Tel.: +86 21 39194100;
fax: +86 21 39194100.
E-mail addresses: liqiming@sinap.ac.cn (Q. Li), wangnaxiu@sinap.ac.cn
(N. Wang).
tute of Applied Physics, Chinese Academy of Sciences (SINAP) in
2011. Several research and development (R&D) projects on MSRs’
technologies were conducted by some countries and institutes
benefiting from the past experience of molten salt technologies
acquired at Oak Ridge National Laboratory (ORNL) (Guo et al., 2013).
The MSRs use liquid salts as both fuel and primary coolant, there-
fore the technology is essentially different from the traditional solid
fueled reactors. It has brought great challenges into the sealing of
flanged connections which is very common in the pressure vessels
and piping systems of nuclear reactors. Prevention against liquid
salt leakage is the primary function for flange assemblies in MSRs,
as the sealing performance is directly related to the overall system
safety, hence the necessity of study in depth.
There are two factors causing the leakage of flange assemblies’
face seal: The one, necessary condition for leakage occurring, i.e.
tiny holes or cracks in the seal face caused by manufacturing or

http://dx.doi.org/10.1016/j.nucengdes.2015.03.003
0029-5493/© 2015 Elsevier B.V. All rights reserved.
 Q. Li et al. / Nuclear Engineering and Design 287 (2015) 90–94
Fig. 1. Cross section of (a) the Freeze-Flange and clamp in MSRE and (b) the modified one, where the clamp is not marked out.
construction error etc. The other, the leaking driven force, i.e. the
differential pressure between the two sides of the leak proof seal. In
general, the fluid leakage may be stopped or suppressed by means
of eliminating or reducing either factors above (Fu and Zhang,
2011). Since the seal structure is one of the key factors affect-
ing the sealing behavior, many researchers have studied the seal
structural optimization using finite element analysis (FEA). Izumi
et al. (2009) investigated the double-nut tightening method and
the spring washer to resist self-loosening with three dimensional
FEA, while Choi and Seo (2011) conducted the shape optimiza-
tion of a torus seal under multiple loading conditions with the
linearized stress constrains on the basis of the stress categories in
the AMSE code section III. The failure mode of mechanical seals
and the various cause events to help in taking appropriate steps to
improve sealing reliability are discussed by Singh et al. (2012). All
these studies have basically regarded the traditional bolted flange
joints. Nevertheless, mathematical design techniques concerning
decreasing leak momentum have rarely been taken into account.
Historically, an innovative flange assembly termed Freeze-Flange
was firstly proposed and developed to undertake in an experimen-
tal MSR (MSRE) at the ORNL (Robertson, 1965). Since then, although
many research and development in the molten salt related tech-
nologies are carried out, few studies on the Freeze-Flange were
reported.
In this study, a modified prototype flange assembly was
designed based on the finite element thermal analysis and verified
by experimental results. The ideal-gas thermodynamic theory was
used to analyze the influence of the flanged connections with more
than one gasket on leakage prevention, and the passive reduction in
the impulse of leakage owing to double or more gaskets introduced
in the present study will be discussed.
2. Freeze-Flange assembly configuration
The Freeze-Flange designed in the MSRE is described in Fig. 1(a)
(Robertson, 1965). It consists of a male flange, a female flange,
clamps, a gasket and a salt screen. The flange is about 580 mm
in outside diameter and roughly 38 mm thick when measured
through the thickest portion of the flange face. Two semicircu-
lar clamps are compelled around the outer edge of the flanges
to assemble the male and female flanges together. A 1 mm wide
gap containing a salt screen is provided between the flanges. One
91
groove with base diameter of about 527 mm in the face of both
the male and female flanges is machined to accommodate the ring
gasket. The salt screen primarily provides a convenient way to
withdraw the frozen salt as a whole cake when the connection is
disassembled.
A modified Freeze-Flange assembly selected as an example is
shown in Fig. 1(b), where the clamp is not marked out. The follow-
ing modifications are made for the new Freeze-Flange. Increasing
the gasket groove number from one to two or more, accordingly
more annulus chambers are created. The metal seals, the silicon
carbide seals or the graphite gaskets etc., which can be used for
high temperature wear and high corrosion resistance, are feasible
candidate for the gaskets.
3. Analysis and discussion
In the process of sealing reliability evaluation, the steady heat
transfer behavior analysis is firstly performed since the tempera-
ture profile plays a significant role in the sealing performance of
the Freeze-Flange. Then theoretical analysis of the thermodynamic
properties of the modified flange assembly is carried out by virtue of
the formulation of ideal-gas and the mathematical methods. Based
on the analytic solutions, the sealing performance of the structural
redesigned flanged connection will be evaluated.
3.1. Temperature profile of the Freeze-Flange
Boundary conditions are described below. With the finite ele-
ment commercial software of ANSYS 14.0 (ANSYS Inc., 2012), the
uninsulated Freeze-Flange is modeled. It is cooled by loss of heat
to the atmosphere with a natural convective heat transfer coeffi-
cient of 10.0 W/m2 K and a surface normal total emissivity of 0.7
(Bergman et al., 2011) accompanied by an ambient temperature
of 22 ◦ C without any other sources of forced cooling. Additionally,
some conditions of constant temperature such as 873 K and 973 K
are independently applied to the inner surface of pipe to simu-
late different typical steady thermal cycles. The Freeze-Flange is
made of the Hastelloy® N alloy, of which the thermal conductivities
varying with temperatures are given in Table 1 (Davis, 2000).
It is easy to know that, although the number of the gaskets is
changed, the cooling conditions and the thermal conducting mech-
anism of the modified Freeze-Flange are the same as the one in the
92 Q. Li et al. / Nuclear Engineering and Design 287 (2015) 90–94

Table 1 3.8
Thermal conductivities of the Hastelloy® N alloy at different temperatures. Originally T=250K
Temperature [K] Approximately
3.3
373 473 573 673 773 873 973

Thermal conductivity [W/m K] 11.5 13.1 14.4 16.5 18.0 20.3 23.6 T=300K
2.8

P' [atm]
MSRE. Fig. 2 reveals the radial temperature profile of the Freeze- T=350K
Flange under different operational conditions. As is shown in Fig. 2,
2.3
the temperatures of the Freeze-Flange fall radially at the same
operational temperature. The salt freeze point is roughly 105 mm
radially from the centerline of the pipe under an operating temper-
1.8
ature of 873 K. This distance rises 23 mm approximately for every
100 K increase of the operating temperature. Experiment results of
MSRE Freeze-Flange at the operating temperature of 973 K are also 1.3
shown in Fig. 2 (Rosenthal et al., 1968). It is worth noting that the 400 500 600 700 800 900 1000
finite element thermal analytic results in this study are remark- T' [K]
ably consistent with the experimental data from the MSRE, where
the temperature drop of the flange progresses slower as the radial Fig. 3. Pressure versus temperature with different calculation methods and ambient
distance becomes larger when the operational temperature is at temperature.
973 K.
where T = T ′ − T , representing the temperature difference, and
3.2. Thermodynamic analysis on the gas closed in the cavity of ˇ is the coefficient of thermal expansion of the Hastelloy® N alloy.
flange assembly Since the real gas behaves as an ideal gas under these conditions
of reasonably low pressures and relatively high temperatures, the
As shown in Fig. 1(b), some volume of gas will be closed by dou- T–P–V behavior of the gas in the cavity may be modeled by the
ble gaskets in the clearance when the flange joints are assembled ideal-gas equation of state (Gengel and Boles, 2005),
at the room temperature. Let us firstly consider that the initial state PV P′V ′
of the gas in the enclosed space A is (T,P,V). Evidently, the pressure = (3)
T T′
P is equal to one atmosphere and the temperature T is the room
Substituting Eqs. (1) and (2) into Eq. (3) we have
temperature of about 300 K. And V, the volume, can be given by the
following relation P′ T ′V T′ (r22 − r12 ) · d
= = ·
V = (r22 − r12 ) · d (1) P TV ′ T (r 2 − r 2 ) · d · (1 + ˇ · T )3
2 1

where r1 and r2 denote the inside and the outside radius of the T′ 1
= · (4)
annulus chamber respectively. And d stands for its height. T (1 + ˇ · T )3
When the system runs stably at some elevated temperature, the
Consequently,
final state of the gas in the cavity A becomes (T ′ , P ′ , V ′ ). The temper-
ature T′ equals the average radial temperature of the Freeze-Flange P 1
P′ = · T′ · 3
(5)
where the cavity locates. Suppose that the sealing performance of T (1 + ˇ · T )
the gaskets satisfies the requirements, we may find that
3
2
where (1 + ˇ · T ) can be transformed approximately into (1 +
V ′ = (r22 − r12 )(1 + ˇ · T ) · d · (1 + ˇ · T ) 3ˇ · T ) by using the one-order Taylor expansion with neglecting
3 higher-order terms. Here, the value of T is not more than 700, and
= (r22 − r12 ) · d · (1 + ˇ · T ) (2)
the ˇ value range is generally between 12.6 × 10−6 and 13.5 × 10−6
with temperature variation from 588 K to 923 K (Davis, 2000; ASME,
1100 2010). Thus, the maximum value of ˇ · T is around 0.01, and so,
Simulation results, T=873K approximately,
Simulation results, T=973K 3
Experimental data, T=973K(ORNL)
(1 + ˇ · T ) ≈ 1 + 3ˇ · T ≈ 1 (6)
950 Therefore, inserting Eq. (6) into Eq. (5) we can obtain
P
Temperature [K]

P′ ≈ · T′ (7)
T
800 It is evident that the value of the PT −1 is a constant. Eq. (7) help-
fully indicates that the value of the pressure P′ varies linearly with
Salt freeze temperature
(727K)
the temperature T′ value. The relationship of the elevated pressure
650 varying with the elevated temperature on the basis of the differ-
ent calculation methods is illustrated in the pressure–temperature
diagram (P–T diagram) of Fig. 3. As can be seen, the approximate
values derived from Eq. (7) are slimly larger than the original val-
500 ues derived from Eq. (5) with the deviation of less than 3.0%, which
0 50 100 150 200 250
is quite acceptable for engineering purposes. Lower temperature T′
Distance from center line of pipe [mm]
results in smaller approximate error.
Fig. 2. Radial temperature profile of Freeze-Flange at different operational temper-
From Eq. (5) we can see, if a material with smaller thermal
ature. expansion coefficient is selected for the Flange, the approximate
 Q. Li et al. / Nuclear Engineering and Design 287 (2015) 90–94
Fig. 4. Schematic of the elastic interaction of the gaskets and the flanges.
error caused by Eq. (7) will be lessened. Practically, so far as the
value of ˇ · T is not large, i.e. not larger than 0.01, the above
conclusion seems to be done.
Furthermore, it is known to all that the room temperature is
usually not constant. If the uncertainty of the room temperature
T is considered, e.g. a reasonable range of 300 ± 50 K will result
in small changes in the values of the PT −1 , the approximate error
will become smaller when the ambient temperature T is higher (as
shown in Fig. 3).
Besides, the volumes stated above do not take into consideration
the elastic interaction of the gaskets and the flanges. It can be easily
noticed that when the seals are installed in the groove and com-
pressed, as described in Fig. 4, it creates some initial interference
force on the groove counter faces and the seals will be compressed
deformation. The free volume of the gas in the cavity will be com-
pressed slightly, and so the pressure will become relatively high.
Evidently, the pressure concluded from Eq. (7) is closer to the actual
value than that results of Eq. (5). Consequently, the approximate
calculation approach is preferable.
Although the above terse conclusions are based on the chamber
A, they also can be applied to the chamber B for their geometric
independence. It seems to be concluded that it is of a universal sig-
nificance. Provided that several annulus enclosed spaces are formed
by some gaskets with ideally seal in the clearance of the flange
assembly, the radial pressure profile of the gases closed in these
chambers (see Fig. 5) corresponds to the radial temperature distri-
bution of the Freeze-Flange (see Fig. 2).
As can be seen from the above, despite the system working
pressure of both the premise and the potential applications are
roughly equal to atmospheric pressure, the mathematical relation-
ships based on a premise of perfect sealing seems to be too idealistic
to come true. Some leakage seems to be inevitable in practice. If ˛,
Fig. 5. Radial pressure distribution of gas in chambers of Freeze-Flange.
93
Table 2
Range of viscosities for various molten salt systems.
Molten salt Viscosity [cP]
LiF-NaF-KF  = 0.04 exp(4170/T (K)), 800 K < T < 1100 K
KF-ZrF4  = 0.0159 exp(3179/T (K)), 800 K < T < 1100 K
NaF-BeF2  = 0.0346 exp(5165/T (K)), 800 K < T < 1100 K
LiF-BeF2  = 0.0116 exp(3755/T (K)), 800 K < T < 1100 K
the volume leakage rate, is considered, after a period of time t, Eq.
(4) becomes
P′ T ′V T′ 1
= = · (8)
P TV ′ T (1 + ˇ · T )3 (1 + ˛ · t)
Consequently,
P 1 1
P′ = · T′ · ·
3 (1 + ˛ · t)
(9)
T (1 + ˇ · T )
If no leakage, ˛ = 0, which expresses a perfect sealing. Eq. (9)
can be rewritten into Eq. (5). As for a specific seal design with a
certain service life of t, provided that the error between Eqs. (9) and
(5) is less than err, the sealing should be conducted with a volume
leakage rate of no more than
t −1
˛= (10)
err −1 − 1
E.g. a specific seal design with a service life of one year, pro-
vided that the error between Eqs. (9) and (5) does not exceed 10%,
the sealing should be conducted with the volume leakage rate of
no more than 1.27 × 10−3 % per hour. Evidently, with an allowed
approximate error, the seal with a longer service life will require a
less leakage rate.
3.3. Sealing reliability evaluation of the modified Freeze-Flange
Based on the solutions above, the sealing reliability of the mod-
ified Freeze-Flange assembly will be discussed.
Firstly, it may improve the performance of the seal as a result
of the active force of leakage reduced. It is well known that the
liquid salts typically run between 800 K and 1000 K with a pres-
sure containment of about two atmospheres in MSRs. As noted in
Section 3.2, the pressure of the closed gas in the chamber A is two
atmospheres or more, which is almost the same as the system oper-
ational pressure. While the differential pressure of the single gasket
on both sides in the conventional structure is the system operating
pressure and the atmospheric pressure. Significantly, the differen-
tial pressure of the gasket on both sides is cut down in the modified
flanged connection contrasted with the old one, so that the possi-
bility of seal failure may be consequently repressed. In addition,
as given in Table 2, the viscosity of the liquid salt will increase as
the temperature decline (Williams et al., 2006) with the radial dis-
tance becoming larger noted in Section 3.1. It may help somewhat
in preventing the molten salt from leaking.
Moreover, it may also improve the sealing capability to adapt
to the operational conditions changes. As we know, retention of
the operating stress within the gasket is important to maintain the
ability to continue to seal. If the system operational temperature
is increased, taking the change from 873 K to 973 K for an exam-
ple, the stress relaxation may take place in the flange assembly
caused by thermal expansion. However, as noted in Section 3.2, the
pressure of the gas in these chambers will also be increased owing
to the temperature elevated. It will compensate the decrement in
the leakage resistance caused by the stress relaxation, so that the
molten salt may be not easy to leak.Additionally, even if all seals fail
due to long time suffering from the high temperature and corrosion,
94 Q. Li et al. / Nuclear Engineering and Design 287 (2015) 90–94
Fig. 6. Cross section of bolted flange joints with double gaskets.
the cavities will serve as buffering chamber for liquid salt leaking.
Furthermore, the small holes between the gaskets and the seal sur-
face are so tiny that they may effectively prevent the molten salt
from moving radially outwards. In this extreme case, the modified
flange assembly will degenerate into the original Freeze-Flange in
the MSRE.
3.4. Sealing reliability evaluation of bolted flange joints with
double gaskets for MSRs
As shown in Fig. 6, the traditional bolted flange joints with dou-
ble gaskets adopted to connect the liquid salt pipe systems of the
MSRs are insulated to prevent cooling, and the temperature of the
gas in the cavity is approximately equal to the system operating
temperature. As noted in Section 3.2, the gas pressure in the cav-
ity can be estimated by Eq. (7) when the system is running. The
pressure in the cavity may be almost equal to or even larger than
the system operational pressure when the fluid salt pipe system is
in the operating conditions of roughly two atmospheres and high
temperature of more than 600 K. Consequently, the liquid salt may
be difficult to leak. As can be seen from Fig. 6, the gasket labeled
1 is exposed to the molten salt corrosion condition, but the dif-
ferential pressure on both sides of the gasket is little. Meantime,
there is a greater pressure difference on both sides of the gasket
labeled 2, which does not contact with the molten salt and thereby
without risk of corrosion. Significantly, the double gaskets sealing
bolted flange connections for MSRs not only have the conventional
sense of the double seal but do not let the gaskets suffer from both
molten salt corrosion and high differential pressure at the same
time, in contrast to the bolted flange joints with single gasket. As
a result, the accident of the leakage occurring may be probably
reduced.
It should be noted that, from Section 3.2, the gas pressure in the
closed cavity is around 2–3 atmospheres with the operational tem-
perature variation from 600 K to 1000 K. Accordingly, the influence
of the double gaskets introduced in the bolted flange joints on the
passive decrease of the leaking driven force performs prominently
in the liquid salt systems which may typically operate at low pres-
sure (less than 4 atmospheres) and high temperature (higher than
600 K), although it may be also present but usually negligible in
the systems of high temperature and high pressure (no less than
1.0 MPa) conditions.
4. Conclusions
The modified Freeze-Flange and the bolted flange connection
with more than one gasket as the sealing components were stud-
ied. An FE thermal analysis was used to describe the temperature
profile of the flanges and the calculation results demonstrated good
agreement with the experimental data of MSRE. The sealing reli-
ability was investigated in terms of the leak motivity diminution
using the ideal-gas thermodynamic principle and mathematical
analysis method. The theoretical results indicate that the modi-
fied Freeze-Flange and the bolted flange joint with double gaskets
will significantly improve the performance of operation under
wider temperature range and the sealing ability of self-adjustment.
The solutions of this study are of great significance not only for
MSR, but also for the Next Generation Nuclear Plant (NGNP) and
other innovative nuclear reactors based on the molten salt coolant
technology, e.g. Fluoride salt cooled Pebble Bed Reactors (FPBR)
also referred as Advanced High Temperature Reactors (AHTR)
(Williams, 2006; Forsberg, 2006; Williams et al., 2006), as well
as other high temperature or low temperature engineering appli-
cations with relatively low internal pressure, e.g. the chemical
industry and the high temperature superconductivity technology.
However, further studies, i.e. mechanics analysis and experiment,
are needed in the future considerations in order to determine
the sealing disposition of the new conceptional flanged connec-
tions.
Acknowledgements
The authors appreciate the support from the “Strategic Priority
Research Program” of the Chinese Academy of Sciences (Grant No.
XDA01020304).
References
ANSYS Inc., 2012. ANSYS 14.0 Workbench User’s Guide.
ASME, 2010. ASME Boiler and Pressure Vessel Code Section II, Part D: Properties
(Metric) Materials, 722.
Bergman, T.L., Lavine, A.S., Incropera, F.P., DeWitt, D.P., 2011. Fundamental of Heat
and Mass Transfer, 7th ed. John Wiley & SONS, pp. 1008.
Choi, W.S., Seo, K.S., 2011. Shape optimization of a torus seal under multiple loading
conditions based on the stress categories in the ASME code section III. Nucl. Eng.
Des. 241 (2011), 2653–2659.
Davis, J.R., 2000. ASM specialty handbook: nickel, cobalt, and their alloys. ASM Int.,
51.
Fu, P., Zhang, D.G., 2011. Handbook of Seal Design. Chemical Industry Press, Beijing
(in Chinese).
Forsberg, C.W., 2006. Developments in molten salt and liquid-salt-cooled reactors.
In: Proceedings of ICAPP06, Paper 629.
Gengel, Y.A., Boles, M.A., 2005. Thermodynamics: An Engineering Approach, 5th ed.
McGraw Hill, pp. 137–153.
GIF-IV, 2002. A technology Roadmap for Generation IV Nuclear Energy Sys-
tems. Technical Report, GIF-002-00, Available from: http://www.gen-4.org/
PDFs/GenIVRoadmap.pdf
Guo, Z.P., Zhou, J.J., Zhang, D.L., Chaudri, K.S., Tian, W.X., Su, G.H., Qiu, S.Z., 2013.
Coupled neutronics/thermal-hydraulics for analysis of molten salt reactor. Nucl.
Eng. Des. 258 (2013), 144–156.
Izumi, S., Yokoyama, T., Kimura, M., Sakai, S., 2009. Loosening-resistance evaluation
of double-nut tightening method and spring washer by three-dimensional finite
element analysis. Eng. Fail. Anal. 16 (2009), 1510–1519.
LeBlanc, D., 2010. Molten salt reactors: a new beginning for an old idea. Nucl. Eng.
Des. 240 (2010), 1644–1656.
MacPherson, H.G., 1985. The molten salt adventure. Nucl. Sci. Eng. 90, 374–
380.
Robertson, R.C., 1965. ORNL-TM-0728, 178-189.
Rosenthal, M.W., Briggs, R.B., Kasten, P.R., 1968. ORNL-4254, 23-29.
Singh, J., Angra, S., Mittal, V.K., 2012. Failure mode analysis of mechanical seals. J.
Eng. Technol. 2 (2012), 129–134.
Williams, D.F., 2006. Assessment of Candidate Molten Salt Coolants for the
NGNP/NHI Heat-Transfer Loop. Technical Report, ORNL/TM-2006/69.
Williams, D.F., Toth, L.M., Clarno, K.T., 2006. Assessment of Candidate Molten Salt
Coolants for the Advanced High-temperature Reactor (AHTR). Technical Report,
ORNL/TM-2006/12.