You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/245195077

Analysis of safety margin of in-vessel retention for AP1000

Article in Nuclear Engineering and Design · August 2010


DOI: 10.1016/j.nucengdes.2010.04.020

CITATIONS READS

69 367

4 authors:

Yapei Zhang S.Z. Qiu


Xi'an Jiaotong University Xi'an Jiaotong University
130 PUBLICATIONS 1,046 CITATIONS 782 PUBLICATIONS 5,599 CITATIONS

SEE PROFILE SEE PROFILE

Guanghui Su XI WEN Tian


Xi'an Jiaotong University Xi'an Jiaotong University
954 PUBLICATIONS 6,746 CITATIONS 675 PUBLICATIONS 4,556 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Thermal-hydraulic analysis of EBR-II Shutdown Heat Removal Tests SHRT-17 and SHRT-45R View project

Space nuclear reactor View project

All content following this page was uploaded by XI WEN Tian on 05 February 2022.

The user has requested enhancement of the downloaded file.


Nuclear Engineering and Design 240 (2010) 2023–2033

Contents lists available at ScienceDirect

Nuclear Engineering and Design


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

Analysis of safety margin of in-vessel retention for AP1000


Y.P. Zhang, S.Z. Qiu ∗ , G.H. Su, W.X. Tian
Department of Nuclear Science and Technology, Xi’an Jiaotong University, Xi’an city 710049, China

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

Article history: In-vessel retention (IVR) of core melt is a key severe accident management strategy adopted by most
Received 23 November 2009 operating nuclear power plants and advanced light water reactors (ALWRs), AP600, AP1000, etc. Exter-
Received in revised form 25 March 2010 nal reactor vessel cooling (ERVC) is a novel severe accident management for IVR analysis. In present
Accepted 13 April 2010
study, IVR analysis code in severe accident (IVRASA) has been developed to evaluate the safety margin
of IVR in AP1000 with anticipative depressurization and reactor cavity flooding in severe accident. For,
IVRASA, the point estimate procedure has been developed for modeling the steady-state endpoint of two
core melt configurations: Configuration I and Configuration II. The results of benchmark calculations of
AP600 by IVRASA were consistent with those of the UCSB and INEEL. Then, IVRASA is used to calculate
the heat transfer process caused by two core melt configurations of AP1000. The results of calculations
of Configuration I indicate that the heat flux remains below the critical heat flux (CHF), however, the
sensitivity calculations show that the heat flux in the metallic layer could exceed the CHF because of the
focusing effect due to the thin metallic layer. On the other hand, the results of calculations of Configu-
ration II suggest that the thermal failure of the lower head at the bottom location is highly unlikely, but
the heat flux in light metallic layer could be higher than that of base case due to the portion of metal
partitioning into the lower head. This work also investigated the effect of the uncertainties of the CHF
correlations on the analysis of IVR.
© 2010 Elsevier B.V. All rights reserved.

1. Introduction from the molten debris in the lower plenum. If the local heat flux
through the vessel wall exceeded the critical heat flux (CHF), vessel
The in-vessel coolability of a core melt is an important issue in failure would be expected.
addressing a postulated inadequate core cooling event in nuclear IVR of molten core debris inside reactor pressure vessel (RPV)
reactors. In such an even, a significant amount of core materials can lower head through external cooling by cavity water has been the
become molten and relocate downward into the lower head of the subject of numerous numerical and experimental investigations
reactor vessel, as happened in the Three Mile Island Unit2 (TMI-2) conducted by Kolev (1993), Theofanous et al. (1994a,b, 1996a,b),
accident. If it is possible to ensure the lower head to remain intact, Asfia and Dhir (1996), Chu et al. (1997), Rempe et al. (1997),
which can retain the relocated core materials, the enhanced safety Loktinonov et al. (1999), Esmaili and Khatib-Rahbar (2004) and
associated with these plants can reduce concerns about contain- Dinh et al. (2004).
ment failure and associated risks. For example, the Westinghouse Henry and Fauske (1993) investigated the capability of the
Advanced 600 MWe PWR (AP600) adopted ERVC with a flooded external cooling of the RPV lower head to prevent the lower head
reactor cavity as the enhanced safety. Subsequent advanced light failure considering the presence of the RPV insulation. Cheung et
water reactors, such as AP1000 and APR14000, also adopted this al. (1997) developed a subscale boundary layer boiling (SBLB) test
enhanced safety. facility to study the boundary layer boiling and critical heat flux
If the reactor cavity is flooded before melt relocated into the phenomena on a downward facing cured heating surface, and to
lower head, the vessel wall would be initially cool and the outer measure the spatial variation of the critical heat flux. Knudson
vessel wall would remain close to the cavity water saturation tem- and Rempe (2002) established IVR model to simulate heat transfer
perature. Boiling crisis is the sufficient and necessary condition for inside the accumulated molten pool, heat transfer from the molten
lower head failure (Theofanous et al., 1996a). Nucleate pool boil- pool to the reactor vessel, and heat transfer from exterior vessel
ing of the cavity water is an efficient mechanism for heat removal surfaces using SCDAP/RELAP5-3D.
Theofanous et al. (1996a) and Rempe et al. (1997) used one-
dimensional mathematical model to calculate the thermal response
∗ Corresponding author. Tel.: +86 29 82665607; fax: +86 29 82665607. of the lower head. They also provided a summary of the various
E-mail address: szqiu@mail.xjtu.edu.cn (S.Z. Qiu). heat transfer correlations for the ceramic pool and the stratified

0029-5493/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.nucengdes.2010.04.020
2024 Y.P. Zhang et al. / Nuclear Engineering and Design 240 (2010) 2023–2033

Nomenclature
ıcr,t oxide crust thickness of the top of the oxide pool, m
A coefficient for estimating heat transfer from the sur- ıw thickness of the vessel wall, m
face of the melt pool, W/m2 K4 ε emissivity
Gr Grashoff number, =gˇTmax H3 / ␯2 ε1 emissivity of the top light metallic layer
g gravity acceleration, m/s2  angle from the bottom center of the lower head,
H height of the metallic layer, vertical size of the cav- degree
ity, m h maximum angle of the heavy metallic layer, degree
H depth of oxide pool, m  kinematic viscosity, m2 s
k thermal conductivity, W/(mK)  Stefan–Boltzmann constant (5.67 × 10−8 W/m2 K4 )
Nu Nusselt number, =(q H )/kTmax Tmax maximum bulk-wall temperature difference, K
Pr Prandtl number, =/˛
Q̇ volumetric heat generation rate, W/m3 Subscript
q average heat flux at pool boundaries, W/(m2 ) CHF critical heat flux
qh,b heat flux from the bottom of heavy metallic layer, cr ceramic crust
W/(m2 ) dn downward (over the hemispherical boundary)
ql,b heat flux from the top oxidic pool crust into the top h heavy metallic layer
l light metallic layer
light metal layer, W/(m2 )
o oxide pool
ql,t heat flux from the top light metallic layer into the
s upper plenum structures
internal atmosphere of the reactor, W/(m2 ) side the side of the oxide pool
ql,w heat flux from the top metallic layer into the vessel up upward (over the flat boundary of the oxidic pool)
wall, W/(m2 ) w vessel wall
qo,dn heat flux towards the bottom of oxidic pool, W/(m2 )
qo,h heat flux from the ceramic pool to the heavy metallic
layer, W/(m2 ) light molten metallic layer. Khatib-Rahber et al. (1996) adopted a
qo,up heat flux towards the top of oxidic pool, W/(m2 ) two-dimensional model. The comparison of the one-dimensional
qves heat flux from the outer surface of the vessel to and two-dimensional models showed that a one dimensional heat
water, W/(m2 ) transfer model of the lower head performed adequately. Esmaili
R reactor vessel, hemisphere radius, m and Khatib-Rahbar (2004, 2005) established a one-dimensional
R ratio of the Nusselt number model to assess the thermal response of the AP1000 lower head
Ra Rayleigh number (metallic layer), = based on two bounding melt configurations (Configuration I and
(gˇTmax H 3 )/(˛) Configuration II). Theofanous et al. (2004) analyzed the thermal
Ra Rayleigh number (oxide pool), = (gˇQ̇ H  5 )/(k˛) margins of AP1000 and provided experimental data to show that
S area of the melt pool, m2 it was reasonable to allow the robust extension of the AP600 IVR
Sdn area of the bottom of the oxide pool, m2 strategy for severe accident management to higher power reac-
Sh,b area of the bottom of the heavy metallic layer, m2 tors, and in particular, to the AP1000 advanced passive design. So
S1,b area of the bottom of the light metallic layer, m2 far, numerical investigations on IVR of AP1000 are few, and this is a
S1,t area of the top of the light metallic layer, m2 necessary part of the analysis in severe accident. Therefore, IVRASA
Sside area of the side of the oxide pool, m2 code was developed to analyze the safety margin of AP1000 during
Sup area of the top of the oxide pool, m2 severe accident in this work.
T temperature, K This paper consists of two main technical parts. The first part
Tbh bulk temperature of the heavy metallic layer, K (Section 2) presents the simple point estimate procedure (IVRASA)
Tbl bulk temperature of the light metallic layer, K based on the existing constitutive relations, originally developed
T1,b temperature of the bottom of light metallic layer, K for AP600 (Theofanous et al., 1996a; Rempe et al., 1997), and com-
T1,t temperature of the top of light metallic layer, K pare the results calculated by IVRASA with those of the University of
To,m melt temperature of oxide pool California–Stanta Barbara (UCSB) (Theofanous et al., 1996a) and the
Tw,m melt temperature of vessel wall Idaho National Engineering and Environmental Laboratory (INEEL)
Ts,i inner surface temperature of the internal structure, (Rempe et al., 1997). The second part (Section 3) applies IVRASA
K to calculate two core melt configurations of AP1000. And more
Ts,o outer surface temperature of the internal structure, details are conducted on the thermal margins of the AP1000 for
K Configuration II.
Tw,i inner surface temperature of the vessel wall, K
Tw,o outer surface temperature of the vessel wall, K 2. IVRASA and verification calculations
V volume of the melt pool, m3
IVRASA model is based on DOE/ID-10460 that the Department
Greek characters of Energy (DOE) contracted the UCSB to produce to support the
˛ thermal diffusivity, m2 /s Westinghouse position on AP600 IVR (Theofanous et al., 1996a)
ˇ thermal expansion coefficient, K−1 and the technical evaluation of AP600 IVR by INEEL (Rempe et al.,
ı the thickness, m 1997). The latter was developed by INEEL independently to verify
ıcr oxide crust thickness of the side of the oxide pool, the results of UCSB.
m The UCSB study regarded Configuration I (Fig. 1a) as the final
ıcr,b oxide crust thickness of the bottom of the oxide pool, steady state, their postulated FInal Bounding State (FIBS). A key
m assumption in the UCSB study was their assumed debris configura-
tion. And the UCSB study argued that their FIBS bounded thermal
Y.P. Zhang et al. / Nuclear Engineering and Design 240 (2010) 2023–2033 2025

Fig. 1. Alternate debris configurations evaluated for AP1000 (a) Configuration I and (b) Configuration II.

loads from any other configurations. However, INEEL’s review sug- equations (equation A-106) has been shown as follow:
gested that the UCSB FIBS did not involve several possible thermal  
Q̇o ı2cr () qo,dn () Q̇o ıw ()
challenges from other debris configurations. So, INEEL gave Config- + ıcr () +
uration II as the final steady state (Fig. 1b). And INEEL recommended 2kcr kcr 2kw
Configuration II as assumed FIBS, which could contain other core
qo,dn ()ıw ()
melt configurations. − To,m + + Tw,o = 0 (5)
Configuration I assumes a molten ceramic pool of UO2 –ZrO2 kw
lying beneath a metallic layer of Fe–Zr, and Configuration II rep-
resents a case where sufficient uranium dissolves into unoxidized Tw,m − Tw,o
zirconium to form a heavier metallic layer of Fe–U–Zr in the bottom ıw () = kw (6)
qo,dn () + Q̇o ıcr ()
and a light metallic layer of Fe–Zr on the top with a molten ceramic
pool between the two metallic layers.
qves () = qo,dn () + Q̇ ıcr () (7)

2.1. IVRASA methodology Eqs. (5) and (6) were solved iteratively to get the value of the
thickness of the ceramic crust, then the heat flux to the water of
IVRASA is a point estimate procedure that is similar to the model the outer surface of the vessel could be obtained.
used in the UCSB report, and IVRASA contains equations similar to Light metallic layer:
the equations reported in the UCSB study. The governing equations
of the IVRASA model can be seen as follows, and more details about Q̇l Vl + ql,b Sl,b = ql,t Sl,t + ql,w Sl,w (8)
the property equations can be seen in UCSB report (Theofanous
kcr Q̇o ıcr,t
et al., 1996a). However, IVRASA was developed in a more general ql,b = (To,m − Tbl ) + (9)
manner so that it could calculate Configuration I and Configuration ıcr,t 2
II, besides Configuration I assumed as UCSB FIBS, and some modifi-
cations of the Rempe model equations have been shown here. The
melt pool Configuration II more bounded thermal loads from other
configurations (Rempe et al., 1997). So, the schematic of the melt
pool Configuration II is shown in Fig. 2, and the schematic of the
melt pool Configuration I is similar with the schematic of the melt
pool Configuration II.
As discussed above, the governing equations of the three lay-
ers are shown as follows (these equations are also suitable for the
Configuration I).

Oxidic pool:

Q̇o Vo = qo,up Sup + qo,dn Sdn (1)

Q̇o Vo
qo,dn = (2)
Sside + Sdn + Sup R
Nuup
R = (3)
Nudn

Q̇o Vo − qo,dn (Sside + Sdn )


qo,up = (4)
Sup
After systematic derivation and dimensional analysis of govern-
ing equations in Rempe report, a modification in the Rempe model Fig. 2. The schematic of the melt pool Configuration II for IVRASA.
2026 Y.P. Zhang et al. / Nuclear Engineering and Design 240 (2010) 2023–2033

4 − T4 ]
[Tl,t Table 1
s,i
ql,t = (10) Comparison of IVRASA with UCSB FIBS.
(1/εl ) + ((1 − εs )Sl,t /εs Ss )
IVRASA UCSB
kw
ql,w = (Tw,m − Tw,o ) (11) Configuration I and Configuration II Configuration I assumed FIBS
ıw Portion of the decay heat in the All the decay heat resides in the
 4 − T4 )  metallic layer ceramic pool
ıs (Tl,t s,i Emissivity from the metallic layer Emissivity from the metallic layer
Ts,o = Ts,i − (12)
ks (Ss /εl Sl,t ) + (1 − εs /εs ) may be represented by a point may be represented by a point
value of 0.29a value of 0.45
 ks   0.25 Some additional heat sources due to No additional heat sources exist
4
Tw = Ts,o − Ts,i − Ts,o (13) oxidation or steel activations exist in the metallic layer
ıs εs in the metallic layerb

ks ıo   UCSB ACOPO molten pool natural UCSB Mini-ACOPO molten pool


Tw = Ts,i − Ts,o + Tw,o (14) convection correlations apply natural convection correlations
kw ıs UCSB ULPU CHF lower bound UCSB ULPU CHF lower bound
correlations apply correlations apply
Other two modifications in the Rempe model equations (equa-
a
tions A-121 and A-122) have been shown as follow: According to the analysis of INEEL, emissivity from the metallic layer was a point
value of 0.29, which is a conservative.
 4 − T4 ) 0.25 b
Power and Behbahani (2004) suggested that there had the potential for inter-
(Tl,t s,i
Tbl = + Tl,t (15) metallic reactions in metallic layer.
(1/Al,t )((1/εl ) + ((1 − εs )Sl,t )/εs Ss )
 0.25
qo,up As shown in Table 1, here, the CHF correlations still adopted the
Tl,b = + Tbl (16) CHF correlations used in UCSB study.
Al,b

Heavy metallic layer: 2.3. Heat transfer correlations in IVRASA


Q̇h Vh + qo,h Sdn = qh,b Sh,b (17)
In DOE/ID-10460, heat transfer correlations were based on
h
Sup data from the UCSB Mini-ACOPO tests. Further testing was subse-
qo,h = qo,dn () sin() cos()d (18)
Sdn 0
quently carried out in the ACOPO facility (Theofanous et al., 1996b,
1997), which was a larger scale version of Mini-ACOPO. Corre-
kw  
qh,b = Tw,i − Tw,o (19) lations obtained from the Mini-ACOPO and ACOPO tests differ.
ıw IVRASA adopted the heat transfer correlations based on ACOPO data
qo,h is the heat flux from the ceramic pool to the heavy metallic (Theofanous et al., 1996b, 1997).
layer, and the definition of the variable is different from the INEEL Heat transfer correlations in molten ceramic pool and metallic
model and the UCSB model. layer in IVRASA were compared with those used in UCSB model
The decay heat in the oxidic pool and the heavy metallic layer is and INEEL calculations, as shown in Table 2. The constitutive set of
calculated in a simple way, that is: empirical heat transfer correlation is used in the same form with
some variations as given in Table 2 for different models.
Q̇o + Q̇h = Pdecay,t (20)

Q̇h Vh mU (270/238) 2.4. Benchmark calculation results of IVRASA


= (21)
Q̇o Vo mUO2
In order to verify IVRASA study, benchmark calculation was con-
where mU is the mass of uranium dissolved into the heavy metallic ducted with the UCSB input and the UCSB-assumed FIBS under
layer, and mUO2 is the mass of mUO2 in the oxidic pool.CHF from the condition of successful reactor coolant system depressuriza-
vessel outer surface: tion and reactor cavity flooding. Benchmark calculation results of
qCHF () = C1 + C2  + C3  2 + C4  3 + C5  4 (22) IVRASA were compared with the UCSB results and INEEL results
as follows. And the UCSB input data were shown in Rempe model
where  is the angle from the bottom center of the vessel lower Appendix C. Results of IVRASA were compared with results pre-
head, and the coefficients C1 through C5 are based on the experi- sented in DOE/ID-10460 Appendix Q and the INEEL calculation
mental results for AP600 (Theofanous et al., 1996a). results. Calculation results of IVRASA were presented as follows.
Those variables reflecting the IVR margins best, such as the heat
The impact of the decay heat in the metallic layers as the sensi- flux, CHF ratios, oxide crust thickness and vessel wall thickness are
tivity study is discussed later. The calculation of the fraction of the given in Figs. 3–6, in which the dash lines represent the IVRASA
decay heat in the metallic layers was same with that in the Rempe results, the dot lines show the INELL results and the solid lines
model (Rempe et al., 1997). display the UCSB results.
It can be seen from these figures that IVRASA correctly inter-
2.2. IVRASA-FIBS preted equations in the UCSB and INEEL analyses and correctly
encoded these equations into IVRASA.
As discussed above, IVRASA was developed in a more general Figs. 3 and 4 indicate that IVRASA, UCSB model, and INEEL pre-
fashion so that it could calculate Configuration I and Configuration dictions differ slightly at the locations adjacent to ceramic pool
II. However, the UCSB study regarded Configuration I (Fig. 1) as the (between 0◦ and ∼76◦ ), and were slightly larger at locations adja-
FIBS, which was a key assumption in the UCSB study. cent to the metallic layer (greater than ∼76◦ ). Particularly, at the
Configuration I is assumed to represent a two-layer melt pool locations adjacent to the metallic layer, IVRASA heat flux prediction
with a light metallic layer on top of a molten ceramic pool as shown is larger than UCSB and INEEL. And INEEL prediction is smallest.
in Fig. 1. According to the technical estimate report by INEEL, some Fig. 5 indicates that oxide crust thickness prediction by IVRASA
important assumptions were modified in IVRASA model. The com- is smallest, and INEEL prediction is largest. Fig. 6 indicates that
parison of IVRASA with UCSB FIBS is shown in Table 1. IVRASA vessel wall thickness prediction is similar to UCSB model,
Y.P. Zhang et al. / Nuclear Engineering and Design 240 (2010) 2023–2033 2027

Table 2
Heat transfer correlations used in different models.

1. Constitutive correlation set used by IVRASA in a steady state


Metal
Side wall:
Churchill and Chu (1975):
0.15 Ra1/3
Nu = 16/27
1+(0.492/Pr)9/16
12
for 0.1 < Ra < 10 and any Pr.

Top and bottom:


Globe and Dropkin (1959):
Nu = 0.069Ra1/3 Pr0.074
for 3 × 105 < Ra < 7 × 109 and 0.02 < Pr < 8750

Ceramic pool
Top:
Theofanous et al. (1996b) ACOPO (1/2 scale):
Nuup =1.95Ra 0.18
for 1010 < Ra <3 × 106 .
Bottom:
Mayinger et al. (1975):
Nudn = 0.55Ra 0.22
Fig. 3. Comparison of IVRASA heat flux with UCSB (DOE/ID-10460) and INEEL heat for 7×106 < Ra <5 × 1014
fluxes assuming UCSB FIBS and input.
2. Constitutive correlation set used by UCSB in a steady state (Theofanous
et al., 1996a)
Metal
Side wall:
Churchill and Chu (1975):
Nu = 0.076Ra1/3 .

Top and bottom:


Globe and Dropkin (1959) modified to Nu = 0.15Gr1/3 .
for 3 × 105 < Ra < 7 × 109 and 0.02 < Pr < 8750.

Ceramic pool
Top:
Theofanous et al. (1996b) Mini-ACOPO (1/8 scale):
Nuup = 0.345Ra 0.233
for 1012 < Ra < 3 × 1014 and 2.6 < Pr < 10.8
Bottom:
Theofanous et al. (1996b) Mini-ACOPO:
Nudn = 0.0038Ra 0.35
for 1012 < Ra < 3 × 1014 and 2.6 < Pr < 10.8

3. Constitutive correlation set used by INEEL in a steady state (Rempe et al.,


1997)
Metal
Side wall:
Churchill and Chu (1975):
Fig. 4. IVRASA and UCSB (DOE/ID-10460) CHF ratios assuming UCSB FIBS and input.
2
0.387Ra1/6
Nu = 0.825 + 8/27
1+(0.492/Pr)9/16

12
for 0.1 < Ra < 10 and any Pr.

Top and bottom:


Globe and Dropkin (1959):
Nu = 0.069Ra1/3 Pr0.074
for 3 × 105 < Ra < 7 × 109 and 0.02 < Pr < 8750

Ceramic pool
Top:
Theofanous et al. (1996b) ACOPO (1/2 scale):
Nuup = 2.4415Ra 0.1722

Bottom:
Theofanous et al. (1996b) ACOPO (1/2 scale):
 H 0.25
Nudn = 0.1857Ra 0.2304 R
for 1010 < Ra <3 × 1016

but different from INEEL at the locations adjacent to the metallic


layer. All these differences in these figures may be due to dif-
ferent correlations adopted in different models. Although some
differences exist among the three models, benchmark calculations
of UCSB-assumed FIBS indicate the applicability and accuracy of
Fig. 5. IVRASA, UCSB (DOE/ID-10460) and INEEL predictions for oxide crust thick- IVRASA and it could be applied to predict the thermal response of
ness assuming UCSB FIBS and input.
the two molten configurations.
2028 Y.P. Zhang et al. / Nuclear Engineering and Design 240 (2010) 2023–2033

Table 4
Fractions of Zr oxidation and UO2 relocation calculated by MELCOR and MAAP.

Model Fraction of Zr oxidation Fraction of UO2 relocation

MELCOR 0.4–0.6 70–80%


MAAP 0.3 70%

26 MW at relocation time 3.2 h for conservative point estimate


calculations.
The other parameters including fraction of Zr oxidation, mass
of UO2 relocation to the lower plenum and the amount of steel
in the lower head determine the corium mass inventories and the
molten configurations. For AP1000, the results of these parameters
calculated by MELOR and MAAP are shown in Table 4.
The amount of steel in the lower plenum was determined by the
mass of UO2 relocation to the lower plenum (Esmaili and Khatib-
Rahbar, 2005). Esmaili and Khatib-Rahbar (2005) contended that
“for the 50–60 m of UO2 in the lower plenum, the steel mass can
vary between 3 and 8 m. . .”, and “as a best estimate, the steel mass
Fig. 6. IVRASA, UCSB (DOE/ID-10460) and INEEL predictions for vessel wall thick- is estimated to vary between 40 and 60 m, for the best estimate
ness assuming UCSB FIBS and input. core UO2 mass of 60–80 m. . .”.
Note that other parameters (metal emissivity, structure emis-
sivity, etc.) remain the same with the AP600 model in the first
3. IVR analysis of AP1000 using IVRASA
section.
In Section 2, the critical heat flux correlation for AP600 was
In Section 2, IVRASA was proved to predict the thermal response
obtained from the ULPU-IIIfacility, and the equation of qCHF,AP600
for AP600 well. In this section, IVRASA will be extended to calcu-
was shown in the previous section. The formation of the CHF cor-
late the safety margins of AP1000 for Configurations I and II. The
relation for AP1000 as follows:
overall containment design of AP1000 below the operating deck
remains essentially the same with AP600, and the reactor vessel
qCHF,AP1000 = factor · qCHF,AP600 (23)
diameter and lower head geometry have been retained from the
AP600. Because the AP1000 is similar to AP600 except for the power The value of the factor was recommended to equate to be 1.44
increase from 600 to 1000 MW, it is convenient to extend IVRASA (Esmaili and Khatib-Rahbar, 2005). The curve of qCHF,AP1000 was
to AP1000 IVR calculations. compared with the experimental data obtained from full-scale
In this section, heat transfer correlations in molten pools and ULPU-V facility for AP1000 (Theofanous et al., 2004) (see Fig. 7). The
metallic layers remain the same (Table 2). The physical processes fitting line (the point dash line in Fig. 7) obtained using the least-
involved in the late stage of severe accidents are very complex and squares method based on the ULPU-V experimental data. Because
remain uncertain. Seiler et al. (2003) developed a theoretical model only five experimental data could be obtained, curve fitting in the
to assess the relocation of the molten core. The initial conditions form of a quadratic equation was not satisfactory. It can be seen
including decay heat, mass of UO2 relocation to the lower head from the Fig. 7 that the factor of 1.44 offers a good estimate for the
and the amount of steel in the lower head are uncertain for Con- critical heat flux overall, but it overestimates the critical heat flux
figurations I and II. Therefore, the determination of the initial and near the top. It also can be seen from Fig. 7 that Eq. (23) used for
boundary conditions is mainly based on the results of plant-specific IVR analysis of AP1000 was conservative, compared to the fitting
MELCOR and MAAP calculations in AP1000 under the condition of curve (see Fig. 7). Therefore, Eq. (23) was selected for IVR analysis
successful reactor coolant system depressurization (Zavisca et al., of AP1000.
2003; Yuan et al., 2003) and some important parameters recom-
mended by Esmaili and Khatib-Rahbar (2004) as follows. Esmaili
and Khatib-Rahbar (2005) developed a one-dimensional model to
assess the thermal response of AP1000 lower head based on the
two core melt configurations.
The decay heat in the oxide pool with energy flow split deter-
mines the heat transfer from the oxide pool to vessel wall and the
metallic layers. So, the quantification of the decay heat in the oxide
pool is very important. Here, all the decay heat is assumed resid-
ing in the oxide pool for Configuration I. For Configuration II, the
decay heat is partitioned between the oxide pool and the metal-
lic layer. For AP1000, results of core relocation time and decay
power calculated by MELCOR are different from those of MAAP,
which are shown in Table 3. The decay power was assigned as

Table 3
Relocation time and decay power calculated by MELCOR and MAAP.

Model Relocation time Decay power

MELCOR 2.6–3.7 h 23–29 MW


MAAP 1.7 h 28.7 MW
Fig. 7. Comparison empirical CHF correlation for AP1000 with experimental data.
Y.P. Zhang et al. / Nuclear Engineering and Design 240 (2010) 2023–2033 2029

Table 5
Mass of corium inventories for base case.

Corium materials UO2 ZrO2 Zr Steel

Mass (kg) 66,266 6211 13,714 37,376

Fig. 10. IVRASA and Esmaili predictions for vessel wall thickness.

input values were consistent with the results of Esmaili. As shown


in Figs. 8 and 9, the heat flux and the heat flux ratio were slightly
higher than those of Esmaili results at locations adjacent to the
Fig. 8. Comparison of IVRASA heat flux with Esmaili heat fluxes.
metallic layer. The reason for these differences is that different heat
transfer correlations were adopted and undocumented differences
3.1. Results of base case (Configuration I) in Esmaili input assumptions. As shown in Figs. 8 and 9, at the bot-
tom of the lower head, the heat flux is lowest, but the CHF is also
This section presents IVRASA calculation results of AP1000 for lowest at this location, the heat flux ratio of qves /qCHF is about 0.2
Configuration I. For base case, heat transfer correlations in molten for two models. At the top of the oxide pool and the metallic layer
pools and metallic layers remain the same (Table 2). The mean val- the heat flux ratio is significantly increased, but remains below
ues of the mass of corium inventories calculated by Esmaili and unity.
Khatib-Rahbar (2004) were adopted by IVRASA as input parameters
for point estimate calculations (Table 5). Some variables reflecting 3.2. Sensitivity analysis (Configuration I)
IVR margins, such as heat flux, ratios of this vessel heat flux to the
CHF, oxide crust thickness and vessel wall thickness were obtained. Although IVRASA predicts that CHF ratios of AP1000 remain
Base case calculation results of IVRASA for AP1000 were compared below unity, Rempe et al. (1997) and Esmaili and Khatib-Rahbar
with the Esmaili results (Esmaili and Khatib-Rahbar, 2005) as fol- (2005) suggested that there were uncertainties in several input
lows. parameters. We consider two kinds of uncertainties that were
Figs. 8–11 show the distributions of the heat flux to water, the reduced metallic melt mass and additional metallic layer sources
heat flux ratios, the vessel wall thickness and the oxide pool crust (due to steel activation, oxidation or the fraction of the fission prod-
thickness as a function of the angle, respectively. In these figures, ucts) presented by Rempe et al. (1997).
the dash lines represent the IVRASA results, and the solid lines dis- IVRASA input parameter uncertainties were assumed except for
play Esmili’s results for comparison. It can be seen from these Figs. the parameters or equations that were modified to assess the phe-
that results of IVRASA point estimate for AP1000 with the mean nomena of interest. Modifications required for each analysis and

Fig. 9. IVRASA and Esmaili CHF ratios. Fig. 11. IVRASA and Esmaili oxide crust thickness.
2030 Y.P. Zhang et al. / Nuclear Engineering and Design 240 (2010) 2023–2033

culated by AP1000 plant-specific MELCOR during severe accident,


CHF ratios calculated by IVRASA reach to 2.5. According to the point
calculations by IVRASA, reduced metallic layer mass could reduce
the safety margins of vessel integrity and even may cause the vessel
failure.
Fig. 13 represents the condition of additional metallic layer
sources. Rempe et al. (1997) and Esmaili and Khatib-Rahbar (2005)
suggested that there was a potential for the metallic layer to contain
additional heat source due to activation of stainless steel and oxi-
dation of materials in the metallic layer. For AP1000 calculations,
IVRASA assumed an additional heat source in the metallic layer
which corresponded to 8% of the core’s power at times between 2
and 6 h after shutdown recommended by INEEL.
Additional heat sources only affect the metallic layer results, as
shown in Fig. 13. It can be seen from Fig. 13 that IVRASA point esti-
mate calculations indicate that this additional heat source increases
heat flux ratios at locations adjacent to the metallic layer by approx-
imately 25%, which reduces the safety margin of IVR. However, this
calculation for AP1000 is very conservative, and the heat flux ratio
Fig. 12. Comparison of CHF ratios for different steel masses in the metallic layer. is below the unity.

results from each analysis are discussed below. The results of the 3.3. Results of melt Configuration II
two conditions are shown in Figs. 12 and 13.
Fig. 12 represents the condition of reduced metallic melt mass The formation of the heavy metallic layer in the lower head
that affects the height of the metallic layer. Because the metal- was proposed to analyze the vessel integrity of AP600 by INEEL in
lic layer has a higher thermal conductivity, the mass of metals 1997. From then on, numerous numerical and experimental inves-
assumed in the metallic layer significantly affects the estimated tigations have been conducted on modeling of the molten pool
magnitude of the heat fluxes focused toward the adjacent vessel structure and the thermal and physicochemical processes occurring
sidewall. If metallic melt mass is reduced, the excess of energy at the interaction of oxidic and metallic melt materials.
which come from the oxide pool and cannot be removed by Experiments performed at OECD MASCA facility (Asmolov et
radiation is focused through the vessel wall causing considerable al., 2004, 2007; Gusarov et al., 2006, 2007; Smirnov et al., 2007;
thermal attack of the vessel. This is so-called focusing effect. This Kiselev et al., 2007), ISTC CORPHAD facility (Bechta et al., 2006a,c,
focusing effect may be a challenge to the integrity of the vessel. 2007a,b) and METCOR facility (Bechta et al., 2004, 2006b) have
For point estimate calculations, IVRASA, respectively, assumed shown the potential for partitioning of the major constituents (U,
the metallic layer contains approximately 37,376 (base case), Zr, Fe) between suboxidized corium melt and steel in the IVR con-
20,000, 15,000 and 3000 kg of AP1000 vessel internal structure ditions. These tests show that the formation of the heavy metallic
steel. layer in the bottom of the vessel head due to component partition-
It can be seen from Fig. 12 that reductions in the steel relo- ing between suboxidized corium melt and steel. This stratification
cation mass only affect metallic layer results, because oxide pool is caused by the increased density of the metallic layer attributed to
heat fluxes are independent of metallic layer mass assumptions. As a transfer of uranium metal that was reduced from uranium oxide
shown in Fig. 12, CHF ratios at vessel locations in contact with the by interaction with Zirconium. Guéneau et al. (1998), Cheynet et al.
metallic layer approach unity for steel masses equal to 15,000 kg. (2002), Almyashev et al. (2002), Mazurin and Gusarov (2002) and
When the steel masses are below this value, CHF ratios are greater Smirnov et al. (2007) have conducted on the thermal and physico-
than unity. This may cause the vessel failure. Particularly, when the chemical processes during the molten pool formation in numerical
steel mass equals to 3000 kg that was the minimum steel mass cal- way. Seiler et al. (2007) proposed a methodology to assess corium
mass inventories and calculate the mass of steel in heavy metallic
layer with different fraction of Zr oxidation for a 1000 MW PWR.
For point estimate calculations of melt Configuration II, the
masses of the corium inventories for base case calculations as
discussed in the previous sections were adopted in this section
(Table 5). The advantage of IVRASA model is that it is convenient
to change the mass of unoxidized uranium in the bottom layer by
altering the value of the fraction of uranium that is in the oxide
form (fU ). The formation of fU is defined as:
mU 270
fU = 1 − (24)
MUO2 238

where MUO2 is the total mass of UO2 , and mU the mass of uranium
in the bottom layer.
Note that the formulas of the masses of uranium, steel and zir-
conium in the bottom layer remain the same with those given by
Esmaili and Khatib-Rahbar (2005). IVRASA calculation results of
Configuration II are shown as follows under different values of fU .
As discussed in previous sections, the reason for the stratifica-
tion is because the density of the bottom metallic layer composed of
Fig. 13. Comparison of CHF ratios with additional heat sources in the metallic layer. U–Zr–Fe is greater than that of the oxide pool. As shown in Table 6
Y.P. Zhang et al. / Nuclear Engineering and Design 240 (2010) 2023–2033 2031

Table 6
Portion of results of IVRASA calculations for Configuration II.

fU 0.85 0.9 0.95


mU (kg) 8,762 5,841 2921
mZr-bot (kg)b 10,143 5,762 1382
mtotal (kg)c 21,905 14,603 7303
h (kg/m3 )d 8,488 (8481)a 8,592 (8584) 8913 (8909)
o (kg/m3 )e 8,353 (8358) 8,372 (8377) 8389 (8392)
a
The values in the parentheses were calculated by Esmaili.
b
mZr-bot represents the total mass of the bottom layer.
c
mtotal represents the total mass of the bottom layer.
d
h represents the density of the bottom layer.
e
o represents the density of the oxide layer.

Fig. 16. Heat flux as function of position on the lower head.

the bottom of the vessel lower head, and to the increase of the mass
of Zr in the light metallic layer (Fig. 15) that slightly mitigates the
focusing effect on the vessel locations adjacent to the light metallic
layer. The heat flux in light metallic layer is higher than that of base
case (Fig. 9) due to the portion of metal partitioning into the lower
head. It can be seen from Fig. 17 that the heat flux ratio for all the
calculations is well below the unity, so the increased focusing effect
in the light metallic layer due to portion of metal partitioning to the
bottom layer is slightly small.
Although IVRASA predicts that CHF ratios remain below unity
Fig. 14. The variation of density with respect to fU .
for Configuration II under different values of fU , there are poten-
tial factors affecting heat loads in the heavy metallic layer, such as
the inter-metallic reactions, oxidation in the heavy metallic layer.
and Fig. 14, we can see that densities of oxide layer and heavy Chevalier et al. (2004) performed a new thermodynamic modeling
metallic layer calculated by IVRASA are in agreement with these of the O–U–Zr system and suggested that the solubility of the oxy-
calculated by Esmaili, and the density of the bottom layer is greater gen existed in uranium–zirconium liquid. However, there are few
than that of the oxide layer. It can be seen from Fig. 14 that the investigations on the impact of the inter-metallic reactions and the
density of the bottom layer increases with an increase of the value oxidation in the heavy metallic layer. It is hard to determine the
of fU . The density of Zr is great smaller than that of uranium, so this quantifications of the impact of the inter-metallic reactions and the
may be due to the decrease of the mass of Zr in the bottom layer, oxidation in the heavy metallic layer, so the impact of the inter-
as shown in Fig. 15. metallic reactions and the oxidation in the heavy metallic layer
As shown in Figs. 16 and 17, we can see that the heat flux at needs additional experimental and analytic studies.
locations adjacent to the heavy metallic layer increases with an
increase of the value fU , and varies inversely with the value of fU in
3.4. Sensitivity analysis of CHF correlations for IVR analysis
the light metallic layer. It is due to the decrease of the total mass in
the heavy metallic layer (Fig. 15) that slightly focuses heat loads to
ERVC for AP1000 may not be successful without an enhanced
insulation design or other enhanced measures. It can be seen from
Fig. 18 that there is a very small margin of IVR calculation of AP1000
for Configuration I using AP600 CHF correlation (Eq. (22)). However,
if modified CHF correlation for AP1000 is used, there is a large ther-
mal margin of IVR calculation. Therefore, different margins of IVR
calculations could be obtained using different CHF correlations.
Theofanous et al. (2003) observed a strong effect of surface
condition in their experiments at the UPLU-V facility for AP1000.
They used the sand particles to roughen the test section sur-
faces modified the surfaces molecules properties, and found that
the molecular deposition of aluminum could be dissolved by de-
ionized water in some of tests, leading to a significant decrease
in the CHF values in subsequent tests. However, this degradation
effect was not observed in those tests using tap water rather than
de-ionized water. Therefore, there exists a large uncertainty in the
UPLU-V experimental data. Rempe et al. (2008) also suggested that
there was a large uncertainty in the CHF data. Moreover, recent
studies argued that nanofluids could be used to enhance the IVR
Fig. 15. The variation of mass with respect to fU . capability in the severe accident management strategy. Buongiorno
2032 Y.P. Zhang et al. / Nuclear Engineering and Design 240 (2010) 2023–2033

AP1000. According to the benchmark calculations of AP600 and


calculations of AP1000, some conclusions are shown as follows.

(1) The results of benchmark calculations of AP600 by IVRASA were


consistent with those of the UCSB and INEEL. This indicated the
applicability and accuracy of IVRASA and it could be applied to
predict the thermal response of various molten configurations.
(2) Then, IVRASA was applied to calculate two core melt config-
urations of AP1000. The calculation results of Configuration I
indicated that the heat flux remained below the CHF. How-
ever, the sensitivity calculations showed that the heat flux in
the metallic layer could exceed the CHF because of the focus-
ing effect due to the metallic layer thinning. On the other hand,
the calculation results of Configuration II suggested that the
thermal failure of the lower head at the bottom location was
highly unlikely, but the heat flux in light metallic layer could
be higher than that of base case due to the portion of metal
partitioning into the lower head. For Configuration II, the heat
Fig. 17. Heat flux ratio as function of position on the lower head. flux ratio for all the calculations is well below the unity, there-
fore the increased focusing effect in the light metallic layer due
to portion of metal partitioning to the bottom layer is slightly
small.
(3) However, there are potential factors affecting heat loads in
the heavy metallic layer, such as the inter-metallic reactions,
oxidation in the heavy metallic layer. However, there are few
investigations on the impact of the inter-metallic reactions and
the oxidation in the heavy metallic layer. It is hard to deter-
mine the quantifications of the impact of the inter-metallic
reaction and the oxidation in the heavy metallic layer, there-
fore the effect of the inter-metallic reactions and the oxidation
in the heavy metallic layer needs additional experimental and
analytic studies.
(4) This work also investigated the effect of the uncertainties of the
CHF correlations on the analysis of IVR. The existing CHF corre-
lation for IVR analysis of AP1000 is based on 2D UPLU-V tests.
However, there exists a large uncertainty in the UPLU-V exper-
imental data. Therefore, 3D ERVC tests should be carried out
for AP1000 CHF correlation under reactor conditions, or some
Fig. 18. Comparison of CHF ratios using different CHF correlations. existing 3D experimental data should be used to IVR analysis of
AP1000.

et al. (2009) suggested that CHF values could be improved by 40%


using a nanofluid for AP1000 design. Modifications of the UPLU-V References
facility also could significantly improve the CHF limits (Theofanous
et al., 2003). The factor in Eq. (23) should be theoretically enlarged Almyashev, V.I., Gusarov, V.V., Mazurin, O.V., 2002. Proceedings of a Scientific and
Practical Workshop on New Achievements in Chemistry and Materials Technol-
under these conditions. Therefore, Eq. (23) used for IVR analysis of ogy, St. Petersburg, p. 62 (in Russian).
AP1000 was conservative. Asfia, F.J., Dhir, V.K., 1996. An experimental study of natural convection in a vol-
However, the UPLU-V facility is a 2D facility and cannot be used umetrically heated spherical pool bounded on top with a rigid wall. Nuclear
Engineering and Design 163, 333–348.
to adequately simulate the 3D heat transfer to the annular channel Asmolov, V.G., et al., 2004. Partitioning of U, Zr and Fe between molten oxidic and
between the reactor vessel and the surrounding insulation struc- metallic corium. In: Proceedings of MASCA Seminar 2004, Aix-en-Provance,
ture. SBLB facility is a 3D facility that can simulate the 3D heat France.
Asmolov, V.G., et al., 2007. Molten corium stratification and component par-
transfer and two-phase boundary layer flow behavior in the annular titioning. In: Proceedings of MASCA Seminar 2007, Cadarache, France.
between the reactor vessel and the surrounding insulation. Cheung Gusarov, V.V., Almjashev, V.I., Khabensky, V.B., Beshta, S.V., Granovsky, V.S.,
et al. (2003) and Yang et al. (2004, 2005) investigated two design 2006. Distribution of components between immiscible melts of a system
under nonisothermal conditions. Glass Physics and Chemistry 32 (6), 638–
modifications that had the potential to enhance the margin for IVR
642.
via ERVC: an enhanced insulation and coatings on reactor vessel Bechta, S.V., et al., 2004. New experimental results on the interaction of molten
external surfaces using SBLB facility for APR1400. corium with reactor vessel steel. In: Proceedings of ICAPP’04, Pittsburgh, PA,
USA, June 13–17, Paper 4114.
Therefore, 3D ERVC tests should be carried out for AP1000 CHF
Bechta, S.V., et al., 2006a. Phase diagram of the ZrO2 –FeO system. Journal of Nuclear
correlation under reactor conditions, or some existing 3D experi- Materials 348 (1/2), 114–121.
mental data should be used to IVR analysis of AP1000, such as data Bechta, S.V., et al., 2006b. Experimental study of interactions between suboxidized
or correlations of SBLB tests. corium and reactor vessel steel. In: Proceedings of ICAPP’06, Reno, NV, USA, June,
Paper 6054.
Bechta, S.V., et al., 2006c. Phase relations in the ZrO2 –FeO system. Russian Journal
4. Conclusions of Inorganic Chemistry 51 (N 2), 325–331.
Bechta, S.V., et al., 2007a. Phase transformation in the binary section of the
UO2 –FeO–Fe system. Radiochemistry 49 (1), 20–24.
A point estimate procedure IVRASA was developed for model- Bechta, S.V., et al., 2007b. Phase diagram of the UO2 –FeO1+x system. Journal of
ing the steady-state endpoint of two core melt configurations of Nuclear Materials 362 (1), 46–52.
Y.P. Zhang et al. / Nuclear Engineering and Design 240 (2010) 2023–2033 2033

Buongiorno, J., Hu, L.W., et al., 2009. A feasibility assessment of the use of nanofluids conditions taking into account the combined processes of the vessel creep
to enhance the in-vessel retention capability in light-water reactors. Nuclear and the molten pool natural convection. Nuclear Engineering and Design 191,
Engineering and Design 239, 941–948. 31–52.
Cheung, F.B., Haddad, K.H., Liu, Y.C., 1997. Critical Heat Flux (CHF) phenomenon on Mayinger, F., Jahn, M., Reineke, H., Steinberner, U., 1975. Examination of thermo-
a downward facing curved surface. NUREG/CR-6507, PSU/ME-97-7321. hydraulic processes and heat transfer in a core melt. Final Report BMFT RS 48/1.
Cheung, F.B., et al., 2003. On the enhancement of external reactor vessel cooling of Technical University, Hannover, W. Germany, As reviewed by F. A. Kulachi, Ohio
high-power reactors. In: Tenth International Topical Meeting on Nuclear Reactor State University, for the US NRC, March 31, 1976.
Thermal Hydraulics, NURETH10, Seoul, Korea, October 5–11, Paper G00403. Mazurin, O.V., Gusarov, V.V., 2002. Physics and Chemistry of Glasses 28, 74 (in
Chevalier, P.Y., et al., 2004. Progress in the thermodynamic modelling of the O–U–Zr Russian).
ternary system. Computer Coupling of Phase Diagrams and Thermo-chemistry Power, D., Behbahani, A., 2004. Density stratification of core debris. In: Proceeding
28, 15–40. of the OECD MASCA Seminar, Paris.
Cheynet, B., Chevalier, P.Y., Fischer, E., 2002. Thermosuite. Calphad 26 (2), 167–174. Rempe, J.L., et al., 1997. Potential for AP600 in-vessel retention through ex-vessel
Chu, T.Y., Bentz, J.H., Slezak, S.E., Walter, F., Pasedag, 1997. Ex-vessel boiling exper- flooding. Technical evolution report, INEEL/EXT-97-00779.
iment: laboratory- and reactor-scale testing of the flooded cavity concept for Rempe, J.L., Suh, K.Y., Cheung, F.B., Kim, S.B., 2008. In-Vessel Retention Strategy for
in-vessel core retention Part||: reactor-scale boiling experiments of the flooded High Power Reactors. Technical report, INEEEL/EXT-04-025621.
cavity concept for in-vessel core retention [J]. Nuclear Engineering and Design Seiler, J.M., Fouquet, A., Froment, K., Defoort, F., 2003. Theoretical analysis for corium
169, 89–99. pool with miscibility gap. Nuclear Technology 141, 233–243.
Churchill, S.W., Chu, H.H.S., 1975. Correlating equations for laminar and turbulent Seiler, J.M., et al., 2007. Consequences of material effects on in-vessel retention.
free convection from a vertical plate. International Journal of Heat Mass Transfer Nuclear Engineering and Design 237 1752–1758.
18, 1323–1329. Smirnov, S.A., et al., 2007. DNS of molten corium pool inductively heated in a cold
Dinh, T.N., Tu, J.P., Theofanous, T.G., 2004. Two-phase natural circulation flow in crucible. In: Proceedings of MASCA Seminar 2007, Cadarache, France.
AP1000, in-vessel retention-related ULPU-V facility experiments. In: Proceed- Theofanous, T.G., et al., 1994a. Critical heat flux through curved, downwards facing
ings of the 2004 International Congress on Advances in Nuclear Power Plants, thick wall. In: Procceding of the OECD/CSNI/NEA Workshop on large Molten Pool
ICAPP’04, pp. 928–938. Hear Transfer, Gernoble, France.
Esmaili, H., Khatib-Rahbar, M., 2004. Analysis of in-vessel retention and ex-vessel Theofanous, T.G., et al., 1994b. Experience from the first two integrated approaches
fuel coolant interaction for AP1000. Energy Research, Inc., ERI/NRC 04-21, to in-vessel retention external cooling. In: Procceding of the OECD/CSNI/NEA
NUREG/CR-6849. Workshop on large Molten Pool Hear Transfer, Gernoble, France.
Esmaili, H., Khatib-Rahbar, M., 2005. Analysis of likelihood of lower head failure and Theofanous, T.G., et al., 1996a. In-vessel cool-ability and retention of a core melt,
ex-vessel fuel coolant interaction energetics for AP1000. Nuclear Engineering DOE/ID-10460, Revised October.
and Design 235, 1583–1605. Theofanous, T.G., et al., 1996b. The first results from the ACOPO experiment. In:
Globe, S., Dropkin, D., 1959. Natural-convection heat transfer in liquids confined by Proceedings of the Topical Meeting on Probabilistic Safety Assessment (PSA’96),
two horizontal plates and heated from below [J]. Heat Transfer 81, 24. Park Soty, Utah.
Guéneau, C., Dauvois, V., Pérodeaud, P., Gonella, C., Dugne, O., 1998. Liquid immis- Theofanous, T.G., et al., 1997. The first results from the ACOPO experiment. Nuclear
cibility in a (U, Zr, O) model corium. Journal of Nuclear Materials 254, 158–174. Engineering and Design 169, 49–57.
Gusarov et al., 2006. Distribution of components between immiscible melts of a Theofanous, T.G., et al., 2003. Limits of Coolability in the AP1000-Related ULPU-2400
system under non-isothermal conditions, Glass Physics and Chemistry 32 (6) Configuration V Facility, Paper G00407. In: Tenth International Topical Meet-
(2006), pp. 638–642. ing on Nuclear Reactor Thermal Hydraulics, NURETH10, October 5–11, Seoul,
Gusarov, V.V., Almjashev, V.I., Khabensky, V.B., Beshta, S.V., Granovsky, V.S., 2007. Korea.
Influence of the temperature difference at immiscibility liquids interface on their Theofanous, T.G., et al., 2004. In-Vessel Retention Technology Development and
phase instability. In: Proceedings of MASCA Seminar 2007, Cadarache, France. Use for Advanced PWR Design in the USA and Korea. Technical Report, FG07-
Henry, R.L., Fauske, H., 1993. External cooling of a reactor vessel under severe acci- 02RL14337.
dent conditions. Nuclear Engineering Design 139, 31–43. Yang, J., et al., 2004. CHF enhancement by vessel coating for external reactor vessel
Kiselev, N.P., et al., 2007. Experimental methodologies and installations used under cooling. In: Proceedings of ICAPP’04, Pittsburgh, PA USA, June 13–17.
the MASCA2 Project. In: Proceedings of MASCA Seminar, Cadarache, France. Yang, J., et al., 2005. Critical heat flux for downward-facing boiling on a coated
Knudson, D.L., Rempe, J.L., 2002. In-vessel retention modeling capabilities of hemispherical vessel surrounded by an insulation structure. In: Proceedings of
SCDAP/RELAP5-3D. In: Procceding of ICONE10-22754, Arlington, VA, USA. ICAPP’05, Seoul, Korea, May 15–19.
Kolev, N.I., 1993. Sicherheitsbericht des geplanten WWER-640 (W-407) Auswer- Yuan, Z., Zavisca, M., Khatib-Rahbar, M., 2003. Impact of the reactor pressure vessel
tung: system für Erhaltung der Schmelze im RDB. Schmelzfänger, KWU insulation on the progression of severe accidents in AP1000. Energy Research,
NA-M/93/016, Project GUS-Kooperation. Inc., ERI/NRC 03-205.
Loktinonov, V.D., Mukhtarov, E.S., Yaroshenko, N.I., Orlov, V.E., 1999. Numerical Zavisca, M., Yuan, Z., Khatib-Rahbar, M., 2003. Analysis of selected accident scenarios
investigation of the reactor pressure vessel behavior under severe accident for AP1000. Energy Research, Inc., ERI/NRC 03-201.

View publication stats

You might also like