Professional Documents
Culture Documents
To cite this article: Chihiro Yanagi, Michio Murase, Yoshitaka Yoshida & Takayoshi Kusunoki
(2015) Prediction of temperature and water level in a spent fuel pit during loss of all
AC power supplies, Journal of Nuclear Science and Technology, 52:2, 193-203, DOI:
10.1080/00223131.2014.939597
ARTICLE Translation
Prediction of temperature and water level in a spent fuel pit during loss of all AC power supplies†
Chihiro Yanagia∗ , Michio Murasea , Yoshitaka Yoshidaa and Takayoshi Kusunokib1
a
Technical Support Project, Institute of Nuclear Technology, Institute of Nuclear Safety System, Inc., 64 Sata, Mihama-cho,
Mikata-gun, Fukui 919-1205, Japan; b Nuclear Power Division, The Kansai Electric Power, Co., Inc., 13-8 Goichi, Mihama-cho,
Mikata-gun, Fukui 919-1141, Japan
(Received 5 June 2014; accepted final version for publication 23 June 2014)
A prediction method for water temperature in a spent fuel pit of a pressurized water reactor (PWR) has
been developed to calculate the increase in water temperature during the shutdown of cooling systems. In
this study, the prediction method was extended to calculate the water level in a spent fuel pit during loss
of all AC power supplies, and predicted results were compared with measured values of spent fuel pools
in the Fukushima Daiichi Nuclear Power Station. The calculations gave reasonable results, but overesti-
mated the decreasing rate of the water level and the water temperature. This indicated that decay heat was
overestimated and evaporation heat transfer from the water surface was underestimated. Results of calcu-
lations with 80% decay heat and 155% (Unit 4 pool) or 230% (Unit 2 pool) evaporation heat flux were in
good agreement with measured values. The data-fitted evaporation heat fluxes agreed rather well with the
evaporation heat transfer correlation proposed by Fujii et al.
Keywords: spent fuel pit; loss of all AC power supplies; decay heat; evaporation; water temperature; water
level; the Fukushima Daiichi Nuclear Power Station
1. Introduction are decay heat and evaporation heat transfer from the
A spent fuel pit (or spent fuel pool, hereafter SFP) water surface to air in the fuel handling building.
in a nuclear power plant is generally equipped with two The normal water level in the SFP is about 7 m above
cooling systems to remove decay heat from spent fuel the top of the spent fuel racks and this is a large amount
assemblies and to keep the SFP water at a low tempera- of water to provide gamma radiation shielding. There-
ture. Above the water surface of the SFP, an air curtain fore, the water temperature increasing rate and water
is formed by air ventilation systems, which are normally level decreasing rate in the SFP are very small, and these
operated by external AC power supplies or by emergency issues have been rarely reported. For decommissioning
diesel generators when external AC power supplies fail. of a nuclear power plant, the USNRC has reported the
Evaluating the increase in the SFP water temperature time limit and the cooling period needed to ensure no
and the time until the upper limit of the SFP water tem- damage to spent fuel assemblies when they are exposed
perature is reached becomes important during the shut- in the air during loss of all AC power supplies [1]. How-
down of cooling systems. On the other hand, evaluating ever, the report calculations focused on the cooling pe-
the water level in the SFP becomes important during loss riod and an adiabatic condition was used. The adiabatic
of all AC power supplies including the emergency diesel condition means that heat losses from the SFP water to
generators. Important uncertainties for prediction of the air and concrete are neglected. The adiabatic condition
water temperature and water level in the SFP during the also means that all decay heat is stored in the water be-
shutdown of cooling systems and air ventilation systems fore reaching 100 ◦ C or all decay heat is consumed in
∗
Corresponding author. Email: yanagi.chihiro@inss.co.jp
1
Present address: Institute of Nuclear Technology, Institute of Nuclear Safety System, Inc., 64 Sata, Mihama-cho, Mikata-gun,
Fukui 919-1205, Japan.
†
This is a translation of an article originally published in Japanese in Transactions of the Atomic Energy Society of Japan
<http://dx.doi.org/10.3327/taesj.J11.050>. Taylor & Francis do not warrant the accuracy of the translation.
C 2014 Atomic Energy Society of Japan. All rights reserved.
194 C. Yanagi et al.
1.4
1.2
1
Temperature increase
(T -T 0)c /(T e -T 0)m
0.8
0.6
0.4
Case1
0.2 Case2
0
0 0.2 0.4 0.6 0.8 1
Time: (t -t 0)c /(t e -t 0)m
Figure 4. Heat transfer model for loss of all AC power
Figure 3. Increase of water temperature during the shutdown supplies.
of cooling systems.
where Qa (kJ) is the enthalpy of air in the fuel handling Table 1. Calculation conditions [5].
building, Qout (kW) is the heat transfer outside the fuel
Initial conditions
handling building, Ta (◦ C) is the average air tempera- Vw,0 (m3 ) Hw,0 (m) Tw,0 (◦ C) Ta,out (◦ C) Tw,in (◦ C)
ture in the fuel handling building, and Va (m3 ) is the vol-
ume of the fuel handling building. The third term on the 1390 11.5 30 10 10
right side is the increasing rate of enthalpy of water. Qout
Note: Vw,0 : initial water volume (m3 ), Hw,0 : initial water level (m),
was calculated using a turbulent natural convection heat Tw,0 : initial water temperature (◦ C), Ta,out : outdoor temperature (◦ C),
transfer outside of the fuel handling building neglecting and Tw,in : temperature of injected water (◦ C).
effects of wind. The function f was obtained from the
relationship between air temperature and its enthalpy
(kJ/m3 ) with 100% relative humidity. Physical properties 4.1. Calculation conditions
were calculated using functions of temperature. Using Table 1 shows the calculation conditions. The initial
the term (QD − QC ) instead of QE in Equation (5) makes water volume Vw,0 and initial water level Hw,0 were given
it possible to calculate the heat balance after the water in the 1F-Report [5]. The initial water temperature Tw,0 ,
temperature reaches the saturated temperature of Tw = outdoor temperature, Ta,out , and temperature of injected
100 ◦ C (heat is not stored in water after boiling). Equa- water, Tw,in , were the same values as assumed in the
tion (5) was introduced into the water temperature pre- 1F-Report [5].
diction system [4] during shutdown of cooling systems. Decay heats predicted by ORIGEN 2.2 were written
for values at the beginning (t = 0) and three months later
(t = 90 days) in the 1F-Report [5]. In this study, decay
heat was calculated using the correlation recommended
3.2. Prediction of water level
by ANS-5.1-1973 [11], in which the coefficient and index
The calculation method of the water level was also were fitted to give decay heat at t = 0, 90 days in the
the same as the method for the shutdown of cooling 1F-Report [5].
systems [4], and the following mass balance equation for In the Unit 4 pool, water levels were measured during
a one-region model was used: the no water injection periods; the calculation accuracy
of decay heat was discussed by comparison of the calcu-
lated water level decreasing rate with the measured value.
d Mw dTw
h f g = − (Q D − QC ) + Cpw Mw , In the Unit 2 pool, the calculation accuracy of evap-
dt dt
oration heat flux, during the period a high water level
Mw
Hw = , (6) was kept by periodical water injection, was discussed by
ρw Aws comparison of the calculated water temperature with the
measured value.
where Hw (m) is the water level, Aws (m2 ) is the water
surface area, and ρw (kg/m3 ) is the water density.
4.2. Unit 4 pool
Decay heats in the Unit 4 SFP were reported [5] as
2.26 MW (t = 0) and 1.58 MW (t = 90 days). In the
4. Validation for prediction of water temperature
calculation for the Unit 4 pool, the water temperature
and level
and level were calculated under the conditions of the in-
The authors could not find any documents describ- tact building and the function of the presented predic-
ing the measurement of the water temperature and level tion system was considered first. Next, calculated wa-
in a SFP during loss of all AC power supplies. Therefore, ter levels during no water injection were compared with
comparison and validation were done by using measure- measured values under the condition that air tempera-
ment results [5] in the Fukushima Daiichi Nuclear Power ture equaled the outdoor temperature of 10 ◦ C for the
Station (1F) (hereafter called the 1F-Report). The objec- damaged building.
tive of calculations with the one-region model for SFPs Figure 5 shows the calculated transient behaviors
in 1F was not to reproduce the thermal-hydraulics but (heat gain, temperatures, water level from the top of
to validate prediction of the water temperature and the the racks, and evaporation heat transfer coefficient) af-
level. In 1F, some buildings were damaged and water in- ter loss of all AC power supplies with the intact build-
jection was done for a long time. ing. After about three days, Tw was almost 100 ◦ C
In the case of a damaged building, QE was calculated (Figure 5(b)), QE rapidly increased due to boiling
under the condition that Ta in Figure 4 equals the out- (Figure 5(a)), and Hw rapidly decreased (Figure 5(c)).
door temperature of Ta,out . Water injection was done pe- QC in Figure 5(a) increased once due to the water tem-
riodically, but in this study, water injection was continu- perature increase and then decreased due to the water
ously assumed at the same time-averaged flow rate and level decrease and increase of TC (QC was calculated for
subcooled heat was subtracted from QD in calculations the SFP volume below the water level, not including the
of the water temperature. volume above the water level). Figure 5(b) shows that the
198 C. Yanagi et al.
Figure 5. Transient behavior after loss of all AC power supplies (Unit 4 pool, intact building).
calculated average water temperatures with and without calculation, but except for the case of small decay heat
heat losses were almost the same, because QD was much (refer to Figure 11(b)) they had only a small effect on the
bigger than heat losses to concrete and air, QC + QE . water level calculation.
Tws values after about three days agreed well with the Figure 6 compares the calculation results during no
measured values. Water level in Figure 5(c) is the level water injection with measured values; the conditions
from the top of racks. Increasing water level in the begin- were that air temperature was equal to the outdoor tem-
ning was due to thermal expansion; water level without perature (Ta = 10 ◦ C) with the damaged building. Wa-
heat losses Hw,a was slightly overestimated, even though ter temperature (Figure 6(a)) was measured by a ther-
during the decreasing water level Hw,a might be generally mocouple that was hung from the water injection crane,
underestimated since evaporation was ignored. Because but the depth of the measuring point was not clear. The
QC was relatively small, calculated water levels with and water temperature distribution by 3D calculation is nec-
without heat losses were almost the same. The calculated essary for estimation of the measuring point location.
water level agreed well with the measured value made Measured water temperatures were between Tw and Tws ,
by visual observation from a helicopter. It took about and so water temperature calculation was considered to
12.7 days until the water level decreased to the top of be reasonable. Heat transfer form was boiling (Tw =
the racks without water injection. The calculated evap- 100 ◦ C) during this period in the calculation, and so it
oration heat transfer coefficient in Figure 5(d) rapidly was hard to validate Equation (2) for evaporation heat
increased due to the increases in average water temper- fluxes. Calculations overestimated the decreasing rate of
ature and water surface temperature, and its value be- the water level by about 20% as shown in Figure 6(b).
came 133 W/(m2 ·K) at Tws = 87 ◦ C. Calculated evapora- The decreasing rate of the water level calculated with
tion heat transfer coefficient and evaporation heat trans- 80% decay heat (0.8 QD ) agreed well with the measured
fer rate had a significant effect on the water temperature value as shown in Figure 7. This showed that ORIGEN
Journal of Nuclear Science and Technology, Volume 52, No. 2, February 2015 199
Figure 6. Transient behavior after loss of all AC power supplies (Unit 4 pool, damaged building).
2.2 overestimated decay heat. The same result was shown culated water temperatures with measured values during
in Case 2 of Figure 3 and for this Case 2, the increasing the period that the water level was fully kept by periodi-
rate of the water temperature calculated with 0.75 QD cal water injection.
agreed well with the measured value. Figure 8 shows the calculated transient behaviors
(heat gain, temperatures, water level from the top of
racks, and evaporation heat transfer coefficient) under
4.3. Unit 2 pool the condition of the intact building. These were almost
Decay heat in the Unit 2 pool was reported [5] as 0.62 the same behaviors as for the Unit 4 pool. Initial decay
MW (t = 0) and 0.52 MW (t = 90 days). This build- heat of the Unit 2 pool was about 27% that of the Unit 4
ing was only partially damaged, so in the calculation for pool. The time until the water level decreased to the top
the Unit 2 pool, evaporation heat transfer was calculated of racks in the Unit 2 pool (Figure 8(c): about 46.6 days)
under the condition of the intact building (evaporation took about 3.7 times longer than the time in the Unit 4
heat transferred to outdoor due to discharge of mixture pool. Decay heat QD (Figure 8(a)) balanced with heat
of steam and air in the building). Equation (2) for evap- losses to concrete and air, QC + QE , at the time when
oration heat fluxes was discussed by comparison of cal- the average water temperature reached about 88 ◦ C and
the water surface temperature (Figure 8(b)) was about
79 ◦ C. Therefore, the boiling status (100 ◦ C) was not es-
tablished.
After 10 days had passed from loss of all AC power
supplies, water was injected periodically, and after
40 days, the pool was evaluated as being at a high water
level. Calculated results after 40 days are shown in
Figure 9. In the calculations, a constant water flow was
continuously assumed to keep the water level high. The
temperature of the injected water was 10 ◦ C. The differ-
ence between decay heat QD and heat losses to concrete
and air, QC + QE , in Figure 9(a) is the subcooled heat of
injected water. The measured temperatures during this
period changed between about 50 and 70 ◦ C due to the
periodical water injection. Water temperature of 50 ◦ C
was evaluated as air temperature [5], i.e. a thermometer
was exposed in the air due to the decrease in the water
level. Maximum temperature (about 70 ◦ C), which was
evaluated as water temperature, is shown in Figure 9(b).
Calculated average water temperatures and water
Figure 7. Water level (Unit-4 pool, damaged building, surface temperatures were higher than measured values.
0.8 QD ). The reasons for overestimating water temperature
200 C. Yanagi et al.
Figure 8. Transient behavior after loss of all AC power supplies (Unit 2 pool, intact building, without water injection).
Figure 9. Transient behavior after loss of all AC power supplies (Unit 2 pool, intact building, with water injection).
Journal of Nuclear Science and Technology, Volume 52, No. 2, February 2015 201
Figure 10. Effects of decay heat (QD ) and evaporation heat flux (qE ) on water temperatures (Unit 2 pool).
were effects of the damaged building, overestimation calculated and measured water temperatures. Calculated
of decay heat, and underestimation of evaporation water surface temperatures agreed well with measured
heat flux. values under the conditions of 0.8QD and 130% evapo-
For the damaged building and the same air temper- ration heat flux (1.3qE ) as shown in Figure 10(a). Calcu-
ature as the outdoor temperature (Ta = 10 ◦ C), the cal- lated average water temperatures agreed well with mea-
culated water temperature decreased by only 4 ◦ C–5 ◦ C. sured values under the conditions of 0.65QD and 1.7qE,
So, the main reasons for overestimating the water tem- as shown in Figure 10(b).
perature were overestimation of decay heat and underes- The magnification number for evaporation heat flux
timation of evaporation heat flux. In the calculation for was 2.3 to obtain good agreement of calculated average
the Unit 4 pool, the water level decreasing rate calcu- water temperatures with measured values under the con-
lated with 80% decay heat (0.8QD ) gave good agreement dition of 0.8QD, as shown in Figure 11(a). In this case,
between the calculated and measured values. However, the water level calculated with heat losses to concrete
overestimation of water temperature could not be clar- and air was lower than that calculated with the adiabatic
ified under the calculation condition of 0.8QD for the condition, as shown in Figure 11(b). Evaporation under
Unit 2 pool. the condition of highly subcooled water caused this early
Figure 10 shows the surveillance results of calcula- decrease in the water level. The smaller decay heat and
tion conditions which gave good agreement between the the bigger evaporation (low water temperature) caused
Figure 11. Effects of decay heat (QD ) and evaporation heat flux (qE ) (Unit 2 pool, 0.8 QD , 2.3 qE ).
202 C. Yanagi et al.
the lower water level. The results indicated that the adi-
abatic calculations were not always conservative.
4.4. Discussion
The decreasing rate of the water level was overesti-
mated in the case of 100% decay heat (1.0QD ) predicted
by ORIGEN 2.2 (Figure 6(b)), but the deceasing rate of
the water level for the 80% decay heat (0.8QD ) predicted
by ORIGEN 2.2 demonstrated good agreement with the
measured water levels (Figure 7).
These results showed that decay heat predicted by
ORIGEN 2.2 was overestimated by 25% (1/0.8 = 1.25)
for long-time cooled fuel assemblies. The magnification
number for evaporation heat flux was 1.55 to obtain
good agreement between calculated average water tem-
peratures and measured values under the condition of
0.8QD, as shown in Figure 12. This result showed that
Figure 13. Comparison of evaporation heat fluxes.
evaporation heat transfer predicted by Equation (2) was
underestimated by 35% (1/1.55 = 0.65).
Prediction error of decay heat differs according to
the specifications and the status (fuel type, burn-up, separate overestimation of decay heat and underestima-
cooling period, etc.) of the spent fuel assembly. If 25% tion of evaporation heat flux quantitatively.
overestimation of decay heat was adopted in the calcula- It was considered that the underestimation of evapo-
tion for the Unit 2 pool, the same as for the Unit 4 pool, ration heat flux (35%–57%) was a reasonable prediction
Equation (2) underestimated evaporation heat fluxes by error under the uncertainty of the correlation for turbu-
57% (1/2.3 = 0.43). Cooling time of spent fuel assem- lent natural convection heat transfer used in the deriva-
blies in the Unit 2 pool was longer than that in the Unit tion of Equation (2) because the measured heat gains in
4 pool; consequently, the prediction error of decay heat forced convection evaporation were about three times of
calculated by ORIGEN 2.2 in the Unit 2 pool would be the values predicted from the analogy [2].
bigger than that in the Unit 4 pool. Under the assump- Figure 13 compares evaporation heat fluxes. The data
tion that decay heat was overestimated by 54% (1/0.65 = marked as 1F in Figure 13 were calculated from the data
1.54) in the Unit 2 pool, as shown in Figure 10(b), Equa- in Figures 10(b), 11(a), and 12, and agreed well with
tion (2) underestimated evaporation heat fluxes by 41% the correlation proposed by Fujii et al. [12]. Fujii et al.
(1/1.7 = 0.59). From these discussions, it is difficult to measured evaporation heat fluxes with surface areas of
0.034 and 0.29 m2 under pressures of 0.1–0.32 MPa in
stagnant air conditions. Steam pressure ratio (Pso /P) and
heat flux values of Fujii et al. were within the conditions
of this study, but water temperatures (Tw ≥ 100 ◦ C) and
steam pressure (Pso ) were high. Therefore, the correla-
tion of Fujii et al. would not be applicable to the calcu-
lation of the water temperature in the SFP. However, it is
interesting that predicted evaporation heat fluxes agreed
well with data-fitted evaporation heat fluxes which were
adjusted with the water temperature of the Uuit 2 and
Unit 4 pools. These data-fitted evaporation heat fluxes
are affected by the uncertainty of the decay heat calcula-
tion. It is necessary to collect more data on evaporation
heat transfer and to improve the correlation.
5. Conclusions
The prediction system that was developed to predict
water temperature in an SFP after the shutdown of cool-
ing systems was extended to calculate the water level in
Figure 12. Water temperatures (Unit 4 pool, damaged the SFP during loss of all AC power supplies. The calcu-
building, 0.8QD , 1.55qE ). lated results were validated by using the measured water
Journal of Nuclear Science and Technology, Volume 52, No. 2, February 2015 203
temperatures and levels in SFPs at 1F. The following re- [3] Yanagi C, Murase M, Yoshida Y, Iwaki T, Nagae T.
sults were obtained: Evaluation of heat loss and water temperature in a
spent fuel pit. J Power Energy Sys. June 2012;6(2):51–
62.
(1) The decreasing rate of the water level with de- [4] Yanagi C, Murase M, Yoshida Y, Iwaki T, Nagae T.
cay heat predicted by ORIGEN 2.2 was overes- [Evaluation of water temperature in spent fuel pit (5)
timated. The decreasing rate of the water level development of a simple calculation method]. 2011 Fall
calculated with 80% decay heat agreed well with Meeting of the Atomic Energy Society of Japan; 2011
measured values. These results showed that de- September 19–22; Kitakyushu, Japan. p. 41. Japanese.
[5] TEPCO. Report on effects of the earthquake in
cay heat predicted by ORIGEN 2.2 was overes- the Northeastern Japan on nuclear facilities in the
timated by 25% for long-time cooled fuel assem- Fukushima Daiichi nuclear power station. The Tokyo
blies. Electric Power Company; September 2011. Japanese.
(2) The correlation for evaporation heat fluxes, [6] Nuclear Safety Research Association, Tokyo, Japan.
which was derived from the analogy between Overview of light water nuclear power stations; January
2010.
turbulent natural convection heat transfer and [7] Ludwig SB, Croff AG. Revision to ORIGEN2 – Version
mass transfer of vapor, overestimated water tem- 2.2, Transmittal memo of CCC-0371/17. TN, USA: Oak
peratures and underestimated evaporation heat Ridge National Laboratory; May 2002.
transfer by 35%–57%. [8] Katakura J, Kataoka M, Suyama K, Jin T, Ohki
(3) Evaporation heat fluxes, which were calculated S. A set of ORIGEN2 cross section libraries based
on JENDL3.3 library: ORLIBJ33, JAERI-Data/Code
to give good agreement between calculated and 2004-015. Ibaragi prefecture, Japan: Japan Atomic En-
measured values, agreed well with the correla- ergy Research Institute; 2004.
tion proposed by Fujii et al. within limited con- [9] Shibata K, Kawano T, Nakagawa T, Iwamoto O,
ditions. Katakura J, Fukahori T, Chiba S, Hasegawa A,
(4) Neglect of heat losses generally had small effects Murata T, Matsunobu H, Ohsawa T, Nakajima Y,
Yoshida T, Zukeran A, Kawai M, Baba M, Ishikawa
on the prediction of the water level decreasing M, Asami T, Watanabe T, Watanabe Y, Igashira M,
rate; however, in the case of small decay heat, Yamamuro N, Kitazawa H, Yamano N, Takano H.
the water level decreasing rate with neglect of JENDL-3.3: Japanese evaluated nuclear data library
heat losses was underestimated because enthalpy version 3 revision-3. J Nucl Sci Technol. November
stored in water was overestimated. 2002;39(11):1125–1136.
[10] Kataoka Y, Fujii T, Murase M, Tominaga K. Experi-
mental study on heat removal characteristics for water
wall type passive containment cooling system. J Nucl Sci
References Technol. October 1994;31(10):1043–1052.
[1] Collins TE, Hubbard G. Technical study of spent fuel [11] American Nuclear Society. Decay energy release rates
pool accident risk at decommissioning nuclear power following shutdown of uranium-cooled thermal reactors.
plants, NUREG-1738. US Nuclear Regulatory Commis- IL, USA: American Nuclear Society; 1973. ANS-5.1-
sion (USNRC); February 2001. 1973.
[2] Yanagi C, Murase M, Yoshida Y, Iwaki T, Na- [12] Fujii T, Kataoka Y, Murase M. Evaporation and con-
gae T, Koizumi Y. [Evaporation heat flux form hot densation heat transfer in a suppression chamber of the
water to air flow]. Trans Jpn Soc Mech Eng B. February water wall type passive containment cooling system. J
2012;78(786):363–372. Japanese. Nucl Sci Technol. May 1996;33(5):374–380.