Introduction
After the Fukushima Daiichi accident, which happened on March 11, 20111)[1], there have been many attempts to estimate fission product source term release to environment2,3)[2]. Figure 1A shows the temporal variation of ambient dose rate measured by a monitoring car at the frontal gate of Fukushima Daiichi site. Figure 1B shows the ambient dose rate measured at Fukushima Prefecture by NaI scintillation detector from March 12 to 17, 2011. Figure 2A shows water level measurement data at Fukushima Daiichi Unit 1. Figure 2B shows pressure measurement data at Fukushima Daiichi Unit 1.
We want to know which reactor unit and which radionuclide generate the peaks in Figure 1. We know already when explosions occurred from the mass media or reports1) [1]. But we don’t know when the fission products start to release to environment from each reactor. We also want to know which radionuclide contributes dominantly to each peak in Figure 1. For this end, fission product release characteristics are evaluated for Fukushima Daiichi Unit 1 as an example case in this study. The Unit 1 event chronology is summarized in Table 1. Chino et al. [3] first estimated atmospheric release of radioactivity to environment by reverse method based on the monitoring results. Tsumune et al. [4] first estimated direct release amount to Pacific Ocean by regional ocean model. Major release to ocean occurred during March 26 to April 6, 2011.
In pages 114–115 of the UNSCEAR 2013 Report [2], the purpose of source term estimation is described and were indicated two kinds of methods to estimate source term from NPP accident to environment impact.
Estimates of the source term (that is the time-dependent release of radioactive material to the environment) were made for two main purposes:
To indicate the amounts of radioactive material released to the environment;
To be used, in combination with models (for example, for atmospheric and marine dispersion), to support for inferring the dispersion and deposition of radionuclides at locations in the environment where measurements were not available or could no longer be made.
Estimates of the release of radioactive material to the atmosphere can be made using two complementary approaches: (a) one based on analyses of how an accident progressed by using severe accident progression analysis code; and (b) the other based on measurements of radioactive material in the environment and using reverse or inverse methods to reconstruct their transport through the atmosphere back to the source of the release. Both approaches are subjects to own limitations and uncertainties.
Source term release to environment of 131I and 137Cs is summarized in Table 2, which is originally summarized in pages 116–117 of UNSCEAR 2013 report.
The first approach, based on analyses of the progression of an accident, uses severe accident simulation codes, such as MELCOR4), ASTEC [5], MAAP5), etc. There are few publications based on this approach. IRSN6), NISA7,8), Hosh and Hirano [6] used accident progression method. Therefore, in this paper a trial estimation based on the severe accident progression analysis code, MELCOR code9), is presented. In-plant thermal hydraulics, core degradation analysis, and fission product transport in plant compartments are calculated with actual time sequences provided in publically available reports. As an illustration purpose, source term to environment is calculated for Fukushima Daiichi Unit 1 during first 184 hours (about 7.7 days) in detail. And then source term release to environment from Unit 2 and Unit 3 cores are also explained briefly at the end. The issues arising from the using of accident progression method are discussed.
The second approach is based on measurements of radioactive material in the environment and the use of reverse or inverse modelling: there are hundreds of publications based on this approach. Typical results are listed here. Chino et al. [3], Terada et al. [7], Katata et al. [8] used this reverse approach to estimate released amount of radionuclide from Fukushima Daiichi nuclear power plants to atmosphere. ZAMG10), TEPCO11), Mathieu et al. [9] Achim et al. [10] and Kobayashi et al. [11] used reverse method also. Stohl et al. [12], Winiarek et al. [13] and Saunier et al. [14] used inverse method to estimate 137Cs and 131I source term from Fukushima accident.
Source term release to environment of other radionuclides is also well summarized in papers of Lin et al. [15] and Steinhauser et al. [16]. The highest estimate of 131I is about 500 PBq, which is estimated by TEPCO12). The highest estimate of 137Cs is about 37 PBq, which is estimated by Stohl et al. [12].
Materials and Methods
A severe accident analysis code MELCOR version 1.8.6 is used to estimate fission product generation in core, transport in reactor pressure vessel (RPV), release to primary containment vessel (PCV), and finally to the environment. Figure 3 shows the MELCOR nodalization for (a) RPV and (b) PCV, reactor building (RB), and environment (ENV). Reactor core and lower plenum are modelled with 4 radial rings and 16 axial levels to describe the core degradation and meltdown phenomena in RPV (Figure 4). Axial level 5 represents the core support plate. The release rate of fission products are heavily related on the in-plant accident progression and boundary conditions such as timing and area of RPV and PCV leakage (or rupture). After the thermal hydraulic and core relocation analyses based on the in-plant geometry and operating conditions of the core cooling systems such as isolation condenser, external water injection, and wetwell venting operation are performed, the fission product release to environment is predicted. The initial inventory of the radionuclides in the core estimated by ORIGEN code13) and summarized in JAEA 2012–018 report [17] is used for this study. The environmental release activity up to 100 hours is estimated by multiplying this initial inventory at core by the leakage fractions to environment up to 100 h estimated by MELCOR code. The thermal hydraulic and core degradation analyses for Units 1 and 2 used for this study are described in Kim et al. [18], Kim et al. [19], and OECD/NEA BSAF Phase-I Report (2015)14,15).
When the RPV pressure reaches the safety relief valve (SRV) opening set pressure, SRV is opened and the steam (and hydrogen and fission product aerosols, if generated) in RPV is released to the suppression chamber (SC) (it also called torus or wetwell). If RPV pressure decreases below the SRV closing set pressure, the SRV is reclosed. The SRV continues to cycle to open and close up to the time of steam exhaustion in PRV. The vacuum breaker is located in vent leg (VL) which is located between the SC and the drywell (DW). If the differential pressure between SC and DW exceeds the set pressure, the vacuum breaker (VB) opens. If the steam released to SC is not completely condensed in the SC; and VB opens, the non-condensable gas such as hydrogen and fission product aerosol can move to the pedestal and drywell regions via VB. PCV is inert by nitrogen (N2) gas filling so that hydrogen burning would not occur during at-power operation.
The control volumes in RPV, PCV, and RB are summarized in Table 3. All the possible transient leakage/rupture flow paths are summarized in Table 4. Normally opened flow paths in RPV and PCV are not shown here except in RB. Beside the normal opening flow paths from RB to TB or from RB to ENV, various transient leakages are assumed due to the increasing temperature and pressure of control volumes and heat structures. One thing to remind here is the modeling of DW head flange leakage when DW pressure increases above 0.75 MPa.
Results and Discussion
1. Thermal hydraulic and core meltdown behaviors
Initial and boundary conditions of Fukushima Daiichi Unit 1 used in MELCOR analysis are summarized in Table 5.
In the plant log as shown in Figure 2B and 6C, the RPV pressure was 7 MPa before 5 hours, and it decreased to 0.6 MPa right after 5 hours. Therefore, it is assumed that the SRV stuck open occurs at 5 hours, when the steam is exhausted in RPV. In the plant log it is recorded that the PCV pressure increased up to 0.9 MPa in 12 hours and SC venting was tried at about 23 hours after the reactor scram. Hydrogen explosion occurs at about 25 hours in the top operating floor (refueling bay) of the reactor building. It is stated that the fresh water injection started at 15 hours and the sea water injection started at 28 hours in the plant log. However, it seems to be difficult for the water to reach to the core because of high pressure in the reactor vessel and also there might be various bypass routes and valves may not work appropriately in the plant due to the strong earthquake and tsunami and loss of electrical power, etc. It was discussed in the biannual BSAF meetings16,17). Therefore, such events as hydrogen explosion at 25 hours, fresh water injection at 15 hours and sea water injection at 28 hours are not modeled explicitly in the MELCOR simulation because of the above reasons. Event timings on the reactor and containment thermal hydraulics, the core degradation progression, and the fission product release to environment analyzed by MELCOR code for Fukushima Daiichi Unit 1 are summarized in Table 6.
Figure 5 shows cumulative leakage flow mass for various leakage flow paths. Figure 5A shows integrated leakage flow mass from RPV to PCV via various leakage flow paths. It is assumed that there was total 140 tons of liquid inside the RPV initially (Table 5). Total 80 tons of liquid (water plus steam) leaks from RPV to PCV by cycling opening and closure of the SRV from 1 to 5 hours. The other leakage mechanisms are SRV stuck open, instrument pipe leak, MSL failure, and SRV gasket failure. Total of these leakages are about 60 tons from 5 to 10 hours. Total leakage from RPV to PCV is 140 tons from 1 to 10 hours. Therefore, the most dominant leakage failure mode from RPV to PCV is identified as SRV cycling open and closure.
Figure 5B shows integrated leakage flow mass from PCV to RB via various leakage flow paths. The most dominant failure mode is identified as DW head flange seal failure due to the pressure increase above 0.75 MPa at 19 hours. The leakage mass via this leakage path is about 23 tons. DW pressure increased to 0.84 MPa at 12 hours in the plant log (Table 5, Figure 2B, Figure 6C, Figure 6D). Such a high pressure in DW head flange region could lift shield plugs up. This lift up of shield plugs, hydrogen and non-condensable gases could be released to the top operating floor (refueling bay) of RB. The second and third leakage mechanism are PCV leakage starting at 30 and 77 hours. These assumptions are made for matching pressure trend with plant monitoring curve in the long term (from 30 to 140 hours) as shown in Figure 6D. SC venting at 23 hours is also assumed but it is not shown here because the leakage flow mass is too small due to the too small failure area (FL921 in Table 3). Total 70 tons of liquid leaked out from PCV to RB during 184 hours (Figure 5D).
Figure 5C shows integrated leakage flow mass from RB to ENV via various leakage flow paths. The most dominant failure mode is identified as leakage from the top operating floor (refueling bay) of the reactor building to environment due to the DW head flange seal failure starting at 19 hours. About 60 tons of liquid leaked out from RB to environment during 184 hours (Figure 5D).
Figure 6 shows the thermal hydraulic and core degradation parameters. Figure 6C shows RPV pressure behavior during first 25 hours. The pressure in RPV is decreased during the first one hour due to isolation condenser (IC) operation. The SRV opens and closes cyclically from 1 to 5 hours. It is assumed that the SRV stuck open at 5 hours. Only two RPV pressure points are recorded in the plant monitoring data set; one is 7 MPa at 5.33 hours and the other is 0.4 MPa at 5.72 hours (Figure 2B, Figure 6C, Figure 6D). Due to the repeated opening of SRV, about 140 tons of water in RPV moves to the SC from 1 to 12 hours (Figure 6B). Reactor vessel failure occurs at about 16 hours.
RPV water level decreased to core bottom at about 5 hours (Figure 6A). From this time core heat up and subsequent core relocation occurs due to the loss of coolant inventory. Hydrogen is generated from zirconium water reaction and from steel water reaction (Figure 6G). Metal water reaction is exothermic in nature. When the cladding temperature increases above 900 K, the metal water reaction starts.
The total heats generated by metal water reaction are 3 to 5 times higher than the core decay heat from 5 to 7 hours (Figure 6E). Fuel temperature increases above 2250 K (Figure 6H). Intact fuels collapse down to lower plenum at 7 hours due to the core support plate failure by over-temperature mode.
After reactor vessel fails at 16 hours, molten core concrete interaction (MCCI) occurs in the reactor cavity due to interaction of high temperature corium and concrete in pedestal and drywell floor. Non-condensable gases (CO, H2, CO2, H2O) are generated in cavities due to MCCI. By following BSAF recommendation, it is assumed that drywell head flange leakage occurs when the PCV pressure exceeds 0.75 MPa. It is occurred at 18 hours in the MELCOR simulation (Figure 6C and 6D). Fission product release to atmosphere occurs at the same time. Note that in the actual plant record, DW pressure reached to 0.84 MPa at 12 hours and hydrogen explosion occurred at about 25 hours (Table 5). The potential leakage of steam and non-condensable gases through drywell head flange to the reactor building roof (operating floor, refueling bay) might be the main reason of the hydrogen explosion of Unit 1.
Core components (UO2 fuel, Zircaloy cladding, and supporting material such as stainless steel) can be molten so that it can be engaged in relocation process when the core is heated. Figure 6E shows temporal change of decay power in reactor vessel, cavity 1 and cavity 2. Two cavity model is used in this analysis. Corium ejection from reactor vessel to cavity occurs between 17.5 to 19 hours (Figure 6F).
2. Fission product generation and movement behaviors
Release of radionuclides can occur from the fuel-cladding gap by exceeding a failure temperature criterion or losing intact geometry, from material in the core using the various CORSOR empirical release correlations18–20) based on fuel temperatures. Release of radionuclides from fuel debris during core-concrete interactions in the reactor cavity is estimated by using the VANESA21) release model. After release to a control volume, masses may exist as aerosols and/or vapors, depending on the vapor pressure of the radionuclide class and the volume temperature. There are three models in CORSOR release correlations: 1) Original CORSOR model, 2) Modified COSOR model (CORSOR-M), and 3) CORSOR-Booth model22).
The original CORSOR model correlates the fractional release rate in exponential form,
where ḟ is the release rate (fraction per minute), A and B are empirical coefficients based on experimental data, and T is the core cell component temperature in degrees Kelvin. Different values for A and B are specified for three separate temperature ranges. The lower temperature limit Ti for each temperature range and the A and B values for that range are defined for each class in sensitivity coefficient array to reflect the release rate from fuel. If the cell temperature is below the lowest temperature limit specified (900 K), no release is calculated. Ti is defined as 900, 1,400 and 2,200K for classes other than class 5 (Te). Ti is defined as 900, 1,600 and 2,000 K for classes other than class 5 (Te). Coefficients A and B are shown in page 5 of CORSOR user’s manual23).
The CORSOR-M model correlates the same release data used for the CORSOR model using an Arrhenius form, ko,
Where, k is the release rate (min−1) at a given temperature T for a particular species, Q is the activation energy for the release process, R is the gas constant, and ko is the so-called preexponential factor. The values of ko, Q, and T are in units of min−1, kcal/mole, and K, respectively. The value of R is 1.987×10−3 in (kcal/mole)K−1. The values of ko and Q for each class are defined in page 7 of CORSOR user’s manual23).
Twelve radionuclide classes are defined in MELCOR code as shown in Table 7. Aerosol transport and deposition in reactor coolant system (RCS) and containment is handled by MAEROS model24). Thermal analysis during the molten core and concrete interaction (MCCI) process in the cavity are handled by CORCON-Mod325).
Release from fuel and release to environment are studied for some selected radioisotopes of volatile radionuclides (Kr, Xe, Cs, I, Te), non-volatile radionuclides (Sr, Mo, Tc, Sb, Ag) and actinides (Am, Cm, Pu) in this paper. Refer to Koo et al. [20] paper for the classification of fission products according to their volatilities.
The fission products generated from the damaged core are released to the suppression chamber (SC) and drywell (DW) at first by SRV cycling open and closures. After the failure of the lower head penetrations at 20 hours, fission products retained in the molten corium or debris beds are released to the pedestal. Finally they are released to the other locations such as reactor building (RB), turbine building and environment (ENV) through various leak paths already existed before the SC venting operation or after the venting. The explosion at the refueling bay might occur due to leakages of hydrogen from drywell head flange seals degradation due to high pressure and temperature in the drywell top region. Initial inventories of Fukushima Daiichi reactor cores at shutdown is summarized in page 118 in the book of Povinec et al.26) Initial inventories for Units 1, 2, and 3 cores at shutdown only for the volatile radionuclides such as 85Kr, 133Xe, 137Cs, 89Sr, 131I, and 132Te are summarized in Table 8. MELCOR code estimates released fraction to environment in according to the class. The elements in each class has the same release mechanism from fuel, transportation, and deposition mechanism in in-plant compartments. Cumulative released radioactivity for each radionuclide is obtained by the multiplication of initial inventory of each radionuclide by release fraction to environment of corresponding class number.
The release characteristics from fuel and the in-plant transport and deposition behaviors in 4 compartments, RPV, SC, DW, and ENV+, are shown in Figure 7 for 12 radionuclide classes defined in Table 7. Class 1 represents noble gases all which are released to the environment finally. Class 2 (Cs), class 4 (I), and class 5 (Te) show similar behavior. Class 2 deposits more than class 4 and class 5 in RPV. Class 3 (Sr) represents semi-volatile about 23% of inventory is released from fuel. Class 7 (Tc), class 11 (Sb), and class 12 (Ag) show similar Trends and less than 60% of the inventory are released from fuel before 100 h. About 0.5% of inventory of class 9 (La) and class 10 (U) are released from fuel. Less than 0.01% of inventory of class 6 (Ru) are released from fuel. About 0.15% of inventory of class 8 (Ce) are released from fuel.
Temporal variation of cumulative release fraction to environment of 5 volatile classes (class no. 1 to 5) are shown in Figure 8. Temporal variation of cumulative released radioactivity to environment for 6 radionuclides is shown in Figure 9. Temporal release rate of radionuclides to environment which is shown in Figure 10, are obtained by the derivative operation of cumulative release curve of Figure 9. For the short-lived radionuclides, such as 133Xe, 131I, and 132Te, decay in the plant is corrected during the course of release rate calculation.
Estimation for the radioactivity release to atmosphere from Unit 2 and Unit 3 is also conducted with similar way to Unit 1. Cumulative radioactivity released to environment up to 184 hours (7.7 days) after reactor trip for Fukushima Daiichi Units 1, 2, and 3 are summarized in Table 9. Core cooling for Unit 3 is performed by RCIC/HPCI up to 36 hours after reactor trip. Core cooling for Unit 2 is performed by RCIC up to 72 hours after reactor trip. Hydrogen explosion occurred at 68 hours in Unit 3. But, hydrogen explosion is not reported in Unit 2.
Source term release timing and radioactivity analysis results for 131I and 137Cs for Units 1, 2, and 3 are summarized in Table 10. The results are also compared with the results in UNSCEAR 2013 report. First radioactivity release timings estimated by MELCOR are 19, 43, and 98 hours for Unit 1, 3, and 2, respectively. For 131I, the results estimated in this study (754 PBq) is greater than the results reported UNSCEAR 2013 report (95-500 PBq). For 137Cs, the results in this study (29 PBq) is well within the results reported in the UNSCEAR 2013 report (6-37 PBq).
Radioactivity release peak timings to atmosphere are recorded at many monitoring posts at Fukushima Daiichi site and Fukushima Prefecture monitoring posts as shown in Figures 11 and 12, respectively. We know that the release rate peaks before 40 hours originated from Unit 1. Remarkable peak timings occurred roughly at 15, 20, 25, and 30 hours after reactor trip in the real monitoring posts. Monitoring result at Oono has about three hours delayed peaks compared to other monitoring results. The release start time to atmosphere is roughly at 20 hours after reactor trip in this study. Even though hydrogen explosion occurs at 25 hours at the plant, actual DW pressure peak occurs already at 12 hours at the plant (Figure 6C, Table 5).
We can also identify from these figures that the peaks during 40 to 95 hours are originated from Unit 3 release. It is also reported that multiple SC vent valve openings were tried at 42, 45, 54, 63, 97, 107, 150 hours in page 357 of reference TEPCO report [1]. The same assumptions are used in our MELCOR analysis. As shown by the multiple peaks in Figure 10B. Vent valve openings in Unit 2 were also tried at 78 and 81 hours. However, it is not assumed in our analysis that the peaks cannot be seen in Figure 10C.
Conclusion
Using the initial and boundary conditions which are provided in open publications and OECD/NEA BSAF project, the fission product release timing and amount are estimated by using the MELCOR code. The timing to release to atmosphere estimated by MELCOR code seems to be reasonable compared to real monitoring post records at accident site or remote locations.
Estimated source terms for volatile radionuclides by forward accident progression analysis based on ORIGEN-MELCOR code frame are compatible with or well within the ranges estimated by international research report such as UNSCEAR-2013 and those estimated or summarized in the technical papers which are usually estimated by reverse or inverse method.
The biggest contributing radionuclides from Fukushima Daiichi accident to environment in early term (within a few weeks) are 133Xe, 131I and 132Te. Short half-lives (a few days) of those radionuclides, however, they diminish their strength very shortly. Thereafter, volatile radionuclides which has half-lives of a few years such as 86Kr, 134Cs, 137Cs, and 133I will dominate in the environment.
This estimate is, however, based on the initial 184 hours duration since reactor trip that release estimate would be much varied according to the conditions of the plant thereafter. One of the aspect which can impact to the estimated result is whether MCCI happen in cavity or not. Ground water intrusion into the plant is also one of the important scenario to the estimation. Reasonable estimation would be very helpful for the future work planning, for instance, decommissioning at the site.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
ScienceCentral
Download Citation 
Print 
