This article was downloaded by: [94.20.20.135] On: 21 March 2014, At: 00:04 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of Nuclear Science and Technology Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tnst20
Release Rate Estimation of Radioactive Noble Gases in the Criticality Accident at Tokai-Mura from Off-Site Monitoring Data a
Shigekazu HIRAO & Hiromi YAMAZAWA
a
a
Department of Energy Engineering and Science , Graduate School of Engineering, Nagoya University , Furo-cho, Chikusa-ku, Nagoya , 464-8603 , Japan Published online: 05 Jan 2012.
To cite this article: Shigekazu HIRAO & Hiromi YAMAZAWA (2010) Release Rate Estimation of Radioactive Noble Gases in the Criticality Accident at Tokai-Mura from Off-Site Monitoring Data, Journal of Nuclear Science and Technology, 47:1, 20-30 To link to this article: http://dx.doi.org/10.1080/18811248.2010.9711924
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Journal of NUCLEAR SCIENCE and TECHNOLOGY, Vol. 47, No. 1, p. 20–30 (2010)
ARTICLE
Release Rate Estimation of Radioactive Noble Gases in the Criticality Accident at Tokai-Mura from Off-Site Monitoring Data Shigekazu HIRAO1; and Hiromi YAMAZAWA1 1
Department of Energy Engineering and Science, Graduate School of Engineering, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan
Downloaded by [94.20.20.135] at 00:04 21 March 2014
(Received May 28, 2009 and accepted in revised form August 26, 2009)
A method of release rate estimation of radioactive noble gases from an accidental nuclear facility was proposed. This method was applied to the criticality accident case at Tokai-mura to evaluate an uncertainty of results. The release rate is estimated by combining the gamma dose rates measured at off-site monitoring posts and those calculated using an atmospheric dispersion model. To analyze the dependence of uncertainty of estimated release rate on that of the wind field, two wind field data were used; one from the meteorological model MM5 and the other from the analysis of measured wind data. The results of comparison that the analyzed wind field was in better agreement with the measured wind data than the MM5 wind field led to the smaller uncertainty in the analyzed wind case than that in the MM5 wind case. A gradual decrease in release rate with time was clearly shown in the analyzed wind case, while it was obscured by a large scatter in the MM5 wind case. The total activities of the released radioactive noble gases evaluated by the present method were 3:0 102 and 2:1 102 TBq for the cases with the retention times of noble gases, that is, the time elapsed before the noble gases are released into the atmosphere, of 5 and 10 min, respectively. The retention time of 10 min was reasonably compared with that evaluated from the total number of the fission reactions with the assumption of the same retention time. KEYWORDS: release rate estimation, air absorbed gamma dose rate, off-site monitoring data, atmospheric dispersion model, MM5, nuclear accident, emergency response system, JCO criticality accident
The most reliable information in deciding and implementing off-site countermeasures is the monitoring data on gamma radiation dose rate, which are available on real-time basis in most cases. However, the data are available only in a limited number of locations sparsely distributed in the emergency planning zone (EPZ), which is defined by the Japanese emergency management schemes as a circular area around a nuclear power plant during an accident with a radius of 8 to 10 km. The sparsely distributed monitoring data are considered to be insufficient to depict a spatial distribution of regions in which intervention levels are exceeded. This is the case where an emergency response system can effectively contribute if the unavailability and uncertainty of source term information are overcome. A straightforward way of overcoming these difficulties is to estimate source term parameters from measured data from monitoring stations. This kind of estimations have been made for past incidences, such as local-scale incidences involving fire and explosion accidents at the bituminization facility in the Tokai Reprocessing Plant4) and the JCO criticality accident,5) and regional- to interregional-scale incidences such as the Chernobyl accident6) and the Algeciras accident.7) From these trials to estimate source terms by combining the monitoring data and the atmospheric disper-
I. Introduction An emergency response system with an atmospheric dispersion model is thought to make an important contribution to decision-making about off-site countermeasures when radioactive materials are released from a nuclear facility during an accident. Several systems such as SPEEDI1) and NARAC2) have been operational for more than 20 years. The quality of the prediction results highly depends on the accuracy of input data such as meteorological data and source term parameters. Some systems use sophisticated meteorological models in combination with online-available meteorological data provided by national or international meteorological agencies to increase the accuracy of prediction results.3) Although many models for meteorology and atmospheric dispersion have been successfully utilized in emergency response systems in the world, it is still a matter of controversy whether source term parameters are timely available in an actual nuclear emergency. The uncertainty of source term parameters is one of the main causes of discrepancies of predicted results from monitoring data.
Corresponding author, E-mail:
[email protected]
ÓAtomic Energy Society of Japan 20
21
Downloaded by [94.20.20.135] at 00:04 21 March 2014
Release Rate Estimation of Radioactive Noble Gases in the Criticality Accident at Tokai-Mura
sion simulation results, a few systematic methods of estimating source terms have been developed for long-range atmospheric dispersion.8,9) However, no method has been developed for the EPZ-scale atmospheric release cases. This is partially because the spatial and temporal distributions of concentrations of radioactive materials and gamma dose rate from them show quite sharp variations in the local scale, and hence the high accuracy of the meteorological field used in the dispersion simulations is required for reasonable estimation of source term parameters. It has been shown that a meteorological model is applicable to providing highly accurate meteorological field to the dispersion model for a several-hundred-kilometer scale.10) However, little is known about the uncertainty of the estimated source term parameters when a meteorological model is applied to the dispersion simulation of the EPZ scale. The purpose of the present study is to estimate the release rate of radionuclides and to discuss uncertainties of estimated release rate. Although there are several source term parameters, such as the rate, location, and period of release, we limit our discussion to the estimation of release rate. The present study demonstrates release rate estimation by comparing calculated air absorbed gamma dose rate (hereafter, gamma dose rate) with measured one, selecting the criticality accident at the JCO facility, as a test case. Given the measurement of gamma dose rate and meteorological information at monitoring posts and stations (hereafter, monitoring posts), we discuss the accuracy of release rate estimation and its dependence on the accuracy of wind field. To do this, two kinds of wind fields are used in this study, one being the wind field calculated using the meteorological model from coarsely gridded meteorological data and the other being the combination of the above-mentioned calculated wind field and locally observed wind data. This paper presents first the descriptions of the model systems and the method for estimating release rate. Then, the method is applied to the JCO case to estimate the release rate of radioactive noble gases from the measured dose rate increase. Finally, the estimated results are compared with the release rate estimated by the JAERI Task Force for Supporting the Investigation of JCO (hereafter, JAERI Task Force) and with the total number of fission reactions estimated from the analysis of the residual solution in the accidental precipitation tank.5)
II. Model Description 1. Atmospheric Dispersion Model A Lagrangian particle random-walk model was used to calculate the transport of airborne radionuclides released from a point source. This atmospheric dispersion model calculates the movement of particles representing radionuclides by using temporally and spatially varying meteorological data. The coordinate system of the atmospheric dispersion model is three-dimensional: the map coordinates for the horizontal direction and the terrain following the z -coordinate for the vertical direction.11) The movement of a particle for a time step with a time interval of t can be written as VOL. 47, NO. 1, JANUARY 2010
xðt þ tÞ ¼ xðtÞ þ ut þ Rx ; yðt þ tÞ ¼ yðtÞ þ vt þ Ry ;
ð1Þ
z ðt þ tÞ ¼ z ðtÞ þ w t þ Rz ; where ðxðtÞ; yðtÞ; z ðtÞÞ and ðxðt þ tÞ; yðt þ tÞ; z ðt þ tÞÞ are the position of a particle at the start and the end of a time step, and ðu; v; w Þ is the mean wind velocity. The horizontal random movements ðRx ; Ry Þ caused by turbulent diffusion are expressed as pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2Þ Rx ¼ Ry ¼ 24Kh tð0:5 Rð0ÞÞ; where Kh is the horizontal diffusion coefficient and Rð0Þ is a uniform random number between 0 and 1. The coefficient Kh is described as Kh ¼
1 dh2 ; 2 dt
ð3Þ
where h is the cross-wind width of the plume and given as a function of down-wind distance and the stability class obtained by using the Pasquill-Gifford chart. The vertical random movement Rz is calculated as sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2ffi @Kz @Kz t ; Rz ¼ t 2Kz t þ ð4Þ @z @z where Kz is the vertical diffusion coefficient at the particle position. In the calculation of gamma dose rate at a monitoring post, a particle is assumed to be a point source. A radioactive plume is composed of more than one nuclide, each emitting multiple gamma rays. In the present study, all gamma rays emitted from one nuclide are contracted as one gamma ray with effective and average energies.12) Based on this method, one particle is assumed to emit gamma rays with the number being equal to the number of nuclides. The gamma dose rate dDi due to the i-th particle consisting of N of nuclides is calculated using N q j E j ðE j Þ exp½ðE j ÞrBððE j Þ; r; hÞ K X ave ave eff en ave dDi ¼ ; 4 j¼1 r2 ð5Þ j j where Eeff and Eave are the effective and average energies of the j-th nuclide, K the dose conversion factor, r the distance between a particle and a monitoring post, en ðEÞ the energy absorption, ðEÞ the linear attenuation coefficient in air, and BððEÞ; r; hÞ the build-up factor as a function of the height h of the particle, which takes into account the decrease in scattered radiation near the surface. The activity of the j-th nuclide, described as q j , is a function of time due to radioactive decay, and the time elapsed for ventilation in the facility and atmosphere is considered. The sum of contributions of all particles provides the gamma dose rate at a monitoring post.
2. Meteorological Data The meteorological input data for the atmospheric dispersion model are the temporal and spatial distributions of wind, vertical diffusion coefficient, and stability class. Atmospheric dispersion calculations were conducted with two different meteorological data sets: 1) wind field calcu-
22
S. HIRAO and H. YAMAZAWA
lated using the meteorological model MM5,13) and 2) the same as before but the wind fields analyzed from observed wind data were embedded in the near-surface layers. In both cases, vertical diffusion coefficient and stability class were derived from MM5 and observation, respectively. MM5 is a nonhydrostatic meteorological model. This model was developed by Pennsylvania State University and the National Center for Atmospheric Research. Alternative wind field data were based on the calculation of inter- and extrapolating wind velocity observed at sites in the domain of the atmospheric dispersion model. The first step of the wind field calculation was the interpolation of the observed wind data onto horizontal grid points as follows N X
ðu10 m ; v10 m Þi; j ¼
Wm ðu; vÞm
m¼1 N X
; Wm ¼ Wm
1 ; rm2
ð6Þ
Downloaded by [94.20.20.135] at 00:04 21 March 2014
m¼1
where ðu10 m ; v10 m Þi; j is wind field at the height of 10 m on the ði; jÞ grid, N the number of observed wind data, ðu; vÞm the wind data at the m-th observation site, Wm the weight coefficient of m-th observed wind data, and rm the distance between the m-th observation site of wind data and the grid. The second step was inter- and extrapolations in the vertical direction. The power law was applied as zk p Uk ¼ U10 m ; U10 m ¼ ðu10 m ; v10 m Þi; j ; ð7Þ z10 m where zk is the height of the k-th model layer above ground and p the power law exponent calculated using the wind velocities observed at two different heights. The final step was the replacement of MM5 wind data by above-mentioned analyzed wind data. The limit of the height for the replacement was set to be 400 m above the ground. In the following section, the wind field data from the result of the combined analysis of the MM5 wind field data and analyzed wind data are named ANL wind field data.
III. Release Rate Estimation The estimation of release rate is based on the idea that the atmospheric dispersion model can be used to calculate a spatial distribution of relative values of gamma dose rate due to the radionuclides accidentally released, but the absolute values are unknown. According to this idea, the ratio of gamma dose rate to release rate can be assumed to be the same for both measurements and calculations Dr Dc ¼ ; ð8Þ Sr t;i Sc t;i where D and S are the gamma dose rate and release rate, respectively. The subscripts t and i denote time and location, respectively. The subscript c represents calculation using the model. Dr is the measured gamma dose rate increase due to the radioactive plume. Release rate can be estimated from this equation by solving it for Sr . However, this relation does not hold strictly mainly due to errors in the calculation. For a given time t, there may be
more than two estimated values of release rate for different monitoring posts. In this case, a geometrical mean was used to have a single value for time as follows: Sr;t ¼
N Y
!1
N
Sr;t;i
:
ð9Þ
i¼1
Equation (9) implicitly assumes that the radionuclides released at time t immediately affect gamma dose rate. Although this assumption is not generally correct, it can be acceptable if we consider that the travel time of plume in the objective domain is on the order of 10 min and that release rate would not have changed rapidly over this time scale. Therefore, the release rate estimated from gamma dose rates for a given time t is considered to be that for the time segment at t.
IV. Model Application 1. Objective Case The criticality accident at the JCO facility in Tokai-mura, Ibaraki Prefecture, Japan on 30 September 1999 was chosen as an objective case. According to the reports, radioactive noble gases were released into the atmosphere during the period of the criticality state continuing from 10:35 JST, 30 September to 06:15 JST, 1 October.5,14) It was reported that the total number of fission reactions was estimated to be 2:5 1018 based on the results of the analysis of the residual solution in the precipitation tank and the release rate of radioactive noble gases to be 8:0 1012 Bq h1 based on the results of the analysis using the atmospheric dispersion model of radionuclides.1,5) Radioactive plume passages were detected as increases in gamma dose rate at the 44 monitoring posts by the former JAERI, the former Japan Nuclear Cycle Development Institute (hereafter JNC), and Ibaraki Prefecture.1,5,15) Gamma dose rate was continuously measured with NaI (TI) scintillation detectors. The 10-min averaged gamma dose rates were used in the present analysis. The surface observations of wind speed and direction at 10 min intervals were available at 22 points. The wind data measured at the height of 70 m and stability class were available at JNC. 2. Gamma Dose Rate Increase due to the Radioactive Plume The main sources of gamma radiation were 1) the natural radioactive materials in the soil, 2) the radon decay products deposited by precipitation, 3) the direct gamma rays from the JCO facility, and 4) the radioactive plume released from the JCO facility. To distinguish the increase in gamma dose rate due to the radioactive plume, the other three variations were subtracted from the data as follows: 1) Normal background dose rate at each monitoring post was calculated by averaging gamma dose rates measured during the period that was not affected by the accident and rainfalls. The averages and standard deviations of measured gamma dose rates were from 30 to 50 nGy h1 and 0.5 to 2.5 nGy h1 , respectively. The gamma dose rate increases due to the plume were defined as the excess above the normal background. Only a gamma dose rate increase of JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY
23
Release Rate Estimation of Radioactive Noble Gases in the Criticality Accident at Tokai-Mura
Downloaded by [94.20.20.135] at 00:04 21 March 2014
more than 10 nGy h1 , which is a 4-sigma of the gamma dose rate variation during the normal period, were used in the release rate estimation. 2) The increases in gamma dose rate due to the radon decay products deposited by precipitation were assumed to be represented by the variation of gamma dose rate measured at Sugaya, about 6.2 km south of the JCO facility. This monitoring post was not affected by the plume during rainfall and sufficiently far from the JCO facility for the direct gamma rays to be neglected. The maximum increase in gamma dose rate during rainfall was 22.3 nGy h1 . 3) The temporal variation of gamma dose rate due to direct gamma rays was calculated by the following relation according to Endo et al.16) rd exp 214 Dd ¼ k f ; ð10Þ 2 rd where Dd is the gamma dose rate due to direct gamma ray (nGy h1 ), k the fitting factor for the measured data, and rd the distance (m) from the JCO facility. The factor f denotes the temporal variation given by f ¼ 2:3 expð0:086tr Þ; f ¼ 1:0; f ¼ 0:38;
(a) Domain 1 Domain 2
Domain 3
(b) Domain 4 Ishigami Kadobe
Funaishikawa
11:00 to 20:45 JST, 30 September; 20:45 to 03:30 JST, 1 October; 03:30 to 06:15 JST, 1 October; ð11Þ
where tr is the elapsed time (h) from 11 JST. Based on Eq. (10), the direct gamma rays could have affected the gamma dose rate at Funaishikawa significantly, which is located about 1.3 km south of the JCO facility. The maximum gamma dose rate increase due to direct gamma rays was calculated to be 20.5 nGy h1 at Funaishikawa. The influences of direct gamma rays on gamma dose rate at other monitoring posts were neglected because of their long distances from the JCO facility. For one time segment, estimation of release rate was carried out only for pairs of measured gamma dose rate increase exceeding 10 nGy h1 and calculated gamma dose rate exceeding a hundredth of the maximum calculated gamma dose rate for all monitoring posts. 3. Calculation Condition The domain of the atmospheric dispersion model is shown as Domain 4 in Fig. 1(b). Domain 4 covers the coastal area of Ibaraki Prefecture. Table 1 shows the parameters used in the atmospheric dispersion model calculation. The domains of the meteorological model MM5 shown in Fig. 1 consist of Domain 1 covering Japan, Domain 2 Kanto region, Domain 3 Ibaraki prefecture, and Domain 4. Other specifications and the physical processes of MM5 used in the present study are summarized in Table 2. In addition to the surface and rawinsonde observations, the regional analysis data (RANAL) from Japan Meteorological Agency (JMA) were used for the initial and boundary conditions and the four-dimensional data assimilation (FDDA) of the meteorological fields in Domain 1. Topography and land-use data were obtained from the United States Geological Survey (USGS) global database. VOL. 47, NO. 1, JANUARY 2010
MP22
Fig. 1 Calculation domains: (a) Domains 1 to 4 for MM5 nested calculations, (b) Domain 4 for the atmospheric dispersion model. Black dots represent the monitoring posts, and the triangle represents the site of accident (the JCO facility).
Table 1 Calculation conditions for the atmospheric dispersion model (Domain 4) Parameter
Value
Dimension of the domain Resolution of the domain Number of layers Top level Time step
76 76 0.667 km 23 5000 m 5s
The vertical z -coordinate is defined as 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 150, 200, 400, 600, 800, 1000, 1500, 2000, 2500, 3000, 3500, 4000, and 5000.
The MM5 calculation was conducted for the period from 09 JST, 28 September to 09 JST, 1 October. The first 48 h of the calculation period is the spin-up time to properly simulate meteorological fields. The output interval of MM5 was 10 min. The released radionuclides were assumed to be noble gases, the half-lives of which are in the range from a few minutes to several days. The composition of the radioactive noble gas mixture was assumed to be in proportion with the cumulative fission yield of each noble gas.17) The change of the composition due to the decay of the noble gases was
24
S. HIRAO and H. YAMAZAWA Table 2 Domain specification and physical schemes for the meteorological model MM5 Scheme
Domain 1
Horizontal coordinate Vertical coordinate Horizontal grid number Horizontal resolution Time step Number of layers Top level Cumulus parameterization Planetary boundary layer scheme Explicit moisture scheme Radiation scheme Surface scheme Nesting
100 100 18 km 50 s
Kain-Fritsch
Domain 2
Domain 3
Lambert conformal Terrain-following sigma-coordinate 100 100 73 73 6 km 2 km 16.667 s 5.556 s 30 100 hPa None Gayno-Seaman PBL Reisner graupel (Reisner 2) RRTM long-wave scheme Five-layer soil model One-way nest
Domain 4
76 76 0.667 km 1.852 s
The vertical sigma-coordinate is defined as 1.00, 0.9975, 0.995, 0.990, 0.985, 0.980, 0.975, 0.970, 0.96, 0.95, 0.94, 0.93, 0.92, 0.91, 0.90, 0.89, 0.87, 0.86, 0.85, 0.80, 0.75, 0.70, 0.65, 0.60, 0.55, 0.50, 0.40, 0.30, 0.20, 0.10, and 0.00.
Downloaded by [94.20.20.135] at 00:04 21 March 2014
Table 3 Radionuclides considered in the calculation
Radionuclide
Half-life
Relative cumulative fission yield
Effective energy (MeV)
Average energy (MeV)
Kr-83m Kr-85m Kr-87 Kr-88 Kr-89 Xe-133m Xe-133 Xe-135m Xe-135 Xe-137 Xe-138 Rb-88 Rb-89 Cs-138
1.82 h 4.48 h 76.3 min 2.84 h 3.15 min 2.19 d 5.243 d 15.29 min 9.14 h 3.818 min 14.08 min 17.78 min 15.15 min 32.4 min
0.014 0.032 0.063 0.090 0.116 0.005 0.169 0.031 0.165 0.154 0.161 — — —
0.003 0.157 0.791 1.954 1.911 0.041 0.047 0.428 0.248 0.190 1.125 0.637 2.072 2.340
0.011 0.164 0.958 1.344 1.120 0.056 0.050 0.447 0.249 0.544 0.851 1.577 1.325 1.197
considered by taking into account the time during which the noble gases were retained in the JCO facility before their being released into the atmosphere and the time elapsed during the atmospheric transport. The decay products of the noble gases were also considered. The retention time was assumed to be 5 min5) because there is no other available information on retention time. The radionuclides considered in the calculation are summarized in Table 3. The dispersion of radionuclides from the JCO facility started at 10:35 JST, 30 September and ended at 06:15 JST, 1 October. The release height was set to be 20 m above the ground. Assuming a constant release rate of 1 Bq h1 , the particles were released at a rate which made the total particle number to be 10 million. The time step was 5 s.
V. Results and Discussion 1. Wind Field The horizontal distribution of the wind fields of MM5 and ANL are shown in Fig. 2. The general features of wind fields
Fig. 2 Time variations of the surface wind fields of the MM5 data (left panels) and the ANL data (right panels) from 12 JST, 30 September to 06 JST, 1 October in 1999. White arrows represent observed wind at monitoring sites.
JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY
25
6 4 2 0 9JST 15JST 30 Sep.
-1
Wind speed (m s )
OBS MM5
21JST
Time
3JST 1 Oct.
8
9JST
OBS ANL
6 4 2 0 9JST 15JST 30 Sep.
21JST
Time
3JST 1 Oct.
Wind direction (deg.)
8
9JST
540
OBS MM5
360 180 0 9JST 15JST 21JST 30 Sep. Time
Wind direction (deg.)
-1
Wind speed (m s )
Release Rate Estimation of Radioactive Noble Gases in the Criticality Accident at Tokai-Mura
540 360
3JST 1 Oct.
9JST
3JST 1 Oct.
9JST
OBS ANL
180 0 9JST 15JST 21JST 30 Sep. Time
Downloaded by [94.20.20.135] at 00:04 21 March 2014
Fig. 3 Comparison of the time series of model-calculated wind speed and direction with the measurement at Funaishikawa
over the area of accident show fair agreement with the observation marked by the white arrows. These features are characterized by 1) onshore wind during daytime, 2) weak wind over the area of accident during nighttime, and 3) strong northeasterly wind during the early morning in the region along the coastline. The MM5 wind field as a whole could capture the synoptic-scale wind field. Comparison of wind speed between MM5 and ANL indicates that the MM5 wind is stronger than the ANL wind. During the transition period from daytime to nighttime, that is, from 18 JST, 30 September to 00 JST, 1 October, the MM5 wind shows good agreement with the observed weak wind in the area of accident. However, after 00 JST, 1 October, MM5 overestimated the offshore wind in the region south of the area of accident. In contrast, the ANL wind shows good agreement with the observed wind. Figure 3 shows the time series of wind speed and direction at 10 min intervals at Funaishikawa, which is near the JCO facility. Comparison between the observed wind and both the MM5 and ANL winds indicates that the period shown in Fig. 3 can be divided into four periods according to the degree of the agreement. For the first period from 09 to 16 JST, the wind speed and direction of the both MM5 and ANL winds show good agreement with those of the observed wind. For the second period from 16 to 19 JST, both MM5 and ANL winds capture the observed clockwise shift of the wind direction. However, MM5 slightly overestimated wind speed. The wind direction shift was not satisfactorily calculated. For the third period from 19 to 03 JST of the next day, wind speed was significantly overestimated by MM5. MM5 also failed to simulate the wide variation in the observed wind direction. In contrast, the ANL wind is in good agreement with the observed wind in this period. For the last period after 03 JST, the wind speed estimated by MM5 is generally larger than the observed wind speed, and the calculated wind direction differs by 90 . The ANL wind speed and direction show fair agreement with the observed data. VOL. 47, NO. 1, JANUARY 2010
2. Gamma Dose Rate The comparison between the gamma dose rate distributions calculated by using MM5 wind and ANL wind is shown in Fig. 4. The difference in wind data between MM5 and ANL as shown in Figs. 2 and 3 was due to that in calculated gamma dose rate. The initial gamma dose rate distribution extending westward at 12 JST, 30 September is similar in both MM5 wind case and ANL wind case. This result demonstrates that the similar wind direction lead to the similar spatial distribution of gamma dose rate. The gamma dose rate distributions in the MM5 wind case at 18 JST, 00 JST, and 6 JST shift more to the east than those in ANL wind. The differences in gamma dose rate distributions correspond to those in wind direction, as shown in Fig. 3. Figure 5 shows the time series of measured gamma dose rates and calculated ones at the monitoring posts in Fig. 1. Increases in calculated gamma dose rate are relatively (not absolutely) compared with those in measured one, because for the calculation of gamma dose rate, the release rate during the entire release period was assumed to be constant at 1 Bq h1 . Compared with the calculated gamma dose rate increases for MM5 wind, those for ANL wind seem to be in better agreement with the measured ones. Both the calculated gamma dose rate increases for MM5 wind and ANL wind around 12 JST at Kadobe in Fig. 5(a) agree with the measured ones. The calculated gamma dose rate increase for MM5 wind around 18 JST at Ishigemi in Fig. 5(b) shows poor agreement with the measured one. This poor agreement is due to the low accuracy of the clockwise wind direction shift calculated by MM5. After 18 JST at Funaishikawa in Fig. 5(c), the calculated gamma dose rate increases for MM5 wind are similar to those for ANL wind. In contrast, the calculated gamma dose rate increases for ANL wind at MP22 in Fig. 5(d) show better agreement with the measured ones than those for MM5 wind. This better agreement at MP22, which is located farther from the JCO facility than Funaishikawa, indicates that the calculated
26
S. HIRAO and H. YAMAZAWA 10000
(a)
Measurement MM5 ANL
1000
1.E-08 1.E-09 1.E-10
100
1.E-11
10000
21JST Time
1.E-12 9JST
3JST 1 Oct.
(b)
Measurement MM5 ANL
1000 Measured gamma dose rate -1 (nGy h )
15JST
1.E-09 1.E-10
100
1.E-11
10 9JST 30 10000
15JST
21JST Time
(c)
Measurement MM5 ANL
1000
1.E-12 9JST
3JST 1 Oct.
1.E-08 1.E-09 1.E-10
100
1.E-11
10 9JST 30 10000
15JST
21JST Time
1.E-12 9JST
3JST 1 Oct.
(d)
Measurement MM5 ANL
1000
1.E-08 1.E-09 1.E-10
100
1.E-11
10 9JST 30
15JST
21JST Time
1.E-12 9JST
3JST 1 Oct.
Fig. 5 Comparisons of 10 min averaged gamma dose rates between calculations and measurement in monitoring posts (a) Kadobe, (b) Ishigami, (c) Funaishikawa, and (d) MP22
Fig. 4 Gamma dose rate distributions calculated by using MM5 wind (left) and ANL wind (right)
gamma dose rate distribution for ANL wind is more reliable than that for MM5 wind. 3. Estimation of Release Rate The release rate of noble gases was estimated for every 10 min time segment over the period from 10:35 JST, 30 September to 6:15 JST, 1 October. The results of the estimation are shown in Fig. 6 and compared with the JAERI Task Force’s result for which the time resolution is 1 h.5) Release rate could not be estimated from 11 to 15 JST because there was no monitoring post that detected the gamma dose rate increase. Release rate estimated in the MM5 wind case varies by two or three orders of magnitude. The variability of the release rate of the ANL wind case is smaller than that of the MM5 wind case. The geometric means of the estimated release rate are 2.9 TBq h1 for MM5 wind case, 2.8 TBq h1 for ANL wind case, and 8.0 TBq h1 for JAERI Task Force’s results.
10000 Release rate (TBq h-1)
Downloaded by [94.20.20.135] at 00:04 21 March 2014
1.E-08
Calculated gamma dose rate -1 (nGy h )
10 9JST 30
MM5 ANL JAERI
100 1 0.01 9JST 15JST 30 Sep.
21JST Time
3JST 1 Oct.
9JST
Fig. 6 Estimation of release rate of every 10 min time segment. The white squares and the black circles represent the results obtained by using the ANL wind data and the MM5 wind data, respectively. Gray diamonds represent JAERI Task Force’s results. Solid line represents the least-squares fitting to the ANL results.
The release rate estimated in the MM5 wind case is larger than that in the ANL wind case in the first half from 11 to 20 JST and smaller than that in the ANL wind case in the second half from 20 to 06 JST of the next day. This large JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY
27
Release Rate Estimation of Radioactive Noble Gases in the Criticality Accident at Tokai-Mura
Downloaded by [94.20.20.135] at 00:04 21 March 2014
S ¼ expð4:61 0:0362te Þ; 10:35 to 20:00 JST, 30 September; S ¼ expð2:29 0:118te Þ;
Frequency
150
n = 319 σ = 3.44
(a) MM5
100 50 σ
0 -3 -2 -1 0 1 2 3 10 10 10 10 10 10 10 R 150 Frequency
variation of the release rate in the MM5 wind case was caused by the low accuracy of the wind field. As shown in Fig. 3, the wind direction calculated by MM5 shows poor agreement with the observation especially at 16, 20, and 00 JST of the next day. In the ANL wind case, it is evident that release rate decreased substantially with time. The rate of the decrease seems to be smaller in the second half than in the first half. In the last stage after 06 JST, release rate was estimated to have dropped rapidly. This trend is similar to the measured temporal changes in the counting rate of the neutron monitor.16) The records of the neutron monitor directly represent the rate of fission reactions during the criticality accident. The production rate of fission nuclides and hence their release rate can be considered to be proportional to the rate of fission reactions. The following empirical formulas were obtained in order to express the temporal variation of the estimated release rate shown in Fig. 6,
n = 551 σ = 2.88
(b) ANL
100 50
σ
0 -3 -2 -1 0 1 2 3 10 10 10 10 10 10 10 R
ð12Þ
Fig. 7 Frequency distribution of the ratios of release rate to geometric mean: (a) MM5 wind case, (b) ANL wind case. Solid curves are the results of log-normal fitting.
where S is the release rate (Bq h1 ) and te the elapsed time (h) from 10:30 JST. The exponents are evaluated by leastsquares fitting.
case, there are certain number of estimates that are more than one order of magnitude smaller or larger than the average. These poor estimates are considered to be caused by errors in the estimation of gamma dose rate distribution especially at the edge of the radioactive plume. In the calculation of gamma dose rate, the low number density of particles at the edge of the radioactive plume could have brought about the calculation error. However, this error is considered to be insignificant because the calculated gamma dose rates used for the release rate estimation are 10-min averaged values and exceed the criteria for selecting pairs of the measured gamma dose rate increase and the calculated one. Another cause of the errors is the sensitivity of the gamma dose rate distribution to complex wind fields. In the case of short-range atmospheric transport phenomena, variations of relatively small horizontal scales in the wind field can significantly affect the concentration distribution and hence the gamma dose rate distribution, because the size of the plume from a point source is small. Although there would be a possibility to improve the performance of release rate estimation by modifying the atmospheric transport calculation results, we leave it for the next step of the study. Figure 8 shows the time series of the geometric standard deviation of Rt calculated in the ANL wind case. The geometric standard deviation could not be calculated for the time segments in which only one monitoring post detected the gamma dose rate increase. The geometric standard deviation of Rt is in the range from 1.7 to 8.3. The variation of the geometric standard deviation implies that the degree of agreement of the gamma dose rate distribution pattern between the actual condition and the calculation varied
20:00 to 06:00 JST, 1 October; S ¼ expð83:3 4:27te Þ; 06:00 to 08:00 JST, 1 October;
4. Uncertainty of Estimated Release Rate To evaluate the uncertainty of estimated release rate, the ratio of the estimated release rate Sti at the i-th monitoring post to the geometric mean St of the time segment t is used as follows: Rti ¼
Sti : St
ð13Þ
If the estimated release rates for a certain time are consistent, that is the case where the estimated release rates are the same in all the monitoring posts, the ratio Rt is 1. Therefore, this parameter would be a good measure for examining the uncertainty of release rate estimated for every time segment. Figure 7 shows the frequency distribution of ratios for all the time segments and for all the monitoring posts. The distribution of the ratio in the ANL wind case is narrower than that in the MM5 wind case. This result is reflected in the geometric standard deviations that are 3.44 in the MM5 wind case and 2.88 in the ANL wind case. The total numbers of the population are 319 for the MM5 wind case and 551 for the ANL wind case. Since release rate estimation was done for the pairs of measured gamma dose rates and calculated ones exceeding the criteria as described in the previous section, the larger number of population of the ANL case implies that the spatial distribution of gamma dose rate was more realistically calculated due to better wind fields in the ANL case than in the MM5 case. However, even in the ANL VOL. 47, NO. 1, JANUARY 2010
28
S. HIRAO and H. YAMAZAWA 104 Calculation (nGy h-1)
-1
Release rate (TBq h )
10000 100 1 0.01 9JST 15JST 30 Sep.
21JST Time
3JST 1 Oct.
10
102 101 100
9JST
Downloaded by [94.20.20.135] at 00:04 21 March 2014
depending on the accuracy of the calculated wind field. The rapid changes in wind field are considered to have led to the discrepancy between the measured and calculated gamma dose rate and hence to the large geometric standard deviation. 5. Error in Gamma Dose Rate Calculation The release rate estimation method depends strongly on the performance of the estimation using the atmospheric dispersion model and gamma dose rate calculation model. By inputting the estimated temporal change in release rate in Eq. (12) to the model, gamma dose rate increases due to the accidental release were calculated to compare them with measurement data. Figure 9 shows the scatterplots of the measured versus calculated gamma dose rates which are the 10-min averaged values for all monitoring posts. Although the range in the scatter is almost the same for both the MM5 wind case and the ANL wind case, it is evident that a larger number of data are plotted near the 1:1 line in the ANL wind case. The fractions of the calculated gamma dose rates falling within a range of a factor 2 of the measured ones are 14% for the MM5 wind case and 35% for the ANL wind case. These results imply that the passage of the radioactive plume was reasonably calculated using the atmospheric dispersion model in the ANL wind case and hence the temporal changes in estimated release rate was considered to be plausible. 6. Comparison with the Former Estimation The geometric mean of release rate had been evaluated to be 8.0 TBq h1 as an average for the whole period by the JAERI Task Force.5) In the present study, the geometric mean of the release rate was estimated as 2.8 TBq h1 in the ANL wind case. These two values were estimated in similar ways by using atmospheric dispersion models. The difference between the estimated release rates is probably due to the difference in the model performance and in the procedure of release rate estimation. If the temporal resolution of gamma dose rate calculation is 1 h, the estimated geometric mean of release rate by our atmospheric disper-
Calculation (nGy h-1)
104
Fig. 8 Geometric standard deviation of the ratio of estimated release rate to geometric mean in the ANL wind case. Black circles and error bars represent the geometric mean and the geometric standard deviation of release rate, respectively, estimated from several monitoring posts. Gray circles represent the results from a single monitoring post.
(a) MM5
3
100 101 102 103 104 -1 Measurement (nGy h ) (b) ANL
103 102 101 100
100 101 102 103 104 -1 Measurement (nGy h )
Fig. 9 Scatterplots of calculated gamma dose rates against measured ones for all monitoring posts: (a) MM5 wind case and (b) ANL wind case
sion model is 3.8 TBq h1 . Furthermore, given the same pairs of the measured gamma dose rates and the calculated ones as the JAERI Task Force’s estimation, the geometric mean of release rate was estimated to be 1.7 TBq h1 . The difference in estimated release rate between the cases with different time resolutions and monitoring posts is smaller than that found between models. The total number of fission reactions has been evaluated to be within the range of 1{4 1018 .5) The activity of each fission product can be calculated by multiplying its fission yield by the decay constant and the total number of fission reactions. By assuming that the total number of fission reactions is 2:5 1018 , the cumulative activities of the produced radioactive noble gases (nuclides listed in Table 3) were calculated to be 1:1 103 TBq. The cumulative activity of the released radioactive noble gases evaluated by assuming the retention time in the JCO facility to be 5 min was 4:7 102 TBq. The integration of the estimated temporal variation of release rate expressed by Eq. (12) becomes 3:0 102 TBq. Uncertainty in retention time is considered to be one of causes for the smaller value obtained in the present study. If the retention time is assumed to be 10 min, the cumulative activities derived from the total number of fission reactions and estimated release rate are 2:4 102 and 2:1 102 TBq, respectively. This retention time of 10 min is estimated by a trial-and-error procedure. This longer retention time seems to be reasonable, considering the transfer pathway of the radioactive noble gases produced by the nuclear fission. These gases were expected to be retained in the precipitation tank just after the production and diffused to the conversion building, and then released from the experimental conversion building through the ventilation system into the atmosphere. The retention time of 5 min JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY
Release Rate Estimation of Radioactive Noble Gases in the Criticality Accident at Tokai-Mura
considers only the performance of the ventilation system at the JCO facility. Therefore, the retention time longer than 5 min seems more likely as the total retention time. JAERI Task Force found that the retention time of the radioactive noble gases in the precipitation tank was about 490 s by using the simulation experiments in TRACY.5) Our results are consistent with this report.
Downloaded by [94.20.20.135] at 00:04 21 March 2014
VI. Conclusions In the present study, the release rate of the radioactive noble gases in the JCO accidental case has been estimated from the combination of the gamma dose rate increase measured at off-site monitoring posts and the calculated ones from the atmospheric dispersion model. To examine the dependence of uncertainty of estimated release rate on that of wind field, two kinds of the wind fields were used: one from the meteorological model MM5 and the other from analysis of the measured wind data. It is shown by the comparison between the two wind fields and the time series of wind data observed at the monitoring post in the vicinity of the accidental site that the analyzed wind field was in better agreement with the measured wind data than the MM5 wind field. The merit of using the better wind field was shown by the smaller geometric standard deviation of 2.88 in the analyzed wind case than that of 3.44 in the MM5 wind case. The differences in the geometric means of estimated release rate between the present study and the JAERI Task Force’s results, that is 2.8 and 8.0 TBq h1 , respectively, indicate that the differences in the configuration of dispersion models may result in relatively large differences in estimated release rate. A gradual decrease in release rate with time was clearly shown in the analyzed wind case, while it was obscured by a large scatter in the MM5 wind case. The similarity in the pattern of temporal changes between the estimated release rate and the measured neutron count rate was considered to indicate the consistency of the estimated results. The integration of the estimated temporal change with time resulted in the total release of 3:0 102 TBq of noble gasses. Although this value is smaller than the total activity of the radioactive noble gases evaluated from the number of fission reactions, this difference can be reduced by using a retention time of 10 min instead of 5 min used in the present analysis. This longer retention time resulted in the estimated release rate of 2:1 102 TBq, which is consistent with the value of 2:4 102 TBq from the fission yield calculation. Based on these results, it was demonstrated that the present procedure provides a plausible release rate if the location and time of release are available, results of gamma dose rate calculation is sufficiently accurate, and the spatial density of monitoring posts is sufficient. The first condition is likely to be met because information from the site of accident will include these fundamental information. As for the second condition, it was shown in this study that the use of observed wind data, which become available simultaneously with gamma dose rate data for the release rate estimation, substantially improved the gamma dose rate calculation. The third condition may be critical when we consider the VOL. 47, NO. 1, JANUARY 2010
29
situations around existing nuclear power plants. The spatial coverage of monitoring posts around many plants seems insufficient because of the existence of seas and mountains. Further consideration is needed on this point for each site.
Acknowledgments This research was partially supported by the Ministry of Education, Culture, Sports, Science and Technology (MEXT), a Grant-in-Aid for the 21st Century COE Program, ‘‘Isotope Science and Engineering from Basics to Applications.’’ References 1) M. Chino, T. Adachi, ‘‘Present status and future prospect on system for prediction of environmental emergency dose information, SPEEDI,’’ J. At. Energy Soc. Jpn., 45[5], 296–301 (2003), [in Japanese]. 2) J. S. Nasstrom, G. Sugiyama, R. L. Baskett et al., ‘‘The national atmospheric release advisory center modeling and decision support system for radiological and nuclear emergency preparedness and response,’’ Int. J. Emerg. Manag., 4[3], 524– 550 (2007). 3) H. Nagai, H. Yamazawa, ‘‘Utilization of mesoscale atmospheric dynamic model PHYSIC as a meteorological forecast model in nuclear emergency response system,’’ J. Nucl. Sci. Technol., 34[8], 835–846 (1997). 4) S. Fujine, M. Murata, H. Abe et al., Cause Finding Experiments and Environmental Analysis on the Accident of the Fire and Explosion in TRP Bituminization Facility, JAERIResearch 99-056, Japan Atomic Energy Research Institute (1997), [in Japanese]. 5) JAERI Task Force for Supporting the Investigation of JCO, JAERI’s Activities in JCO Accident, JAERI-Tech 2000-074, Japan Atomic Energy Research Institute (2000), [in Japanese]. 6) M. Chino, H. Ishikawa, H. Yamazawa et al., Application of the SPEEDI System to the Chernobyl Reactor Accident, JAERI-M 86-142, Japan Atomic Energy Research Institute (1986). 7) H. Yamazawa, ‘‘Long-range dispersion analysis on accidental atmospheric release of cesium-137 at algeciras,’’ J. At. Energy Soc. Jpn., 41[2], 114–116 (1999), [in Japanese]. 8) H. Yamazawa, ‘‘Source term estimation method using longrange inverse atmospheric transport simulation,’’ J. At. Energy Soc. Jpn., 40[11], 885–891 (1998), [in Japanese]. 9) A. Furuno, M. Chino, H. Yamazawa, ‘‘Development of a source term estimation method for nuclear emergency by long-range atmospheric dispersion simulations,’’ J. At. Energy Soc. Jpn., 5[3], 229–240 (2006), [in Japanese]. 10) H. Terada, A. Furuno, M. Chino, ‘‘Improvement of worldwide version of system for prediction of environmental emergency dose information (WSPEEDI), (I),’’ J. Nucl. Sci. Technol., 41[5], 632–640 (2004). 11) M. Chino, H. Ishikawa, ‘‘Dose evaluation model in complex terrain by using particle diffusion method combined with three-dimensional wind field,’’ J. At. Energy Soc. Jpn., 26[6], 526–534 (1984), [in Japanese]. 12) A. Hidaka, M. Kai, ‘‘Use of photon energy released from nuclide for estimation of gamma doses due to radioactive plumes,’’ J. At. Energy Soc. Jpn., 29[11], 1023–1029 (1987), [in Japanese]. 13) G. A. Grell, J. Dudhida, D. R. Stauffer, A Description of the Fifth-Generation Penn State/NCAR Mesoscale Model (MM5),
30
dent in JCO, JNC TN8440 2001-004, Japan Nuclear Cycle Development Institute (2001), [in Japanese]. 16) A. Endo, Y. Yamaguchi, Y. Sakamono et al., ‘‘External doses in the environment from the Tokai-mura criticality accident,’’ Radiat. Protect. Dosim., 93[3], 207–214 (2001). 17) K. Tasaka, J. Katakura, H. Ihara et al., JNDC Nuclear Data Library of Fission Products—Second Version—, JAERI 1320, Japan Atomic Energy Research Institute (1999).
Downloaded by [94.20.20.135] at 00:04 21 March 2014
NCAR Tech. Note NCAR/TN-389+STR, National Center for Atmospheric Research (NCAR), 117 (1994). 14) International Atomic Energy Agency, Repot on the Preliminary Fact Finding Mission Following the Accident at the Nuclear Fuel Processing Facility in Tokaimura, Japan, IAEA-TOAC, International Atomic Energy Agency (1999). 15) N. Miyagawa, H. Watanabe, T. Shimizu et al., The Results of the Environmental Monitoring Related to the Criticality Acci-
S. HIRAO and H. YAMAZAWA
JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY
