Preliminary forensic engineering study on aggravation of radioactive releases during the Fukushima Daiichi accident

Nuclear Engineering and Design 324 (2017) 315–336

Contents lists available at ScienceDirect

Nuclear Engineering and Design
journal homepage: www.elsevier.com/locate/nucengdes

Preliminary forensic engineering study on aggravation of radioactive
releases during the Fukushima Daiichi accident

MARK

Genn Saji
Independent Research and Consultant, Ex-Secretariat of Nuclear Safety Commission (retired), Japan

A B S T R A C T
Even after 6 years since the Fukushima accident, the exact accident progression for each unit and location of core
debris have not been clarified, although solidified low-melting metal debris was identified at the bottom of the
1F3 reactor pressure vessel (RPV) in July 2017. Currently efforts are directed towards robotic inspection with
remote cameras, as well as dose and temperature measurements of the environment inside the Primary
Containment Vessels (PCV). In spite of their effort, the observed environmental temperature distribution data
does not support the existence of a significant radiation heat sources attributable to the molten core at the
bottom of the PCV. At this point the total decay heat of the core debris should be as large as a few hundred
kilowatts after 2000 days since the initiation of the accident as summarized in Annex A. In addition, the temperature of the accumulated water inside PCVs is 10–20 °C even with a reduction of the water injection to
3.0 m3/h. The RPVs appear to be holding the heating core debris with water, implying that “in-vessel retention”
of core debris has been achieved thanks to effective accident management. This should help greatly in the
retrieval of the core debris by removing the top head of RPV.
Under these circumstances the author has conducted a forensic engineering study (i.e., different fields of
science working together to collect and integrate independent evidences) to clarify the most likely accident
scenarios of the Fukushima Daiichi accident. Through this study the author identified that a large portion of the
land contamination observed at the north-west direction is mostly the result of the accident that occurred at Unit

2.
In this unit its blowout panel, provided for over-pressure protection against a main steam pipe breach in the
secondary confinement building, was inadvertently activated before a leakage of radioactive effluent from the
PCV. Its activation is believed due to the hydrogen explosion in Unit 1 which occurred on March 12, next day of
the Fukushima accident initiation. By losing the confinement function, the radioactive effluent leaked from the
1F2’s PCV and would have been discharged without mitigation. This accident scenario explains the series of
leakage events identified in two of the 24 monitoring posts which had been installed by the Fukushima
Prefectural Government. A series of six large releases were repeated between March 15 and 16, behaving like a
periodical actuation of the safety valves for the PCV. Such multiple release events were very likely induced by
the overpressure release of the PCV due to leakage of the dry well flange. This leakage should have been induced
through discharging steam and hydrogen due to the activation of Safety and Release Valves (SRV) into the
suppression pool (SP) water. Unfortunately the wet well atmosphere must have been that of air, since there was
no nitrogen charge line to the atmosphere of the SP surface water. The resultant air-hydrogen mixture resulted in
an “internal hydrogen explosion” which should have deformed the flange. The recent robotic inspection inside
PCV revealed that a gigantic water splash appears to have occurred at the bottom of the PVC dislodging the
gratings installed over the platform.
Next to the series of large releases from Unit 2, Unit 1 also induced two large releases on March 12.
Fortunately, these releases left more than 3 orders of magnitude less soil contamination compared with the series
of releases from Unit 2. Unit 3 also released a significant amount of radioactive species twice on March 13,
resulting in a very small soil contamination. The main constituent of the radioactivity is likely radioactive noble
gases in this unit.

E-mail address:
/>Received 28 February 2017; Received in revised form 31 July 2017; Accepted 7 August 2017
Available online 21 September 2017
0029-5493/ © 2017 The Author. Published by Elsevier B.V. This is an open access article under the CC BY license ( />

Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

Nomenclature

R/B
RCIC
RHRS
RPV
SBO
S/C
S/P
SGTS
SRV
SPDS
NPP
PCV
TEPCO
T/H
W/W

1F1∼1F4 Fukushima Daiichi Unit 1∼4
BFL
basement-floor level
DBA/E design basis accident/event
DID
defense-in-depth
D/W
drywell
ECCS
emergency core cooling system
FDA

Fukushima Daiichi accident
HPCI
high pressure coolant injection
I&C
Instrumentation and Control
LOCA
loss of coolant accident
LPCI
low pressure coolant injection
LUHS
loss of ultimate heat sink
PCV
Primary Containment Vessel

reactor building
reactor core isolation cooling system
residual heat removal system
reactor pressure vessel
station blackout (loss of all AC power)
suppression chamber
suppression pool
standby gas treatment system
Safety/Release Valve
safety parameter display system
nuclear power plant
Primary Containment Vessel
Tokyo electric power company
Turbine Hall
wet well (suppression pool)

(LUHS)” and further aggravated through the “loss of I & C (Instrumentation and Control) power.” By losing all safety provisions to
control the troubled reactors, a series of environmental release events
followed during the active phase of the accident which took place
during the course of the first week. Unfortunately, the amount of
radioactive species and timing of the large environmental releases are
still not known, since all of the environmental monitoring stations
surrounding the site boundary were wiped out due to the “loss of I & C
power”. An overall accident scenario is illustrated in Fig. 1 in the eventtree formalism.

1. Introduction
1.1. Global accident sequence
The accident at the Fukushima Daiichi nuclear power station in
Japan is one of the most serious in operating history for a commercial
nuclear power plant (Nuclear Emergency Response Headquarters, June
and September 2011: National Diet of Japan, 2012; (National
Government's) Investigation Committee, 2012; TEPCO's Investigation
Report, 2012; Atomic Energy Society of Japan, 2015; RJIF, 2014). The
author has also published Post Accident Safety Analysis Report of the
Fukushima Accident – Future Direction of Evacuation: Lessons Learned
(Saji, 2013).
The tsunami, which arrived at 15:37 on March 11, 2011 brought the
plants into an unprecedented severe accident status of prolonged SBO
(NUREG/CR-5850, 1994) combined with the “loss of ultimate heat sink

1.2. Breakdown of fundamental safety approach – Significance of the
Fukushima severe accident
Assuming a LOCA as its DBA following the well-accepted hypothesis
from the 1970s, the fundamental approach for safety assurance of the

Fig. 1. Simplified accident sequence for Fukushima Daiichi Unit 1 to Unit 3. Note: The disaster was triggered by the gigantic earthquake which induced the loss of offsite power. Although

DGs started as designed, they all failed 51 min later due to the tsunami, which submerged seawater pumps disrupting the discharge of residual heat to the ocean and induced LUHS.
Resultant degradation of core cooling and loss of the hydrogen removal function induced a hydrogen explosion which devastated the reactor buildings.

316


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

determine progressive causes of the event. Since many law suits are
being raised after the Fukushima Daiichi Accident (FDA) against
TEPCO, the Government Investigation Committee reports should have
considered such forensic engineering investigation. Instead, the testimony collected during such investigation is not being disclosed in order
to weigh more on an investigation of the truth of the accident details.
For example, nine months after the accident, the Fukushima Nuclear
Accident Independent Investigation Commission was established by a
unanimous resolution of both the House of Representatives and the
House of Councilors of the National Diet, which represent the people of
Japan (National Diet of Japan, 2012). In the preface of the final report,
Dr. K. Kurokawa explains the objectives of the committee which includes the following statement. “To investigate what was at the center
of this accident, we could not but touch upon the root of the problems
of the former regulators and their relationship structure with the operators.” In line with the committee’s indication the new regulatory
framework (i.e., Nuclear Regulatory Agency (NRA)) was established by
abolishing the previous Nuclear Safety Commission (NSC). Note that
the forensic engineering studies are not in the basis for the investigation
committee’s survey.
However Dr. J.O. Henrie’s approach in investigating the TMI accident is impossible to apply in the study on FDA, since the process data
are not available unlike in the case of TMI accident. In addition, Dr.
Henrie focused on H2 generated by the reaction of zirconium with

water, by stating that “H2 generated by radiolysis was probably insignificant”. The author has theoretically investigated the root cause of
hydrogen generation during the FDA and found that the “radiationinduced electrolysis” is more likely than radiolysis (Saji, 2016). It
clarifies that the hydrogen generation may not have been from the high
temperature zirconium– steam reaction. A short overview of the hydrogen generation mechanisms is summarized in Annex B. Rather it is
more likely due to “radiation-induced electrolysis” occurring with a
“different radiation cell” configuration (Saji, 2017). With this mechanism, a large amount of hydrogen is generated before the loss of the
reactor water level. It also explains the root cause of the hydrogen
explosion that occurred in Unit 4, where all of the fuel assemblies in the
reactor core were evacuated to the spent fuel pool for special maintenance2 at the time of the accident.
Since there is a high possibility that hydrogen generation may neither be through the zirconium-steam reaction nor radiolysis, more robust evidence of the footprints for the accident is necessary. In view of
this, diverse evidence were collected and integrated in this forensic
engineering study, although the author does not intend to go into the
legal implications. Typical data included:

Fukushima Daiichi has been deployed. In this approach, the DBA-LOCA
should envelope a spectrum of accidents induced by the malfunction of
equipment and human errors. This practice was imported by referring
to the many plants with a firm construction and performance record
developed in the Mid- to Eastern US where the seismic events are not
dominant. “Design Basis Events (DBE) are defined as conditions of
normal operation, including anticipated operational occurrences, design basis accidents, external events, and natural phenomena for which
the plant must be designed to ensure functions.“ (NRC 50.49).
Historically “defense-in-depth (DID)” is the basic approach for
prevention of the occurrence of DBEs, as well as for protection from the
further evolution of the events and mitigation of their consequences.
The concept of the DID has been fortified by incorporating lessons
learned from accidents and operational experiences. One of the respected textbooks on DID is the IAEA’s INSAG-10 which classifies five
levels of defense as extracted below (IAEA, 1996):
Level 1: Prevention of abnormal operation and failures
Level 2: Control of abnormal operation and detection of failures

Level 3: Control of accidents within the design basis
Level 4: Control of severe conditions including prevention of accident progression and mitigation of the consequences of a severe
accident
Level 5: Mitigation of the radiological consequences of significant
external releases of radioactive materials
In referring to this ranking, the first 3 layers related to the DBE were
wiped out upon the tsunami’s arrival. Nevertheless there should have
been some means to control severe accidents and/or to mitigate their
consequences in relation to Level 4 of DID. This is the main objective of
this report. It is to provide feedback as to why the TEPCO’s commendable accident management effort could not sufficiently prevent
gross release of radioactive species to the environment. This induced
prolonged off-site counter measures (Level 5) resulting in prolonged
evacuation of the general public1 and more than 2000 “disaster-related
premature early deaths” due to stresses and deterioration of Quality of
Life of the evacuees, although there were no radiological health death
toll.
Due to the difficulties of predicting natural phenomena, occurrence
frequency and prediction of consequences from earthquakes and tsunamis, the Fukushima Daiichi did not specify a well defined design base
for tsunamis. Instead, TEPCO referred to historical experiences for
tsunamis which indicated OP + 5.7 as the maximum tsunamis, although they raised the level to 6.1 m on 2006. The ground level of the
site was set at 10 m above average sea level (OP. Onahama Pail).
However, the OP + 13.1 m tsunami ran up the site and flooded the D/G
and electrical panels, resulted in unprecedented prolonged SBO. This
indicates the difficulties in predicting the natural phenomena and
protecting the site against such phenomena. The author published a
new approach for incorporation of risk for earthquakes and tsunamis
(Saji, 2014).

• Water samples acquired from the T/H. Detailed characteristic data
•

•
•

1.3. Forensic engineering studies on severe accidents
Recently forensic engineering practices are being compiled especially in the field of civil engineering (Terwel et al., 2012). Formally
forensic engineering also involves testimony on the findings of these
investigations before a court of law or other judicial forum, when required. A similar approach for the evaluation of the 1979 Three Mile
Island Unit 2 accident was performed by James O. Henrie without legal
implication (Henrie, 1989). Unlike a computer simulation of an event,
forensic engineering is the evaluation of the recorded data and damages
as well as examining the surviving components after an event to
1

•
•

obtained by TEPCO and in which chemical and radiological analyses
were performed by JAEA soon after the accident.
Independent radiation monitoring data obtained around the periphery regions of the Daiichi site by the Fukushima Municipal
Government.
Remote aerial 131I measurement data taken by USDOE and recovered by JAEA in the soil contamination data.
Chronological data compiled by TEPCO from March 11–16. The
data contains detailed operator actions, local indications of instrumentation and observations of the actual state of components.
TEPCO’s robotic inspection data soon after the accident as well as
the recent robotic inspection data taken inside the PCVs and 1F1
torus room.
Identification of “core debris” inside the Unit 3 PCV by introducing a
submarine type robot in July 2017.

2

Replacement of its reactor core shroud which is a stainless steel cylinder surrounding
a nuclear reactor core. It helps by directing this cool water towards the reactor core,
providing stability to the nuclear reactions.

Even after 5 yeas there remain as many as 128,000 evacuees in Fukushima.

317


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

• Muon imaging data taken and analyzed jointly by several research
institutes.
• Hamaoka Unit 1 accident analysis reports.
• The author’s study on “radiation-induced electrolysis” (Saji, 2016,

RPVs, the injected water should leak down to the PCV. Since the flow
rate of the injected water is still as much as 3 m3/h (TEPCO, 2017b), the
PCVs should have been filled with the injected water. However, such a
difficulty has not been experienced. In addition, the temperature of the
accumulated water in the PCV that is at an ambient temperature. This
observation is not consistent with the melt-through scenario, since the
decay heat, estimated by TEPCO is 70/90/90 kW for Units 1–3, respectively (TEPCO, 2017b). Their water temperatures are around 17 °C
whereas the temperature at the bottom of the RPV is 21.5/27.3/24.8 for
Units 1–3 respectively.
More recently, starting July 19, 2017, TEPCO released a series of
results of investigation of Unit 3 PCV by employing a newly developed
submarine type (ROV) robot (TEPCO, 2017c). The most remarkable

finding is that they finally identified “core debris” drooping down from
the bottom head of RPV like a stalactite and then piled up inside of the
accumulated water at the bottom of PCV. In their photos it is remarkable to observe that there is no steaming, although the estimated total
decay heat is as large as 127 kW in 1F3 after 2000 days since the reactor
trip. If the debris contains some fraction of the core materials, steam
bubble should be visible. This observation suggests that the currently
observed debris should consist of solidified molten metals. More detailed discussion is included in Section 5.3.
Under this situation, what is needed is an unfolding method to
identify the root cause of the accident based on the footprints left by the
accident. This is the basic motivation for writing this paper.

2017).

2. Unique features of the Fukushima Daiichi accident
2.1. Fukushima disaster: un-experienced type of severe accident
The Fukushima accident is an un-experienced type of severe accident with the loss of the ultimate heat sink (LUHS) and prolonged SBO
which was further aggravated by the loss of I & C power. With these unexperienced accident initiators, the author believes that attention
should not be focused mainly on the current core status based on the
TMI experience. Rather it is essential to investigate whether the various
safety provisions, notably containment vessels and ECCS, which are
designed based on DBA-LOCA, have some residual safety functions even
under LUHS and prolonged SBO. This accident is very different from
both TMI (small LOCA aggravated by inappropriate operator actions)
and Chernobyl (reactivity accident due to intrinsically unsafe graphite
moderated light water cooled reactor configuration with a positive void
coefficient). The extrapolation of the knowledge on the degraded core
status obtained from these earlier severe accidents may not be applicable.
Currently a “melt-down” scenario has been hypothesized and is
practically the consensus in Japan. Although it has not been internationally defined as such by IAEA, this appears to mean that the reactor fuels have molten together with in-core structural materials
forming a volume of “corium.” Due to high decay heat during the active

phase of the severe accident (as long as 5 days in Daiichi), the corium is
considered to have melted through the lower head of the RPVs and
relocated at the bottom of PCVs.
Based on this hypothesis, most of TEPCO’s decommission activities
have been directed towards identifying the location and properties of
corium, which should be a lava-like mixture of nuclear fuel and other
structural materials first observed after the Chernobyl accident. In spite
of their effort, TEPCO has not been able to confirm its existence at the
bottom of the PCVs, where the corium should have melted through the
lower heads of the RPVs.
The “melt-down” scenario was developed through mechanistic
analyses by running severe accident analysis codes (e.g., MELCORE,
MAAP and SAMPSON). Unfortunately no data from the SPDS (safety
parameter display system) was available to verify the code predictions
during the active phase of the FDA. The SPDS was developed based on
the TMI experience and installed in most NPPs around the world. It was
also installed at the Fukushima Daiichi, however it was not available for
accident management and little data was recorded due to the loss of the
I & C’s power and failure of its local detectors (Toshiba, 2012).
Nevertheless, since these codes predicted a series of melt through of
the lower reactor vessel heads, current efforts are directed towards
robotic inspection of the bottom of the heads with remote cameras as
well as dose and temperature measurements of the environment inside
of the PCVs. In spite of their effort, the observed data does not support
the existence of the large radiation heat sources attributable to corium.
In addition recent cosmic ray muon imaging results from 1F2 concluded that most of the core debris should still be retained in the lower
head as well as on the Reactor Support Plate. These results were released on July 26, 2016 by a team from TEPCO, IRID, KEK, Tsukuba
University and Tokyo Metropolitan University (TEPCO, 2016). The results imply that the mechanistic analyses applying the severe accident
analysis codes may not be reliable without plant data when performing
the reverse (unfolding) analysis.

However, the “meltdown” hypothesis should be reconsidered in
view of the recent robotic inspection results inside of the PCVs (TEPCO,
2017a). If the core debris has molten through the bottom head of the

2.2. Unit-specific accident progression
During the Fukushima Daiichi accident, which occurred at 15:37 on
March 11, 2011, each unit revealed its own peculiar accident progression as summarized below.

• 1F1:
•
•
•

Very early leakage of radioactive species into the reactor
building, as early as 21:51 on March 11, followed by the initial
hydrogen explosion, which occurred the next day, March 12. This
motivated many analysts to assume the very early onset of a zirconium-hydrogen reaction.
1F2: the largest leakage of highly contaminated water into the
basement of the Turbine Hall (T/H). This unit is likely responsible
for the severe soil contamination with radioactive cesium deposited
in a northwest direction from the plant. It also contaminated the
near field with radioactive iodine to the south of the plant.
1F3: the most severe hydrogen explosion among these 4 units:
however, there is no trace of radioactive contamination attributable
to this unit. Nevertheless un-heating solidified debris has been
identified recently.
1F4: hydrogen explosion occurred even though its reactor core had
been evacuated to the spent fuel pit. No serious radioactive releases.

Even after 6 years since the accident, the exact accident scenarios

that explain these morphologies remain unknown.
2.3. GE’s assessment on performance of Mark I containments at Fukushima
Daiichi
Fukushima Daiichi Units 1–4 are BWRs were constructed by importingthe technical bases of Mark I containments (GE Report, 03/19/
2011). The following statements are quoted from GE’s summary page,
although the author is not necessarily in agreement with their entire
statement.
“Early reports regarding Units 1–3 stated plant operators used safety
relief valves to relieve pressure in the reactor pressure vessel. In
addition, when the fuel rods became uncovered, hydrogen formed in
the core (due to zirconium/water reaction) and was also transported
into the wet well when the reactor vessel safety relief valves opened.
The combination of steam and hydrogen flowing into the wet well
318


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

increased the temperature and pressure. Since there was no on- or
off-site power available, there was no means of cooling the wet well
water. Over time, the pressure in the primary containment rose over
the design pressure. To avoid a containment breach, venting became
necessary. Upon venting it is believed that vented hydrogen gas
caused the explosions at these units.”

Table 1
Results of nuclide analysis by JAEA (Bq/ml).
Sample

Sampling date: mm.dd
Start of Counting:
mm.dd
Duration of Counting
(s)
I-131(Bq/
8.04d
ml)
Cs-134
2.06a
Cs-137
30.0d
Ba-140
12.7d
La-140
40.3 h
Sr-89
50.5d
Sr-90
29.12a

GE noted the following points:

• Concurrent long-term loss of both on-site and off-site power for an
•
•

extended period of time is a beyond-design-basis event for the primary containment on any operating nuclear power plant.
The Mark I containment vessels appeared to have held pressure to
well above the design pressure.

The response of the reactor pressure vessel and reactor in general
agree with the severe accident management studies performed in
the 1980s and early 1990s.

1F1 T/B BFL
Accum. Water
2011.3.24.9.40
2011.4.13.12.50

1F2 T/B BFL At
S. Stairway
2011.3.27.20.40
2011.4.13.18.34

1F3 T/B BFL
Leaked in Water
2011.3.24.21.00
2011.4.14.21.00

2000

2000

2000

3.00E+04

2.00E+06

1.60E+05

1.20+05
1.60+05
560
300
5.70E+01
2.10E+01

2.60E+06
2.80E+06
2.40E+05
2.20E+05
7.00E+05
1.40E+05

1.40E+05
1.60E+05
1.50E+04
1.70E+04
8.60E+04
1.50E+04

containing radioactive species, accumulated in the basements of the R/
B as well as the T/H. The injected water carried the radioactive species
and leaked out of the RPV into the PCV, until finally accumulating into
the basements of T/Hs. The continued water injection gradually raised
the water level, on top of the tsunami water, both of which flooded the
basements of these buildings. When the accumulated water purification
systems were constructed and began operating around June 17, a
portion of the water volume as well as radioactive species were removed. There is a further complexity to the volume of accumulated

water, due to an unidentified amount from an underground water
source which seeped into the original amount.
In spite of these uncertainties, it should be possible to estimate the
leakage of radioactive species from the PCV by knowing the concentration of radioactive species and total volume of the accumulated
water on the same sampling date. The amounts of release not accounted
for are: (1) direct release from the R/B to the environment, such as
through the hydrogen explosions, venting events through the operator’s
actions, as well as several “spontaneous venting” events; (2) an amount
of radioactivity deposited inside of the R/B but not in the accumulated
water in the basement of the T/H; (3) sludge of the insoluble species
(e.g., refractory species, such as Zr, Mo, Ce, Np, Pu, Cu, U and intermediate species such as Sr, Ru, Ba) likely deposited at the bottom of the
basement.
On May 22, 2011, TEPCO released “Results of Analysis of
Accumulated Water in the Turbine Building (JAEA)” (TEPCO, 2011a) as
summarized in Table 1. The sampling was made under a very high
radiation field, with contact dose rates of nearly 1 Sv/h. Although it
may not be representative with just one set of samplings, the data are
very precious.
In principle, by knowing the total volume of accumulated water, the
total amount of radioactive species leaked in this pathway can be estimated. Unfortunately, the levels of the accumulated water only became available months after sampling. For this reason, the total amount
of accumulated water was assumed to be close to the estimated total
amount of injected water up to the sampling dates as shown in Table 2.
By multiplying the values in Tables 1 and 2, and adjusting the results to the shutdown activity on March 11, the total radioactive inventory in Bq are summarized in Table 3.
This table is converted to a fraction of radioactive species with respect to the shutdown inventory in Table 4.
The results indicate that the release fraction (i.e., the ratio of released/shutdown inventory of radioactive species) into the

However, a similar statement of “a beyond-design-basis event”
made by TEPCO was thoroughly criticized by the Japanese public and
has not exonerated them from blame for the accident. The author intends to clarify why the radiation exposure to the public could not have
been evaded, in spite of TEPCO’s commendable accident management

activities under very restrictive situations.
2.4. Bypass leak in Unit 2 through drywell flange leakage
GE’s early assessment as quoted above indicates that the Mark I
containment vessels appeared to have held pressure to well above the
design pressure. However, the environmental monitoring data which
will be discussed in Chapter 4 indicates that Unit 2’s dry well flange
leaked several times on March 15–16, behaving like a mechanical safety
valve resulting in a series of puff releases (i.e., several short duration
releases due to overpressure in the PCV) which were not filtered nor
confined for removal of aerosol through precipitation on the walls.
Such a release during the severe accident was preventable as stated
in the IAEA’s Safety Guide No. NS-G-1.10 (IAEA, 2004). The following
basic requirement is quoted from Section 6.4 of IAEA’s Safety Guide.
“For existing plants, the phenomena relating to possible severe accidents and their consequences should be carefully analyzed to identify
design margins and measures for accident management that can be
carried out to prevent or mitigate the consequences of severe accidents.” Additionally, the blowout panel of the R/B was inadvertently
opened which was said to be due to the hydrogen explosion in nearby
Unit 1, which occurred on March 12. With the large hole in the R/B, it
was unable to serve as a secondary confinement function resulting in a
bypass leakage directly from the reactor vessel. Note that the SGTS,
which was supposed to filter the effluent leakage before discharging
from the stack, was not available due to SBO.
With an unfiltered bypass leakage, the main cause of the environmental contamination to the NWN direction from the plant induced
three orders of magnitude more severe land contaminations brought on
by the events at Units 1 and 3. Apparently the dry well flange did not
have the necessary safety margins for the FDA.
3. Amount of radioactive releases
3.1. Radioactivity leaked into the basement of T/H
The forward analysis to predict the amount released to the environment is extremely difficult, however an indirect estimation can be
made from the sampling data from the accumulated water as explained

in this section. The pure-water/seawater injection was initiated during
the early phase of the accident and continued during the removal of
decay heat. Since the injected amount exceeded the necessary amount
for decay heat removal with evaporation, a large portion of the water,

Table 2
Estimated volume of accumulated water.
Location
3

Total injection (m )

319

1F1 T/H BFL

1F2 T/H BFL

1F3 T/H BFL

2633

9824

4225


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

Table 3
Total amount of radioactive species in accumulated water.

Table 5
Cs-137 deposition and areas.

Species

Half Life

1F1 T/H BFL

1F2 T/H BFL

1F3 T/H BFL

Zone

Cs-137 deposition

Area (km2)

Cs-137(PBq)

I-131
Cs-134
Cs-137
Ba-140
La-140

Sr-89
Sr-90

8.04d
2.06a
30.0d
12.7d
40.3 h
50.5d
29.12a

1.36E+15
3.16E+14
4.21E+14
8.93E+12
6.52E+17
2.36E+11
5.53E+10

3.38E+17
2.55E+16
2.75E+16
1.43E+16
1.78E+21
1.08E+16
3.69E+14

1.16E+16
5.92E+14
6.76E+14

3.84E+14
5.93E+19
5.72E+14
3.95E+13

VIII
VII
VI
V
IV
III
II
I
0
Total

> 3E+06 (Bq/m2)
3E+06 – 1E+06
1E+06 – 6E+05
6E+05 – 3E+05
3E+05 – 1E+05
1E+05 – 6E+04
6E+04 – 3E+04
3E+04 – 1E+04
< 1E+04
1E+04 – 6E+06

76
304
208

292
2048
769
2101
6862
NA
12,660

0.34
0.61
0.17
1.31
0.41
0.06
0.10
0.14
NA
3.13

Table 4
Ratio of released/shutdown inventories.
Species

Half Life

1F1 T/H BFL

1F2 T/H BFL

1F3 T/H BFL

I-131
Cs-134
Cs-137
Ba-140
La-140
Sr-89
Sr-90

8.04d
2.06a
30.0d
12.7d
40.3 h
50.5d
29.12a

1.02E−03
3.57E−04
3.76E−03
3.41E−06
NA
1.38E−07
5.57E−07

1.47E−01
1.67E−02
1.42E−01
3.16E−03
NA

3.67E−03
2.15E−03

5.07E−03
3.88E−04
3.49E−03
8.50E−05
NA
1.94E−04
2.31E−04

Daiichi to the Emergency Response Center to assist in performing quick
dispersion calculations by SPEEDI for accident management. However,
due to the SBO’s creating further adverse conditions which were induced by the “loss of I & C power”, neither the environmental radiation
monitoring nor plant data from process computers were available.
Due to this lack of info related to the event, one of the most precious
data was obtained via remote monitoring by the US DOE/NNEA, where
the first survey data was disclosed on March 22 (US DOE/NNEA, 2011).
More detailed data were collected by fixed wing and helicopter survey
flights at altitudes ranging from 150 to 700 meters. The cesium deposition was determined from aerial and ground-based measurements.
Using US technology, the Ministry of Culture, Education and Sports
(MEXT) produced wider land contamination maps, which were verified
with detailed in-situ soil measurements. Table 5 has been updated by
using the new set of data for the integration of the total amount of 137Cs
deposited on the land. The results show approximately a factor of 0.4
less than the author’s previous results (Saji, 2013). The difference was
traced down to the dose range of Zone VIII (previously Zone V), where
the upper bound was reduced by an order of magnitude. Due to the
uncertainties in the remote monitoring soon after the accident with the
strong gamma spectra from radioactive iodine isotopes, it is likely that

the DOE assumed a large safety margin in the most highly contaminated
zone. The in-situ soil sampling data now revealed that the highest dose
rate in the most contaminated region was 3000 kBq/m2.
In updating Table 5, the recently obtained wider (> 100 km from
the Fukushima Daiichi) land contamination map3 was used to include
above 10 kBq/m2 zones, although some decrease in dose rate due to
weathering (approximately 40%/y) is anticipated. Because of this
concern, the earlier August 30, 2012 data4 were used for the region
within 100 km.
With detailed land contamination density data now available, it was
a straightforward process to estimate the total amount of the cesium
simply by calculating the areas of each zone. The estimated gross land
contamination of approximately 3PBq in 137Cs indicates that the environmental release is much less than 75–86 PB (1996 estimation) reported for the Chernobyl accident (UNSCEAR, 2000). The estimated
environmental release fraction of 137Cs, which is the ratio of the environmental releases to the shutdown core inventory, amounts to only
1.5% of the1F1 core inventory or 1.2% for 1F2 and 1F3. The best estimate for the environmental release is 1–2% of the shutdown core
inventory of 137Cs, considering uncertainties in the land contamination
density maps.
This value should not be confused with a fuel failure rate since the
containment system had a large decontamination factor. The actual fuel
failure can be an order or so larger in the reactor vessel. These values
indicate the following land contamination characteristics:

accumulated water ranges from 14% for 1F2 to a fraction of a percent
for “volatile” species (i.e., I and Cs) in 1F1 and 1F3. Those for “intermediate” species (i.e., Ba, La, and Sr) are even smaller in magnitude.
This suggest that the major pathway of radioactive release from the fuel
cladding should be from a puncture in the fuel cladding due to high
temperature ballooning failure. The ballooning failure occurs through
the softening of fuel cladding at a high temperature combined with
overpressure induced with gaseous and volatile species.
The large release fraction for 1F2 indicates that the release passage

of 1F2 may be different from both 1F1 and 1F3. The difference suggests
an involvement of the S/C where the steam, containing the radioactive
species, is discharged from the RPV into the S/C pools. The radioactive
species are decontaminated in the pool water by a factor of two orders
of magnitude of reduction. In addition, the large explosive sound occurred near the S/C between 6:00∼6:10 on March 15 may indicate
another “internal hydrogen explosion” event in the S/C, which resulted
in the leaking of the highly contaminated suppression pool water into
the R/B through degradation of the flange of PCV. Further discussion
will be made in Section 5.2.
Also, the behavior of the release fraction of Cs-134 and Cs-137 are
strange since the former depends strongly on the burn-up through a fuel
management sequence whereas the latter is relatively independent. In
Fukushima Daiichi, the reactor cores contain fuels with a variety of fuel
management histories, some of which date back to more than 10 years.
There is a high possibility that radioactive releases are more from the
aged fuel assemblies, many of which may have punctured due to their
higher internal fission gas pressure. It is necessary to calculate the activation data of each assembly, instead of a core average, when the
status of an individual fuel assembly would be confirmed years from
now.
The release fractions indicate that a large portion of volatile species
(as well as rare gases) have been released from the core into the reactor
water at least in 1F2. Although the estimation of radioactive species
into the accumulated water may become a source term for the marine
environment release, it does not provide any source term information
for the atmospheric releases that produced widespread land contamination around the Fukushima Daiichi.

(1) The accident resulted in a severe land contamination (zone VII and
3.2. Aerial releases of radioactive cesium
The Japanese regulatory body was expecting that the environmental
radiation monitoring data could have been transmitted from Fukushima

3
4

320

/> />

Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

Fig. 2. Wind direction observed at the Fukushima Daiichi site during
the active phase (March 11–16, 2011).

VIII) covering an area of 380km2, where rehabilitation will be
prohibitive without proving a substantial reduction in the dose rate
of greater than a factor of 10.
(2) The total area of high concentration (> 100 kBq/m2) covers an area
of up to 2900 km2, although most of the contaminated area is located in the Abukuma Mountain Chains.

the reliable source of information, however their data were not available for accident management purposes due to SBO. As a quick measure, the on-site data was collected by TEPCO’s monitoring car5, which
showed the changing wind directions every few hours during the first
week of the accident as shown in Fig. 2 (TEPCO, 2011b).
This plot was made by extracting the wind direction data from
TEPCO’s archived monitoring data, recorded from March 11–31, 2011.
Due to the wake (i.e., a track of atmospheric turbulence) of on-site
buildings, this wind direction may not be the true direction of the
elevated release through the hydrogen explosions.
The wind direction is shown in a clockwise azimuth with 0 (or 12)

pointing to the North and 3 o’clock pointing to the East. The original
data was obtained by TEPCO’s two monitoring cars, one mostly parked
near the Front Gate located to the WSW of the Daiichi and the other
near MP-4 in a WWN direction. The wind direction indicated by each
dot is an average value from the previous dot, since the original data
were recorded every 2 ∼ 5 ∼ 10 ∼ 30 ∼ 60 min depending on the
change in dose rates. When the wind direction crosses the 12 o’clock
direction, additional data points were included before and after. This
smoothing of the trend graph was performed to indicate a global trend
for the hourly wind direction. Even with this processing of the data, the
wind direction changed too frequently to identify preferential wind
directions that were stable for a few hours.
When the wind direction value is larger than 6 o’clock, the wind is
blowing from land to the Pacific Ocean. At the time of the hydrogen
explosion in 1F1 (15:36 on March 12), the wind direction was from the
ocean to land in a SSE direction. This wind direction is consistent with
the TV news video where a semi-spherical cloud is moving in a northerly direction. Another “internal hydrogen explosion” occurred in 1F2
with the suppression chamber pressure “down-scaled” (off-scaled to
zero, at 06:14 on March 15). The wind direction was also from the
ocean to land. Prior to this event, at 11:01 on March 14, the hydrogen
explosion occurred in the Reactor Building of 1F3. At the time of this
event, the recorded wind direction was from land to sea. However, this
information is not necessarily consistent with video coverage at the
time of the explosion, as it shows that a tall mushroom cloud was traveling in a southern direction.
In addition to TEPCO’s monitoring posts, the Fukushima Prefectural
Government had 25 monitoring posts scattered around the Fukushima
site. Their data was also not available for accident management purposes due to the tsunami or failure of their telemetry system, in spite of
their battery and engine power backups. Fortunately, most continued to

4. Environmental monitoring of radioactive species

4.1. Analysis of environmental monitoring data
One of the unique features of the Fukushima Daiichi accident is in
the simultaneous evolution of its accident sequences, which resulted in
the environmental radiation releases from one unit after another. The
resultant releases left significant land contamination especially to the
northwest of the damaged plants. In order to appreciate the scientific
implication of the areas marked with the footprints of land contamination, it is necessary to clarify whether they were left with one
dominating release event or the superposition of multiple releases from
different units at different times. The land contamination maps indicate
that the radioactive plumes passed over the area thereby contaminating
the soil. Therefore people who resided in the area at the time of the
plume passage may have inhaled the radioactive effluent, especially in
the form of 131I and 133I.
Although an atmospheric dispersion assessment has been used in the
investigation of the environmental release and consequence assessment
studies (Chino et al., 2011), this approach is not applicable in the case
of the FDA. The reason why an atmospheric dispersion assessment
would not be dependable in the case of the FDA is due to the following
limitations:
(1) No actual source term data was available to begin with.
(2) No release height data was known.
(3) No local wind velocity and on-site wind direction data could be
transmitted for assessment due to SBO.
(4) The terrain of the nearby Abukuma Mountain range is not incorporated in the dispersion model with a flat landscape assumption. For example, Iidate-mura is located 500 meters above sea level
and is one of the most highly contaminated highland stretching in a
NW direction from the Daiichi. The village looks like a gorge
through which the plumes released at different times passing
through this flat. With such a terrain, it is necessary to consider a
change in the atmospheric pressure due to the height which may
have induced the radioactive rainfall.

5
Data from the monitoring car substituted the eight failed stationary monitoring stations by periodically measuring doses near the location of each station. In addition, a
portable survey meter was placed near the front gate, located to the west of the Daiichi.
There was also meteorological equipment in the car.

TEPCO’s monitoring stations installed along a semicircular
boundary of the site (MP-1 ∼ MP-8) and on the stack should have been
321


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

release mechanism. Further explanation will be provided in Section 4.3.

work silently and automatically stored the data measured during the
active phase of the accident on their own hard drives. On September 24,
2012, the Fukushima Prefectural Government recovered the radiation
monitoring data and posted it on their homepage only in Japanese
(Fukushima Prefectural Government, 2012).
These data were plotted for each monitoring station by the author
and displayed in a form of a “mandala” (pictorial disambiguation)6 as
shown in Fig. 3. The figure was highlighted with the following
screening criteria to select eight representative points from the original
23 data sets released from the Fukushima Prefectural Government
home page, namely;

4.2. Chronology of the Fukushima accident

The chronology as summarized in Table 6 is extracted from reference (TEPCO's Investigation Report, 2012) to highlight those events
directly related with the environmental releases.
(1) March 11: Dose rate increased in 1F1 R/B only 6 h after the arrival
of the tsunami.
(2) March 12: Two large peaks in dose rate curve recorded by the
monitoring station located at Yamada station (Fig. 4), occurred
from that morning until noon. They were also recorded at other
monitoring stations located to the NW (#19 Kamihatori, 5.6 km
from Daiichi) and NWN (#22 Namie, 8.6 km from Daiichi) as shown
in Fig. 3. These releases induced severe land contamination. Unit 1
is very likely responsible for these events. These peaks occurred
before the venting operation, indicating leakage from PCV’s flange
without the scrubbing effects of the suppression pool water.
(3) March 13: Two large releases (Fig. 4) were from 1F3, although they
did not leave significant land contamination, since the step-wise
increase in ground shine is only 1.0 nSv/h. They were the result of
the venting from S/C in which scrubbing of the effluent should have
been effective.
(4) March 14: Hydrogen explosion in Unit 3. It left no remarkable
contamination (Figs. 3 and 4), since they are not correlated with
either the venting operation or the explosive phenomena. It means
that the aerosol was deposited onto the inner wall of the reactor
building by the time of the hydrogen explosion occurred. TEPCO’s
on-site meteorological data was indicating that the wind was towards the in-land direction; therefore the plume should have been
detected, if it reached one of the monitoring posts.
(5) March 15: Three large peaks in the dose rate curve (Fig. 5) are
recorded in the monitoring station located in the WNW (#16 Yamada, 4.1 km from Daiichi), leaving behind severe land contamination. Unit 2 is very likely responsible for these events. In
particular the spontaneous depressurization of D/W, which occurred around noon, appears to have been the result of effluent
releases without scrubbing by the suppression pool water. The
leakages occurred three times since step-wise increases in the

ground shine after the plume passages are repeated three times.
These phenomena are likely due to leakages from the D/W flange of
the PCV.
(6) March 16: Following the large releases that occurred on March 15
from 1F2 (Fig. 5), two additional large leakages continued through
the next day as shown in Fig. 4. The amount of these releases are
among the largest, and likely from 1F2. Unfortunately TEPCO’s
timeline does not cover March 16 and no data is available for further study.
(7) The exact timing of the hydrogen explosion in 1F4 is not known
although the large explosive sound and quaking was felt at 06:14 on
March 14 at the Unit 4 side ceiling of Units 3/4's common control
room. It is likely due to this event, although, damage to the 5th
floor of the R/B was visually confirmed at 06:55 on March 15.

• One representative location from each of the 8 sectors, covering the
western half of the Daiichi.
• Including hourly data at least until March 15, 2011.
These representative data were plotted to cover the most active
phase of the accident over six days from March 12 to 18.
A zoom-in view of Fig. 3 is shown in Fig. 4, by focusing on the
Yamada Station, Futaba-machi (#16; 4.1 km WNW from the Daiichi),
since it contained the most representative overall dose information
recorded during the plume passage as well as the ground shine after
land contamination in the WNW to NW direction.
Since Fig. 4 fails to show that another large release event occurred
at midnight to the early morning of March 15 in a SWS to S direction,
Fig. 5 is also selected to supplement the missing information.
In the following explanation, the terminologies (e.g., plume passage
dose, ground shine and land contamination) are illustrated in Fig. 6.
In these figures, a sharp increase in the dose rate curve represents a

plume passage dose which decays quickly after its passage. After
reaching the peak, the dose rate decreases with an asymptotic decay
curve which levels off at a larger background than before the plume
passage. This increase represents a ground shine dose due to land
contamination. This step-wise increase in the ground shine represents a
contamination of soil due to radioactive materials. The initial two large
releases from 1F1 occurred on March 12 left an order of magnitude
larger ground shine than prior to the arrival of the plume.
The first release is likely due to the overpressure leakage through
the dry well (D/W) flange of the PCV’s top head, whereas the second
release is due to the venting performed at 9:15 on March 12. The issue
of the dry well flange leakage is discussed in Section 5.5. There is also a
possibility that the first peak is due to the “internal hydrogen explosion”
at the wet well vent pipe induced a small crack since a leakage of water
has been identified in the suppression pool room as explained in Section
6.2. The effect of venting is indicated only by a plume passage dose
without an increase in ground shine. The heavily contaminated corridor
stretching to the northwest is likely the superposition of the large releases from 1F1 on March 12 (shown in Fig. 4) and from 1F2 during
March 15–16 (shown in Figs. 4 and 5). The latter induced an additional
increase in contamination by 3 orders of magnitude on top of the
ground shine left from Unit 1. These correlations are identified by
collating with the chronology of the accident as summarized in Table 6
in comparison with the environmental contamination maps.
The large release occurred on March 15 at the time when the explosive sound occurred in Unit 2’s suppression pool as shown in Fig. 5.
This graph is connected with the three large release events recorded in
Fig. 4 on March 16. The wind direction changed from SWS on March 15
to WNW on March 16. Unit 2 appears to have released effluent repeatedly for more than a day, thereby heavily contaminating the near
field of the Daiichi. The radioactive iodine was the main constituent of
the release. This is a unique feature of the releases indicating a different

4.3. Reconstruction of land contamination densities due to iodine
In general, retrospective reconstruction of 131I map is very difficult
due to its short half-life. However on June 27, 2013, JAEA (Japan
Atomic Energy Agency) gave a press release on their successful reconstruction of the 131I land deposition maps based on the spectral data
taken by the US DOE through their aerial remote monitoring operation
performed from April 2–4, 2011. (Torii, 2013).
In their analysis, Dr. T. Tori performed a reverse (unfolding) Monte
Carlo Simulation by incorporating the attenuation of the gamma ray
from the source to the detector as well as the detector characteristics

6
The term “mandala” is a religious picture often used in Buddhism to illustrate the
spiritual world as in the case of a Russian icon in Christianity. In the picture, Buddha is
usually surrounded with the images of saints.

322


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

Fig. 3. Survey map of environmental radiation monitoring data recovered by the Fukushima Prefectural Government.

It has often been assumed that the radioactive iodine was released
in the form of CsI. The author has confirmed the chemical stoichiometry
by investigating the numerous reported soil contamination samples
taken in the highly contaminated region stretching in the NW direction
from the Daiichi. However the 131I map created from the newly developed method indicates that iodine is not necessarily released as CsI.
Dr. Torii’s 131I map is more reasonable since the estimated shutdown

inventories of 131I/134Cs at the time of the accident are 2292/170 PBq,
respectively, for both Units 2 and 3. The elemental iodine is scarcely
soluble in water, although it is a strong oxidizing agent. Therefore, the

since the peak energy (365keV) was barely detected. Fig. 7 is copied
from their press release. The figure shows a comparison of the land
contamination density (Bq/m2) of 131I (left) and 134Cs (right) as of June
16, 2011. In comparison with the 134Cs map the 131I map indicates:

• Large near field deposition of I occurred in the southern direction
as far as 20 km away from the Daiichi;
• Cs induces highly contaminated corridor stretching from ap131

134

proximately 20–30 km NW from the Daiichi more heavily than 131I.

323


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

Table 6
Simplified chronology directly related to environmental releases.
Day

Unit

Time and Events

3/11

All
All
1

14:46 Earthquake. 14:48 Reactor trip.
15:37 Tsunami induced SBO.
21:51 Dose rate increased in 1F1 R/B.

3/12

1
1
1
1
1

02:30 D/W pressure reached 840 kPa (abs).
04:23 0.59 µSv/h at front gate. D/W head flange leakage started(?).
09:15 Vent valve (MO) was manually opened.
14:30 D/W confirmed depressurized.
15:29 1015 μSv/h at MP4
15:36 Hydrogen explosion in 1F1 R/B

3/13

3

3
3
3
3
3
3

08:41
08:56
09:08
09:20
14:15
14:31
21:10

3/14

4

06:14 Large explosive sound and shaking of the ceiling at the Unit 4
side of Unit 3/4 common control room.
07:20 D/W pressure stabilized at 0.5 MPa (abs)
10:30 High dose rate in 1F4 prevented worker entry to the R/B.
11:01 Hydrogen explosion in 1F3 R/B
11:15 RPV pressures 0.195/0.203 MPa (abs) at Chanel (A/B). D/W
Pressure 0.330 MPa (abs) and S/C pressure 0.390 MPa (abs). Both
RPV and PCV were judged sound
23:46 D/W pressure 750 kPa (abs)

Fig. 4. Aerial doses at Yamada, 4.1 km WNW of Fukushima Daiichi.

3
4
3
3

2
3/15

Fig. 5. Aerial doses at Matsudate, 14.2 km SWS of Fukushima Daiichi.

2
2
4
2
2
2
3

vent line configured to the rupture disk
882 μSv/h at MP4 (8:41 venting started)
RPV depressurized and water injection through DDFP.
DW is judged likely vented
905 μSv/h at MP4
Dose rate inside R/B 100 ∼ 300 mSv/h.
Confirmed pressure decrease in D/W

00:05 D/W Pressure 740 kPa (abs). Unable to vent. (D/W head
flange leakage started?)
06:14 Large explosive sound and floor quaking. D/W pressure

130 kpa(abs), S/C pressure 0 kPa. Evacuation of staff.
06:55 Damage on 5th floor of R/B confirmed
09:30 11,980 μSv/h at Front Gate
11:20 D/W pressure 730 kPa (abs)
11:25 D/W pressure 155 kPs. S/C pressure 0.
16:05 ∼ S/C venting interrupted but recovered. This activity
repeated several times through early April.

plumes traveled in accordance with the thermo-hydraulic effects (i.e.,
temperature of the initial released steam containing radioactive species
and release height as well as wind direction and velocity) of atmosphere
and not a simple dispersion mechanism applying Gauss’s atmospheric
turbulence model.
Fig. 6. Mechanism of changing dose rate curves during the plume passage at each
monitoring station.

4.4. Summary of environmental releases in correlation with chronology
The review of the environmental dose rate correlated with the
chronology of each unit clarified the following points:

radioactive release as CsI may indicate that the release is from a water
environment, such as from the suppression pool water, in which CsI has
been dissolved. The release of mostly iodine appears to indicate that it
is directly from the D/W of the PCV without going through the suppression pool water.
By correlating the 131I land deposition map (Fig. 5) with the
chronology (Table 6), the internal hydrogen explosion occurred in Unit
2 S/P at 06:14 on March 15 which induced a large release of 131I to the
SWS direction of the Fukushima Daiichi site. The potential flow of the
plume with the high concentration of 131I in a southern direction from
the Daiichi is a new finding and important to public safety. Also the

new 131I land contamination map is puzzling from the point of view of
its release mechanism. The highly contaminated corridor stretching in a
NW direction from the Daiichi is likely the result of numerous plumes
released from both Unit 1 and 2 and occurring at various times. This is
implied by the near field (< 5 km) 134Cs land contamination map as
shown in Fig. 8. In the figure, the routes of the numerous plume passages were more clearly reproduced. These footprints indicate that the

(1) A series of venting operations after scrubbing with the suppression
pool water (W/W venting) left minor soil contamination in 1F1 and
1F3.
(2) A series of “spontaneous venting” events appear to have occurred in
Unit 2 after the hydrogen explosion involving S/C. The explosion
deteriorated the dry well flange seals leading to the effluent
leakage. When the overpressure in the PCV is reduced through the
spontaneous venting, it leads to the reduced pressure boiling of the
suppression pool water releasing a large amount of dissolved
radioactive species contained in the S/P.
(3) Although the devastation of the secondary containment system (i.e.
R/B) should have been prevented, it did not result in a large release
upon the explosion in 1F3. It is likely due to the precipitation of
particles in the aerosol before the explosion. As a matter of fact the
rubble of the building was reported to be highly contaminated.

324


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

Fig. 7. Comparison of land contamination density (Bq/m2) of

131

I (Left) and

134

Cs (Right) as of June 16, 2011 including soil sampling data taken at each spot.

5. Hydrogen generation and explosions

closed configuration, this portion of the pipe is left cold, allowing
condensation of steam from the RPV during normal operation. The
condensation separated the dissolved hydrogen and oxygen gases generated due to the radiological decomposition of the reactor water. Upon
commencement of the routine ECCS test, the mixed gas exploded and
shattered the OD 165 mm pipe with an 11 mm wall thickness. Since the
isolation valves were closed automatically 30 s after the breach, the
accident was terminated without developing into a small LOCA. In spite
of the extensive follow-up experiments performed with closed pipe
containing mixed gas and steam, only 4 cases exploded out of 104 tests
performed with various noble metal concentrations at the surface. The
author suspects that radiation may become an ignition source, since the
high energy charged particles produce “spurs” which is an agglomeration of the secondary charged particles before being thermalized. Inside
the “spurs” the temperature should be extremely high at the molecular
scale.
The Hamaoka hydrogen explosion issue was left unresolved at the
time of the FDA. The regulatory body (NISA at that time) instructed the
owners of the 14 BWRs with similar ECCS designs to remove the accumulated gas and water before conducting the monthly ECCS tests. For
the removal of gas it was necessary to cool down the piping by closing

the isolation valves. This preparatory operation usually took several
days but the ECCS operation without the removal of the accumulated
hydrogen was feared to be risky. It is a strange directive since accidents
in need of the activation of ECCS cannot be scheduled. At the time of
the FDA, the circulation of reactor water was terminated, which should
have induced a more favorable situation for the accumulation of hydrogen and oxygen gas such as in the RCIC and HPCI steam turbine
piping.
Unfortunately, direct evidence of this mode of pipe rupture has yet
to be confirmed in 1F1. Nevertheless a steam jet ejection was observed
on June 4, 2011 from the pipe sleeve penetrating the floor near the wall

5.1. Early pipe break and hydrogen release (1F1)
As listed in Table 6, an increase in dose rate inside of the 1F1 R/B
prevented the entry of workers as early as 21:51 on March 11, approximately 6 h after the arrival of the tsunami. Entry to the R/B was
restricted at 23:05 in consideration of the high dose rate. This resulted
in a serious restriction for accident management purposes such as
venting by manually opening the vent valve. Nevertheless TEPCO’s staff
entered the R/B for venting, resulting in a dose > 100 mSv even when
wearing an “air set” (respirator). This phenomenon motivated some
Japanese scientists to suspect an earthquake-induced pipe break or
failure of the Isolation Condenser (I/C) leading to a very early core melt
down event. However this author believes that it is due to an “internal
hydrogen explosion” which is preceded by the earlier event at the
Hamaoka Unit 1 in Japan followed with Brunsbüttel BWR in Germany,
both of them occurred in 2001. This mode of pipe break could have
been prevented if more in-depth studies had been completed as explained below. A brief summary of this accident is appended in the
Annex C. The Hamaoka accident is important due to the fact that a
hydrogen explosion inside of the primary steam-water coolant may
induce an “internal hydrogen explosion” when dissolved hydrogen is
separated from the primary coolant. This separation mechanism should

have also existed in the W/W vent pipe during the course of the accident in which the suppression pool water which reached temperatures
as high as 160 °C.
The Hamaoka accident occurred in a part of the ECCS (RHRS) at the
top portion of the steam piping from the RPV. During operation, this
portion of the stand pipe is open to the steam of the primary circulation
water at its inlet side, however, valves are kept closed at the exit to the
RHRS Heat Exchangers which is at an ambient temperature. Due to this
325


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

130 kpa(abs), S/C pressure 0 kPa. The author believes that this event
should be due to an “internal hydrogen explosion” in the suppression
pool (S/P).
At the exact same moment, another large explosive sound and
quaking of the floor was observed in the common control room of Units
3 and 4. Since no down scaling of the S/P pressure was reported in Unit
3, the author judges that the event occurred in 1F2 but not 1F3.
5.2.1. Hydrogen generation through radiation-induced electrolysis
Since the scientific cause for a series of hydrogen explosions during
the FDA has not been established, the author investigated his basic
theory named “radiation-induced electrolysis (RIE)” by applying the
estimation of the amounts of H2 generation during the active phase of
the FDA (Saji, 2016). The author’s theory was originally developed by
including Faraday’s Law of electrolysis into the basic time-dependent
material balance equation of radiation-chemical species for his study on
accelerated corrosion phenomena (Saji, 2017). As such this theory applies to the early phase of the accident before the loss of water levels in

the reactor cores, although the following simulations were performed
from the time of the seismic reactor trip to the hydrogen explosions,
since the water levels were recovered by water injection.
Through this mechanism as much as 58,000 m3-STP of hydrogen gas
is estimated (Fig. 9) to have accumulated inside the 1F2 PCV prior to
the “internal hydrogen explosion” which occurred at 06:14 on March
15. With this large volume of hydrogen gas the explosion was a viable
possibility.

Fig. 8. Near field land contamination maps of 134Cs. (Note) The 134Cs map was shown
here instead of 137Cs since the three distinct highly contaminated streaks are more clearly
distinguished, due to the grouping of the contamination density data.

of the TIP (Travelling In-vessel Probe) room in 1F1. The contamination
level on the floor was 4.7 Sv/h, one to two orders of magnitude larger
than from nearby areas. These data were acquired during the robotic
survey inside the R/B. On November 14, 2012 the floating robotic inspection inside the S/C Room (the room housing the S/C) identified
leakages from one of 8 Sand Cushion Drain Pipes and from the upper
portion of the S/C Vent Pipe. This indicates that there would have been
a pipe break somewhere in the upper part of the RPV (TEPCO Handout,
2013a). The leakage rate was as large as tap water coming out from a
garden hose (TEPCO Handout, 2013b). All of the leaked water initiated
from the upper part of the RPV was collected in these pipes. However
potential damage of piping in the upper part of PCV has not been
performed yet, since there is not enough space to insert a robotic
camera.

Fig. 9. Accumulated H2 in Unit 2 (m3-STP).

The calculation is made to estimate the total hydrogen inventory

within the primary water pressure boundary, which includes its recirculation piping and the RPV. The top half of the RPV was assumed
initially filled with steam. The hydrogen generated in the irradiated
reactor water through the “radiation-induced electrolysis” mechanism
is released mostly in the top half of the RPV. The pressure increase in
the reactor vessel is released to the suppression pool through safety
valves (spring action) and release valves (nitrogen gas-driven).
Therefore most of the hydrogen gas should have been accumulated in
the S/C.
The damage to the 1F2’s blowout panel, which is said was induced
by the hydrogen explosion7 of 1F1 at 15:36 on March 12, facilitated the
release of hydrogen gas which likely prevented a severe explosion in the
R/B of 1F2. However, the “internal hydrogen explosion” of the Suppression Pool resulted in the leakage of a large amount of radioactive
species through this damaged blowout panel thereby severely

5.2. Hydrogen explosion in 1F2 suppression pool
The series of hydrogen explosions, devastating the reactor buildings
one after another, are the most dramatic events during the Fukushima
disaster; however the 1F2 R/B evaded this form of hydrogen explosion.
At this unit an outstandingly large amount of radioactive leakage to the
basement of the T/H has been detected from March 24–27, 2011 suggesting a leakage of the suppression pool water.
Also, as presented in Table 6, a large explosive sound and quaking of
the floor were observed in the suppression pool area of 1F2 on March
15, 06:14. The control room staff observed the D/W pressure was at

7
The exact timing and cause for activation of the blowout panel is not known. It is also
said to have been activated through the hydrogen explosion in 1F3 which occurred at
11:01 on March 14. In spite of these uncertainties, the implication of the following discussion is unaffected.

326

Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

suppression chamber the vent pipes exhaust into a toroidal vent header which extends circumferentially all the way around the inside of
the suppression chamber. Extending downward from the vent header
are ninety-six down-comer pipes which terminate about three feet
below the suppression pool minimum water level”. The down-comer
pipes opened under the pool water should have prevented the nitrogen
gas covering the suppression water surface.
In this basic configuration, the atmosphere of the “free air volume of
the Suppression Pool” was likely air sealed in from the time of the last
pool water exchange for maintenance and inspection. The periodic repainting maintenance is indispensable since the suppression chamber is
constructed of carbon steel in which a large volume of pool water is
stored. Under this situation, as much as 6900 N-m3 of oxygen could
have existed in the “free air volume” of the S/C. This amount is not far
from the chemical stoichiometry compared with 20,000 N-m3 of hydrogen injected through the SRVs into the S/C through the pressurized
“feed and bleed” operation.
In addition, since the suppression pool water was not degassed,
there would be as much as 19.4 N-m3 of dissolved oxygen when the
water is saturated with atmospheric oxygen at room temperature, although not comparable with the oxygen in the cover gas. Therefore, a
more than sufficient volume of oxygen is available for the “internal
hydrogen explosion”. This explosion would have induced a distortion of
the flange and failure of the seal due to the over pressure from the
explosion as illustrated in Fig. 11. The mixed gas of hydrogen and
oxygen should have existed equally in Units 1–3, however the hydrogen
explosion in S/C appears to have occurred only in Unit 2.
When the “internal hydrogen explosion” occurred in the “free air

volume” of the S/C in Unit 2, the resultant high overpressure induced
down pressure on the surface of the suppression pool water. This down
pressure forced the suppression pool water to flow back through the
“eight large vent pipes (81″ in diameter)” to the bottom head of the
PCV. The resultant huge high speed water jets collided at the bottom
head and should have induced violent splashing of water. The splashing
dislodged the gratings embedded in the platform installed for exchange
of control rods as illustrated in Fig. 11. This phenomenon appears to
have actually occurred at the time of FDA as observed during the recent
robotic inspection performed down to the pedestal region of the PCV
through the CRD exchange rail (TEPCO, 2017a).

contaminating the NW direction of Fukushima Daiichi. Most of the
releases were from the wet well where the decontamination factor of
the suppression pool water is as large as two orders of magnitude.
However, a series of the dry well head flange leakages occurred without
this decontamination process. Thus a severe environmental contamination occurred in the northwest direction of the Fukushima
Daiichi.
5.2.2. Failure of RCIC and “feed and bleed” operation
The decay heat removal in 1F2 was established by the RCIC which
was manually started at 15:39 of March 11, soon after the tsunami’s
arrival. It was configured by injecting water from the Condensate
Storage Tank (CST) which contained pure water at the ambient temperature. In order to save the water inventory of the CST, the water
source was switched to S/C at 2:55 on March 12 followed with seawater
injection initiated at 19:54 on March 14 in 1F2. In spite of these water
injection efforts, the decay heat was removed through bleeding of steam
vapor since no heat sink was available due to LUHS thus “feed and
bleed” cooling should have been effective. Due to the RPV pressure
increase, Safety and Release Valve (SRV) started to bleed the steam into
the S/C, resulting in a gradual decrease of the water level as shown in

Fig. 10. The safety valve function of SRV is spring action and works
without high pressure I/C nitrogen gas.
Upon failure of RCIC, the water injection operation was performed
after depressurization of the RPV. The “feed” operation was performed
with fire trucks. Therefore, “feed and bleed” cooling was the main
mechanism for decay heat removal due to the LUHS during the active
phase of the FDA.
It should also be noted that even when the water level decreased to
the bottom of the active fuel, there was a large volume of water left in
the lower plenum of the RPV. This water should have contributed to the
cooling of the core to a certain extent, however its effect in preventing
core melt is still controversial. This question may not be answered until
the actual status of the fuel and core debris are confirmed years from
now
5.2.3. Hydrogen release into suppression chamber
At the time of the RCIC failure, approximately 5.8 × 104 N-m3 of
hydrogen accumulated in the total primary water (water in the RPV and
in recirculation lines) of approximately 900 m3. From this volume,
approximately 2 × 104 N-m3 of hydrogen gas is estimated to be released into the S/P with a water volume of 3800 m3 in Mark 1 containment. Although the D/W atmosphere is nitrogen, there is no specification of the suppression pool atmosphere. This implies that the air
atmosphere could have existed in the S/P’s atmosphere (GE Technology
Systems Manual). However, the GE Report explains “during normal
operation, the drywell atmosphere and the wet well atmosphere is inert
(filled with nitrogen), and the wet well water is at the ambient temperature” (GE Report, 03/19/2011). This report describes the Mark I
containment design currently in use at the 23 U.S. reactors and its
ability to fulfill its safety function in containing fission product releases
under design basis conditions. This practice may not have been followed at the Fukushima Daiichi. As a matter of fact, there is no nitrogen
gas charge line connecting directly to the S/C, although the charge lines
are found connected to the D/W in TEPCO’s illustration (TEPCO’s Press
release, 2015) which is also posted by the Nuclear Regulatory Agency.
The 1F2 S/P atmosphere was changed to nitrogen gas two years after

the accident by way of the oxygen analysis rack line.
The S/C is designed to release the steam-water mixture to the suppression pool by postulating a DBA-LOCA. The following explanation is
given in GE’s Technology Systems Manual. “The interconnecting vent
network is provided between the drywell and suppression chamber to
channel the steam and water mixture from a LOCA, to the suppression
pool and allow non-condensing gases to be vented back to the drywell.
Eight large vent pipes (81″ in diameter) extend radially outward and
downward from the drywell into the suppression chamber. Inside the

Fig. 10. RPV water level decrease with “feed and bleed” operation.

327


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

Fig. 11. “H2 Explosion” in S/P → Flange Leakage.

trip. If the debris contained some fraction of the core materials, steam
bubbles should be visible. However the observed debris appear to be
cold in temperature.
This observation suggests that the debris should contain solidified
molten metals. They should consist of low melting temperature steel
structures such as control rods, control rod housings and stub pipes, all
of them are made of stainless steal. It is likely that the fuel rods were
standing vertical or broken into pieces and then theses were gradually
melted in-core low melting point structural metals. A part of the fuel
debris should have been piled up at the bottom of RPV since its density

is higher than the molten steel. The heated molten steal structure
eventually broke the reactor pressure boundary and drooped down into
the water, either solidified or deposited at the bottom of PCV. During
this process, the initially hot molten stainless steel droplets induced a
gigantic turbulence of the accumulated water, which dislodged the steel
grating placed on the platform of the PCV. This scenario is consistent
with the current ambient temperature (23.4∼25.6 °C) of the accumulated water at the bottom of PCV.

5.3. Core melt and the most serious hydrogen explosion in 1F3
The most severe hydrogen explosion during the FDA occurred in
1F3 at 11:01 on March 14. However this event left no footprint in any of
the radiation monitoring records as shown in Figs. 4 and 5.

5.3.1. Identification of non-heating core debris in 1F3
Starting July 19, 2017, TEPCO released a series of results of investigation of Unit 3 PCV by employing a newly developed submarine
type (ROV) robot. The most remarkable finding is that they finally
identified “core debris” drooping down like a stalactite from the bottom
head of RPV and then piled up inside of the accumulated water at the
bottom of PCV (TEPCO, 2017c). Some of their photos are copied as
shown in Fig. 12.
Since these photos were released as a TEPCO’s press handout, detailed explanations were not included. In spite of lack of details, these
pictures are very important in contemplating what occurred in 1F3
during the active phase of the FDA. In the photos it is remarkable to
observe that there are no steam bubbles, although the estimated total
decay heat is as large as 127 kW in 1F3 after 2000 days since the reactor

Fig. 12. Photo images of core debris found at the bottom of 1F3 PCV (TEPCO, 2017c).

328

Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

boundary but solidified to plug the crack.
Therefore initiation of the core melting should have been triggered
by the loss of RPV pressure occurred after March 12 at 13:05, just one
day after the reactor trip. This early loss of reactor pressure indicates an
occurrence of a “small break LOCA” by accelerating penetration of preexisting stress corrosion cracking in some in-vessel structures, such as
the CRD housings and stub tubes. Unfortunately this type of hidden
defects were just started to come to the attention of the nuclear industries such as by EPRI (EPRI, 2000). Also in Japan a guideline for inservice inspection of RPV was compiled in (JANTIVIP06 2008, only in
Japanese). It is likely such inspection has never been performed before
the FDA, since there were no TEPCO’s press release dealing with the ISI
for the in-vessel components although the industries’ concern was focused on replacement of reactor core shrouds just before the FDA. Recently the author published the basic mechanism of accelerated corrosion of in-vessel components (Saji, 2017). With this mechanism, the
reactor core behaves as a gigantic cathode whereas the out of core region becomes anodic, where SCC is accelerated due to de-passivation.
Although the hydrogen explosion in Unit 3 was triggered due to the
gigantic tsunami, it was a coincidence that the small leak from the small
break LOCA occurred simultaneously while operating RCIC. The author
sincerely request nuclear industries to perform ISI of in-vessel components of BWR with the similar design.

5.3.2. Early termination of water injection with RCIC
Despite SBO, 1F3’s battery power survived unlike in the case of both
1F1 and 1F2. By saving consumption of the battery power, RCIC and
HPCI were utilized for keeping its water level above the top of the
active fuel zone. The RCIC was manually started at 15:05 on March 11,
however it automatically tripped at 15:25 due to the high water level in
the RPV. In order to save the battery power, the operators adjusted the
flow rate settings and manually adjusted the flow rate to avoid tripping
and restarting cycles. With this manipulation, RCIC continued to inject

water until March 12 at 11:36. The RCIC became inoperable due to low
RPV steam pressure, which drives the steam turbine of RCIC. The RCIC
system operates on high-pressure steam from the reactor itself, and thus
is operable with no electric power other than battery power to operate
the control valves. TEPCO operators tried to maintain the reactor water
levels within ± 75% of top and bottom of the active fuel level.
At 12:35 of March 12, the HPCI automatic started when detecting
the low water level set point in RPV. Since the HPCI had higher capacity
than the RCIC it consumed more steam for driving the HPCI turbine.
Since the HPCI was used at pressure lower than the minimum operation
pressure (0.69 MPa), its operation was continue to be unstable. On
March 13 at 9:25 the RPV was depressurized and HPCI was switched to
DDFP to splay cool the suppression pool water. The RPV pressure decreased to such a low level that inhibited further operation of both RCIC
and HPCI at 11:36, with only 20 h of total operation time. In contrast
with the Unit 3, the Unit 2’s RCIC appears to have worked at least until
its depressurization of RPV at 18:02 on March 14, with the total operation time of 75 h.
This comparison indicates that there should have been a leak from
the “Small-Break LOCA” in the 1F3 RPV. In order to confirm this hypothesis, changes in the 1F3’s RPV pressure and water level are plotted
in Fig. 13.
Note that RPV pressure was lost on March 12 at 13:05 but temporally recovered at March 13 on 5:00. It is very likely its water level was
around the bottom of active fuel level (BAF), although the reactor water
level gauges were not available until March 13 at 4:00. (The water level
gauge failed on March12 at 20:36) Therefore, the active part of the
reactor fuel was exposed to steam while its lower portion was mostly
kept under water for days. With the fuel uncovered after 3/12 13:05, it
is likely that the low melting point core metals should have been molten
down to the bottom head of RPV and penetrated the reactor pressure

5.3.3. Environmental releases from 1F3
The vent line was configure at 8:41 on March 13 up to the rupture

disk which is installed between the final vent line and exhaust stack.
Soon after that, at 8:56 on March 13, a dose rate of 882mSv/h was
detected by a monitoring car which stayed near MP 4. This indicates
that the venting was successful although the status of the rupture disk is
still unknown. The rupture disk is designed to burst at 427 kPa (gauge),
which appears to protect the PCV whose design pressure is approximately 0.38MPa (gauge). During the venting operation there was no
record that the D/W pressure reached that level.
The containment leak rate test is performed at 0.26 MPa (gauge) in
which the allowable leak rate is 0.348%/d for 1F1. The design philosophy for installing the rupture disk on top of the final vent valve is not
appropriate. As a matter of fact TEPCO falsified this test and was examined by organizing an independent outsider investigation committee
(Investigation Committee Report, 2002). In the report, there are several
Fig. 13. 1F3 RPV pressure (MPa) and water level (m).

329


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

statements on installing “diaphragms” at the flanges among the measures to avoid leakage from valves. Although no further statement is
reported, it is very likely the “diaphragm” means that the rupture disk
was installed to comply with the regulatory leak rate in the other Fukushima Daiichi units.
Because of this particular design configuration, the venting operations appeared to have performed below the burst pressure of the
rupture disk, resulting in a release of hydrogen rich steam and radioactive effluent throughout the R/B by way of the SGTS ventilation duct
network. As a matter of fact, the decrease of the D/W pressure was
confirmed at around 9:20–9:35 on March 13. At 14:45 on the same day,
the dose rate inside of the R/B surged to 100–300 mSv/h. The workers
judged that the hydrogen explosion of the R/B was imminent and
started to evacuate the field crew but the crew resumed their activity

before 17:00. Similar tense situations continued for the crew continued
until the hydrogen explosion occurred at 11:01 on March 14.
Nevertheless, there remained no footprint from the radioactive releases
recorded by monitoring stations even at the time of the hydrogen explosion as indicated in Figs. 4 and 5. It is likely that most of the
radioactive effluent released inside the R/B deposited on the walls and
ventilation filters resulting in no serious environmental contamination.
In fact, when estimated in a similar method as done in Section 5.2
approximately 13,000 N-m3 of hydrogen gas is released into the R/B.
The temperature of the suppression pool water is assumed to have been
150 °C with a water volume of 2980 m3. The total air volume of the R/B
is assumed to have been at 180.000 m3. The estimated hydrogen concentration is 7%, which is well above the lower hydrogen explosion
limit in air. With these results it is not surprising to have experienced
the severest hydrogen explosion in 1F3 but not sufficient in hydrogen
volume enough to induce the explosions in both 1F3 and 1F4. This issue
is discussed in the next section.

still left unresolved whether the hydrogen explosion occurred at 06:14
on March 15 was in 1F2 or 1F4. However the largest release of radioactive inventory into the basement of the 1F2 T/H supports the failure
of the S/C in the 1F2 unit. In addition, the explosion left no remarkable
land contamination around 1F4. The amount of hydrogen generation
through “the radiation-induced electrolysis” has been reported in (Saji,
2016). Perhaps a part of the hydrogen is due to an unintended flow
from 1F3 R/B and a part is due to the radiation-induced electrolysis
from 1F4’s Spent Fuel Pool.
5.5. Performance of drywell flanges during the accident
Mohamed M. Talaat and his colleagues recently conducted a stateof-the-art thermo-mechanical finite element analysis (Talaat, 2014). A
previous overpressure evaluation conducted in 1990 identified leakage
at the D/W flange connection as the controlling containment failure
mode. The previous evaluation was based on hand calculations assuming an approximate structural model for analysis by formula. It did
not include the effects of temperature variation through the wall

thickness and the rotation of the D/W head flange due to the sporadic
bolt clamping force. Their conclusions include the following results. At
temperature levels of 150 °C and 205 °C, the silicone rubber O-rings are
still effective in resisting leakage, and the median overpressure capacities are 0.86 and 0.72 MPa (125 and 104 psi), respectively.
In spite of these theoretical estimations there is a little known of the
leakage test results performed at the Brunswick Mark I containment in
the 1970s. During the test workers found that the containment pressure
of 70 psi (482 kPa) (gauge) pushing upward against the inner dome of
the drywell head lifted it off the drywell flange enough to provide a
pathway for air to leak from the containment.
At Fukushima NPPs the design pressure of the PCV is 0.43 MPa
(gauge) for 1F1 and 0.38 MPa (gauge) for 1F2 ∼ 1F4. It is mandatory to
perform the leakage rate test every year. In 1F1 the regulations required
that the leak rate be 0.348%/d at 0.26 MPa (gauge). Due to the large
volume of PCV, the leak rate test with a nitrogen atmosphere at the
design pressure was judged to be impractical. However, such a leak rate
test at nearly half of the design pressure does not guarantee the safety
function of the dry well head flange during an accident. A potential
leakage of the double O-ring seal is checked by a sampling pipe in 1F1.
The head flange leakage had likely occurred in 1F1 as indicated in
the first large peak observed during the later part of the morning on
March 12 (Fig. 4). At 2:30 on March 12 D/W pressure reached 840kPa
(abs) whereas the design pressure is 530kPa (abs). TEPCO’s staff
manually opened the Large Vent Valve (motor driven) at 9:15 on March
12, however no evidence could confirm its success until installing a
portable air compressor to open the Vent Valve (AO) at 14:30, by then
they found that the D/W pressure had already reduced. This indicates
that the “spontaneous venting” should have occurred by way of the
head flange leakage.
This is also the case in 1F2 until 06:14, March 15 when a large

explosive sound and floor quakes were experienced. The D/W pressure
was 130 kpa(abs) and S/C pressure went down to 0kPa. This “internal
hydrogen explosion” should have resulted in a distortion of the D/W
flange inducing a series of puff releases which continued through March
16 as shown in Figs. 4 and 5. Although the Mark I PCVs appears to
perform well during the FDA, the hydrogen explosion inside the S/C
destroyed the containment function in 1F2.
Lastly no footprint of the flange leakage was identified in 1F3 via
the monitoring data. It is due to the fact that the PCV pressure never
reached its design pressure due to repeated venting. This venting activity repeated several times throughout early April, while the seawater
injection continued during this period

5.4. Hydrogen explosion in 1F4
The discussions in Section 5.3 on the spread of the vented hydrogen
gas into the 1F3 R/B explains a potential root cause for a portion of the
hydrogen gas also released into the 1F4 R/B. In 1F4 the entire core
loading of fuel assemblies were evacuated to the SFP, therefore no reactor configuration has existed at the time of the FDA.
TEPCO has been explaining the root cause of the hydrogen explosion in 1F4 is due to the venting of 1F3 through their common vent
stack. This route is unlikely due to the rupture disk installed in the vent
line prior to the common stack. However, the vented hydrogen gas
spread by way of the duct network and should have exhausted through
the common R/B exhaust stack, which measures 120m in height. The
stack is also shared with 1F4 for exhaust of ventilation for the R/B.
Before discharging the potentially contaminated room air, these exhausts are filtered with SGTS filter banks which are installed both in
1F3 and 1F4 ventilation exhausts.
TEPCO found in their post-accident survey that the dose rate of the
filter bank is 6.7 mSv/h at the outlet side and 0.1 mSv/h at the inlet
side, indicating that the radioactive effluent should have flowed from
the 1F3 exhaust to 1F4 R/B. Although the deposition of radioactivity on
the outlet side of 1F4 SGTS filter may be due to the hydrogen explosion

in 1F3 but may not support continued inflow of hydrogen gas from 1F3.
The hydrogen explosion in 1F3 occurred at 11:01 on March 14, whereas
the hydrogen explosion of 1F4 is said to have occurred at 06:14 on
March 15 almost one day later. The leakage of hydrogen gas from 1F3
to 1F4 should have been terminated upon the hydrogen explosion in
1F3.
The exact timing of the hydrogen explosion in 1F4 has not been well
established. The staff staying in Units 3 and 4’s common control room
heard a large explosive sound and quaking of the ceiling on the 1F4 side
at around 06:14 on March 15. Although TEPCO’s seismic wave analysis
suggests that this quaking should have occurred at 1F4, there is no
exact evidence that was due to the hydrogen explosion at this unit. It is

6. Discussions and lessons learned
Isolation of radiation from the public by providing containment
330


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

systems is the fundamental requirement of nuclear safety. This fundamental safety provision was seriously degraded during the FDA. The
containment system consists of primary and secondary containment
where the latter was devastated in all 4 Units. Unit 1’s primary containment vessel appears to be damaged due to the overpressure leakage
of the D/W flange in spite of death-defying manual venting operation in
the high radiation environment as early as 6 h after the arrival of the
tsunami. TEPCO’s staff were exposed to over 100mSv irradiation in very
short work time inside the R/B. In Unit 2, the hydrogen explosion in the
suppression chamber appears to have induced the series of D/W head

flange leakage events.
The most serious damage experienced to safety provisions, which
are provided to impede the accident progression, can be summarized as
below:

valves should have been arranged in parallel in consideration of the
single failure criteria. Such a parallel isolation valve configuration is
actually used in the BWR/2 Core Spray System, however it was changed
to a serial arrangement with one valve left closed in each train of valves
in the BWR/2 ECCS design (GE Technology Systems Manual – NRC).
This is also logical but having a configuration where two isolation
valves are normally open is not comprehensible. Unfortunately the
detailed drawings are not disclosed, the author is unable to clarify
where this issue is rooted. The author was told during a symposium that
TEPCO‘s position is that these two accidents are not correlated.
Nevertheless, the risk of an “internal hydrogen explosion” should have
existed in Unit 1.

(1) “Internal hydrogen explosion” likely occurred in 1F2’s Suppression
Pool which resulted in the failure of PCV. Together with the loss of
the Secondary Containment function, the radioactive effluent was
directly released into the environment and induced the most serious
environmental contamination.
(2) Devastation of the reactor buildings, which comprise a part of the
secondary containment boundary to halt the environmental releases of the radioactive effluent. The loss of the secondary containment system was experienced in 1F1, 1F3 and 1F4. This mode
of failure was evaded in 1F2 through an inadvertent actuation of its
blowout panel but resulted in the loss of its containment function.

Fukushima Daiichi NPPs were constructed by applying the concept
of Mark I Containment which utilizes a toroidal shaped suppression

pool. These primary containments are enclosed by a secondary containment consisting of a reactor building and refueling bay (MK I and
MK II designs), a shield and auxiliary and enclosure buildings. Most of
the auxiliary systems which deal with the primary water (e.g., ECCS)
are located inside of the secondary containment boundary.
Although the safety role of the secondary containment was of concern as early as 1988 by S.R. Green (Greene, 1988), their purpose was
assumed to minimize the ground level release of radioactive material
for a spectrum of traditional design basis accidents. Green was already
concerned that the deflagration of this hydrogen within the secondary
containment would result in pressure loadings which might threaten
the structural integrity of the secondary containment. The FDA was not
contained within the DBA-LOCA. It was an un-experienced beyond the
design basis accident due to LUHS combined with long-term SBO. The
mitigation strategies contemplated at that time depended heavily on
the SGTS, which was not available due to long-term SBO.
However, the FDA revealed that the reactor building is a kind of prefabricated structure without safety enhancements. The design basis leak
rate is 50–100% per day at 30 mm Aq, which is a de facto industry
standard for the heating and ventilation system design for ordinary
concrete buildings without windows such as in the case of warehouses.
Therefore how hydrogen was generated and leaked into the secondary containment is one of the key points of the FDA. As to the hydrogen generation during the severe accident, the author has theoretically demonstrated that a large volume of hydrogen can be generated
through “radiation-induced electrolysis” even without a core melt. The
theory integrates Faraday’s law of electrolysis into the radiation chemical material balance equation (Saji, 2016).
As to the second point, how hydrogen leaked into the reactor
building, it is partly through the leakage of the vented hydrogen into
the general ventilation duct network of the reactor building rooms.
According to reference (TEPCO's Investigation Report, 2012), there are
two vent lines, one for D/W venting and the other for W/W venting.
These two vent lines joins together into a single line towards their
common vent stacks (shared between Units 1 and 2. Units 3 and 4 share
another common vent stack). However, before leaching to the stack, it
is also connected to the general ventilation duct network of the reactor

building in which SGTS is installed. It is likely that the designer expected mitigation of effluent through the decontamination factor of the
SGTS which has both a re-combiner and aerosol filters. However, due to
SBO, SGTS was not working and moreover, the isolation valves (dampers) failed to close due to SBO. Rather the duct work helped in
spreading hydrogen gas throughout the entire reactor building. Separation and independence of the “hardened vent line” (USNRC, 20122015) was not integrated in the Fukushima Daiichi.
In addition to this mechanism, there is a high probability of leakage
from the flanges of the PCV due to over pressure while waiting for
permission to vent at Fukushima. As a matter of fact as early as 23:50
on March 11, TEPCO’s 1F1 staff confirmed a pressure of 600 kPa(abs)

6.2. Devastation of the reactor buildings

The apparent cause of this phenomenon is undoubtedly the hydrogen explosion. As reported in reference (Saji, 2016, 2017), generation of hydrogen during severe accidents may not be limited to a high
temperature zirconium-steam reaction. Rather the “radiation-induced
electrolysis” mechanism should have induced a large amount of hydrogen generation in the BWR water chemistry. Unlike in the case of
PWRs, hydrogen is not dosed to suppress the radiological hydrogen
generation above the “critical hydrogen concentration” (Elliot and
Bartels, 2009).
6.1. Pipe break through “internal hydrogen explosion”
As summarized in the Annex C, the hydrogen explosion occurred
during a routine ECCS test at Hamaoka Unit 1 in 2001. The pipe ruptured as a part of ECCS (at the top portion of the steam piping from the
RPV) in which the steam was condensed separating the hydrogen and
oxygen. A similar pipe rupture incident occurred one month later at the
Brunsbüttel 1 NPP in Germany. The cause of the ignition was not clearly
identified but it is suspected to have been induced by minuscule particles of noble metals (in the case when using GE’s “noble metal water
chemistry”) deposited on the surface of the piping. However, only 4 out
of 104 tests ignited in a mixture of hydrogen and oxygen contained in a
closed cylinder into which steam was injected from the side.
Due to uncertainties, the regulatory body (NISA) instructed the
power companies to remove the accumulated gas and water before
conducting the monthly ECCS test with a similar design in the 14 units.

Power companies followed this instruction by isolating the ECCS several days before conducting such a test to cool the system since the
steam temperature and pressure is as high as 270 °C, 7 MPa.
Unfortunately no fundamental solution was developed before the FDA.
The author however recently revisited this unsolved issue and found
that the configuration of isolation valves for the RHRS system in
Hamaoka (BWR/4, Mark-1) as shown in Figure A2, as well as
Fukushima Unit 1 (BWR/3, Mark-1), is strange. The two isolation valves
at the PCV boundary are arranged in a series, indicating that these
valves should normally be in the open configuration when considering
the single failure criteria. With the normally open valve, the steam
should condense in the RHRS piping which is cold during normal operation, separating the dissolved hydrogen and oxygen. These isolation
331


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

and have filters for Dry Well venting. There should be no rupture
disk in the vent line.

whereas the D/W design pressure is 430kPa(gauge). They could
manually start venting at 9:15, only on March 12, since they had to wait
for the nearby population to evacuate before being able to begin. A
series of “spontaneous venting” occurred both in 1F1 and 1F2. This
behavior indicates that the flange joint was behaving like a safety valve
for a pressure vessel. When the effluent leaks from the flanges of the
PCV, it means there is a functional failure. Recall that an intrinsic safety
factor of 3 is incorporated into Section 3 for “design by analysis” of the
ASME Boiler and Pressure Vessel Code for structural integrity. The PCV

should likewise be designed with an intrinsic safety factor of 3 for
functional integrity against leakage from the flange joint.

Without these vulnerabilities, the environmental release could have
been at least three orders of magnitude less, combined with TEPCO’s
excellent accident management mitigation taken during the active
phase of the FDA. Such a large reduction in ground shine should have
been feasible if the internal hydrogen explosion was prevented which
likely occurred in the air volume of suppression pool of 1F2.
However, the author has some reservation as to the core melt scenario introduced in Subsection 5.3.1, since the recent interim report of
the muonic imaging report states the following conclusion: “The evaluation at present shows the possibility that some fuel debris remain in
the core and at the lower area of RPV, but massive and high density
material has not been found. “

7. Conclusions and lessons learned
Almost all of the investigation reports published to date considered
that the gigantic earthquake and tsunami induced an unprecedented
beyond design basis accident and therefore the resultant consequences
were unavoidable. The author does not share this view, since design
fortification in consideration of the prevention and protection against
design basis events alone is insufficient. He believes that the environmental contamination and public exposure could have been substantially mitigated had the vulnerabilities as identified in this forensic
engineering study been removed.

Acknowledgement
Much of the contents of this report are extracted from the author’s
periodical e-mail updates on the FDA, so far distributed with the following records:
• Daily updates, starting March 12 with Earthquake (1) ∼ July 25
with Earthquake (135).
• Twice a week updates, starting July 26 with Earthquake (136) ∼
October 14 with Earthquake (160, Oct 11–14).

• Once a week updates, starting October 22 with Earthquake (161,
Oct 14-21) ∼ July 27, 2012 with Earthquake (203, July 20–27, 2012).
• Biweekly, starting August 10 with Earthquake (204, July 27 - Aug
10, 2012) to March 15 with Earthquake (248, Mar 21 - Apr 4, 2014).
• Once a month, starting April 1, 2014 with Earthquake (249, Apr 4 30, 2014) ∼ Earthquake (288, July 1–31, 2017).
The Earthquake series is intended to provide scientific minutes of
the Fukushima Daiichi nuclear disaster. The minutes have been distributed widely in English to the author’s international colleagues.
Some of them kindly provided comments during the reviewing process
of this paper.
Appreciation is directed to Ms. Dana Pandolfi for her effort of
technical editing, although the entire context is the author’s responsibility.

(1) The threat of hydrogen generation through “radiation-induced
electrolysis” especially in BWRs.
(2) Potential threat of an “internal hydrogen explosion” in the suppression pools. The cover gas of the suppression pool water should
have been nitrogen.
(3) The potential threat of an “internal hydrogen explosion” in pipes
where the steam condensation and accumulation of hydrogen and
oxygen gases might occur as in the case of the Hamaoka Unit 1
accident. This might be limited to those plants with similar PCV
design.
(4) Leak rate of the PCV should have been testable at its design basis
pressure. The intrinsic safety factor of the containment flanges
against overpressure effluent leakage should have been 3 for the
functional integrity of the PCV.
(5) Spread of the hydrogen gas from the vent lines through duct works
connected to SGTS. The hardened vent line should be independent
Annex A: Current decay heat and “melt-down” scenario

This Annex is developed to clarify whether the corium (the lava-like mixture of fissile material created during the nuclear accident) has relocated

down to the bottom of the Primary Containment Vessel (PCV) by melting through the bottom head of the reactor pressure vessel (RPV). Recent
representative temperature data inside both the RPV and PCV are available from TEPCO’s Web Site (TEPCO, 2011b). By referring to the temperatures
measured on 2017/06/26, Table A-1 indicates that both the RPV and PCV are stabilized thanks to adequate water flow rates of water injection into
the RPVs.
If the “melt-through” of the RPV bottom heads had occurred, most of the injected water should flow down to the PCV, which would have
eventually resulted in overflowing; however such events have not been reported by TEPCO. In addition, the relocation of the corium should indicate
that the large decay heat of the corium should exist at the bottom of PCVs. In the above mentioned handout, the following decay heat estimated by
TEPCO were included. The author verified TEPCO’s results by referring to previous fuel management records to incorporate the effects of long-lived
nuclear species and operation cycle durations accumulated in aged fuel. The results indicate that TEPCO substantially under estimated the results as
summarized in Table A-2. This comparison indicates that TEPCO’s decay heat was likely estimated by using some reference decay curve heat (e.g.
ANSI/ANS 5.1) multiplied by the nominal thermal power of each unit.
The estimated decay power is not consistent with the “melt down” scenario. Instead it more likely indicates that “in-vessel retention of core
debris” was likely achieved, thanks to the commendable severe accident management efforts during the active phase of the Fukushima Daiichi
accident.
Table A-1
Current representative water temperature of the PCV and RPV.
Units

1F1

1F2

1F3

RPV (°C)
PCV (°C)

22.1 ∼ 22.3
22.1 ∼ 29.5

28.1 ∼ 28.5
28.8 ∼ 28.7

23.7 ∼ 25.7
23.4 ∼ 25.6

332


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

Table A-2
Verification of TEPCO’s decay heat.
Units

1F1

1F2

1F3

Decay Power (TEPCO-2016)
Saji (2000 days)

70 kW
153 kW

90 kW

229 kW

90 kW
127 kW

Annex B: Overview of hydrogen generation mechanisms
B.1. Brief overview of high temperature zirconium-steam reaction
The root-cause for a series of hydrogen explosions is one of the least known although almost all of the investigation reports explain the accident
scenario as a tsunami-induced SBO (station blackout, loss of all AC power) which resulted in the core “melt-down” accidents. Through the core melt a
high temperature Zr-steam reaction generated a large amount of hydrogen gas, which leaked into the Reactor Building and then exploded. However,
TEPCO has not been able to identify any evidence demonstrating the existence of large heat generating corium (molten fuel debris) despite their
robotic investigation of the reactor pressure vessel inside the containment primary vessel even six years after the accident.
It has been widely explained that the hydrogen gas is initially generated by the rapid oxidation of the zirconium alloy tubes (“fuel cladding”) that
surround the sintered uranium-dioxide fuel pellets in commercial reactors. Some scientists insist that the zirconium cladding in a water or steam
environment would undergo a rapid and self-sustaining heat generating (exothermic) oxidation reaction. Although it is an exothermic reaction, such
a self-sustaining reaction can be excluded by referring to the experiment performed as early as in 1954 by W. A. Bostrom (Bostrom, 1954). During the
course of his investigation several samples were heated above the melting point. The sample consisted of a flat disk 3/4″ in diameter and 1/2″ thick.
A carbon sample holder was used and the entire assembly was placed about 6″ under water. This sample was held above the melting point for about
10 s. The reaction proceeded quite rapidly but not with great violence, indicating that the oxidation of Zircaloy-2 with water is not self-sustaining for
the specimen dimensions employed even at temperatures somewhat above the melting point. Several other samples were melted under water
inadvertently, and it was observed that although the reaction proceeded more rapidly, it did not become violent or autocatalytic in nature. It has also
been observed during other experiments and also during arc-melting that molten zirconium dropping into water does not react violently.
Since the time of these early studies, a number of investigations have been performed as reviewed by L. Baker, Jr. (Baker, Conf-830816). The
most significant point of the succeeding studies is that the kinetics of the high temperature steam reaction of Zircaloy alloys, uranium and austenitic
stainless steels have been found to be consistent with parabolic law behavior often observed in such experiments. This phenomenon is induced by
quickly forming the protective oxide films on the surface of metals, retarding the corrosion phenomena. The rate of hydrogen generation from
zirconium-steam reaction is not so violent as widely contemplated for the Fukushima accident. The conceptual chemical reaction channel assumed in
these hypotheses is through Zr + 2 H2O → ZrO2 + 2H2. However, it is questionable to anticipate the water or dissolved hydrogen molecules which
diffuse through the protective oxide film covering the zirconium surface in case of the Fukushima accident since the fuel rods were almost always
covered with the primary water, except for a few hours when the water injection was interrupted.

B.2. “Radiation-induced electrolysis” (RIE)
Since the scientific cause for a series of hydrogen explosions has not been established, the author investigated his basic theory named “radiation-

Fig. B-1. Configuration of Fukushima Daiichi Unit 1–3.

333


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

induced electrolysis (RIE)” by applying the estimation of the amounts of H2 generation during the active phase of the Fukushima accident. The
author’s theory was originally developed by including Faraday’s Law of electrolysis into the basic time-dependent material balance equation of
radiation-chemical species for his study on accelerated corrosion phenomena. As such this theory applies to the early phase of the accident before the
loss of water levels in the reactor cores, although the simulations were performed from the time of the seismic reactor trip to the hydrogen
explosions.
The RIE configuration assumed in estimating the hydrogen generation in the primary water is illustrated in Fig. B-1. In such an open configuration of RPV, the generation H2 is not suppressed as pointed out in the Spinks and Wood’s textbook (Spinks and Woods, 1990). Thus accumulated
hydrogen and air mixture in the free air volume of the suppression pool resulted in the explosion, since there were no direct charge line for nitrogen
gas from its nitrogen gas generator to the suppression pool.
Through this mechanism as much as 29,400 m3-STP of hydrogen gas is estimated to be accumulated inside the PCV just prior to the hydrogen
explosion which occurred one day after the reactor trip in 1F1. With this large volume of hydrogen gas and air the explosion was a viable possibility
upon the “venting” operation.
In view of this observation, hydrogen generation from the spent fuel pools was also estimated by applying the RIE mechanism. With a mix of
different levels of radioactivity of spent fuel, a variance in the absorbed dose rate of water through γ-decay heat should have existed. This configuration induced an electrochemical potential difference between the highly radioactive region where there was spent fuel stored by evacuating the
core and less radioactive fuels stored for several years. The spent fuel was stored in racks placed at the bottom of the pool where the wall was covered
with a stainless steel lining. The metallic contacts enabled electric conduction between the highly radioactive fuel assemblies and the cooled spent
fuel. The contemplated RIE mechanism is illustrated in Fig. B-2.
The author searched for a potential radiation chemical mechanism for the hydrogen explosion in Unit 4 of the Fukushima Daiichi during the
accident by changing the pool water temperature and flow velocity in the spent fuel. During the trial calculations SBO was found to have induced a

rapid initiation of electrolysis when the pool water temperature reached approximately 40 °C.(Figs. C-1 and C-2)
The present estimation of the hydrogen generation rate is still large enough to have induced the explosion in 1F4 SFP since as large as 1000 Nm3/
d is estimated when the pool water temperature exceeds approximately 40 °C. It also revealed that the behavior of the radiation chemical process is
much more complicated than simply the dependence on temperature. It depends on the management of spent fuels in the SFP, absorbed dose rate
and volume of irradiated- and mixing volume of water as well as its flow velocity (i.e. residence time of the water staying in the highly active region
of spent fuel).
In order to abide to these complexities the author proposes the simple solution of inserting a ceramic insulator to prevent direct metallic contact
of the spent fuel racks to the SFP liner thereby disconnecting the flow of electrons from the anodic cooled fuel assemblies. However the author has
reservations regarding the current results as our knowledge is extremely limited as to the chemical characteristics of the cooling water especially in
the core region. In particular the application of the author’s basic approach for his RIE corrosion study to the severe accident situation has been
established essentially without experimental data to verify. However the observed phenomenology of a series of hydrogen explosions during the
Fukushima accident is not contradictory to the author’s prediction. Obviously understanding the phenomena occurring through radiation-induced
electrolysis in the Fukushima accident is not complete, and the theoretical framework of radiation chemistry applicable to severe accidents in LWRs
has to be more firmly established.

Fig. B-2. Radiation-induced electrolysis in 1F4 SFP.

334


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

Appendix C. Annex C: A fact sheet of Hamaoka Unit 1 accident
This appendix summarizes the hydrogen-explosion/pipe-rupture accident that occurred at the Hamaoka Unit 1 as shown in Fig. C-1.
The pipe rupture occurred at a section of the ECCS (at the top portion of the steam piping from the RPV) in which the steam was condensed
separating the hydrogen and oxygen as illustrated in Fig. C-2.
Outline of the Hamaoka Unit 1 accident

• Rated power: 540 MWe, BWR
• Occurred: 17:02 of November 7, 2001
• Plant State: Steady state operation, during a scheduled ECCS test for manual startup of HPCI pump.
• Water Chemistry: Hydrogen water chemistry, supplemented with the noble metal™ water chemistry.
Detection of the accident

• Operators’ observation: Explosive noise observed by the operators of the central control room as well as at local panels.
• Trip sequence: HPCI trip followed with a containment isolation resulting in the automatic closure of the isolation valves in approximately 30 s.
• Environmental monitoring: 30–40 nGy/h (no change)
• Confirmation: A maintenance crew accidentally went inside of the reactor building, at 17:20 and identified that the 1F and 2F floors were wet
with water. It was concluded that the fire alarms were triggered due to steam leakage.

Fig. C-1. Pipe Rupture at the Steam Condensate System of RHRS (Pipe size: OD = 165 m/m, WT = 11 m/m).

Fig. C-2. Schematic diagram of RHRS Line B. The pipe
rupture occurred at a section of the ECCS (at the top portion
of the steam piping from the RPV) in which the steam was
condensed separating the hydrogen and oxygen.

335


Nuclear Engineering and Design 324 (2017) 315–336

G. Saji

Post-accident inspection

• Estimated leakage: approximately 2 tons of steam. The estimated radioactive leakage was 8ì10 Bq.
ã Causes of the explosion: Hydrogen burn induced from hydrogen gas accumulated at the top portion of the Steam Condensate piping system. The

hydrogen was transported with the steam which condensed in the cold pipe at the top thereby separating the hydrogen gas.
• Hydrogen concentration: 0.6% volume in the similar riser pipe location of RHRS A and 19% of O gas. Similarly in Unit 2 RHRS A, H = 46%,
O = 23%: RHRS B, H = 27%, O = 23%.
• Estimated hydrogen accumulation after 8 months of operation: 6–8 meter from the condensed water surface, with H = 66% and O = 33%,
8

2

2

2

2

2

2

2

N2 = 1%.

Follow-up test results

• Cause of ignition: It is suspected to have been induced by minuscule particles of noble metals deposited on the surface of the piping. However,
only 4 out of 104 tests ignited in a mixture of hydrogen and oxygen contained in a cylinder into which steam was injected from the side.
• Ignition tests: Self-ignition at 340–370 °C, 5–8 MP with dry mixture of hydrogen and oxygen. No self-ignition with steam. Some cases of selfignition with a noble-metal catalysis.
• Combustion to Detonation transition: At 1–2 m from the ignition point.
• Maximum plastic deformation of pipes: greater than 23%
• Brunsbüttel Accident: A similar pipe rupture incident that occurred on December 14, 2001 at the Brunsbüttel l NPP in Germany. In this plant, the

noble metal chemistry is not used.

Instruction by the regulator (NISA)

• Remove accumulated gas and water before conducting the monthly ECCS test with a similar design in the 14 units.
• Power companies followed this instruction by isolating the ECCS several days before conducting such a test to cool the system since the steam
temperature is as high as 270 °C, 7 MPa. No drastic solution developed before the Fukushima disaster.

Saji, G., 2016. Root cause study on hydrogen generation and explosion through radiationinduced electrolysis in the Fukushima Daiichi accident. Nucl. Eng. Des. doi http://dx.
doi.org/10.1016/j.nucengdes.2016.01.039.
Saji, G., 2017. Radiation-induced electrolytic phenomena with differential radiation cell
in water-cooled nuclear reactors. Nucl. Eng. Des. Doi information: />10.1016/j.nucengdes.2017.03.022.
Spinks, J.W.T., Woods, R.J., 1990. An Introduction to Radiation Chemistry-third ed. John
Wiley & Sons.
Talaat, M.M., 2014. Overpressure fragility evaluation of a Mark I drywell using thermalmechanical finite element analysis. In: Proceedings of the ASME 2014 Pressure
Vessels & Piping Conference. PVP2014 July 20–24, 2014, Anaheim, California, USA.
TEPCO, 2011. Results of Analysis of Accumulated Water in the Turbine Building. http://
www.tepco.co.jp/nu/fukushima-np/images/handouts_110522_04-j.pdf (in Japanese).
TEPCO, 2011. Archive (monitoring data update for 2011/2/11-2011/3/21. in Japanese)
/>TEPCO, 2012. Investigation Report of the Fukushima Nuclear Accidents, 2012. http://
www.tepco.co.jp/en/press/corp-com/release/2012/1205638_1870.html.
TEPCO Handout, 2013. Results of Investigation around Lower Parts of Unit 1 Vent Pipes
at Fukushima Daiichi NPS (Second Day). />TEPCO Handout, 2013 (in Japanese). />handouts/2013/images/handouts_131113_15-j.pdf. />docs/ML0230/ML023020246.pdf.
TEPCO, 2015. />TEPCO, 2016. Locating fuel debris inside the unit 2 reactor using a muon measurement
technology at Fukushima Daiichi Nuclear Power Station. />en/nu/fukushima-np/handout/2016/images/handouts_160728_01-e.pdf.
TEPCO, 2017. Unit 2 Primary Containment Vessel Investigation at Fukushima Daiichi
Nuclear Power Station (Investigation results by the self-propelled investigation device). />handouts_170216_01-e.pdf.
TEPCO, 2017. Reduction of the flow rate of the water injection in Unit 1∼3. http://www.
tepco.co.jp/nu/fukushima-np/handouts/2017/images1/handouts_170522_05-j.pdf.
TEPCO, 2017. Progress of Unit 3 PCV internal investigation (Preliminary report of July 21

investigation) (PDF 351KB) />handouts/2017/images/handouts_170722_01-e.pdf.
Terwel, K., et al., 2012. An initial survey of forensic engineering practices in some
European countries and the USA. In: ASCE 6th Congress of Forensic Engineering,
November 1-3, 2012, San Francisco, California, USA.
Toshiba, 2012. Issues in I & C system in Unit 2 and 3 during the Fukushima Daiichi
Nuclear Power Station. />Torii, T., 2013. Enhanced analysis methods to derive the spatial distribution of 131I deposition on the ground by airborne surveys at an early stage after the Fukushima
Daiichi Nuclear Power Plant accident. Health Physics: August 2013 – Volume 105 Issue 2 - p 192–200. doi: />US DOE/NNEA, 2011: />USNRC, 2012-2015. Hardened Vents and Filtration (for Boiling Water Reactors with Mark
I and Mark II containment designs).

References
Atomic Energy Society of Japan, 2015. The Fukushima Nuclear Accident – Final Report of
the AESJ Investigation Committee. ISBN 978-4-431-55159-1. Springer.
Baker, L., Conf-830816. Hydrogen-generating reactions in LWR severe accidents. www.
iaea.org/inis/collection/NCLCollectionStore/_Public/15/003/15003080.pdf.
Bostrom, W.A., 1954. The high temperature oxidation of zircaloy in water. WAPD-104/.
Chino, M., et al., 2011. Preliminary estimation of release amounts of 131I and 137Cs
accidentally discharged from the Fukushima Daiichi nuclear power plant into the
atmosphere. J. Nucl. Sci. Technol. 48 (7), 1129–1134.
Elliot, A.J., Bartels, D.M., 2009. The Reaction Set, Rate Constants and g-Values for the
Simulation of the Radiolysis of Light Water Reactors, 153-127160-450-0001.
EPRI, 2000. BWR Reactor Pressure Vessel Inspection and Flaw Evaluation Guidelines
(BWRVIP-74NP). />Fukushima Prefectural Government, 2012. />monitoring/etc/post4-20120921.pdf.
GE Technology Systems Manual – NRC., General Electric Advanced Technology Manual
Chapter 6.5 Primary Containments.
GE Report, 03/19/2011. Mark I Containment Report. http://files.gereports.com/wpcontent/uploads/2011/10/NEI-Mark-1-White-Paper.pdf.
Greene, S.R., 1988. Role of BWR Secondary Containments in severe accident mitigation;
issues and insights from recent analyses“ CONF-8806153-1.
Henrie, J.O., 1989. Timing of The Three Mile Island Unit 2 Core Degradation Nuclear
Safety as Determined by Forensic Engineering. EGG-M-88439.
IAEA, 1996, INSAG-10, Defense in depth in nuclear safety.

UNSCEAR 2000, Annex J (Sources and Effects of Ionizing Radiation, Volume II: Effects,
Annex J, Exposure and effects of the Chernobyl accident).
Investigation Committee Report (in Japanese), 2002. www.tepco.co.jp/cc/press/betu02_
j/images/1211c.pdf.
National Diet of Japan Fukushima Nuclear Accident Independent Investigation Commission,
2012. />(National Government's) Investigation Committee on the Accident at the Fukushima
Nuclear Power Stations, 2012. />IAEA, 2004. Design of reactor containment systems for nuclear power plants. Safety
standards series No. NS-G-1.
Nuclear Emergency Response Headquarters, June 2011. Report of Japanese Government
to the IAEA Ministerial Conference on Nuclear Safety – The accident at TEPCO’s
Fukushima Nuclear Power Stations.
Nuclear Emergency Response Headquarters, September 2011. Additional Report of the
Japanese Government to the IAEA - The accident at TEPCO's Fukushima Nuclear
Power Stations.
NUREG/CR-5850, 1994. Analysis of Long-Term Station Blackout Without Automatic
Depressurization at Peach Bottom Using MELCOR (Version 1.8).
RJIF, 2014. The Independent Investigation Commission on the Fukushima Nuclear
Accident. ISBN-13: 978-0415713962. Routledge.
Saji, G., 2013. A Post Accident Safety Analysis Report of the Fukushima Accident – Future
Direction of Evacuation: Lessons Learned, ICONE21-16526.
Saji, G., 2014. Safety goals for seismic and tsunami risks: lessons learned from the
Fukushima Daiichi disaster. Nucl. Eng. Des. 280 (2014), 449–463.

336