A PRA Practioner looks at the Great East Japan Earthquake and Tsunami

Transcription

1 A PRA Practioner looks at the Great East Japan Earthquake and Tsunami From a PRA practioner s point of view: What went wrong? How likely was it? Woody Epstein Manager Risk Consulting, Japan Scandpower

2 April 29 th, 2011 The information in this presentation is from the White Paper written for the Tokyo Institute of Technology, A PRA Practioner looks at the Great East Japan Earthquake and Tsunami. For a copy of the full paper, please send a request to Woody Epstein sep@scandpower.com

3 Alea iacta est. At 14:46 the the Great East Japan Earthquake caused loss of offsite power to all units of Fukushima Daiichi, Fukushima Daini, Onagawa, and Tokai Daini. The diesel generators started as planned. Fukushima Daiichi was struck by three tsunami waves. The first wave arrived at the plant at 15:27, 41 minutes after the earthquake. The first tsunami wave passed over a breakwater that was 5.7 meter above the sea surface, disabling the seawater pumps, one of which was carried out to sea along with a diesel fuel tank. At 15:45, the third tsunami wave, between 10 and 15 meters above the sea surface struck the turbine buildings with an impact estimated at 100T/sqm, completely submerging the facilities' doors. The velocity of the waves sent seawater around the turbine buildings to the far side of the reactor buildings, inundating the containment.

4

5 Immediately Unit 1 went into station blackout. Unbeknownst to the operators, the force of the earthquake, the force of the tsunami, or both had damaged the reactor enough that an unrecoverable severe accident sequence had begun. The claim by TEPCO that only the tsunami caused damaged to Fukushima Daiichi Unit #1 has been challenged:

6 One worker, a maintenance engineer who was at the Fukushima Immediately complex Unit 1 on went March into station 11, said: blackout. I personally saw Unbeknownst pipes that came to the apart operators, and I assume the force that of the there were earthquake, many more the that force had of the been tsunami, broken or throughout both had the plant. damaged There s the no reactor doubt enough that the that earthquake an unrecoverable did a lot severe of damage accident inside sequence the plant. had begun. There were definitely leaking pipes, but we don t know which pipes The claim that by has TEPCO to be that investigated. only the tsunami I also saw caused that part damaged of the wall to Fukushima of the turbine Daiichi building Unit #1 for has Unit been 1 had challenged: come away. That crack might have affected the reactor.

7

8 Immediately Unit 1 went into station blackout. Unbeknownst to the operators, the force of the earthquake, the force of the tsunami, or both had damaged the reactor enough that an unrecoverable severe accident sequence had begun. The claim by TEPCO that only the tsunami caused damaged to Fukushima Daiichi Unit #1 has been challenged:

9 One worker, a maintenance engineer who was at the Fukushima Immediately complex Unit 1 on went March into station 11, said: blackout. I personally saw Unbeknownst pipes that came to the apart operators, and I assume the force that of the there were earthquake, many more the that force had of the been tsunami, broken or throughout both had the plant. damaged There s the no reactor doubt enough that the that earthquake an unrecoverable did a lot severe of damage accident inside sequence the plant. had begun. There were definitely leaking pipes, but we don t know which pipes The claim that by has TEPCO to be that investigated. only the tsunami I also saw caused that part damaged of the wall to Fukushima of the turbine Daiichi building Unit #1 for has Unit been 1 had challenged: come away. That crack might have affected the reactor.

10 A second worker who was also on site at the time of the One earthquake, worker, said. a maintenance It felt like engineer the earthquake who was hit in at two the Fukushima waves, Immediately the first complex Unit impact 1 on went was March into so station intense 11, said: blackout. you I personally could see saw the Unbeknownst building pipes that shaking, came to the apart operators, the and pipes I assume the buckling, force that of and the there within were minutes, earthquake, many I saw more the pipes that force had bursting. of the been tsunami, broken Some or throughout fell both off had the the plant. wall. damaged Others There s the snapped. no reactor doubt enough I that was the pretty that earthquake an sure unrecoverable that did some a lot of severe the of damage oxygen accident inside tanks sequence stored the plant. had on site begun. There had were exploded definitely but I didn t leaking see for pipes, myself. but Someone we don t yelled know that which we all pipes needed The claim to that evacuate by has TEPCO to be and that investigated. I was only good the tsunami with I also that. saw caused But that I was part severely damaged of the alarmed wall to Fukushima of because the turbine Daiichi as I was building Unit leaving #1 for has I Unit was been 1 told, had and challenged: I come could away. see, that That several crack might pipes have had cracked affected the open, reactor. including what I believe were cold water supply pipes. That would mean that coolant couldn t get to the reactor core.

11 Alea iacta est. The core melt of Unit 1 bagan at approximately 19:30 on March 11 th. The shock wave from tsunami impact or the earthquake s PGA damaged pipes causing the radioactive steam reported at 21:00. It is estimated that core melting was completed by 07:00 on March 12 th. Professor Ninokata on March 14 th : The accident was written irreversibly by 20:00 on March 11th. Mark Reinhart, on March 18 th : based on reported radiation readings outside the plant at the reactor building and at the site boundary as well as radioactive isotopes being released; e.g., cesium and iodine, a reasonable proposed conclusion is as follows: One or more reactor cores are melted. One or more reactor pressure vessels are compromised. One or more containments are compromised. In addition, one or more Spent Fuel Pools are dry or have very low water level; the fuel in the pool has or has had zirconium fires. There is just no other way that there could be such high radiation levels outside the NPP.

12 Delay and aversion.. Venting and cooling the reactors may have prevented the hydrogen explosion. Why did the government and TEPCO delay in beginning these operations? Risk aversion and unwillingness to imagine disastrous scenarios are endemic in Japan. The president of TEPCO and the CEO were both unavailable at headquarters. There is a cultural reticence of Japanese people to take action unless they are 100% sure of its correctness. For a person to be wrong, to make a mistake, is a grave social and professional error in Japan. Because seawater may damage the reactors and venting entails release of radioactive materials into the air, no one would take these step without a direct order from the top.

13 Sōtegai: Unforeseen events In Japan, nuclear professionals at utilities have always said that a situation where nuclear reactors lost all their power sources for more than 12 hours would be "unthinkable, or sōtegai no matter how bad the disaster. During the first month of the accident, they continued to believe this. TEPCO in an early press release: "The accident was caused by the violence of nature, a tsunami caused by an unforeseeable earthquake, and it is regrettable the crisis has escalated to such an extreme state of affairs. A TEPCO employee said that during the initial hours of the crisis at the Fukushima No. 1 plant that they couldn't foresee the hydrogen explosion: "The possibility of an explosion occurring if hydrogen fills the building was beyond the reach of our thoughts.

14 Sōtegai: Unforeseen events In Japan, nuclear professionals at utilities have always said that a situation where nuclear reactors lost all their power sources for more than 12 hours would be "unthinkable, or sōtegai no matter how bad the disaster. During the first month of the accident, they continued to believe this. Were the earthquake and tsunami sōtegai? Were they truly unforeseeable events? TEPCO in an early press release: "The accident was caused by the violence of nature, a tsunami caused by an unforeseeable earthquake, and it is regrettable the crisis has escalated to such an extreme state of affairs. A TEPCO employee said that during the initial hours of the crisis at the Fukushima No. 1 plant that they couldn't foresee the hydrogen explosion: "The possibility of an explosion occurring if hydrogen fills the building was beyond the reach of our thoughts.

15 Japan Meteorological Agency, January 2011 Earthquake predictions within 30 years Probability of an earthquake Greater than Shindo 6 During the Next 30 Years NPP Prefecture Probability in % Tomari Hokkaido 0.4 Higashi Dori Aomori 2.2 Onagawa Miyagi 8.3 Kashiwazaki Niigata 2.3 Fukushima Daiichi Fukushima 0.0 Fukushima Daini Fukushima 0.6 Tokai Daini Ibaraki 2.4 Hamaoka Shizuoka 84.0 Shika Ishikawa 0.0 Tsuruga Fukui 1.0 Mihama Fukui 0.6 Ooi Fukui 0.0 Takahama Fukui 0.4 Shimane Shimane 0.0 Ikata Ehime 0.0 Genkai Saga 0.0 Sendai Kagoshima 2.3 Monju Fukui 0.5

16 Japan Meteorological Agency, January 2011 Earthquake predictions within 30 years Probability of an earthquake Greater than Shindo 6 During the Next 30 Years NPP Prefecture Probability in % Tomari Hokkaido 0.4 Higashi Dori Aomori 2.2 Onagawa Miyagi 8.3 Kashiwazaki Niigata 2.3 Fukushima Daiichi Fukushima 0.0 Fukushima Daini Fukushima 0.6 Tokai Daini Ibaraki 2.4 Hamaoka Shizuoka 84.0 Shika Ishikawa 0.0 Tsuruga Fukui 1.0 Mihama Fukui 0.6 Ooi Fukui 0.0 Takahama Fukui 0.4 Shimane Shimane 0.0 Ikata Ehime 0.0 Genkai Saga 0.0 Sendai Kagoshima 2.3 Monju Fukui 0.5

17 Japan Meteorological Agency, January 2011 Earthquake predictions within 30 years Probability of an earthquake Greater than Shindo 6 During the Next 30 Years NPP Prefecture Probability in % Tomari Hokkaido 0.4 Higashi Dori Aomori 2.2 Onagawa Miyagi 8.3 Kashiwazaki Niigata 2.3 Fukushima Daiichi Fukushima 0.0 Fukushima Daini Fukushima 0.6 Tokai Daini Ibaraki 2.4 Hamaoka Shizuoka 84.0 Shika Ishikawa 0.0 Tsuruga Fukui 1.0 Mihama Fukui 0.6 Ooi Fukui 0.0 Takahama Fukui 0.4 Shimane Shimane 0.0 Ikata Ehime 0.0 Genkai Saga 0.0 Sendai Kagoshima 2.3 Monju Fukui 0.5

18 Japan Meteorological Agency, January 2011 Earthquake predictions within 30 years Probability of an earthquake Greater than Shindo 6 During the Next 30 Years NPP Prefecture Probability in % Tomari Hokkaido 0.4 Higashi Dori Aomori 2.2 Onagawa Miyagi 8.3 Kashiwazaki Niigata 2.3 Fukushima Daiichi Fukushima 0.0 Fukushima Daini Fukushima 0.6 Tokai Daini Ibaraki 2.4 Hamaoka Shizuoka 84.0 Shika Ishikawa 0.0 Tsuruga Fukui 1.0 Mihama Fukui 0.6 Ooi Fukui 0.0 Takahama Fukui 0.4 Shimane Shimane 0.0 Ikata Ehime 0.0 Genkai Saga 0.0 Sendai Kagoshima 2.3 Monju Fukui 0.5

19 Japan Meteorological Agency, January 2011 Earthquake predictions within 30 years Probability of an earthquake Greater than Shindo 6 During the Next 30 Years NPP Prefecture Probability in % Tomari Hokkaido 0.4 Higashi Dori Aomori 2.2 Onagawa Miyagi 8.3 Kashiwazaki Niigata 2.3 Fukushima Daiichi Fukushima 0.0 Fukushima Daini Fukushima 0.6 Tokai Daini Ibaraki 2.4 Hamaoka Shizuoka 84.0 Shika Ishikawa 0.0 Tsuruga Fukui 1.0 Mihama Fukui 0.6 Ooi Fukui 0.0 Takahama Fukui 0.4 Shimane Shimane 0.0 Ikata Ehime 0.0 Genkai Saga 0.0 Sendai Kagoshima 2.3 Monju Fukui 0.5

20 Japan Meteorological Agency, January 2011 Earthquake predictions within 30 years Probability of an earthquake Greater than Shindo 6 During the Next 30 Years NPP Prefecture Probability in % Tomari Hokkaido 0.4 Higashi Dori Aomori 2.2 Onagawa Miyagi 8.3 Kashiwazaki Niigata 2.3 Fukushima Daiichi Fukushima 0.0 Fukushima Daini Fukushima 0.6 Tokai Daini Ibaraki 2.4 Hamaoka Shizuoka 84.0 Shika Ishikawa 0.0 Tsuruga Fukui 1.0 Mihama Fukui 0.6 Ooi Fukui 0.0 Takahama Fukui 0.4 Shimane Shimane 0.0 Ikata Ehime 0.0 Genkai Saga 0.0 Sendai Kagoshima 2.3 Monju Fukui 0.5

21 Japan Meteorological Agency, January 2011 Earthquake predictions within 30 years Probability of an earthquake Greater than Shindo 6 During the Next 30 Years NPP Prefecture Probability in % Tomari Hokkaido 0.4 Higashi Dori Aomori 2.2 Onagawa Miyagi 8.3 Kashiwazaki Niigata 2.3 Fukushima Daiichi Fukushima 0.0 Fukushima Daini Fukushima 0.6 Tokai Daini Ibaraki 2.4 Hamaoka Shizuoka 84.0 Shika Ishikawa 0.0 Tsuruga Fukui 1.0 Mihama Fukui 0.6 Ooi Fukui 0.0 Takahama Fukui 0.4 Shimane Shimane 0.0 Ikata Ehime 0.0 Genkai Saga 0.0 Sendai Kagoshima 2.3 Monju Fukui 0.5

22 Japan Meteorological Agency, January 2011 Earthquake predictions within 30 years Probability of an earthquake Greater than Shindo 6 During the Next 30 Years NPP Prefecture Probability in % Caveat Risk Practioner! Zero is a dangerous number. Tomari Hokkaido 0.4 Higashi Dori Aomori 2.2 Onagawa Miyagi 8.3 Kashiwazaki Niigata 2.3 Fukushima Daiichi Fukushima 0.0 Fukushima Daini Fukushima 0.6 Tokai Daini Ibaraki 2.4 Hamaoka Shizuoka 84.0 Shika Ishikawa 0.0 Tsuruga Fukui 1.0 Mihama Fukui 0.6 Ooi Fukui 0.0 Takahama Fukui 0.4 Shimane Shimane 0.0 Ikata Ehime 0.0 Genkai Saga 0.0 Sendai Kagoshima 2.3 Monju Fukui 0.5

23 Japan Meteorological Agency, January 2011 Earthquake predictions within 30 years Probability of an earthquake Greater than Shindo 6 During the Next 30 Years NPP Prefecture Probability in % Caveat Risk Practioner! Zero is a dangerous number. Tomari Hokkaido 0.4 Higashi Dori Aomori 2.2 Onagawa Miyagi 8.3 Kashiwazaki Niigata 2.3 Fukushima Daiichi Fukushima 0.0 Fukushima Daini Fukushima 0.6 Tokai Daini Ibaraki 2.4 Hamaoka Shizuoka 84.0 Shika Ishikawa 0.0 Tsuruga Fukui 1.0 Mihama Fukui 0.6 Ooi Fukui 0.0 Takahama Fukui 0.4 Shimane Shimane 0.0 Ikata Ehime 0.0 Genkai Saga 0.0 Sendai Kagoshima 2.3 Monju Fukui 0.5

24 Japan Meteorological Agency, January 2011 Earthquake predictions within 30 years 30 year events Probability of an earthquake Greater than Shindo 6 During the Next 30 Years NPP Prefecture Probability in % Tomari Hokkaido 0.4 Caveat Risk Practioner! Zero is a dangerous number. Higashi Dori Aomori 2.2 Onagawa Miyagi 8.3 Kashiwazaki Niigata 2.3 Fukushima Daiichi Fukushima 0.0 Fukushima Daini Fukushima 0.6 Tokai Daini Ibaraki 2.4 Hamaoka Shizuoka 84.0 Shika Ishikawa 0.0 Tsuruga Fukui 1.0 Mihama Fukui 0.6 Ooi Fukui 0.0 Takahama Fukui 0.4 Shimane Shimane 0.0 Ikata Ehime 0.0 Genkai Saga 0.0 Sendai Kagoshima 2.3 Monju Fukui 0.5

25 Japan Meteorological Agency, January 2011 Earthquake predictions within 30 years 30 year events Probability of an earthquake Greater than Shindo 6 During the Next 30 Years NPP Prefecture Probability in % Tomari Hokkaido 0.4 Caveat Risk Practioner! Zero is a dangerous number. Higashi Dori Aomori 2.2 Onagawa Miyagi 8.3 Kashiwazaki Niigata 2.3 Fukushima Daiichi Fukushima 0.0 Fukushima Daini Fukushima 0.6 Tokai Daini Ibaraki 2.4 Hamaoka Shizuoka 84.0 Shika Ishikawa 0.0 Tsuruga Fukui 1.0 Mihama Fukui 0.6 Ooi Fukui 0.0 Takahama Fukui 0.4 Shimane Shimane 0.0 Ikata Ehime 0.0 Genkai Saga 0.0 Sendai Kagoshima 2.3 Mean Values Monju Fukui 0.5

26 Japan Meteorological Agency, January 2011 Earthquake predictions within 30 years 30 year events Probability of an earthquake Greater than Shindo 6 During the Next 30 Years NPP Prefecture Probability in % Tomari Hokkaido 0.4 Caveat Risk Practioner! Zero is a dangerous number. Higashi Dori Aomori 2.2 Onagawa Miyagi 8.3 Kashiwazaki Niigata 2.3 Fukushima Daiichi Fukushima 0.0 Fukushima Daini Fukushima 0.6 Tokai Daini Ibaraki 2.4 Hamaoka Shizuoka 84.0 Shika Ishikawa 0.0 Tsuruga Fukui 1.0 Mihama Fukui 0.6 Ooi Fukui 0.0 Takahama Fukui 0.4 Shimane Shimane 0.0 Ikata Ehime 0.0 Genkai Saga 0.0 Sendai Kagoshima 2.3 Mean Values Monju Fukui 0.5

27 Japan Meteorological Agency, January 2011 Earthquake predictions within 30 years 30 year events Probability of an earthquake Greater than Shindo 6 During the Next 30 Years NPP Prefecture Probability in % Tomari Hokkaido 0.4 Caveat Risk Practioner! Zero is a dangerous number. Higashi Dori Aomori 2.2 Onagawa Miyagi 8.3 Kashiwazaki Niigata 2.3 Fukushima Daiichi Fukushima 0.0 Fukushima Daini Fukushima 0.6 Tokai Daini Ibaraki 2.4 Hamaoka Shizuoka 84.0 Shika Ishikawa 0.0 Tsuruga Fukui 1.0 Mihama Fukui 0.6 Ooi Fukui 0.0 Takahama Fukui 0.4 Shimane Shimane 0.0 Ikata Ehime 0.0 Genkai Saga 0.0 Sendai Kagoshima 2.3 Mean Values Monju Fukui 0.5

28 Dr. Robert Geller of Tokyo University in Nature, Vol. 472, pg. 408 The regions assessed as most dangerous are the zones of three hypothetical scenario earthquakes (Tokai, Tonankai and Nankai). However, since 1979, earthquakes that caused 10 or more fatalities in Japan actually occurred in places assigned a relatively low probability. This discrepancy the latest in a string of negative results for the characteristic earthquake model and its cousin, the seismic-gap model, strongly suggests that the hazard map and the methods used to produce it are flawed and should be discarded. It is time to tell the public frankly that earthquakes cannot be predicted, to scrap the Tokai prediction system. All of Japan is at risk from earthquakes, and the present state of seismological science does not allow us to reliably differentiate the risk level in particular geographic areas. We should instead tell the public and the government to prepare for the unexpected and do our best to communicate both what we know and what we do not. And future basic research in seismology must be soundly based on physics, impartially reviewed, and be led by Japan s top scientists rather than by faceless bureaucrats.

29 Location and Layout of the Tsunami Walls Water Entrapment Bath tub effect False Sense of Security Tsunami Direction No Tsunami Defense in Depth Existed The tsunami wall was built in The placement of the turbine building in front of the reactor building was done by Babcock & Wilcox.

30 Location and Layout of the Tsunami Walls Water Entrapment Bath tub effect False Sense of Security Tsunami Direction The site was originally 35m above sea level, but was brought down to 10m for construction ease and for bedrock foundations. No Tsunami Defense in Depth Existed The tsunami wall was built in The placement of the turbine building in front of the reactor building was done by Babcock & Wilcox.

31 Location and Layout of the Tsunami Walls Water Entrapment Bath tub effect False Sense of Security No Tsunami Defense in Depth Existed Tsunami Direction The primary, undisputable cause of the The site was originally 35m above sea level, but was brought down to 10m for construction ease and for bedrock foundations. Fukushima Daiichi accident was the placement of the turbine building. The tsunami wall was built in The placement of the turbine building in front of the reactor building was done by Babcock & Wilcox.

32 Was there a tsunami PRA? How were the design heights determined?

33 JSCE Method for Near Field Tusnami Parametric study of the tsunami source location, strike, depth, dip angle, slip angle; Numerical model for near field tsunami Non-linear long wave theory for shallow water, staggered mesh, leap frog method; A model is nothing more than a model until it can be verified by accurate prediction. A simulation is just a simulation, subject to mathematical and programming errors, and needs independent verification. Model, parameter, and simulation uncertainties dominate predictions always.

34 The notion of a design tsunami

35 Time History of the Design Tsunami???

36 Time History of the Design Tsunami??? But which tsunami was the design tsunami for near field? TEPCO stated in July, 2011 that the 1938 Great Fukushima Earthquake (7.9M W ) was used to determine that the design basis height of 5.7m was safe. Was this earthquake chosen to make the data fit the design?

37 Time History of the Design Tsunami??? But which tsunami was the design tsunami for near field? TEPCO stated in July, 2011 that the 1938 Great Fukushima Earthquake (7.9M W ) was used to determine that the design basis height of 5.7m was safe. Was this earthquake chosen to make the data fit the design?

38 Time History of the Design Tsunami??? But which tsunami was the design tsunami for near field? TEPCO stated in July, 2011 that the 1938 Great Fukushima Earthquake (7.9M W ) was used to determine that the design basis height of 5.7m was safe. Was this earthquake chosen to make the data fit the design?

39 TEPCO, from a 2008 presentation about Fukushima Daiichi tsunami defenses : Amazingly, the tsunami wall, built in 1966 with a height of 5.7m, was high enough to conform to the guidelines published in 2002.

40 Historical Earthquakes in the Sendai/Sanriku Area Greater than 8.0M W Year Magnitude Interval in Years

41 Historical Earthquakes in the Sendai/Sanriku Area Greater than 8.0M W Year Magnitude Interval in Years

42 Historical Earthquakes in the Sendai/Sanriku Area Greater than 8.0M W Year Magnitude Interval in Years

43 Historical Earthquakes in the Sendai/Sanriku Area Greater than 8.0M W Year Magnitude Interval in Years earthquake recurrence may be a non-ergodic stochastic process (instationary), therefore the use of Bayes theorem is to some extent misleading - stationary data distributions are not applicable [but] a temporary stable state of the stochastic non-ergodic process of earthquake occurrence in Japan is applicable for about 400 years - but this is only temporary.

44 Historical Earthquakes in the Sendai/Sanriku Area Greater than 8.0M W Year Magnitude Interval in Years earthquake recurrence may be a non-ergodic stochastic process (instationary), therefore the use of Bayes theorem is to some extent misleading - stationary data distributions are not applicable [but] a temporary stable state of the stochastic non-ergodic process of earthquake occurrence in Japan is applicable for about 400 years - but this is only temporary. Dr. Jens-Uwe Klügel KKG

45 Assumptions for the Bayesian Analysis: 1. Historical data is consistently given in M W values. We will assume that all earthquakes greater than 8.0M W have a Shindo of greater than 6; 2. We will only look at earthquakes which affected the Sendai coast; 3. We will begin with the prior probability distribution known as the noninformed prior, which means that we have no prior knowledge as to the actual recurrence frequency; 4. To be conservative, since we do not have evidence of an earthquake with Shindo 6 or greater before 869, we will start the analysis with the Jōgan earthquake and we will do a 1-step update with for both the ergodic and non-ergodic models: 6 events in 1411 years ( ) and 5 events in 399 years ( ). One step updates take into account long time intervals with no events. 5. We will assume that the range of the frequency can be between 1/10,000 years and 1/100 years for the non-informed prior.

46 Bayesian Analysis of Earthquake Greater than Shindo 6 We should consider these results as bounds to our knowledge, with simple yearly recurrence fractions of 1/250 and 1/129. Probability Density 1.60E E E E E E E E E+00 Sanriku/Sendai Earthquake > Shindo 6 Ergodic 1.00E E E E+00 Frequency/year Name Value/Year Recurrence/yr % Recurrence/30 yr Non-ergodic Mean 4.88e-3 1/ % 5th Percentile 2.32e-3 1/ % 50th Percentile 4.70e-3 1/ % 95th Percentile 8.10e-3 1/ % Range Factor 1.87

47 Historical tsunami with run-ups greater than 8m in the Hamadori area of Fukushima Prefecture Year Estimated Maximum Run-up Height Interval in Years 790 BCE 8m 5 CE 8m m 865 On the Sanriku coast (150km north) all three of these tsunami had run-up heights of greater than 18m, and the 1896 and 1933 tsunami had run-up heights of 38m and 29m, respectively

48 Historical tsunami with run-ups greater than 8m in the Hamadori area of Fukushima Prefecture Year Estimated Maximum Run-up Height Interval in Years 790 BCE 8m 5 CE 8m m 865 On the Sanriku coast (150km north) all three of these tsunami had run-up heights of greater than 18m, and the 1896 and 1933 tsunami had run-up heights of 38m and 29m, respectively

49 Historical Tsunami Sources and Heights

50 Historical Tsunami and Tsunami Stones

51 Bayesian Analyses of Tsunami >=8m Probability Density 1.60E E E E E E E E E+00 Sendai Plain Tsunami >= 8m Ergodic Model 1.00E E E E E+00 The non-ergodic thesis would indicate the recurrence frequency of an 8m tsunami in the last 400 years equal to zero. But by using Bayesian analysis, which takes into account NO events in 399 years, we can estimate a distribution. Sendai Plain Tsunami >= 8m Non-Ergodic Model Frequency/year Name Value/Year Recurrence/yr % Recurrence/30 yr Mean 1.43e-3 1/ % 5th Percentile 4.89e-4 1/ % 50th Percentile 1.31e-3 1/ % 95th Percentile 2.77e-3 1/ % Range Factor 1.58 Probability Density 4.50E E E E E E E E E E E E E E E E+00 Frequency/year Name Value/Year Recurrence/yr % Recurrence/30 yr Mean 2.32e-3 1/ % 5th Percentile 2.75e-5 1/ % 50th Percentile 1.59e-3 1/ % 95th Percentile 6.65e-3 1/ % Range Factor 29.03

52 Bayesian Analyses of Tsunami >=8m, Earthquake > Shindo E E-02 Sendai Plain Tsunami and Shindo +6 Ergodic Model We need to consider the dual event of large earthquake AND tsunami to account for the destruction of the local infrastructure. Probability Density 1.20E E E E E E E E E E E E E E-02 Frequency/year Name Value/Year Recurrence/yr % Recurrence/30 yr Mean 8.14e-4 1/ % 5th Percentile 3.87e-4 1/ % 50th Percentile 7.83e-4 1/ % 95th Percentile 1.35e-3 1/ % Range Factor 2.01 Probability Density 1.50E E E-03 Sendai Plain Tsunami and Shindo +6 Non-ergodic Model 0.00E E E E E E E E E+00 Frequency/year The analysis bounds show that the mean value of such an event is between 3.92e-5 and 8.14e-4. Name Value/Year Recurrence/yr % Recurrence/30 yr Mean 3.92e-5 1/25, % 5th Percentile 3.96e-6 1/252, % 50th Percentile 3.59e-5 1/25, % 95th Percentile 7.97e-5 1/12, % Range Factor 4.48

53 Limits of Bayesian Analyses and Simulation Results Bayesian analyses reflect assumptions and state of knowledge from historical events; We use both ergodic and non-ergodic models to bound our knowledge; JSCE values reflect assumptions and state of knowledge given by the numerical simulation and the modeling methods; All information needs to be considered when making a judgment under uncertainty; Information released to the public ignored presentation of the historical evidence and presented the results from the simulation as fact.

54 To keep in mind: Conservatism is an earmark of good risk assessment when faced with large uncertainties in models, simulations, and data. Numerical simulations and Bayesian analyses should be used as bounding studies to inform risk management. Earthquake prediction is almost impossible.

55 Bayesian Analyses of Tsunami >=8m, Earthquake > Shindo 6 The analysis bounds show that the mean value of such an event is between 3.92e-5 and 8.14e-4. What I want to stress is that it is not the probability of CDF only, to be compared with regulation value of 1e-4/y, it is the probability of uncontrolled releases, i.e. PSA Level 2 and not only Level 1, since given the analyzed scenario there was no defence against radioactive releases. Given core melt there will be penetration of the primary loop and also damage to the reactor containment, both due to the forces of tsunami but also as a consequence of hydrogen explosions that will follow melting of the fuel. What I am trying to say is that the calculated value should be compared with a regulation value of uncontrolled releases to the environment, or with LERF, in the range of 1e-5/y - 1e-6/y (I m not sure what value is stipulated in Japan) and not 1e-4/y. --- Jerzy Grynblat

56 Moreover, from a presentation in July at the ANS Conference in the USA, this slide was presented (red lines are mine): This probabilistic method was due for acceptance in 2011.

57 and it was not unforeseen (August 25, 2011)

58 Tsunami Lesson On March 11, experts had not imagined an angle of subduction so acute, the effect of which was the large tsunami. No tsunami defense in depth existed. The turbine building was extremely vulnerable because of its location between the reactor and the sea. NPP operators in Japan should prepare for the maximum tsunami run-up height which could happen within a large costal area of the plant, make extensive changes in plant layout, and create defense in depth capabilities.

59 Somehow conservatism with tsunami safety was lacking. Japanese NPP structures are built to withstand twice the anticipated maximum M W. During V&V of the tsunami simulation, claims were that the design tsunami exceeded the scenario tsunami by a 2 times height, on average. Given that from reliable historical sources that 8m tsunami within a 50km radius had been encountered, TEPCO should have built a tsunami wall capable of at least an 16m run-up (2 times 8m).

60 Physician, Heal Thyself: The Inadequacy of PRA Considerations Tsunami initiating events; Multi unit impacts; The double initiator: earthquake and tsunami; Partial core exposure or hydrogen explosions; Ground separation or subsidence by earthquakes; Accidents with durations greater than 24 hours or procedures for events lasting longer than 24 hours; Severe aftershocks and their impact on weakened facilities or equipment; Systems to mitigate station blackout; Extended fuel storage or fires impacting the storage; Accident management and emergency recovery human factors in our PRA.

61 Unconsidered (But Likely) Impediments to Recovery at Daiichi Radioactive releases hampering activities; Effects of hydrogen explosions; The difficulty recovering from simultaneous damage to multi-unit sites; Severe damage to the surrounding local infrastructure; Uncertainty in the recovery roadmap.

62 As of April, 29 th, consequences to People, Plant & Performance We have multi-unit, continual high radiation; We have contaminated water and sludge to dispose of from the emergency cooling measures; The infrastructure and management are still unprepared for accidents during recovery; There is no defense in depth for the economy of Japan; There is increased seismic vulnerability of recovery equipment and structures; There are +100,000 people evacuated from radiation zones when the surrounding area is overloaded with tsunami evacuees; There are no level 2 or level 3 analyses available on site; Sub-contractors and TEPCO staff are stressed, underfed, overworked, sleep deprived doing dangerous work in radiation suits; We have inadequate risk communication to the public and other countries; The actions of TEPCO, the regulators, and the Japanese government have caused fear and mistrust in Japanese society and the world community, and continues daily.

63 Conclusion: Nuclear Power Risk Assessment in Japan The larger questions: 1. Do risk assessment professionals in Japan understand risk assessment? 2. Is risk assessment undertaken only for showing regulators that safety goals have been attained? 3. Are risk professionals willing to ask and to answer the difficult questions to themselves, the regulators, and the public?

64 Afterword Our generation who has, consciously or unconsciously, approved the construction of the Fukushima nuclear power plants and enjoyed the benefits of the vast supply of electricity generated, in particular those of us who hailed the slogan that Nuclear Power is Safe should be the first to join a Skilled Veteran Corps to install or repair the cooling system. This is the duty of our generation to the next generation and the one thereafter. Y. Yamada, 71 years old, former Sumitomo Metals Engineer, in an open to his colleagues.

65 For more information: Woody Epstein Manager Risk Consulting, Japan