I_P said:
TEPCO has considerable stake in claiming that the meltdowns were entirely the result of an unforeseeable, rare event (huge tsunami) rather than being initiated by earthquake damage from shaking that was within or just barely exceeded the plant design basis.
I assume that the judgement that the shaking was within or just barely exceeded the design basis was based on a publication by TEPCO of the
Seismic Data measured at Fukushima (
http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110516e27.pdf). From the table, 18 maximum response accelerations were reported for Dai-Ichi, one each for each direction (i.e. north-south, east-west and up-down) for each of the 6 reactors. 3 of the 18 exceeded design parameters by 25% (unit 2, E-W), 15% (unit 3, E-W) and 21% (unit 5, E-W), respectively. However, 17 of the 18 observations are marked with an asterisk showing that only about 2-3 minutes of the earthquake was recorded, due to a glitch in the software but TEPCO said that the error probably wasn't that significant.
Notwithstanding this report, it is clear that this was a huge earthquake and that strong shaking lasted between 3 and 5 minutes (
http://www.scientificamerican.com/article.cfm?id=fast-facts-japan). When earthquakes are that long, simply measuring peak acceleration may be an incomplete statistic. What impact does earthquake duration have on structural damage? First, I looked at
Effect of Earthquake Duration on Structural Reliability, Lindt & Goh, 2004, but they only looked at durations of 30 to 90 seconds (page 1591). Earthquakes of 3 to 5 minutes are apparently quite outside the norm.
I then looked at a FEMA publication called
Structural Dynamics of Linear Elastic Single-Degree-of-Freedom (SDOF) Systems (http://www.nibs.org/client/assets/files/bssc/Topic03-StructuralDynamicsofSDOFSystemsNotes.pdf ). Among other things, it covers the important issue of resonances induced by earthquakes in structures. It is well worth reading to understand the next point, unless you are familiar with the subject. On page 88, there is a "Four-way Log Plot of (the) Response Spectrum (Plotted vs. Period)." It would seem that this is the same type of plot in TEPCO's publication on page 5. TEPCO's pdf is of very good quality and one can zoom in on the two 4-way log plots to see them better. Figure 2-2 is of ground motion at the GN4 Observation Point. If you look carefully at the zoomed-in 4-way plot, it looks like the plot of the South Point exceeds 500 gals (peak ground acceleration) for wavelengths (TEPCO calls it a period) between about 0.06 seconds to about 0.22 seconds. The plot appears to approach
1000 gals at a wavelength of just over 0.1 seconds in the E-W direction. If you call it 950 gals, that would be more than twice the approximately 450 gals design basis. The N-S plot (Fig. 2-4) also appears to significantly exceed 500 gals for a wavelength of 0.1 seconds. (I am looking at the lines on the plot that are in the SW to NE direction and that are labeled "50," "100," "200," "500," "1000," and "2000." Outside the top right of the plot there is a label shown diagonally in the same direction as "cm/s2," which would be gals.)
On page 6 of TEPCO's report, Fig. 4-1 seems to show the acceleration response spectra for the base mat of unit 1 exceeding
1000 gals for a wavelength of just over 0.5 seconds. Likewise, Fig. 4-2 shows the response spectra for unit 2 as substantially exceeding
1000 gals for a wavelength peaking at about 0.3 seconds. Similarly, Fig. 4-3 shows a peak for unit 3 of about
1500 gals for a wavelength peaking also at 0.3 seconds. Fig. 4-4 for unit 4 is lower but still has peaks that clearly appear to be above 500 gals. Fig. 4-5 is for unit 5 and shows a peak exceeding
1000 gals also at a wavelength of about 0.3 seconds. Fig 4-6 is for unit 6 and shows a peak approaching
1000 gals at a wavelength of about 1.25 seconds. These are all in the E-W direction.
In the N-S direction, Fig. 4-9 for unit 3 shows a peak almost at
1000 gals for a wavelength of about 0.45 seconds. Figures 4.7 (unit 1) and 4.8 (unit 2), show peaks clearly above 500 gals and maybe closer to 800 - 900 gals at a wavelength also of about 0.45 seconds. Figures 3-1 through the top plot of 3-6 clearly show the data recording stopping abruptly.
Unless I am completely misunderstanding these plots, it would seem that the peak accelerations at Fukushima Dai-Ichi greatly exceeded the design basis of about 450 gals at least at specific wavelengths.
So, I looked further to see if there were other measurements of earthquake intensity. I looked at
Analysis of Cumulative Absolute Velocity (CAV) and JMA Instrumental Seismic Intensity (I_JMA) Using the PEER-NGA, Strong Motion Database, Campbell and Bozorgnia, 2010 (
http://peer.berkeley.edu/publicatio...s_2010/webR_PEER10_102_Campbell_Bozorgnia.pdf).
On pdf pg. 14,
Cumulative absolute velocity (CAV) is defined as the integral of the absolute value of an acceleration time series... (and) ...includes the cumulative effects of ground motion duration. This is a key advantage of CAV over other peak ground motion and response-spectral parameters and is one of the reasons that EPRI (1988) found it to be the ground motion parameter that best correlated with structural damage out of the many ground motion parameters that it investigated.
The authors then defined a slightly different measure that they termed standardized CAV in which accelerations below a threshold value are not counted (pdf pg. 16) on the basis that small magnitude earthquakes of long duration (which may cause no damage whatsoever) could have a large CAV. On that same page, they note that
(t)he USNRC (U.S. Nuclear Regulatory Commission 1997) uses standardized CAV as one of the ground motion measures to determine whether a nuclear power plant must be shut down after an earthquake when the operating basis earthquake (OBE) ground motion is exceeded.
On page 22:
The CAV check is exceeded if anyone of the three components of the standardized CAV from the free-field ground motion is greater than 0.16 g-sec. If both the response spectrum check (see pdf pp. 21 & 22 for details) and the CAV check are exceeded, the OBE is considered exceeded and plant shutdown is required.
The paper then goes into detail regarding the JMA seismic intensity scale (I_JMA) (JMA = Japan Meteorological Agency). The rest of the paper is interesting but can be skipped for the purposes of this discussion.
I then looked at
Damage Indicating Parameters and Damage Modes of Mechanical Components by K. Ochiai (Japan Nuclear Technology Institute), K. Kobayashi (TEPCO) and A. Chigama (IAEA) (http://www.jnes.go.jp/seismic-symposium10/presentationdata/4_sessionC/C-24.pdf ). On pdf page 3, there is a table that compares I_JMA and standardized CAV. Damage measured by I_JMA is due to a "Large Effective Inertia Force" and is related to "First Excursion Damage." In other words, I-JMA best measures the damage caused by a transient large amplitude deflection. Damage measured by standardized CAV is due to "Much Energy Accumulation" and is related to "Cumulative Damage (Fatigue/Ductility Exhaustion)." In short, standardized CAV is a measure of fatigue due to repeated cycling of structural elements and is related to the total energy absorbed by the structure.
Page 5 has a chart for several earthquakes plotting standard CAV on the y-axis and I_JMA on the x-axis. The S-CAV scale is from 0 to 4.5. Page 6 shows peak accelerations at Kashiwazaki-Kariwa NPS No. 7 unit of 673 gals N-S and 1007 gals E-W at the turbine pedestal top. Page 7 shows several data points for K-K, but none exceeding a S-CAV of 3 (on the right hand side where the units for S-CAV are in (g-sec)). Now, please skip to page 11. On the lower right is a chart again plotting S-CAV against 2 JMA scales. on that chart is a shaded box labeled "Design Base" which has a maximum S-CAV of about 2.25. Let's focus on that number.
I then looked at a Preliminary Report done by ITER consult on the accident at Fukushima Dai-Ichi (
http://www.iter-consult.it/ITER_Report_Fukushima_Accident.pdf) dated May 2011. On page 13, it states that
Daiichi units 2, 3 and 5 exceeded their maximum response acceleration design basis in E-W direction by about 20%. Recording was over 130-150 seconds,
which is in agreement with TEPCO's report referenced above.
However, the ITER report goes on to say:
Various parameters have been proposed in the literature for estimation of the destructive power of an earthquake. Among these parameters, the CAV (cumulative absolute velocity) has been recently proposed. Using the data recorded in the Tohoku event, the CAV can be evaluated in
10
, whereas in Kashiwazaki-Kariva earthquake of 2007 the CAV was equal to 2 with a recorded Peak Ground Acceleration (PGA) much higher tha(n) in Fukushima. This is apparently due to the exceptional duration of the Tohoku event.
Recalling the report on Damage Indicating Parameters, the charts showed standardized CAV with a maximum of 4.5 and the "Design Base" at about 2.25, yet here they are mentioning a CAV of 10
for the Tohoku (March 11, 2001) earthquake. It is not clear whether the 10 is a CAV or a standardized CAV. See Fig. 2.1 of the Campbell and Bozorgnia paper on pdf page 15 for a chart showing the difference between CAV and S-CAV. Even if the 10 is CAV, that would still give a very large value for recomputed standardized CAV for the Tohoku earthquake.
Now referring back to page 9 of the Damage Indicating Parameters presentation, they report a piping vibration test with a maximum acceleration of 1,877 gals that resulted in a crack during the "5th excitation" (repetition of the test). The S-CAV was reported as 23.2 g-sec or about double the CAV of the Tohoku earthquake. On page 10, they show the "Damaging Excitation Motion" test run (looks like a seismogram) and compare it to an actual seismogram from Unit 1 (presumably at K-K) at the same scale with a peak acceleration of 884 gals and an S-CAV of 2.4 g-sec. They did not have a crack in a pipe until the 5th repetition. In other words, a pipe break (which would cause a LOCA) did not take place in the test until the piping had been subjected to 5 simulated earthquakes of more than twice the intensity (measured by CAV) each of the Tohoku earthquake.
If the CAV for the Tohoku earthquake reported by ITER was calculated at Fukushima Dai-Ichi (it isn't clear), then it means that the earthquake that hit that power plant was truly huge, but yet maybe unlikely to have caused a break in piping, if the test at K-K is comparable to Fukushima.
One caveat on the ITER report as on page 2 they report:
In fact 12 out of 13 back-up D/Gs on site, located in the basements of the turbine buildings, were disabled. Only one air-cooled D/G (all others were seawater-cooled) was able to supply electrical power to units 5 and 6, which remained under full control after some initial troubles.
I had made that statement in a prior post and we all now know that is wrong, but I knew I had read it somewhere and I hadn't made it up. So the possibility exists that the ITER report has other errors in it.
In conclusion, it isn't altogether clear that "the shaking (at Fukushima) was within or just barely exceeded the design basis." The intensity as measured by CAV or S-CAV may have greatly exceeded the design basis.
