SPECTRAL RESPONSE FEATURES USED IN LAST IAEA STRESS TEST TO NPP CERNAVODA (ROMANIA) BY CONSIDERING STRONG NONLINEAR BEHAVIOUR OF SITE SOILS GHEORGHE MĂRMUREANU, ELENA-FLORINELA MANEA # , CARMEN ORTANZA CIOFLAN, ALEXANDRU MĂRMUREANU, DRAGOS TOMA-DANILĂ National Institute for Earth Physics, P. O. Box MG-2, RO-077125 Bucharest – Magurele, Romania E-mail: [email protected]# Corresponding author: [email protected]Received February 17, 2017 Abstract. Devastating – and, in some sense, unforeseen – earthquakes in Nepal, Japan, New Zealand, Haiti, Sumatra and elsewhere have triggered in last time a heated debate about the legitimacy and limitations of Probabilistic Seismic Hazard Assessment (PSHA). This method is a pure numerical creation, as it was developed from mathematical statistics and not based on earthquake physics. An important source of errors came from one of its key component: the Ground Motion Prediction Equation (GMPE), which describes a relationship between a ground motion parameter (PGA etc.), magnitude M, distance R., without taking into account the nonlinear behavior of site soils during strong earthquakes. In order to capture these effects, the spectral amplification factors (SAF) were computed for the last recorded Vrancea intermediate-depth events showing a decrease with the increase of earthquake magnitude of earthquakes and these values are far of those given by R.G.1.60 of the U. S. Atomic Energy Commission and IAEA Vienna. These SAF were used by NIEP, as alternative analysis, for NPP Cernavoda (Romania) in last “2011 STRESS TEST” asked by IAEA Vienna after Japan strong earthquake (M w = 9.0). Key words: Seismic hazard analysis, strong Vrancea earthquakes, nonlinear soil behavior. 1. INTRODUCTION Estimating seismic ground shaking is an important step in anticipating earthquake effects on people and structures. Ground motion at a particular site can be influenced by three main elements: source, seismic wave’s travel path, and local site conditions. If the first describes how the size and nature of the earthquake source controls the generation of earthquake waves, then the second describes the effect of the earth on these waves as they travel (at some depth) from source to a particular location (site), and the third describes the effect of the uppermost several hundred meters of rock and soil and the surface topography at that location on the resultant ground motion produced by the emerging or passing earthquake waves. Romanian Journal of Physics 62, 822 (2017)
11
Embed
SPECTRAL RESPONSE FEATURES TEST TO NPP CERNAVODA BY ... · spectral response features used in last iaea stress test to npp cernavoda (romania) by considering strong nonlinear behaviour
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
SPECTRAL RESPONSE FEATURES USED IN LAST IAEA STRESS
TEST TO NPP CERNAVODA (ROMANIA) BY CONSIDERING
STRONG NONLINEAR BEHAVIOUR OF SITE SOILS
GHEORGHE MĂRMUREANU, ELENA-FLORINELA MANEA#, CARMEN ORTANZA
CIOFLAN, ALEXANDRU MĂRMUREANU, DRAGOS TOMA-DANILĂ
National Institute for Earth Physics, P. O. Box MG-2, RO-077125 Bucharest – Magurele, Romania
In all extra-Carpathian area (Fig. 1) the geophysical profile composition
passes from clay to sandy clays or sands, from marl to sandy marl or sand lenses
etc. The sedimentary cover, relatively thick (exceeding 5 km), is the result of four
major cycles of sedimentation: Paleozoic, Permian-Triassic, Jurassic-Cretaceous
and Upper Miocene – Quaternary [1]. A soil is of basic type sand or gravel (termed
coarse soils), silt or clay (termed fine soils) etc. For example, in south of Bucharest
there are layers of 520 m of dense sand and in the north the thickness of all
sedimentary layers is around 1,480 m [2, 3, 4]. Also, the sedimentary thickness
reaches 15–20 km [5] below Focsani Depression, located in front of Eastern
Carpathian Curvature (Fig. 1) etc.
Fig. 1 – The VRANCEA99 seismic refraction experiment: The crustal structure beneath the
southeastern Carpathians and the Moesia Platform from a seismic refraction profile in Romania
between the cities Bacau and Bucharest, traversing the Vrancea epicenter region in NNE–SSW
direction [5, 6].
2. GROUND MOTION AND BASIC SEISMIC HAZARD
Seismic wave attenuation can be thought of as consisting of two major
elements, geometric spreading and absorption (sometimes called damping). The
methodology used in most probabilistic seismic hazard analysis was first defined
3 Spectral response used in last IAEA stress test to NPP Cernavoda (Romania) Article no. 822
by Cornel [7] in 1968. At last ICTP Advanced Conference on Seismic Risk
Mitigation and Sustainable Development from Trieste (Italy) on 10–14 May, 2010,
Professor Z. Wang from University of Kentucky, USA [8] questioned: What is
Probabilistic Seismic Hazard Analysis (PSHA)?
where γ(y) is the annual probability of ground motion “y” being exceeded. In other
form:
where: S = the number of seismic areas; ν = the expected frequency, per time
period, per seismic area, – of earthquakes of magnitude at least mo; Φ'(…) = the
„standard normal complementary cumulative distribution function (CCDF)” which
is based on the usual assumption that the ground motion parameter is a lognormal
aleatory variable. The ground motion distribution is possibly truncated. This
function has a value less than unity; fM(…) = probability density function (PDF) of
the magnitude distribution; fR(…) = probability density function (PDF) of
distances, from the site, of the locations of earthquakes, given an earthquake
occurs in the seismic area; λ(a) = annual rate of events with site ground motion
level a or more. Under the additional assumption that the events in every source
follow independent Poisson processes, the mean rate λ(a) can be used to compute
the probability of exceeding, in any time interval of length t:
P [A > a in time t] =1–e–λ(a)t
(3)
in which P[.] is the probability of the event. Consequently, double integral Φ'(…)
can’t exceed value “1”. PSHA was developed from mathematical statistics
(e.g. Benjamin and Cornell, 1970 [9] under four fundamental assumptions [7]):
(a) Constant in time average occurrence rate of earthquakes; (b) Equal likelihood of
earthquake occurrence along a line or overall area source (single point);
(c) Variability of ground motion at a site is independent; (d) Poisson (or memory
lees) behavior of earthquake occurrence. In summary, the probabilistic seismic
hazard analysis (PSHA model) is flaw: (1) it is not based on earthquake science
(invalid physical models; point source; Poisson distribution); (2) invalid
mathematics; (3) miss interpretation of annual probability of exceeding or return
period. PSHA become a pure numerical creation [8]. A key component for seismic
hazard assessment including all methods developed so far is the ground motion
attenuation prediction equation (GMPE) which is a statistical tool in this
)1(dd)()()}(lnd]2
)ln(lnexp[
2
11{)()(
2
ln,
2
0 ln,
rmrfmfyyy
vyYvPy RM
y
mr
y
y
)2()](exp[––1dd)()/(),(–ln
’exp(––1)(1
amrmfmrfrmga
vaAPii MR
S
i
i
Article no. 822 Gheorghe Mărmureanu et al. 4
methodology. More, in last strong Vrancea earthquakes there are many peak
ground accelerations larger than epicenter values [3]. Estimating seismic ground
shaking is an important step in anticipating earthquake effects on people and
structures.
3. NEW ALTERNATIVE OF SEISMIC HAZARD EVALUATION FOR NPP CERNAVODA
The authors, in order to make quantitative evidence of large nonlinear
effects, used/ introduced and developed the concept of the nonlinear spectral
amplification factor (SAF) as ratio between maximum spectral absolute
acceleration (Sa), relative velocity (Sv), relative displacement (Sd) from response
spectra for a fraction of critical damping (ζ %) at fundamental period or any period
and peak values of acceleration (amax), velocity (vmax) and displacement (dmax),
respectively, from processed strong motion records , that is: (SAF)a= Smax
a /amax ;
(SAF)v = Smax
v /vmax ; (SAF)d = Smax
d /dmax; amax = y (t)max ; vmax = x (t)max; dmax =
= x(t)max. In Tables 1 are given the data recorded on Cernavoda NPP seismic station
(Table 1) and spectral amplification factors (SAF) for last strong Vrancea
earthquakes (Table 2). All data are obtained in the same conditions. The concept
was used for last Stress Test asked by IAEA Vienna for Romanian Cernavoda NPP
after strong 2011 Tohoku earthquake (MW = 9.0) on March 11, when NIEP, and
INCERC [10] recorded last three deep strong Vrancea earthquakes: August
30,1986 (MW = 7.1), May 30 (MW = 6.9) and May 31,1990 (MW = 6.4).
Fig. 2 – The absolute values of the variation of dynamic torsion modulus function (G, daN/cm2)
and torsion damping function (D%) of specific strain (γ%) for marl samples obtained in Hardin &
Drnevich resonant columns (USA patent) from NIEP, Laboratory of Earthquake Engineering
[11, 12, 13].
5 Spectral response used in last IAEA stress test to NPP Cernavoda (Romania) Article no. 822
The Spectral Amplification Factors (SAF) were finally computed for
absolute accelerations at 5% fraction of critical damping (ζ = 5%) in two seismic
stations on NPP Cernavoda site for last strong and deep Vrancea earthquakes:
August 30, 1986 (MW = 7.1 and h = 131.4 km); May 30,1990 (MW = 6.9 and
h = 90.9 km) and May 31,1990 (MW = 6.4 and h = 86.9 km). The geophysical
profile for NPP Cernavoda site is as follows: first 5.00 m of fractured limestone
with shear modulus Gmax
= 7,000 daN/cm2, internal damping, Dmin = 3.7% and
density, ρ = 2,3 t/m3; next 7.00 m of fractured limestone with clay with G
max =
= 6,000 daN/cm2, Dmin = 3.6% and ρ = 2,1 t/m
3; next 34.00 m of marl with G
max =
= 4,470 daN/cm2, Dmin = 4.2% and density ρ = 2,1 t/m
3. The marl is going down
more than 270.00 m [11, 13]. In Tables 1 is given the data recorded at Cernavoda
NPP seismic station and spectral amplification factors (SAF) for last strong
Vrancea earthquakes (Table 2). All data are obtained in the same conditions.
Fig. 3 – Dependence of dynamic torsion modulus function (G, daN/cm²) and of torsion damping
function (D%) and with shear strains (γ%) and frequency (ω) for clay samples obtained in Hardin &
Drnevich resonant columns (USA patent) from NIEP, Laboratory of Earthquake Engineering from
NIEP. Between 1 and 10 Hertz, shear modulus – G and damping – D are constant, domain used
in design of civil engineering structures [11, 12].
In Tables 3, 4 and Fig. 4, spectral amplification factors are given where the
effect of the nonlinearity in Cernavoda NPP site is characterized by the coefficient
“c”. The coefficient “c” is the ratio of SAF for May 31,1990 Vrancea earthquake to
SAF for each stronger earthquake. Sa*(g) and a*(g) are the maximum spectral
acceleration and, respectively, maximum acceleration if the system would have a
linear response (behavior) to the fundamental period. For Vrancea earthquake of
May 31, 1990 (MGR = 6.1) the response can be assumed to be in elastic range and
we have the possibility to compare the nonlinear effects with those predicted by a
linear model.
Article no. 822 Gheorghe Mărmureanu et al. 6
If we maintain the same amplification factor (SAF = 5.7851) as for relatively strong earthquake of May 31, 1990 with magnitude MW = 6.4 then at Cernavoda NPP Seismic Station for earthquake magnitude of May 30, 1990 (Mw=6.9) the peak acceleration has to be a*max = 0.1241(+21.6%), while the recorded values were only, amax = 0.102g. Similarly, for Vrancea earthquake of August 30,1986 (MW = 7.1), the peak acceleration has to be a*max = 0.0907g (+41.8%), instead of actual acceleration value of 0.064g recorded at Cernavoda NPP Seismic Station (Table 3). The present analysis indicates that the effect of nonlinearity could be very important and from Table 3 is 41.87%; from Table 4 is 49.1% and for stronger earthquakes it will be larger. In addition, these spectral amplification factors are function of the earthquake Vrancea magnitude and soil structure.
Table 1
Data recorded on Cernavoda NPP Seismic station and peak values of acceleration,
velocity and displacement
Nr. Vrancea earthquakes Component amax (cm/s2) vmax
(cm/s)
dmax
(cm)
1
August 30, 1986,
MW = 7.1; MGR = 7.0;
h =131.4 km
N-S
V
E-W
50.62
63.28
62.85
4.33
2.71
4.05
0.83
0.65
0.69
2
May 30,1990,
MW = 6.9; MGR = 6.7;
h = 90.9 km
N-S
V
E-W
107.11
53.10
100.43
9.83
7.28
9.80
3.37
4.26
2.71
3
May 31,1990,
MW = 6.4; MGR = 6.1;
h = 86.9 km
N-S
V
E-W
49.74
14.29
50.03
2.88
1.31
2.88
0.63
0.29
0.34
From Tables 3 and 4 we can see that there is a strong nonlinear dependence of the spectral amplification factors (SAF) on earthquake magnitude [7] for all records made on NPP Cernavoda Site for last strong Vrancea earthquakes. The amplification factors are decreasing with increasing the magnitudes of deep strong Vrancea earthquakes and these values are far of that given by Regulatory Guide 1.60 of the U. S. Atomic Energy Commission [13]. The spectral amplification factors (SAF) and, in fact, the nonlinearity, is functions of Vrancea earthquake magnitude. The amplification factors decrease as the magnitude increases. If we will use for NPP Cernavoda site a relation of form: SAF = a MW
2 + b MW + c and if
we introduce data from Table 3 for MW =7.1; 6.9 and 6.4 we have: a = –1.7857; b = 21.6898 and c = –59.976 and
SAF = –1.7857 MW2 + 21.6898 MW – 59.876 (4)
and for MW = 7.2, SAF = 3.72; MW = 7.3, SAF = 3.29; MW = 7.4, SAF = 2.87;
M = 7.5, SAF = 2,35 etc. Regulatory Guide 1.60 of the U. S. Atomic Energy
Commission and IA.EA Vienna are using, all time, a constant value, SAF= 3.13.
7 Spectral response used in last IAEA stress test to NPP Cernavoda (Romania) Article no. 822
Table 2
Spectral amplification factors (SAF). Cernavoda NPP site
Earthquake Damping
(ξ%) Comp. (g)
(g)/amax
Svmax Sv
max/vmax
August 30,1986;
MW = 7.1; h = 131.4
km
2% N-S V
E-W
0.277 0.295 0.366
5.36 4.57 5.71
35.443 34.442 34.221
8.18 12.70 8.44
5% N-S V
E-W
0.193 0.203 0.261
3.74 3.14 4.07
24.102 22.583 23.601
5.56 8.33 5.82
10% N-S V
E-W
0,131 0.149 0.191
2.53 2.31 2.98
17.482 15.031 15.440
4.03 5.54 3.81
20% N-S V
E-W
0.087 0,107 0.133
1.68 1.65 2.07
13.053 9.796
11.059
3.01 3.61 2.73
May 30,1990,
MW = 6.9; h = 90.9 km
2% N-S V
E-W
0.652 0.259 0.765
5.97 4.78 7.47
50.231 24.168 57.917
5.11 3.32 5.91
5% N-S V
E-W
0.398 0.153 0.485
3.65 2.83 4.75
31.722 20.112 38.166
3.23 2.76 3.89
10% N-S V
E-W
0.306 0.116 0.340
2.80 2.14 3.32
22.581 15.824 25.885
2.30 2.17 2.64
20% N-S V
E-W
0.222 0.090 0.217
2.03 1.66 2.12
15.624 12.042 16.114
1.59 1.65 1.64
May 31,1990,
MW = 6.4; h = 86.9 km
2% N-S V
E-W
0.259 0.068 0.430
5.11 4.67 6.53
17.825 15.646 32.644
3.09 11.94 11.33
5% N-S V
E-W
0.173 0.050 0.293
3.41 3.43 5.78
12.433 12.510 23.353
2.15 9.55 8.11
10% N-S V
E-W
0.115 0.035 0.217
2.27 2.40 3.30
8.639 8.984
15.784
1.50 6.86 5.48
20% N-S V
E-W
0.074 0.024 0.145
1.46 1.65 2.20
6.454 5.944
11.240
1.12 4.54 3.90
Article no. 822 Gheorghe Mărmureanu et al. 8
Table 3
Cernavoda NP Plant NIEP Seismic Station [11]; Φ = 44.340 N; λ = 28.030 E