*5.2.1. Seismic hazard*

Full-scope site geological, geophysical, seismological and geotechnical investigation and evaluation has been performed with subsequent probabilistic seismic hazard assessment (PSHA). The methodology is described in (Tóth et al., 2009).

Seismic Safety Analysis and Upgrading of Operating Nuclear Power Plants 107

seismic input, since the increase in the amplitudes in ground motion response spectra due to relatively small increase (one order of magnitude) in the exceedance probability is

The parameters of a 10-2/a non-exceedance level earthquake have also been defined. The PGA is equal to 0.087g in this case. This information is used for certain fatigue type analyses. The response spectrum and cumulative absolute velocity criteria are used for the definition

There is Pleistocene layer of 25-30 m covering the site, the upper 12-15 part of which originates from floods and consists of fine structure, well classified sand, while its lower part consist of sandy gravel and gravel. Under the Pleistocene layers, there are various upper Pannonian layers, which are irregularly divided by sandstone ridges. These ridges are cemented to various extents and can be regarded as semi-rock. The 25-30 m saturated young soft soil (~300m/s shear-wave velocity) covering the eroded Pannonian surface at the site is susceptible to liquefaction at the depth 10-15 m. The probabilistic liquefaction analysis performed in 1995 has shown that the best estimate return period of the occurrence of liquefaction exceeds 10000 years therefore the liquefaction was not considered as a design

For the seismic PSA purposes, the evaluation of the site effects was extended to very low probabilities (10−4÷10−6/a). According to the seismic PSA the building relative settlement due to the liquefaction is the dominating effect contributing to the CDF just below the design basis probability level. This experience triggered a state-of-the-art analysis of the liquefaction. It was also observed that the uncertainty of the analysis is essential due to

Liquefaction susceptibility can be expressed in terms of factor of safety *FSliq* against the

����� <sup>=</sup> ���

where *CRR* is the cyclic resistance ration and the *CSR* is cyclic stress ratio, see Regulatory Guide 1.198 (NRC, 2003). The cyclic stress generated by the given earthquake is as follows

� = ���� � ����

���

*v0* and 

overburden stresses, respectively, and *rd* is a nonlinear stress reduction coefficient that varies

*av* is the equivalent shear stress amplitude, *amax* is the peak horizontal acceleration at

� �������

��� = ��� ��� ��� (9)

� ���� (10)

*'v0* are the total and effective vertical

of OBE exceedance; see US NRC Regulatory Guide 1.166 (NRC, 1997a).

accounted.

*5.2.2. Geotechnical conditions* 

basis phenomenon.

occurrence of liquefaction as,

(Seed and Idriss 1971):

where 

with depth.

uncertainty of soil parameters and the methods.

ground surface, *g* is the acceleration of gravity,

The design base earthquake is defined on the 10-4/a non-exceedance level, taken on the mean hazard curve. Recent Hungarian regulatory requirement is: the design base event has to be defined on the median hazard curve at 0.005 non-exceedance probability for the total lifetime of the plant, which means exactly a 10-4/a frequency for a 50 years operational lifetime (or approximately 10-5/a for a new built).

The horizontal and vertical peak ground accelerations (PGA) are equal to 0.25g and 0.2g respectively. (The PGA correlated to the original design basis seismicity was 0.025g.)

The uniform hazard response spectra were defined for the Pannonian surface (as for a virtual outcrop) below the site. The ground motion response spectra (GMRS) are calculated taking into account the nonlinear features of the soft soil layer covering the site.

The results are shown in Figure 4.

**Figure 4.** Accounting site mean spectral ratios and mean bedrock and surface UHRS at 5% damping for three different probability levels

The acceptability of the obtained ground motion response spectra for the design base was justified earlier (in 1995) by comparison with deterministically defined 84% response spectra (on the basis of US NRC Draft Guide 1032 issued later as Regulatory Guide 1.165) and recently per U.S. NRC Regulatory Guide 1.208 (NRC, 2005) and ASCE/SEI 43-05 procedure (ASCE, 2005). The latter ensure the avoidance of the cliff-edge effect with respect to the seismic input, since the increase in the amplitudes in ground motion response spectra due to relatively small increase (one order of magnitude) in the exceedance probability is accounted.

The parameters of a 10-2/a non-exceedance level earthquake have also been defined. The PGA is equal to 0.087g in this case. This information is used for certain fatigue type analyses. The response spectrum and cumulative absolute velocity criteria are used for the definition of OBE exceedance; see US NRC Regulatory Guide 1.166 (NRC, 1997a).

#### *5.2.2. Geotechnical conditions*

106 Nuclear Power – Practical Aspects

*5.2.1. Seismic hazard* 

**5.2. The seismic design basis** 

The results are shown in Figure 4.

three different probability levels

(PSHA). The methodology is described in (Tóth et al., 2009).

lifetime (or approximately 10-5/a for a new built).

Full-scope site geological, geophysical, seismological and geotechnical investigation and evaluation has been performed with subsequent probabilistic seismic hazard assessment

The design base earthquake is defined on the 10-4/a non-exceedance level, taken on the mean hazard curve. Recent Hungarian regulatory requirement is: the design base event has to be defined on the median hazard curve at 0.005 non-exceedance probability for the total lifetime of the plant, which means exactly a 10-4/a frequency for a 50 years operational

The horizontal and vertical peak ground accelerations (PGA) are equal to 0.25g and 0.2g

The uniform hazard response spectra were defined for the Pannonian surface (as for a virtual outcrop) below the site. The ground motion response spectra (GMRS) are calculated

**Figure 4.** Accounting site mean spectral ratios and mean bedrock and surface UHRS at 5% damping for

The acceptability of the obtained ground motion response spectra for the design base was justified earlier (in 1995) by comparison with deterministically defined 84% response spectra (on the basis of US NRC Draft Guide 1032 issued later as Regulatory Guide 1.165) and recently per U.S. NRC Regulatory Guide 1.208 (NRC, 2005) and ASCE/SEI 43-05 procedure (ASCE, 2005). The latter ensure the avoidance of the cliff-edge effect with respect to the

respectively. (The PGA correlated to the original design basis seismicity was 0.025g.)

taking into account the nonlinear features of the soft soil layer covering the site.

There is Pleistocene layer of 25-30 m covering the site, the upper 12-15 part of which originates from floods and consists of fine structure, well classified sand, while its lower part consist of sandy gravel and gravel. Under the Pleistocene layers, there are various upper Pannonian layers, which are irregularly divided by sandstone ridges. These ridges are cemented to various extents and can be regarded as semi-rock. The 25-30 m saturated young soft soil (~300m/s shear-wave velocity) covering the eroded Pannonian surface at the site is susceptible to liquefaction at the depth 10-15 m. The probabilistic liquefaction analysis performed in 1995 has shown that the best estimate return period of the occurrence of liquefaction exceeds 10000 years therefore the liquefaction was not considered as a design basis phenomenon.

For the seismic PSA purposes, the evaluation of the site effects was extended to very low probabilities (10−4÷10−6/a). According to the seismic PSA the building relative settlement due to the liquefaction is the dominating effect contributing to the CDF just below the design basis probability level. This experience triggered a state-of-the-art analysis of the liquefaction. It was also observed that the uncertainty of the analysis is essential due to uncertainty of soil parameters and the methods.

Liquefaction susceptibility can be expressed in terms of factor of safety *FSliq* against the occurrence of liquefaction as,

$$F\mathbb{S}\_{liq} = \frac{CRR}{\mathbb{CSR}}\tag{9}$$

where *CRR* is the cyclic resistance ration and the *CSR* is cyclic stress ratio, see Regulatory Guide 1.198 (NRC, 2003). The cyclic stress generated by the given earthquake is as follows (Seed and Idriss 1971):

$$\text{CSR} = \frac{\tau\_{av}}{\sigma\_{vo}'} = 0.65 \cdot \left(\frac{\sigma\_{vo}}{\sigma\_{vo}'}\right) \cdot \left(\frac{a\_{max}}{g}\right) \cdot r\_d \tag{10}$$

where *av* is the equivalent shear stress amplitude, *amax* is the peak horizontal acceleration at ground surface, *g* is the acceleration of gravity, *v0* and *'v0* are the total and effective vertical overburden stresses, respectively, and *rd* is a nonlinear stress reduction coefficient that varies with depth.

Depending on the method used the value of safety factor varies in rather wide range. The methodologies (Seed & Idriss, 1971, 1982; Tokimatsu & Yoshimi, 1983) used for Paks site resulted in a relative low margin, while the analysis via effective stress method provides much larger margins (Győri et al, 2002).

Seismic Safety Analysis and Upgrading of Operating Nuclear Power Plants 109

fixes in the over-crowded by equipment and piping gallery building. If it the case, the systems for regular heat removal placed in the turbine hall would be available for heat removal after an earthquake, if their re-qualification is performed. Meantime, in 1995 the site seismic hazard evaluation has been completed which resulted in the DBE with 0.25g PGA. Response and stress calculations made for the newly defined DBE have shown that essential part of the mechanical equipment and pipelines can sustain the DBE demand and the reinforcement of the systems and structures necessary for seismic safety is feasible

Theoretical considerations have been made for the evaluation of upgrading effort required for fixing the pipelines and components required for heat removal via systems "as usual", i.e. systems dedicated for emergency cases. It has been assumed that the "as is" seismic capacity of the pipe segments can be treated as a random variable; its value can be expressed by total design capacity multiplied by several factors representing the randomness of the actual design features, floor-response, etc. If it is the case, the calculated "as is" HCLPF values of pipe segments have to be lognormal distributed. If the distribution is known, the parameters of the distribution can be defined on the basis of HCLPF calculations for "as is" conditions and the number of pipe segments requiring fixes can be evaluated. The distributions of "as is" HCLPF values of pipe segments presented in Figure 5 justified the assumptions and made possible the evaluation of

**Figure 5.** Distribution of "as is" HCLPF values of pipe segments (units 1-4 and units 3-4)

great number (more than 100/unit) of fast closing valves.

Based on the assessment of fixes of the piping and components, the cost of these fixes turned to be cheaper than the (automatic) isolation of the unreinforced parts of the systems by a

0.01 0.1 1 10

HCLPF

with reasonable effort.

upgrading needs.

0.0

0.2

0.4

probability

0.6

0.8

1.0

The building settlement caused by earthquake can affect the underground communications (service water piping and emergency power supply cables) due to relative displacements. This effect will be amplified if liquefaction occurs. The dominant failure mode in the acceleration ranges higher than the design basis is due to the relative building caused by the soil liquefaction. This makes it necessary to re-qualify the underground lines and connections jeopardized by the settlement of the main building or, if it is necessary, to modify them to make their relative movement unimpeded. An advanced probabilistic liquefaction and relative building settlement analysis is going on using an amended soil parameter database in relation to the investigation of beyond design base vulnerabilities performed for severe accident management reasons (Győri et al, 2011).
