Impact of seismicity on Nice slope stability—Ligurian Basin, SE France: a geotechnical revisit

Abstract

The shallow Nice submarine slope is notorious for the 1979 tsunamigenic landslide that caused eight casualties and severe infrastructural damage. Many previous studies have tackled the question whether earthquake shaking would lead to slope failure and a repetition of the deadly scenario in the region. The answers are controversial. In this study, we assess for the first time the factor of safety using peak ground accelerations (PGAs) from synthetic accelerograms from a simulated offshore Mw 6.3 earthquake at a distance of 25 km from the slope. Based on cone penetration tests (CPTu) and multichannel seismic reflection data, a coarser grained sediment layer was identified. In an innovative geotechnical approach based on uniform cyclic and arbitrary triaxial loading tests, we show that the sandy silt on the Nice submarine slope will fail under certain ground motion conditions. The uniform cyclic triaxial tests indicate that liquefaction failure is likely to occur in Nice slope sediments in the case of a Mw 6.3 earthquake 25 km away. A potential future submarine landslide could have a slide volume (7.7 × 10⁶ m³) similar to the 1979 event. Arbitrary loading tests reveal post-loading pore water pressure rise, which might explain post-earthquake slope failures observed in the field. This study shows that some of the earlier studies offshore Nice may have overestimated the slope stability because they underestimated potential PGAs on the shallow marine slope deposits.

Appendix

Materials and methods

Cyclic Shear Stress

Any arbitrary earthquake signal can be translated into a uniform cyclic loading signal defined by a CSR_eq and an equivalent number of uniform cycles (Cetin and Seed 2004; Liu et al. 2001; Seed and Idriss 1971). The maximum cyclic shear stress was calculated at ~ 23 mbsf. The total vertical stress was calculated with an average bulk density of 1800 kg m⁻³, which is representative for the slope sediments (Kopf and Cruise Participants 2008), and a Mediterranean water density of 1035 kg m⁻³. The stress reduction factor accounts for the damping of the soil as an elastic body (Seed and Idriss 1971). Thus, it considers the variation of cyclic shear stresses with depth and was calculated according to a modified equation after Cetin and Seed (2004). The stress reduction factor is based on four descriptive variables:

$$ {r}_d=\frac{\left(1+\frac{-23.013-2.949\times {a}_{max}+0.999\times {M}_w+0.0525\times {V}_{s12m}}{16.258+0.201\times {e}^{0.341\times \left(-20+0.0785\times {V}_{s12m}+7.586\right)}}\right)}{\left(1+\frac{-23.013-2.949\times {a}_{max}+0.999\times {M}_w+0.0525\times {V}_{s12m}}{16.258+0.201\times {e}^{0.341\times \left(0.0785\times {V}_{s12m}+7.586\right)}}\right)}-0.0046\times \left(d-20\right) $$

(10)

where a_max is the peak ground acceleration, d is the depth of the sediment, V_s12m is the mean shear wave velocity in the upper 12 m of sediment, and M_w is the moment magnitude of the earthquake. Table 2 summarizes our input parameters to calculate the seismic demand (in terms of cyclic shear stress) at depth.

The cyclic shear stress is induced by cyclic vertical loading and unloading on a cylindrical sediment sample at constant lateral stress. The maximum cyclic shear stress τ_cyc in the triaxial sample is:

$$ {\tau}_{cyc}=\frac{q_{cyc}}{2} $$

(11)

The samples were loaded in harmonic compression-extension mode (i.e., q_min < 0 < q_max and |q_min| = |q_max|). The loading signal was applied with a frequency of 1 Hz. Both the loading pattern and the loading frequency are standards in earthquake engineering (ASTM Standard D5311/D5311M − 13 2013; Kramer 1996). The vertical displacement, principle stresses, deviator stress, and excess pore water pressure were recorded at 100 Hz during cyclic loading. Prior to each experiment, the samples were vacuum saturated to a Skempton B-value ≥ 0.92 (Skempton 1954) with deionized, deaerated water.

Seismic waves passing a sediment are associated with complex strain and stress paths near the ground surface, where the principle stresses change in direction and magnitude (El Shamy and Abdelhamid 2017). Thus, Seed et al. (1978) investigated the impact of multidirectional loading conditions and suggested a strength reduction factor of 10% for uniaxial loading. We corrected the CRR by 10% to account for the unidirectional loading during the triaxial tests.

$$ {CRR}_{0.9}= CRR\times 0.9 $$

(12)

Sample Preparation

Most triaxial tests were conducted on reconstituted samples (of the original sediment) to make sure that (i) there are no mineralogical differences from one sample to another, (ii) the samples are homogenous, and (iii) we could perform as many tests as needed without running out of sample material. Reconstituted samples were prepared from a slurry following the approach from Bradshaw and Baxter (2007). The samples were prepared by mixing soil and water to a slurry with a water content of 33%, which is 2% higher than the liquid limit (Fig. 4). The slurry was filled in a cylindrical mold and tamped to remove air bubbles. The samples were one-dimensionally pre-consolidated to 100 kPa vertical stress. After pre-consolidation, the samples were set up in the triaxial cell and vacuum saturated for at least 2 h. In the DTTD, the samples were isotopically consolidated, with a ramp sufficient small to allow the sample to drain, to an effective confining stress of 170 kPa. This sample preparation procedure allowed us to create comparable homogenous samples with a small scatter in void ratios (Table 1):

$$ e=\frac{V_V}{V_S} $$

(13)

where V_V is the volume of voids and V_S is the volume of solids.

In contrast, core samples were carefully extracted from the core via a metal cylinder to maintain the in situ fabric as good as possible. We used for core and reconstituted samples the same consolidation procedure. By comparing core and reconstituted samples under identical loading conditions, the influence of remolding on the cyclic shear strength was evaluated.