You are on page 1of 6

Proceedings of the 14th Australia and New Zealand Conference on Geomechanics, Cairns 2023 (ANZ2023)

Site-Specific Probabilistic Seismic Hazard and Ground Response Analysis on


Snowy 2.0 Pumped-hydro Storage Project

Ching Dai1 Usama Fauzi2


1 Senior Principal geotechnical engineer, PhD, Tetra Tech Coffey, Sydney Australia, ching.dai@tetratech.com
2 Former senior geotechnical engineer, PhD, Tetra Tech Coffey

ABSTRACT: The current Australia Seismic Design Code (AS1170) is limited to a PGA 2,500-year return period. To satisfy the
project requirement of 1 in 10,000 years seismic return period, a project-wide site-specified Probabilistic Seismic Hazard Assessment
(PSHA) has been carried out on the Snowy 2.0 project to derive the earthquake loading for its structures, including the intakes,
multiple tunnels and portals, the underground powerhouse, buildings, slopes, surge chamber and gate shafts. The assessment was
based on the site-mapped paleo-faults crossing the ~27km long Snowy 2.0 tunnel alignment between the Tantangara and Talbingo
reservoirs and the nearby neotectonic faults in the National Fault Source Model. The site soil/rock strength and shear wave velocity
have been included in the analysis. Results of the site-specific PSHA are used to develop the 5% damped peak ground acceleration
and response spectra for a 475-year return period, a 2,000-year return period and a 10,000-year return period. The analysis develops
the different design levels for horizontal peak ground acceleration (PGA) values. Deterministic Seismic Hazard Analysis (DSHA)
was evaluated and compared with PSHA. Vertical de-amplification by 1-D seismic ground response analysis was carried out. A de-
amplification factor of 0.5 is recommended for deep underground structures. The site-specific PSHA and ground response analyses
achieve a rationale for seismic loading for the Snowy 2.0 project. The paper presents the challenges of the seismic hazard analysis
beyond the current Australia Code limitation and also sets out the innovations incorporated during the detailed design analysis.
KEYWORDS: Site-specific seismic hazard, spectrum, ground response

1. INTRODUCTION

The 2,000 MW Snowy 2.0 Project (Figure 1) links the two pumping capabilities work in a ‘closed’ pumped-storage system,
existing Snowy Scheme reservoirs, Tantangara and Talbingo, i.e. water is recycled between the upper reservoir (Tantangara)
through 10.9m diameter tunnels and an underground power and a lower reservoir (Talbingo), so the same water can be used
station with pumping capabilities. Hydropower will be generated to generate power more than once, making the most of available
with ~1 km pressure head-spinning reversible turbines, which water to meet peak power demands.
can also pump water in the opposite direction. Snowy 2.0’s

Figure 1. Snowy2 project – Power waterway long section and interpreted geology long section

The seismic design of Snowy 2.0 requires compliance with


the following standards and guidelines:
• Australian Standard AS 1170.4 – 2007 Part 4: Earthquake 2. SEISMIC SOURCE MODEL
actions in Australia – Earthquake Actions in Australia, and
• ANCOLD Guidelines for Design of Dams and Appurtenant PSHA starts with the identification of seismogenic source zones
Structures for Earthquake (May 2019); (SSZ) within snow mountain areas. Each SSZ is characterised by
However, the maximum return period of earthquakes its geometry, earthquake potential and probability distribution of
considered in AS1170.4:2007 is 2500 years (given in Table 3.1 potential rupture locations within the SSZ.
of the standard). ANCOLD guidelines require that the maximum Geoscience Australia products, and contributors from the
return period of earthquakes is 10,000 years. The Snowy 2.0 wider Australian seismology community, have produced a
Project requires a site-specific seismic hazard assessment to National Seismic Hazard Assessment (NSHA18). The NSHA18
determine the earthquake loading levels. It is understood that the project conducted the complete Probabilistic Seismic Hazard
project must demonstrate that it has met the information and Analysis and generated the Hazard Curve for all capital cities in
design requirements in ANCOLD. Australia.
Proceedings of the 14th Australia and New Zealand Conference on Geomechanics, Cairns 2023 (ANZ2023)

The existing shallow crustal faults and area seismic sources • GMPEs with associated weights (refer to Table 5 in
throughout Australia have been characterised and incorporated Allen et al. (2018b).
into the NSHA18. Details of the data for the NSHA are described • Seismic Source Models as shown in Table 3 and
in detail by Allen et al. (2018b, 2018a). Seismic hazard Table 4 in Allen et al. (2018b).
parameters presented in this paper are based on the NSHA18 • Fault source parameters: magnitude frequency
seismic source models (i.e. fault, area source, smoothed gridded distribution and the potential periodic or episodic recurrence
seismicity). The subduction zone in NSHA18 was not behaviour.
incorporated in our model as these sources are located more than • Area source (regional and background) parameters:
1000km north and should not affect the hazard at Snowy 2.0. Hypocentral depths as shown in (Table 9 in Allen et al. (2018b)
For the analysis of the SNOWY 2.0 project, the NSHA and maximum magnitude Mmax as shown in Table 4 in Allen et
seismic source model approximately 500km from Snowy 2.0 was al. (2018a).
used. The model was validated by comparing the PGA hazard
curve from this analysis with the NSHA18 model at Canberra.
3.2 The Site-Specific PSHA Results

The PSHA was carried out for the reference Site Sub-soil Class
B – Rock as defined in AS 1170.4 – 2007. The locations are
Tantangara intake, Talbingo intake, and powerhouse.

3.2.1 Site-Specific Hazard Curve

Hazard curves developed from the PSHA illustrate the variations


in PGA as a function of the annual exceedance probability
(AEP). The AEP is the reciprocal of the return period.
Figure 3 shows the PGA (5%-damped) hazard curve
developed for return periods ranging from 10 to 10,000 years at
different design locations.
Figure 2. Neotectonic fault traces and locations used in NSHA18 Table 1 lists the PGA value for selected return periods of
engineering interest.
The nearby crustal fault around the site is Tantangara Fault,
see Figure 2. A prominent, sharp, west-facing scarp suggests that Table 1 Site-Specific PGA (5%-Damped) on the ground surface
for Subsoil Class B (Vs30 = 760 m/s) at Different Return Periods
this fault has been active in the Cenozoic period. The Fault is
mapped as being continuous with the Jindabyne Thrust
(Tantangara and Berridale 1:100k geological sheets). However,
the topography does not support a connection in terms of the
continuous scarp.
In total, 20 alternative seismic source models have
contributed to NSHA18, consisting of eight developed by
Geoscience Australia (Burbidge et al., 2012; Griffin et al., 2016;
Leonard, 2008) and eight third-party contributors (Cuthbertson,
2016; Dimas et al., 2016; Dimas and Venkatesan, 2016; Hall et
al., 2007; Mote et al., 2017; Sinadinovski and McCue, 2016). The
seismic source models are uploaded to open-source Seismic
Hazard Analysis (SHA) software (OpenQuake 2.1.0). Detailed * min. 0.08 as required by AS1170.4-R2018
information on the recurrence parameters of each source model
can be found in Appendix B of Tetra Tech Coffey (2020) report. The seismic hazard result (Figure 3 and Table 1) shows that:
• For Tantangara intake, the seismic hazard is dominated
by the Tantangara fault, which is located less than 500m
3. PROBABILISTIC SEISMIC HAZARD ANALYSES from the site, resulting in higher hazards compared to
Powerhouse and Talbingo Intake.
3.1 Logic tree analysis
• For Powerhouse and Talbingo intake, the Tantangara
This section presents the results of the site-specific SHA for fault is located far away > 17km from the site.
Tantangara intake (148.6523°E, The longitude and 35.7926°S Powerhouse and Talbingo Intake seismic hazard is
latitude), Talbingo intake (148.3776°E longitude and mainly contributed to by Tumut Pond Fault and
35.768741°S latitude), and the tunnel between those two intakes. Khancoban Yellow Bog Fault located 12km from the
The method initially formalised by Cornell (1968) is adopted site.
in the analysis. The logic tree was developed through expert
elicitation workshops. It was weighted four independent regional
source models (Griffin et al., 2018). Experts were invited to
constitute peer-reviewed seismic source models and Ground
Motion Prediction Equations (GMPE) models for consideration,
resulting in 20 seismic source models being proposed. To weigh
the candidate seismic source and GMPE used in NSHA18, 18
experts in seismic hazard assessment, representative of the
collective expertise of the Australian earthquake hazard
community, were invited to two workshops held at Geoscience
Australia in March 2017. The logic tree results in the workshop
are used in NSHA18. The details are described below:
Proceedings of the 14th Australia and New Zealand Conference on Geomechanics, Cairns 2023 (ANZ2023)

Following the definitions above, it was concluded that the


neotectonic faults that may have moved 5 – 10 million years ago
are not active, and none of the faults in the Snowy area would be
regarded as active faults. This includes the Tantangara Fault
which is in the reservoir close to the Tantangara intake is
included as a neotectonic fault with a very small slip rate in the
NSHA1018 model.
In the region of very low to low seismic activity, our
knowledge of faults and their activity is derived from NSHA 18,
and there are no active faults. Therefore, we conducted a DSHA
check for reservoir-induced seismicity (Mw = 3.5 and R = 1km)
and a reasonable nearby earthquake (Mw = 5.5 and R = 25km).
The results show that 84th of DSHA is less than PSHA at 2,000
years, which does not govern the design.

5. ASSESSMENT OF SITE SUBSOIL/ROCK-CLASS

The PSHA results of this study are for the Site Subsoil Class B
Figure 3 Hazard Curve at different design locations (Rock) condition. A soil amplification factor is needed to adjust
the PGA from Subsoil Class B conditions to the actual site
conditions. Soil amplification factors are included in AS 1170.4
3.2.2 Uniform Hazard Response Spectra - 2007. Our assessment of the Site Subsoil Class for each of the
Snowy 2.0 hydro main structures is based on the site geological
The 5%-damped uniform hazard horizontal acceleration mapping, rock outcrop descriptions and investigations.
response spectra for Class B – Rock (Vs30 = 760m/s) are shown
in Figure 4. Talbingo Intake
The Ravine Beds at the Talbingo Intake and tailrace tunnel are
described as a shallow marine deposit of shale, slate, siltstone
and conglomerate overlain by the Devonian Boraig and Byron
Groups around Ravine. When fresh, the Ravine bed rocks
appear to be generally strong. Weathering which weakens the
rock may extend a few 10s of m below the surface. As the intake
structures are shown as 40 to 80m beneath the surface, and in
fresh rock, in our assessment, the Site Sub-soil Class will be A -
strong rock or B rock. Until further information from drilling
investigations is available for additional evaluation, we have
assigned Site Sub-soil Class B - Rock to the Talbingo Intake
structures.

Tantangara intake and boreholes (BH1114 to BH1117)


The underlying rock encountered in all boreholes drilled at
Tantangara Reservoir intake is described as porphyritic dacite,
Figure 4 Uniform Hazard Spectra dark blueish/greenish grey with white, green, grey phenocrysts.
Point load tests (PLT) were carried out at about every metre
The above PSHA results are for the Site Subsoil Class B of core in both axial and diametral directions. Four UCS tests
(Rock) condition. Soil amplification factors are discussed in from BH1115 have an average UCS of 35MPa with a minimum
Section 5 this paper. of 26MPa and a maximum of 44MPa. There were 100 PLTs
carried out at BH1115, with a calculated average UCS of
34.64MPa. The maximum and minimum calculated UCS values
4. DETERMINISTIC SEISMIC HAZARD ANALYSES were 125.1MPa and 5.2MPa, respectively.
CHECK If the UCS/Is(50) multiplier from BH1115 is applied to the
PLTs from BH1116 (138 PLTs valid results) and BH1117, then
In cases where active faults are present in the vicinity of the site, the average calculated UCS for BH1116 is 34.9MPa with a
the deterministic method can be used. maximum of 82.3MPa. The PLTs at BH1117 (based on 58 valid
DSHA uses available active fault data to generate single- results) have an average of 38.88MPa, with a maximum of
valued estimates of earthquake ground motions which might 97.5MPa. Based on these data with an average UCS <50 MPa,
occur at a site. The terminology of active fault is from ANCOLD Site Sub-soil Class B can be assigned to the Tantangara intake
Guidelines for Design of Dams and Appurtenant Structures for structures.
Earthquakes which is adapted from ICOLD (2016):
• An active fault is defined as a fault, reasonably Powerhouse Complex
identified and located, known to have produced historical The power station cavern complex is in the Ravine Beds West
earthquakes or showing evidence of movements in Holocene rock mass, with the surge shaft also in this unit. The inclined
time (i.e. in the last 11,000 years) and large faults that have pressure shaft starts in the Ravine Beds, then passes through the
moved in the latest Pleistocene time i.e. between 11,000 and Boraig Group and then re-enters the Ravine Beds. The Boraig
5,000 years ago). Group was formed during a period of explosive volcanism
• A Neotectonic Fault is a fault, not active as defined predominantly appearing as rhyolite with rhyodacite, tuff, lapilli
above, that experienced displacement under conditions imposed tuff and feldspathic sandstone. It unconformably overlies the
in the current crustal stress regime and hence may move again in Ravine Beds. The drilling data from BH5209, BH5206 and
the future. BH5207 indicates the predominant lithology encountered in the
Proceedings of the 14th Australia and New Zealand Conference on Geomechanics, Cairns 2023 (ANZ2023)

power station cavern complex was Ravine Beds West, generally


comprising siltstone, with minor beds of conglomerate and
sandstone. At approximately 450m AHD the planned power
station cavern complex is about 750m below ground level, ans
the rock mass is fresh and of very high strength with variable
fracturing. Based on this data, we assigned Site Sub-soil Class A
– Strong Rock to the powerhouse complex.
Table 2 lists the UCS value from BH5209, BH5206 and
BH5207, assessed site class and recommended amplification
factor for the Snowy 2.0 hydro sites.

Table 2 Recommended amplification factors for PGA

(a) Deaggregation of PSHA (all sources)

* Refer to the next section of De-amplification factor assessment

Considering that the de-amplification is significant, a wave


propagation analysis was performed in order to determine the
spectrum in the under-ground structures based on the spectrum
at the outcrop. Section 6 below presents the analysis details.

6. DE-AMPLIFICATION SEISMIC GROUND RESPONSE

The hazard curve by all seismic source models at the Powerhouse


is presented in Figure 5.
The goal of the seismic ground response analysis (SGRA) is
to decrease the uncertainties in soil and rock de-amplification
factors. In the geology cross-section, the Powerhouse is carefully
located in a good quality rock mass at 750m below ground level (b) Deaggregation of selected time histories
(bgl) or Site Subsoil Class A. At present, the de-amplification Figure 6 Deaggregation of PSHA at period, T, of 0s or PGA for the
factors based on AS 1170.4 and the recent Pacific Earthquake 10,000-year return period
Engineering Research (PEER) study by Stewart and Seyan
(2013) for Site Subsoil Class A are 0.8. However, due to the The scope of SGRA work included:
quality of the rock and the depth of the powerhouse, de-
amplification of the ground response could be significant. As 6.1 Step 1: De-aggregation
such, a site-specified SGRA was performed to determine the Hazard deaggregation was used to identify the seismic sources
response spectrum in the underground structures. that are the most significant contributors to the total probabilistic
hazard at a given return period and PGA. In addition to the hazard
curve that shows the combined effect of the earthquake
magnitude and source-to-site distance from the seismic source,
we disaggregated the hazard by calculating the fractional
contribution to the total hazard from various combinations of
magnitude and distance.
De-aggregation of PSHA (Figure 5) was conducted to identify
the relative contribution of earthquakes of various magnitudes
and site-to-source distances. The PSHA was de-aggregated at
period T of 0s. It should be noted that PSHA has a 20 seismic
sources model. Each seismic source model has been de-
aggregated. Figure 6(a) shows the hazard de-aggregation results
for all sources and indicates there are two main groups of seismic
sources that contribute to the seismic hazard. These are
Figure 5 hazard curves at Powerhouse by all seismic source models
background and nearby neotectonic faults.
Proceedings of the 14th Australia and New Zealand Conference on Geomechanics, Cairns 2023 (ANZ2023)

6.2 Step 2: Selection of horizontal acceleration time histories


We have selected a suite of eleven acceleration time histories to
represent the hazard. We searched the PEER NGA2 and NGA
East databases (PEER, 2013 and 2014), and selected time
histories and linear scaling factors to reasonably match the target
spectra. The selected time histories are consistent with the
rupture scenarios that control the hazard at the site, as shown in
Figure 6(b).

Figure 8 The adopted shear wave velocity profile

6.5 Step 5: Deconvolution


Figure 7 Comparison of matched time histories response spectra and the
target spectra DEEPSOIL 7.0 (Hashash, 2016) provides options for the
deconvolution of time histories at a depth of 750m bgl via a
6.3 Step 3: Spectral match frequency-domain equivalent linear analysis approach. The
The acceleration time histories for the SGRA were spectrally deconvolution was performed by inputting spectrally matched
matched to the target acceleration response spectra using the acceleration time histories at the ground surface. The
computer program SeismoMatch™ version 2020. corresponding rock motion at 750m bgl is then computed and
The spectral matching results were considered acceptable provided. The analysis output is presented in the next section.
when:
• The matched acceleration response spectra agree
7. RESPONSE SPECTRA AT POWERHOUSE
reasonably well with the target spectrum.
• The non-stationary characteristics of the initial time One-dimensional Seismic Ground Response Analysis (SGRA)
series are well maintained. was carried out for the ground profile and time history to assess
• The Arias intensity of the initial time histories and the PGA and spectral acceleration (Sa) deamplification factor for
spectrally matched time histories agree reasonably well. the underground site. The de-amplification factor is calculated as
Figure 7 shows a comparison of the response spectra of the ratio of peak acceleration between the surface and the base of
individually matched time histories to the target spectra. Figure the 1D ground profile. It was estimated that the predominated site
7 indicates that the response spectra of individual matched time period as 0s or PGA based on T = 4H/Vs, where H is the
histories agree reasonably well with the target spectra within the thickness of rock/soil.
period range of interest from 0s to 4s. Figure 7 also indicates that Figure 9 shows the computed ratio of Spectrum at 750m bgl
the geometric mean spectra of the eleven matched time histories Spectrum to that at the surface from the nonlinear and equivalent
are close to the target spectra within the same period range of linear analyses using the eleven times histories. The PGA is at
interest. period zero from frequency analysis. The deamplification factor
is the ratio of PGAs at the underground point and at the surface
6.4 Step 4: Develop site-specified dynamic rock properties point. Figure 9 indicates the PGA is the value that approaches the
Dynamic rock properties for the seismic ground response period of zero.
analyses were derived from the geotechnical data collected at the
site. Vs is a fundamental parameter for seismic ground response
analyses.
We used existing site geotechnical data (Geotechnical and
Geological Profile, SG-GE01-G-DRG-0002, Tetra Tech Coffey
2020) and nearby shear wave velocity (Vs) survey (BH5113C).
The selected Vs profile to represent the site is shown in Figure 8
below. The potential variability of the rock was further
considered by allowing a variation of +/- 25m/s to the assumed
shear wave velocity.

Figure 9 Spectrum ratio (Spowerhouse/Ssurface)


Proceedings of the 14th Australia and New Zealand Conference on Geomechanics, Cairns 2023 (ANZ2023)

Generally, the results are consistent with previous experience


with SGRA. The de-amplification factor of 0.5 at PGA is lower
in amplitude than those provided in AS 1170.4 Table 6.4 (de- or similar conditions (a de-amplification factor of 0.8).
amplification factor of 0.8). Figure 10 below is the plot of the de- Therefore, it was recommended that a de-amplification factor of
amplification factor vs the depth from the ground surface from 0.5 should be applied to the PGA in the Powerhouse. The detailed
the SGRA analysis and the recommendation of the recommendation of PGA based on PSHA and SGRA to comply
Deamplification Factor of 0.5. It results a cost-effective design with the design criteria is listed in Table 3.
solution at the Powerhouse.
Table 3 Recommended PGAs for Powerhouse and Tunnel

Spectral Acceleration (g)


Location OBE DBE SEE & MDE SEE & MDE SEE & MDE
(High C (Equivalent to ULS by (Extreme
Consequence) AS1170.4) Consequence)
(145 (475
years) years) (2,500 years) (10,000 years)
(2,000 years)

Powerhouse 0.02g** 0.04g** 0.065g* 0.075g* 0.17g


0.12g (on the 0.13g (on the surface) 0.27g (on the
Talbingo surface) 0.065g* (80m below surface)
side 0.06g* (80m below the surface) 0.135g* (80m below
the surface) the surface)
Tunnel 0.02g** 0.04g**
0.15g (on the 0.18g (on the surface) 0.5g (on the
Tantagara surface) 0.09g* (80m below surface)
side 0.075g* (80m below the surface) 0.25g* (80m below
the surface) the surface)
* The Deamplification Factor of 0.5 is adopted
** The min. 0.08 are required by AS1170.4-R2018

9. ACKNOWLEDGEMENT

This paper is in memory of our deceased colleague, geologist


Richmond Beetham. He undertook the geological part of the
PSHA work for this project. The authors recognised and
appreciated his contribution to the project.
The project was undertaken under the DJV, a Design Joint
Venture composed of Tractebel, Lombardi and Tetra Tech
Coffey. It also closely works with the cooperation of the
Contractor Future Generation Joint Venture. The authors
appreciate their cooperation and support. It is also appreciated
the project owner - Snowy Hydro Limited has approved the paper
for publication.
Figure 10 De-amplification factor vs the depth from the ground surface

10. REFERENCES
8. CONCLUSIONS
Allen, Trevor Ian, Clark, D., Griffin, J., Geoscience Australia, 2018a.
A site-specifc seismic hazard analysis (SHA) was completed for The 2018 National Seismic Hazard Assessment for Australia: model
the proposed Snowy 2.0 site located at the Snowy Mountains in input files.
New South Wales, Australia for the Tantangara intake Allen, Trevor Ian, Griffin, J., Ghasemi, H., Leonard, M., Clark, D.,
(148.6523°E longitude and 35.7926°S latitude), Talbingo intake Geoscience Australia, 2018b, The 2018 National Seismic Hazard
Assessment for Australia: model overview.
(148.3776°E longitude and 35.768741°S latitude), and the
Allen, T.I., Leonard, M., Ghasemi, H., Gibson, G., 2018. The 2018
tunnels and powerhouse located between those two intakes. SHA National Seismic Hazard Assessment for Australia – earthquake
were completed for a rock condition (Subsoil Class B, Vs30 = epicentre catalogue. Geoscience Australia
760 m/s) and based principally on the 2018 National Seismic https://doi.org/10.11636/Record.2018.030
Hazard Assessment for Australia (NSHA18). C.A. Cornell, 1968, Engineering seismic risk analysis, Bulletin of the
Results of the site-specific PSHA were used to develop the Seismological Society of America, 58(5), 1583–1606.
5% damped peak ground acceleration and response spectra with Tetra Tech Coffey, Dec. 2020, Seismic Hazard Analysis Report – Snowy
a 145-year return period, a 475-year return period, a 2,000-year 2 Hydro Project, S2-GEO-GN-REP-1004_E_GE06
and a 10,000-year return period. The ANCOLD ANCOLD, 2019. Guidelines for Design of Dams and Appurtenant
Structures for Earthquake, May 2019. ed.
(2019)/ICOLD148 guidelines are to be applied to recommend the
AS1170.4-R2018, Structural design actions, Part 4: earthquake actions in
design level for horizontal peak ground acceleration (PGA) Australia
values listed in Table 3 below. The PSHA results of this study
are for the Site Subsoil Class B (Rock) condition.
A Seismic ground response analysis (SGRA) has been
completed for the proposed Powerhouse (148.4421 E longitude
and -35.7846 S latitude) of the Snowy 2.0 project. The SGRA
comprised the development of a site-specific ground profile,
review and the de-aggregation of the PSHA conducted,
development of input acceleration time histories, deconvolution,
and conducting 1D total stress nonlinear and equivalent linear
SGRA to assess the PGA and Sa de-amplification factors.
The computed de-amplification factors are approximately 0.5
for the 10,000-year return period. The computed amplification
factors are lower in amplitude than those provided in AS 1170.4

You might also like