Part 7 - Simplified Performance-Based Liquefaction Hazard Analysis with the CPT

Summary

The lecture presents a simplified, map-based way to approximate performance-based liquefaction hazard analyses for CPT data. The goal is to make results similar to a full probabilistic analysis available for everyday engineering projects, without requiring engineers to run complex probabilistic calculations or specialized software.

Why use a simplified method?

A full performance-based analysis can involve millions of iterative calculations and may be impractical for routine consulting. The lecturer compares the proposed approach to the USGS’s simplified seismic-hazard workflow: engineers obtain a mapped value, then adjust it for local site conditions. The mapped value is not itself the site-specific answer; it is a reference that is corrected using information about the site.

Reference maps and their meaning

The method assumes a generic “reference” soil profile and performs analyses at many grid points across a region. Results are then mapped for selected return periods, including 475, 1,033, and 2,475 years.

These are liquefaction loading or reference-parameter maps, not conventional liquefaction-hazard maps. They do not represent the actual soil conditions at a location. Instead, they provide a reference value for a generic soil profile, which engineers adjust for their site’s soil and stress conditions.

For the CPT procedures discussed, the reference soil layer is at 6 m depth. The specific reference profile was chosen to remain broadly consistent with earlier work, with some adjustments made during the research.

Simplified CPT Liquefaction-Triggering Procedure

The lecturer focuses on the Boulanger and Idriss (2014) triggering model. The method uses cyclic stress ratio (CSR) rather than the “required CPT resistance” parameter used in earlier SPT-focused work, partly because engineers are more familiar with CSR.

The core idea is that a map provides a reference CSR, which is adjusted to estimate the CSR for each layer in the site’s actual soil profile.

Application steps

1. Obtain the reference CSR from the map. The maps express CSR in percent—for example, a mapped value of 25 represents a CSR of 0.25.

2. Calculate layer-specific correction terms. These account for differences between the reference profile and the site, including:

   • Site amplification, based on the site class
   • Depth-reduction effects
   • Total and effective soil stresses
   • Magnitude scaling
   • Overburden effects

3. Use an appropriate earthquake magnitude estimate. For the depth-reduction calculation, the lecturer recommends using the mean magnitude from the deaggregation at the return period of interest, rather than the modal magnitude.

4. Apply the recommended magnitude-scaling factor. The researchers found that the 2014 Boulanger and Idriss magnitude-scaling factor produced problematic results for their simplified method. They therefore used the earlier 2012 version, which depends on earthquake magnitude rather than relative density.

5. Combine the corrections with the mapped value. Summing the correction terms produces an estimate of the median CSR for that soil layer.

6. Correct for a nonlinear bias related to PGA. The research found a bias correlated with peak ground acceleration (PGA). Separate corrections are used for low, moderate, and high PGA ranges; no correction is needed for PGA at or above 0.2 g.

7. Calculate liquefaction safety or probability. Use the estimated CSR with the conventional median cyclic resistance ratio (CRR) relationship to calculate a factor of safety or probability of liquefaction. The lecturer recommends a standard deviation of 0.506 for probability calculations, intended to account for both input-parameter uncertainty and model uncertainty.

The triggering method was compared with full performance-based analyses across multiple cities, soil profiles, and return periods. The simplified results generally tracked the full-analysis results closely.

Post-Liquefaction Settlement Procedure

For free-field volumetric strain and settlement, the method similarly maps the result for a generic reference layer, then corrects it for site-specific conditions.

Application steps

1. Obtain the reference volumetric strain from the map.
2. Calculate a site correction factor. The researchers found that a ratio based on a pseudo-probabilistic calculation could closely approximate the ratio between site-specific and reference strains.
3. Estimate volumetric strain using the approximate relationships discussed earlier in the series.
4. Adjust the strain for PGA-related bias. Different adjustments are used below 0.2 g and at or above 0.2 g.
5. Calculate settlement by layer. Multiply each calibrated strain by its layer thickness, sum the layer contributions, and apply the specified bias correction.

The settlement estimates generally approximated the full performance-based results, though with more scatter than the triggering estimates. Results were particularly reliable for predicted settlements below roughly 10 cm at lower PGA, and up to about 30 cm at higher PGA. Greater departures were possible beyond those ranges. The lecturer notes that the simplified results still showed less scatter than conventional pseudo-probabilistic methods.

Lateral-Spread Displacement Procedure

For lateral spreading, the mapped reference parameter is maximum shear strain rather than displacement. The method then adjusts that strain for site conditions before estimating displacement.

Application steps

1. Obtain the reference shear strain from the map.
2. Calculate a strain correction factor using the ratio of site-specific to reference shear strain, estimated with the described pseudo-probabilistic procedure.
3. Estimate site-specific maximum shear strain and apply the stated limits: zero for nonpositive values, the calculated value within the permitted range, and a cap of 51.2%.
4. Correct the strain for PGA-related bias.
5. Calculate the lateral displacement index (LDI) using the referenced method.
6. Estimate lateral-spread displacement using the appropriate relationship for either a ground-slope case or a free-face case.

The method showed good agreement with full performance-based analyses, especially for Boulanger and Idriss-based factors of safety. The approximation was strongest up to about 5 m of predicted displacement for that method. Results using Robertson and Ride showed more scatter, though average predictions remained useful.

Overall Conclusions

• Reference parameter maps and correction equations can provide practical approximations of full performance-based liquefaction analyses.
• The procedures covered in this lecture address CPT-based liquefaction triggering, post-liquefaction settlement, and lateral-spread displacement.
• The maps are not direct maps of actual liquefaction hazard: they represent generic reference conditions and must be corrected for local soil and site conditions.
• At the time described in the lecture, maps existed for states that funded the research. The lecturer said funding had since been secured to develop CPT maps for the entire United States using the 2014 USGS seismic source model, followed by SPT maps.
• The next and final lecture in the series is described as a walkthrough of the spreadsheet developed to apply the methods.

Speakers and Sources Featured

Speaker

• One unnamed lecturer, speaking on behalf of a Brigham Young University research group. No other speakers are heard in the subtitles.

Researchers, methods, or organizations cited

• USGS National Seismic Hazard Mapping Project (NSHMP)
• Cramer and Mayfield
• Mayfield, Cramer, and Huang
• Boulanger and Idriss (2012 and 2014)
• Cetin, Seed, and others (2004)
• Robertson and Ride
• Ku and others (2012)
• Kramer and others (2014)
• Ishihara and Yoshimine
• Zhang and others
• Juang and others (2013)
• Russell Green (Virginia Tech), mentioned in connection with published concerns about the 2014 magnitude-scaling factor
• Brigham Young University research group and the state transportation departments that sponsored the research