Updated presentation on improved analyses of earthquake-induced liquefaction and settlement using 1D nonlinear effective stress analyses

1. Improved Analyses
of Earthquake-Induced Liquefaction
and Settlement
By Robert Pyke Ph.D., G.E.
Robert Pyke, Consulting Engineer, Walnut Creek, CA, USA

2. This presentation is based on:
• Pyke, R., “Improved analyses of earthquake-induced liquefaction and settlement”,
Proc. 7th International Conference on Earthquake Geotechnical Engineering,
Rome, June 2019
• Pyke, R. and North, J., “Shortcomings of simplified analyses of earthquake-induced
liquefaction and settlement”, Proc. 7th International Conference on Earthquake
Geotechnical Engineering, Rome, June 2019
• Crawford, C., Tootle, J., Pyke, R. and Reimer, M., “Comparison of simplified and
more refined analyses of seismic settlements”, Proc. 7th International Conference
on Earthquake Geotechnical Engineering, Rome, June 2019

3. This version has been annotated to provide the continuity
that would be given orally in a live presentation.

4. Simplified Analyses
• Began with Seed and Idriss (1969) when running cyclic laboratory tests and
site response analyses were very specialized activities
• 50 years ago! When 3 people in the world could run site response analyses
and input motions were passed around as punched card decks!
• (Subsequently) obtain cyclic stress ratio causing liquefaction from
penetration resistance rather than from laboratory tests
• (Later still) obtain settlements as a function of factor of safety against
liquefaction – e.g. Pradel (1998), Zhang et al. (2002)

5. The next slide shows the key elements of the simplified procedure
graphically. These figures are taken from Boulanger and Idriss (2014).
Basically the procedure uses a simple formula to computed the induced
cyclic shear stresses from the ground surface peak acceleration and uses
penetration resistance and case histories to evaluate the cyclic shear
stresses causing liquefaction.

7. Issues with Simplified Analyses
• Duration varies with type of source and distance as well as with magnitude. Site
specific motions are much more accurate than averages of worldwide data.
• Ground surface acceleration and induced shear stresses are strongly impacted by
the development, redistribution and dissipation of excess pore pressures
• Penetration resistance is a poor indicator of soil behavior under cyclic loading
• Settlements following liquefaction obtained from Japanese data are fine for layers
that reach initial liquefaction but are wildly exaggerated for all other layers
• See Boulanger et al. (2016,2018), Pyke and North (2019), for more details

8. Particular Issues Regarding Thin Layers
• If only seen in a single boring or CPT they may be lenses rather than layers, and the
induced stresses in a lenses may be lower than in the surrounding material – see
Pyke (1995)
• Both the SPT and the CPT “have a nose”. They can detect softer layers even before
they get to them. See the following slide taken from Boulanger and DeJong
(2018). A sand layer underlain by a soft clay has to be in the order of one meter
thick in order that the true peak tip resistance be recorded. Similarly, an SPT
sampler needs to have several feet of the same material below the tip to record a
meaningful blowcount.

10. More Fundamental Issues
• See Pyke (1995, 2003, 2016) and Semple (2015)
• Be aware of both local geology and history of the occurrence of
liquefaction and settlement in the same tectonic environment
• Tom Holzer (USGS): “I must have studied 50 sites where
liquefaction has occurred, and in all but one case, the area that
exhibited liquefaction could be defined by the depositional
environment”

11. Beware of automation!
• Therefore, it is not only poor practice, but can be dangerous, to automate
the process of processing field data and conducting analyses without
human oversight and intervention!
• See my companion presentation on “Improved Analysis of Potential
Lateral Spreading Displacements in Earthquakes”, for an excellent
example of the need for human intervention and interpretation of a soil
profile that is taken from Youd (2018), or better yet, read the Youd paper!

12. First Case History:
This case history involves an elementary school site in Alameda, CA,
close to the Hayward fault, which can be seen in the next slide. The slide
after that shows a close up of the site which is situated on a made made
island with about 15 feet of hydraulically placed sand fill. The sand fill is
obviously susceptible to liquefaction, but all sand fills are not the same
and this one showed no signs of liquefaction and settlement in the 1989
Loma Prieta earthquake.

13. Hayward Fault

15. As part of a required seismic safety evaluation a local geotechnical
consultant evaluated the potential for liquefaction and seismic
settlement using CPT data and the program C-LIQ*. Neither this
application nor the subsequent slide, which is taken from a poor-quality
.pdf that is part of the public record, does justice to C-LIQ, which is a
nice program for looking at data but contains some methods of analysis
which are questionable and /or have to be used with great care. The
method of Zhang et al. (2002), which is included in C-LIQ, and the
alternate method of Pradel (1998) simply tend to be very conservative
for estimating earthquake-induced settlements.
* https://geologismiki.gr/

16. As already noted, the next slide is hard to read but it shows a typical CPT
sounding that was pushed at the direction of the geotechnical
consultant. No borings were made, or samples taken. However, it is
known from accumulated local experience that the profile basically has
five strata:
1. Sand fill
2. young Bay Mud, lightly OC clayey silts and silty clays
3. Merritt Sand, late Pleistocene wind-blown sands
4. Old Bay Clay, OC clayey silts and silty clays
5. Lower Alameda Formation, very dense sands and gravels

18. However, the previous slide shows liquefaction in four different strata
down to a depth of more than 60 feet and an estimated 10 inches of
settlement of the ground surface. In reality, only one of these materials,
the hydraulically placed sand fill, is susceptible to liquefaction and
seismic settlement. This is a big problem!

19. However, because the structural engineer advised that is was
uneconomical to retrofit the school buildings to accommodate 10 inches
of settlement, the school was closed, and the children are being bussed
to other schools, instead of walking to their neighborhood school.
Thus, over-conservatism led not to increased safety but to adverse
social consequences.

20. The subsurface profile is idealized in the following slide. The ranges of
normalized SPT blowcounts are interpreted from the CPT data and the
shear wave velocities are estimated by the author based on his
extensive local experience. On the basis of this data and local
knowledge, the sand fill might be susceptible to liquefaction and
settlement, but the other layers are not.

22. The profile on the previous slide is simplified in that it averages soil
properties over each depositional unit, but it captures the essential
depositional units and site response in earthquakes will be dominated
by the average properties. It will now be used in a site response analysis
using the new computer program TESS2. TESS2 uses the same explicit
finite difference scheme as the earlier program TESS, but it has been
rewritten to incorporate many new features including the ability to
simultaneously analyze the response of a soil column to two horizontal
input motions. Some further details regarding TESS2 follow presentation
of the results.

23. The following slide provides an indication of the amplitudes of the input
motions, which were fitted to the target spectrum, labelled Site Class C.
Five pairs of horizontal motions representing a magnitude 7+
earthquake on the Hayward fault were input at the top of the Lower
Alameda Formation at a depth of 100 feet.
It can be seen that the ground surface motions are significantly reduced
as a result of both damping in the mud layers and the development of
excess pore pressures and softening of the sand fill.

24. Input motions
Ground surface motions

25. And the next slide shows the printed output from TESS2 with three added
columns. The basic results are just for one horizontal component of one
motion, but the excess pore pressures and the settlements are the sum of
the contributions from each of the two input motion components, and
are necessarily the same for both components. The column headed
Rumax is the maximum excess pore pressure ratio, and the column
headed Rufinal is the excess pore pressure ratio at the end of the
specified input motion. Note that liquefaction only occurs in Layer 5.
Layer 4 also has very high excess pore pressures, but it does not quite
liquefy because of simultaneous dissipation of excess pore pressures
towards the water surface.

26. TESS2 Results

27. Column 3 of the previous slide shows the peak shear stresses generated
by this component of motion, and column 9 shows the peak shear
stresses that would have been computed using Boulanger and Idriss
(2014). The simplified method values are approximately twice the actual
values. This is one of the reasons that the simplified analysis of
liquefaction is conservative. Because the liquefaction analysis is
conservative, the evaluation of potential seismic settlements using the
method embodies in C-LIQ is very conservative. This run, plus the 4
other runs, suggest seismic settlements of the ground surface in the
order of an inch or two. Not ten inches! This, more correct, finding
would have made an enormous difference to the feasibility of
retrofitting the school buildings and avoided the need to bus the
children to other schools.

28. TESS2 - bi-directional, nonlinear, effective stress site
response analyses
• The same explicit finite difference solutions for response and
redistribution and dissipation of excess pore pressures as TESS
• Simple hyperbolic soil model – see Pyke (1979, 1993, 2004,2020)
• Excess pore pressures following Seed, Martin and Lysmer (1976)
• Settlements following Pyke (1973) and Seed, Pyke and Martin (1978)
• Runs two horizontal components simultaneously and adds excess
pore pressures and settlements or latent settlements

29. Ease of Use
• Users only have to specify shear wave velocity, soil type and undrained
shear strength or apparent relative density
• Selection and modification of site-specific input motions now much easier
thanks to PEER database etc.
• But the user still has to think about the depositional environment and how
to subdivide the profile into layers
• The effect of ageing on liquefaction potential is built in. The user has to
specify the age of the deposit or the program will not run. Thus the user is
forced to think about the problem, rather than the analysis procedure being
totally automated.

30. Settlements in non saturated sand layers and latent settlements in fully saturated
sand layers in TESS2 are computed using the methodology of Martin, Finn and Seed
(1975), Seed, Pyke and Martin (1978) and data from Pyke (1973), which is supplied
as a default. Users can substitute their own site-specific data if desired. The results
of a typical cyclic simple shear test are shown on the following slide. There are two
points worth noting:
1. The settlement per cycle clearly decreases with an increasing number of cycles;
2. The shear modulus increases with the number of cycles
[While this test was intended to be at a constant cyclic shear strain, because the sample becomes
stiffer under cyclic loading and there is compliance in the testing device, the cyclic strains are not
entirely constant and the results have to be reduced on a cycle by cycle basis. Also, similar effects of
the number of cycles were also seen in bi-directional shaking table tests so the decrease in
settlement per cycle and the increase in shear modulus is not the result of unidirectional shaking.]

32. A summary of the data obtained for freshly pluviated Monterey No. 0
sand is shown on the following slide. It may be seen that even for this
“baby sand”*, the settlement is unlikely to exceed about 0.5 % of the
layer height, even for loose sands and strong shaking. Thus, one may
conclude that for natural soils, the settlements, even for loose sands
and strong shaking, are unlikely to exceed about 0.5% of the layer height
unless liquefaction occurs.
* A washed and graded sand freshly deposited in the laboratory without over-consolidation, pre-
straining or ageing.

34. The key to the Martin and Pyke procedure for computing settlements
caused by irregular cyclic loadings is to know the settlement per cycle as
a function of cyclic shear strain and the accumulated settlement or
latent settlement to that point, as shown for one relative density on the
following slide.

35. Basis for calculation of settlements

36. In TESS2, if any layer reaches 100 percent excess pore pressure, the
latent settlements for fully saturated layers by default jump to the post-
liquefaction settlements suggested by Ishihara and Yoshimine (1992).
However, prior to that point, the settlements estimated by the Seed,
Pyke and Martin (1978) procedure and using the default data from Pyke
(1973) are generally lower than the Ishihara and Yoshimine values. The
user can of course over-ride the default data with site-specific values if
these have been obtained.

37. Ishihara and Yoshimine (1992):

38. Second Case History – Pyke and North (2019):
This case history involves a typical site in “Silicon Valley”, on the margins
of San Francisco Bay. The eight CPT soundings shown on the next slide
all showed some sand, but always at different depths, consistent with
late Pleistocene alluvial fan deposits. Thus the deposit is lensed, not
layered. There is no record of such materials ever liquefying, and if any
excess pore pressures develop in one of the sand lenses, that will cause
softening and the lens becomes a soft inclusion and carries less load.
See Pyke (1995).

40. Detailed results for CPT-2 are shown on the following two slides using
the methods for evaluating liquefaction based on Boulanger and Idriss
(2014) and for estimating seismic settlements based on Zhang et al.
(2002), as implemented in C-LIQ .
The results shown apply the standard option that is built into C-LIQ for
eliminating transitions and assume an “Ic cutoff” of 2.6, which is
standard boundary between “clayey” and “non-clayey” soils. Occasional
liquefaction is shown to a depth of 28 m and the accumulated seismic
settlements at the surface are 10 cm.

43. The next slide shows the results of a limited parametric study using
C-LIQ and CPT soundings 2 and 5. Result of the estimated seismic
settlements are shown for four cases.
Without detection and elimination of transitions and an Ic cutoff of 2.6
With elimination of transitions and an Ic cutoff of 2.6
With elimination of transitions and an Ic cutoff of 2.05, the limiting value for clean
sands, which is more applicable to the correlations used internally in C-LIQ
With an “ageing factor” of 1.5, which produces some odd results in this case

44. Settlements in cm computed using C-LIQ:
Assuming: CPT-2 CPT-5
No transitions
17 25
Default transitions
10 23
Ic cutoff = 2.05
4 1
Ageing factor = 1.5
4 2

45. The most correct answer for the estimated seismic settlement at this
site, even under strong ground motions, is likely zero. The above results
suggest that the simplified methods of analysis implemented in C-LIQ
can give reasonable results if the user makes good choices regarding the
input parameters, but that if used blindly, these methods can give very
conservative results.

46. The following slide shows a summary of the results obtained using bi-
directional, nonlinear, effective stress analyses using TESS2.
Conservatively assuming that the sand layers are horizontally
continuous, there is some excess pore pressure development in the
cleaner sand layers, the peak accelerations at the ground surface are
reduced from 0.55 g to approximately 0.3 or 0.4 g depending on the
values assumed for the hydraulic conductivities, and the estimated
settlements are reduced to 0.6 mm, close to the likely correct answer of
zero.
Again, this is still a conservative approximation, but in this case and in
many others, it is not worth doing the necessary site investigations and
a 2D or 3D analysis that models the lensed deposit in detail.

47. Settlements in cm for CPT-2 calculated using TESS2

48. Third Case History – River Islands
• On Stewart Tract within City of Lathrop, CA
• Mostly underlain by Pleistocene Sands
• But one patch, outlined on the following slide, was reworked about
3000 years ago and shows lower penetration resistances and implied
densities

50. Preliminary analyses, as shown on the next slide, using the methods of
Boulanger and Idriss (2014) and Zhang et al. (2002), as implemented in
C-LIQ, indicate liquefaction for the full depth of the reworked sands and
suggest settlements in the order of 1 foot or more. This would mean
that expensive ground improvement would be required for the planned
high-tech office and R&D use, making that use non-competitive with
alternate locations.

52. Thus, the developer elected to fund improved site investigations,
laboratory testing and analyses in order to ascertain whether site
improvement was really required. This involved pushing new CPTs with
shear wave velocity measurements, measurement of SPT blowcounts,
and use of a piston sampler in adjacent borings. Some of the key data is
shown on the following slide.

54. Samples were drained in the field and then carefully transported to the
laboratory. Stress-controlled, constant height, cyclic simple shear tests
were then conducted in order to determine the cyclic stress ratios
causing liquefaction using a new cyclic simple shear device designed by
Dr Michael Riemer of the University of California, Berkeley. With very
careful handling and practice it was possible to get relatively
undisturbed samples into the test device.

60. The results of the cyclic simple shear tests are summarized on
the following slide. These results justified the use of a cyclic
resistance ratio higher than that which would have been
obtained from the CPT or SPT data and back-calculation of case
histories.

62. Potential seismic sources and design response spectra are shown on the
following slide. It was decided to conservatively adopt the use of median
+ one standard deviation ground motion generated by a magnitude 7
earthquake on Segment 7 of the Great Valley fault for this evaluation.

64. Selected results from analyses using TESS2 are shown on the following
slides. Results are shown for just one of the five pairs of input motions
that were used and for two different values of the hydraulic
conductivity. Run No. 60202 used a hydraulic conductivity in the
reworked sands of 10-2 cm/sec and Run No. 60203 used 10-3 cm/sec.
Although it had been intended that site-specific data on seismic
settlement be obtained by running strain-controlled cyclic simple shear
tests, the schedule did not allow that, and so settlements were
estimated using the default data built into the program based on
extensive tests run on Monterey No. 0 sand by Pyke (1973). This is
believed to be conservative.

66. Note that in the previous slide, liquefaction occurs only in one layer of
the profile and not for the full 40 feet depth of the saturated reworked
sands. For the best estimate of CRR10 of 0.24 used in these runs, the
difference in the values of hydraulic conductivity made little difference
And note in the following slide that the peak induced shear stresses are
only about two-thirds of those obtained using
C-LIQ and Boulanger and Idriss (2014).

69. As in the previous case histories, the estimated seismic settlements are
much less than those obtained using C-LIQ for multiple reasons, but
most notably because the induced cyclic stresses are less than those
estimated using the simplified method and because less layers are found
to liquefy.

71. The results of limited parametric study are summarized on the following
slide. For the more conservative value of CRR10 and for a hydraulic
conductivity of 10-3 cm/sec, two layers in the model liquefy and the
estimated settlement of the ground surface is about 3 inches. The more
likely estimate of the maximum ground surface settlement is 2 inches.
Because redistribution and dissipation of excess pore pressures can be
significant, measurement or assumption of appropriate values for the
hydraulic conductivity is important. However, the default values that are
built into the program as a function of soil type will be adequate for
most routine analyses.

72. Relative Density 50 % 50 % 60 % 60 %
CRR10 0.20 0.24 0.20 0.24
K = 10-2cm/sec 1.8 1.8 1.7 1.7
K = 10-3cm/sec 3.4 1.8 2.8 1.7
Limited parametric study - settlement in inches

73. Conclusions
• Simplified analyses at best should only be used for screening analyses
• Depositional history and historic performance may actually be a better
screening tool
• If you have to show a calculation, or if there are real economic or
safety issues, do a decent calculation
• It’s not that hard to do a decent calculation these days (if you have the
right tools)

74. Supporting publications
• In addition to the references listed below, there were two excellent invited papers
presented at the recent 7th International Conference on Earthquake Geotechnical
Engineering held in Rome that illustrate how nonlinear effective stress analyses are
required to understand the development and dissipation of excess pore pressures
in real soil profiles. These are:
• Keynote Lecture 08, “The use of numerical analysis in the interpretation of
liquefaction case histories”, by Steve Kramer, and
• Keynote Lecture 09, “Key aspects in the engineering assessment of soil
liquefaction”, by Misko Cubrinovski

75. References 1
• Boulanger, R.W., and Idriss, I.M., “CPT and SPT Based Liquefaction Triggering Procedures”, Report No. UCD/CGM-14/01,
University of California, Davis, 2014
• Boulanger, R. W., and DeJong, J. T., “Inverse Filtering Procedure to Correct Cone Penetration Data for Thin-layer and
Transition Effects.” Proc., Cone Penetration Testing 2018, Hicks, Pisano, and Peuchen, eds., Delft University of Technology,
The Netherlands, 2018
• Boulanger, R.W., et al., “Evaluating Liquefaction and Lateral Spreading in Interbeddded Sand, Silt and Clay Deposits Using the
Cone Penetrometer”, Geotechnical and Geophysical Site Characterization 5, Australian Geomechanics Society, Sydney,
Australia, 2016
• Ishihara, K., and Yoshimine, M., “Evaluation of Settlements in Sand Deposits Following Liquefaction During Earthquakes”,
Soils and Foundations, Vol.32, No.1, pp.173-188, March 1992
• Martin, G.R., Seed, H.B., and Finn, W.D.L., “Fundamentals of Liquefaction under Cyclic Loading”, Journal of the Geotechnical
Engineering Division, ASCE, Vol.101, No. GT5, May 1975
• Lewis, M.R., et al., “Site Characterization Philosophy and Liquefaction Evaluation of Aged Sands”, Geotechnical Earthquake
Engineering and Soil Dynamics IV Congress, ASCE GSP 181, 2008
• Pradel, D., “Procedure to Evaluate Earthquake-Induced Settlements in Dry Sandy Soils”, Journal of the Geotechnical and
GeoEnvironmental Division, ASCE, Vo. 124, No. 4, 1998
• Pyke, R., "Settlement and Liquefaction of Sands Under Multi-Directional Loading," Ph.D. Thesis, University of California,
Berkeley, 1973
• Pyke, R., "Non-linear Soil Models for Irregular Cyclic Loadings," Journal of the Geotechnical Engineering Division, ASCE,
Volume 105, No. GT6, June 1979.

76. References 2
• Pyke, R., 2004, “Evolution of Soil Models Since the 1970s.”, Opinion Paper, International Workshop on Uncertainties in
Nonlinear Soil Properties and their Impact on Modeling Dynamic Soil Response, Sponsored by the National Science
Foundation and PEER Lifelines Program PEER Headquarters, UC Berkeley, March 18-19, 2004.
• Pyke, R., “Evaluating the Potential for Earthquake-Induced Liquefaction in Practice”, 6th International Conference on
Earthquake Geotechnical Engineering, Christchurch, New Zealand, November 2015.
• Pyke, R., et al., “Modeling of Dynamic Soil Properties”, Appendix 7.A, Guidelines for Determining Design Basis Ground
Motions, Report No. TR-102293, Electric Power Research Institute, November 1993
• Seed, H.B., and Idriss, I.M., “Simplified Procedure for Evaluating Soil Liquefaction Potential”, Journal of the Soil Mechanics
and Foundations Division, ASCE, Vol.97, No.9, September 1971
• Seed, H.B., Martin, P.P., and Lysmer, J., “Pore Pressure Changes During Soil Liquefaction”, Journal of the Soil Mechanics and
Foundations Division, ASCE, Vol. 102, No.GT4, April 1976
• Seed, H.B., Pyke, R., and Martin, G.R., "Effect of Multi-directional Shaking on Pore Pressure Development in Sands," Journal
of the Geotechnical Engineering Division, ASCE, Vol. 104, No. GT1, January 1978.
• Semple, R., “Problems with Liquefaction Criteria and Their Application in Australia”, Australian Geomechanics, Vol. 48, No. 3,
pp 15-48, September 2013
• Youd, T.L., “Application of MLR Procedure for Prediction of Liquefaction-Induced Lateral Spread Displacement”, Journal of
the Geotechnical and GeoEnvironmental Division, ASCE, Vo. 144, No. 6, 2018
• Zhang, G., et al., “Estimating Liquefaction-Induced Ground Settlements from CPT for Level Ground”, Canadian Geotechnical
Journal, Volume 39, pp.1168-1180, 2002

77. The End!
If you would like copies of the final papers as accepted by the 7ICEGE or have any
comments or questions, write to me at: bobpyke@attglobal.net