Appendix B - EQC Earthquake Commission

Transcription

FINDINGS FROM THE GROUND IMPROVEMENT PROGRAMME
Appendix B
Technical Papers
from the Science Trials
69
Appendix B
Technical Papers
from the Science Trials
• Development of Horizontal Soil Mixed Beams as a Shallow Ground
Improvement Method Beneath Existing Houses
R. Hunter, S. van Ballegooy, J.R. Leeves, T. Donnelly
• Horizontal Soil Mixed Beam Ground Improvement as a Liquefaction
Mitigation Method Beneath Existing Houses
M. Wansbone, S. van Ballegooy
• Rammed Aggregate Pier Ground Improvement as a Liquefaction
Mitigation Method in Sandy and Silty Soils
K.J. Wissmann, S. van Ballegooy, B.C. Metcalfe, J.N. Dismuke, C.K. Anderson
• Large-Scale Testing of Shallow Ground Improvements using
Controlled Staged-Loading with T-Rex
S. van Ballegooy, J.N. Roberts, K.H. Stokoe, B.R. Cox, F.J. Wentz, S. Hwang
• Large Scale Testing of Shallow Ground Improvements using
Blast-Induced Liquefaction
F.J. Wentz, S. van Ballegooy, K.M. Rollins, S.A. Ashford, M.J. Olsen
• Utilizing direct-push crosshole testing to assess the effectiveness
of soil stiffening caused by installation of stone columns and
Rammed Aggregate Piers
L.M. Wotherspoon, B.R. Cox, K.R. Stokoe II, D.J. Ashfield, R.A. Phillips
70
Development of Horizontal Soil
Mixed Beams as a Shallow Ground
Improvement Method Beneath
Existing Houses
R. Hunter, S. van Ballegooy, J. R. Leeves and T. Donnelly
71
Development of Horizontal Soil Mixed Beams as a Shallow Ground
Improvement Method Beneath Existing Houses
R. Hunter1, S. van Ballegooy2, J. R. Leeves2 and T. Donnelly3
1Tonkin
& Taylor Ltd., Environmental & Engineering Consultants, 33 Parkhouse Rd, Wigram, Christchurch; PH
+649 355 6000; FAX +64 9 307 0265.
2Tonkin & Taylor Ltd., Environmental & Engineering Consultants, 105 Carlton Gore Rd, Newmarket, Auckland 1023;
PH +649 355 6000; FAX +64 9 307 0265.
3Contract Landscapes Ltd, 14 Wookey Lane, Kumeu, Auckland; PH +649 412 7048.
ABSTRACT
Following the 2010-2011 Canterbury Earthquake Sequence (CES), the vulnerability of residential
houses in some areas of Christchurch to liquefaction-induced damage was realised. As a result of the
ground surface subsidence caused by the CES, the liquefaction vulnerability has also increased in some
parts of Christchurch (Russell et al. 2015). The liquefaction-induced damage resulted in a large number
of residential houses in Christchurch that were uneconomic to repair. They are being demolished and
rebuilt on stiffer and stronger foundation systems and in some areas which are particularly vulnerable
to liquefaction, the stiffer and stronger foundation systems are being used in conjunction with shallow
ground improvements. There are also a large number of houses that have liquefaction-induced damage,
but are economic to repair. Until recently there was no practical ground improvement solution that could
be economically constructed beneath existing repairable residential houses to decrease their
liquefaction vulnerability. However, during a shallow ground improvement trial research project,
commissioned by the New Zealand Earthquake Commission (EQC) in 2013, a method was developed
to improve ground beneath residential houses, known as Horizontal Soil Mixing (HSM). HSM involves
the mechanical mixing of injected grout into in-situ soils using a modified directional drill and a
specifically designed soil mixing tool to construct a series of HSM beams to improve the thickness and
stiffness of the non-liquefying crust and decrease the vulnerability of the existing house to future
liquefaction-induced damage. This paper describes the development of the HSM construction
methodology, including constraints and issues that were encountered and overcome.
Keywords: Earthquakes, Liquefaction, Ground Improvement, Horizontal Soil Mixed Beams, HSM
1
INTRODUCTION
The Canterbury region of New Zealand (NZ) has been affected by a series of earthquakes and
aftershocks with the four most significant earthquakes occurring on 4 September 2010 (Mw 7.1), 22
February 2011 (Mw 6.2), 13 June 2011 (Mw 5.6 and 6.0 separated by 80 minutes), and 23 December
2011 (Mw 5.8 and 5.9 separated by 80 minutes). The earthquake shaking from these events triggered
localised-to-widespread minor-to-severe liquefaction in Canterbury. The damage caused by the
liquefaction was severe in several Christchurch suburbs. Liquefaction ejecta, liquefaction-induced
differential settlement, and lateral spreading were the principal ground deformation modes that damaged
residential dwellings in the Canterbury region. The liquefaction-induced damage is well documented in
Cubrinovski and Green (2010), Cubrinovski et al. (2011), Wotherspoon et al. (2011), Green et al. (2012),
Tonkin & Taylor (2013), van Ballegooy et al. (2014b) and van Ballegooy et al. (2015), among others.
Between 6,000 and 10,000 residential houses on the flat land, in the areas that are being repaired and
rebuilt, have been assessed by the private insurers as uneconomic to repair due to the liquefaction
related damage and are likely to be rebuilt. In the aftermath of the CES, greater consideration was given
to the importance of supporting houses on robust, stiffened foundations capable of resisting the
damaging effects of liquefaction (i.e., angular distortion, lateral stretch, loss of ground support). In some
areas which are particularly vulnerable to liquefaction damage the stiffer and stronger foundation
systems are being used in conjunction with shallow ground improvements (MBIE, 2012).
As a result of the CES the land has subsided due to tectonic subsidence, liquefaction-induced volumetric
densification, ejection of liquefied material and lateral spreading. However, the groundwater elevations
across Christchurch have generally not changed (van Ballegooy et al. 2014a). Hence, in some areas
the groundwater levels are now closer to the ground surface and for some areas of Christchurch, the
shallower groundwater surface results in increased liquefaction vulnerability in future moderate to strong
earthquakes (Russell, et al. 2015). EQC has determined that the Increased Liquefaction Vulnerability
(ILV), as a result of ground subsidence caused by the CES, at 100 year return period levels of
earthquake shaking, is a form of land damage covered by its insurance. Approximately 5,000 properties
in the areas that are being repaired and rebuilt, are likely to qualify for the ILV land damage
compensation. Settlement of the insurance liabilities for the ILV land damage on each property is based
on the cost to repair the land damage on an individual property basis, or damage based valuation
methodology, whichever is least, up to the capped maximum liability provided for by the 1993 EQC Act.
Therefore, EQC funded two work streams of ground improvement trials; (1) to evaluate the efficacy
(science) of shallow ground improvement methods; and (2) to determine the cost of the shallow ground
improvement methods by undertaking full scale construction trials on residential properties. The aim the
trials was to investigate and determine the efficacy, construction practicality and cost of various shallow
ground improvement methods which are effective in reducing liquefaction vulnerability and which can
be also be used for repairing ILV land damage. This is both for “cleared land” cases where the damaged
houses are uneconomic to repair and hence will be rebuilt, as well as “repair cases” where the damaged
houses are economic to repair and hence will not be rebuilt.
While there were a number of cleared land shallow ground improvement methods available for testing,
there were limited options available for improving the soils underneath existing and repairable houses
without requiring either the removal or demolition of a house which is otherwise considered economically
repairable. Permeation grouting using cement grout was initially trialled, which then lead to the
development of a new and novel approach to shallow ground improvement, known as Horizontal Soil
Mixing (HSM). The method involves the mechanical mixing of injected grout into in-situ soils using a
modified directional drill and specifically designed soil mixing tool to construct a series of HSM beams
to improve the thickness and stiffness of the non-liquefying crust beneath an existing house, decreasing
the vulnerability to future liquefaction-induced damage (Ishihara, 1985).
Extensive field testing was undertaken on the HSM beam ground improvement method to prove the
efficacy, including CPT, crosshole Vs/Vp geophysical testing, truck mounted vibroseis (T-Rex) shake
testing and blast-induced liquefaction testing, supplemented by numerical modelling. Following the
ground improvement trials, a ground improvement pilot project was undertaken to construct full scale
shallow ground improvements on a number of residential sites across Christchurch including three repair
case properties where HSM beams were constructed beneath three existing houses. The development,
testing, observations, results, implementation and areas of potential improvement for the HSM beam
method are discussed in this paper.
2
DEVELOPMENT OF THE HSM GROUND IMPROVEMENT METHOD
The concept of the HSM beam ground improvement method was developed by Contract Landscapes
Limited (CLL). The concept is relatively straightforward; using a directional drill, drill a horizontal pilot
hole underneath an existing house (refer to step 1 in Figure 1), dig a trench on the opposite side of the
house, attach a soil mixing tool to the end of the drill string (refer to step 2 in Figure 1) and pull the soil
mixing tool back underneath the house while injecting cement grout to form a non-liquefying horizontal
soil-cement ‘beam’ (refer to step 3 in Figure 1). This process can then be repeated to construct a number
of horizontal beams in an arrangement that sufficiently stiffens and thickens the non-liquefying crust
(refer to step 4 in Figure 1), increasing the potential for the house to suffer less liquefaction induced
damage and be more readily repairable.
Developing the HSM ground improvement methodology involved the refinement of a number of
elements, such as:
•
•
•
•
•
•
•
The tracking of the drilling head in the pilot hole beneath a house;
The development of a suitable soil mixing tool;
Soil mixing methodology including the rate of drill rotation and linear pull back speed (and how this
needs to vary in different soil conditions);
A suitable grout mixture including the required additives;
Cement dosage rates to achieve the required target strength of the HSM beams and a process for
monitoring the volume of injected grout while constructing the HSM beams;
Configuration and vertical positioning in the ground of the HSM beams; and;
The order in which to best construct the beams.
While the majority of these elements were determined during the initial stages of the ground
improvement trials in 2013, refinements continued during the ground improvement pilot project
undertaken in 2014. A brief summary of the development of these elements is discussed below. Further
refinements to optimise the process are discussed in Section 4.
Figure 1: Step by step process of constructing HSM beams beneath a residential house
2.1
Directional Drilling and Locating and Tracking the Drill Head
During the construction of the first four test panels in the ground improvement trials, a 65 hp (48 kW)
Vermeer D16x20A directional drill was used. Based on initial results it was evident that this directional
drill was underpowered and accordingly an 85-hp (63 kW) Ditch Witch JT2020 mach 1 was used for the
construction of remaining test panels and HSM Beams under the houses. This rig performed adequately
in the ground improvement trial and pilot programme, however additional power would be beneficial in
denser or variable ground where soil mixing becomes more difficult. Experience showed that, in general,
the denser sandier soils and stiffer silt soils result in more strain on the directional drill rig and engine,
compared to the looser sandier soils and softer siltier soils.
Working in the residential setting provided challenges in terms of the limited amount of space available.
As a result of these limitations, high performance drill rods with high spec joints and threads were utilised
to enable steep entry and exit angles from the ground. These rods were capable of bending though 33
degrees per 3 m length of rod and are also capable of tolerating the increased torque that was applied
as a result of pulling the mixing tool through the soil, which generally does not occur during conventional
directional drilling.
Initially a DigiTrak F2 System was used for tracking the drill head position beneath the ground surface
during the directional drilling phase of the horizontal pilot hole (refer to step 1 in Figure 1). This system
of operation required two people, one to operate the drill rig and another to use a locating device at the
ground surface to locate and track the drill head as it progressed, while communicating with the drill rig
operator through a portable radio headset so that he could steer the drill head. While this system was
suitable for the construction of test panels (discussed later), it was not practical for locating drilling
underneath the houses where the locating positions was limited to either end of the house. For this
reason, a F5 DigiTrak System was adopted, which enabled the drill operator to locate and steer the drill
head independently by positioning the locator box on the ground at the target location, which would send
signals of the drill head position to a screen mounted on the drill rig. This system, along with a highly
experienced dill operator, enabled drilling to occur within the ± 50 mm tolerances required. This
tolerance was verified by exhuming the HSM beams in the ground improvement trials and confirming
location by survey. Experience has shown that a competent drill rig operator is key factor in ensuring
correctly positioned HSM beams are constructed. When less experienced operators were used the
location of HSM beams was more variable and sometimes lay outside the required tolerances.
2.2
Grout Mixtures and Target Cement Dose Rates
Experimentation of grout mixtures was undertaken in a field laboratory to determine appropriate grout
mixtures. The materials and quantities that were used varied. They included variations in the use of
ordinary Portland cement (OPC), flyash, plasticisers and the volume of water. For each mixture, the
specific gravity, bleed and viscosity was measured.
The grout mixer and pump unit used during the ground improvement trials and the pilot programme was
an STA 5M3 + T12. Batching grout began with the addition of the required ingredients into an agitated
mixing bowl, before being sucked through a turbo mixing unit and into an agitated storage tank. From
here the grout was pumped through grout lines and the drill string (drill rods) exiting from orifice ports
on the mixing tool. The pump was capable of pumping 60 L/min at a maximum of 45 bar. For the ground
improvement trials and pilot project, typical grout flow rates ranged between 30-40 L/min, while
maintaining a pressure ranging between 2-10 bar at the pump.
During the initial development phase of the HSM beams construction methodology, grout with a water
cement ratio of 2:1 was used. An initial HSM beam with varying cement content was constructed,
ranging between 5 to 15% cement (by estimated dry density of the in-situ soil). During this process
‘hydraulic lock’ was experienced for the portion of the beam constructed with 15% cement content. This
is where the volume of grout being injected was sufficiently large to increase the pressure in the HSM
beam annulus to a point where the drill was unable to turn the mixing tool any further. On this occasion,
an excavation pit was required to release the pressure to free the mixing tool. The importance of injecting
small volume of grout became apparent to prevent hydraulic lock from occurring. This process also
significantly reduced any potential ground heave. Further laboratory batching led to the development of
a grout mixture with a 0.35:1 water cement ratio including the addition of 3% plasticiser by volume, which
would enable smaller volumes of grout to be injected. This grout mixture has proven to be satisfactory
and typically requires between 30-40 L of grout to be injected per 1 m length of HSM beam to give a
cement dosage of approximately 15% by dry density of the in-situ soil.
The target dose rate of 15% cement (by dry density of the in-situ soil) was more than sufficient to achieve
a minimum target strength of 1 MPa (equivalent to weak rock), based on unconfined compressive
strength testing (discussed in Section 3). While a lower dose rate of 10% cement (by dry density of the
in-situ soil) could be used to achieve the minimum 1 MPa target strength, the target dose rate was set
higher than required because grout injection rates and drill pull back speeds were manually controlled,
resulting in a varying cement dosage rate along the length of each HSM beam between ± 5%.
Improvements to reduce the variability of the cement dosage rate along the length of the HSM beams
are further discussed in Section 4.
2.3
Soil-Cement Mixing
The soil mixing tool that was used to construct the HSM beams in both the ground improvement trial
and pilot projects, consisted of six mixing blades at three graduated intervals with a 0.5 m diameter.
Initially the mixing tool had four orifice ports for the grout to exit from, however this was later reduced to
two. The reduction from four to two orifice ports increased the pressure and velocity in which the grout
exited and improved the consistency of grout being mixed through the soil.
Creating well mixed soil-cement beams is important to ensure that the HSM beams have adequate
strength. As part of the ground improvement trial test programme a number of HSM beams were
exposed using a 5 tonne excavator, supplemented by hand digging between the beams. Figure 2 shows
the photo of some exposed HSM beams which were constructed beneath one of the houses. In the
initial stages of the development of the HSM beam construction methodology, the exposing of the HSM
beams allowed them to be dissected to reveal if the grout was mixing suitably through the entire diameter
of the beam. These observations enabled suitable linear pull back speeds as well as rotational mixing
tool speeds to be optimised for different soil types. Typical rotation speeds varied between 150 and 250
rpm with pull back speeds ranging from 0.5 to 1 m per minute, depending if the soil is silty or sandy, soft
or stiff, or loose or dense.
The quality assurance procedure used in the ground improvement pilot for the three houses was to first
construct one of the HSM beams outside of the house footprint, then excavate down to the beams to
undertake an inspection of mixing quality and recover samples (for UCS laboratory testing) to asses if
a suitable standard of soil mixing was achieved. If the HSM beams were found to have an unsuitable
standard of mixing, changes were made to the rotation speeds and/or the linear pullback speeds until a
suitable level of soil mixing was achieved at each site before the HSM beams were then constructed
beneath the house footprint.
Figure 2: Exposed HSM beams which were constructed beneath one of the houses at the ground
improvement trial site. The HSM beams were exposed after the blast-induced liquefaction testing was
completed
Experience showed that a more consistent mixing of grout into the soil through the full diameter of the
HSM beams occurred in sandier soils, compared to silty soils. When soil mixing in siltier soils a slower
pull back speed was used to ensure consistent mixing. Pull back speeds through sandy soils were
typically 1 m per minute at 150 rpm, while in siltier soils this was slowed to approximately 0.5 to 0.75
meters per minute at 250 rpm.
2.4
HSM Beam Layout, Configuration and Vertical Positioning in the Ground
As part of the ground improvement trials, the effectiveness of both single rows of HSM beams and
double rows of HSM beams was examined through extensive in-situ testing and supplemented with
numerical modelling (refer to Section 3). The single row configuration did not significantly reduce the
liquefaction vulnerability because of the relatively small increase in the non-liquefying crust thickness,
whereas the double row of HSM beams was effective in reducing liquefaction vulnerability by thickening
and stiffening the non-liquefying crust. Figure 3 shows a long-section and cross-section of the typical
layout, configuration and vertical positioning in the ground of the HSM beams that were constructed in
the ground improvement pilot project.
The depth that HSM beams were positioned in the ground was dependent on the depth of the soils
which were assessed to liquefy at the ultimate limit state levels of earthquake shaking (defined in the
MBIE, 2012 guidelines). This depth typically coincides with the depth to the groundwater surface in the
areas in Christchurch that were severely affected by moderate to severe liquefaction-induced damage
from the CES. It is noted that the depth to the groundwater fluctuates throughout the year (van Ballegooy
et al, 2014a). For the ground improvement trials and pilot project, the centre of the top row of HSM
beams were positioned at the 85 percentile water table elevation, which is typically between 1-2 m below
the ground surface.
Figure 3: Long-section and cross-section showing the typical layout, configuration and vertical
positioning in the ground of the HSM beams beneath an existing house
2.5
Ground Surface Heave
Ground surface heave directly above the areas where the HSM beam construction was undertaken was
not visually observed and neither was it measured using a laser level accurate to ± 5 mm on bare earth
during the ground improvement trials. However, during the ground improvement pilot project, cracking
of brittle concrete surfacing occurred at the first property while constructing the HSM beams. Prior to
commencing HSM construction works on each property, eight surveying nails were installed into the
perimeter concrete ring beams of the existing house. These were surveyed to ± 0.4 mm accuracy prior
to commencing and also at the end of the HSM construction works, as well as when mixing tool passed
beneath each survey location. Results show that the perimeter concrete ring beams typically heaved by
2-5 mm while the soil mixing tool passed beneath the survey locations. Post construction surveying
showed that the perimeter concrete ring beams settled back down to their original elevations ± 1 mm.
As a result of this heave, hairline cracks to internal wall lining and ceilings also occurred.
3
HSM BEAM GROUND IMPROVEMENT EFFICACY TESTING
The ground improvement trial testing programme for the HSM beam ground improvement method
included the construction of two 7 m by 7 m test panels at three different sites in residential areas that
are not being repaired or rebuilt in Christchurch. Each site had different soil conditions. At each site a
test panel with a single row of HSM beams and another panel with a double row of HSM beams was
constructed. In addition, a single layer and a double layer of beams were constructed under two
residential houses that will not be repaired or rebuilt. The testing included crosshole geophysical testing,
Scala Penetrometer testing, exhuming, sampling and unconfined compressive strength (UCS) testing
as well as truck mounted vibroseis (T-Rex) testing and blast-induced liquefaction testing.
Soil-cement samples from HSM beams were collected from the different test sites for UCS testing. The
results showed that 1-5 MPa strengths were achieved after 28 days for the beams with 15% cement (by
in-situ dry soil density) when the soil was suitably mixed. It is noted that laboratory batched soil-cement
samples, also of 15% cement by in-situ dry soil density, achieved UCS strengths ranging between 3-6
MPa after 28 days, showing that the in-situ mixing is more variable. Scala Penetrometer testing on the
HSM beams commonly resulted in penetration resistances greater than 20 blows per 100 mm
penetration after 7 days, and refusal after 28 days. The Scala Penetration testing was found to be a
useful cost effective verification test method to confirm that the HSM beams were constructed in the
correct locations and were also a useful proxy to confirm that the HSM beams were achieving their
minimum target strength of 1 MPa.
The T-Rex shaker was used to examine the dynamic effectiveness of the HSM beams to evaluate
whether they provide sufficient stiffness to reduce the cyclic shear strains, caused by earthquake
shaking, in the ground between the HSM beams. By reducing the cyclic shear strains, the potential for
development of excess pore water pressure and liquefaction is reduced, thereby increasing the
thickness of the non-liquefying crust. In addition, blast-induced liquefaction testing was undertaken on
the HSM beam ground improvement test panels and houses as a practical means of undertaking largescale trial evaluation of their performance in the absence of an actual earthquake. While it is recognised
that blast-induced liquefaction is somewhat different mechanistically from liquefaction induced by
earthquake ground shaking, the behaviour of the liquefied soil is believed to be sufficiently similar to
allow a suitable assessment of the performance of the HSM beam ground improvement. The results
from the T-Rex shake testing and blast-induced liquefaction testing as well as the supplementary
numerical modelling will be published in the near future. The preliminary conclusions are:
•
•
•
•
•
4
The double row of HSM beams limited development of cyclic shear strain in the soil between the
beams at low to moderate levels of shaking (PGA = 0.1-0.3g) based on the results of the T-Rex
shake testing as well as 2-D dynamic numerical modelling. Conversely, the single row of beams
did not appear limit the development of cyclic shear strain in the soil between the beams at low to
moderate levels of shaking. At strong levels of shaking (PGA >0.3g), the cyclic shear strain between
the beams became larger and excess pore water pressures developed, causing the soil in between
the beams to liquefy.
The results from the blast-induced liquefaction testing indicate that the house with the double row
of HSM beams beneath it had less differential settlement / flexural distortion than the house with a
single row of beams as well as the two adjacent houses on natural (unimproved) ground.
The double row of HSM beams appeared to have prevented ejecta from coming up within the beam
footprint. The mechanism for this is not yet been ascertained but may be due to the stiffening of
the non-liquefying crust and also the disruption of pre-formed ejecta pathways through existing
defects in the upper non-liquefying soil layers. This disruption which may have occurred during the
soil mixing stage of the HSM beam construction.
By extending the HSM beams beyond the perimeter of the house footprint, the formation of sand
ejecta pathways was forced further away from the house, reducing the potential for ground loss
beneath the house, resulting in a reduction in potential for differential settlement over the building
footprint.
Based on the results of 3-D numerical modelling, the amount that the HSM beams reduce the
magnitude of differential ground surface settlement is dependent upon the orientation of the beams
relative to the soil conditions which are likely to cause the differential settlement and flexural
distortion. Compared to a 1.6 m thick compacted gravel raft, the 3-D numerical modelling indicates
that the HSM beams are typically 50-60% as effective at reducing the relative magnitude of
differential settlement and flexural distortion. For particularly favourable beam orientations, where
differential ground surface settlement occurs along the long axis of the beams, performance may
be better than the compacted gravel raft, whereas for particularly unfavourable beam orientations,
where differential ground surface settlement occurs perpendicular to the long axis of the beams,
performance is likely to be as low as 10% of the 1.6 m thick compacted gravel raft.
FUTURE POTENTIAL IMPROVEMENTS TO THE HSM CONSTRUCTION METHODOLOGY
At the time of writing this paper, a trial with a new retractable soil mixing tool is being undertaken. If
successful in its intended functionality, the soil mixing blades can be opened up underground and this
will eliminate the need for the receiving trench to attach the soil mixing tool to the drill string (shown in
Figure 1 and 3). Currently the flow of grout is manually controlled and requires practised communication
and control between the drill rig operator, grout pump operator and quality assurance technician
recording data. Further investment and development of a computer controlled system could allow for
the grout to be injected automatically depending on the linear pull back rate and rotation of the soil
mixing tool. This system would allow for a higher level of quality assurance, reduce the required labour
and may also allow the cement dosage rates to be reduced. In addition, technical improvements that
may provide better performance of the HSM beams could include; reinforcement of the beams, tying the
ends of the beams together and constructing the bottom and top rows of beams at right angles to one
another. Some of these improvements may also provide lateral stretch resistance of the building footprint
in areas with lateral spreading potential.
5
CONCLUSION AND APPLICATION
HSM beams as a shallow ground improvement method under existing houses is a novel approach that
has been developed and shown to improve the performance of the ground by improving the thickness
and stiffness of the non-liquefying crust, thereby decreasing the vulnerability of the existing houses to
future liquefaction-induced damage. The HSM beams have been successfully constructed beneath
three repairable houses as part of a ground improvement pilot project. The pilot project has proven that
this method can be successfully implemented as a shallow ground improvement method provided there
is sufficient working area around the house and that the house foundations do not extend down in the
zone of the HSM beams. After the HSM beams are constructed some reinstatement works are required
when using the current methodology. There are various improvements to the rig and methodology that
could occur to improve the productivity and effectiveness of the HSM beams, and to reduce the amount
of reinstatement works at the site. It is also noted that although the purpose of developing the HSM
beam ground improvement method was to reduce the vulnerability of existing houses to liquefactioninduced damage from future earthquakes, HSM beams also have potential for the remediation of
commercial sites and for geotechnical purposes other than liquefaction mitigation, for example; house
re-levelling, settlement control and temporary retention.
6
ACKNOWLEDGEMENTS
This work was funded by the New Zealand Earthquake Commission. This project would not have
occurred without the innovation, flexibility and hard work that was undertaken by Contract Landscape
Limited. The authors also acknowledge and recognise the significant contribution by many people
involved in this project, in particular Miles Stretton from Stretton Consulting Limited and Hugh Cowan
and Kaya Yamabe from the Earthquake Commission.
7
REFERENCES
Cubrinovski, M. and Green, R.A. (eds.). (2010). Geotechnical Reconnaissance of the 2010 Darfield (Canterbury) Earthquake,
(contributing authors in alphabetical order: J. Allen, S. Ashford, E. Bowman, B. Bradley, B. Cox, M. Cubrinovski, R. Green,
T. Hutchinson, E. Kavazanjian, R. Orense, M. Pender, M. Quigley, & L. Wotherspoon), Bulletin of the New Zealand Society
for Earthquake Engineering, 43(4), p. 243-320.
Cubrinovski, M., Bradley, B., Wotherspoon, L., Green, R., Bray, J., Wood, C., Pender, M., Allen, J., Bradshaw, A., Rix, G., Taylor,
M., Robinson, K., Henderson, D., Giorgini, S., Ma, K., Winkley, A., Zupan, J., O’Rourke, T. DePascale, G., and Wells, D.,
(2011) “Geotechnical aspects of the 22 February 2011 Christchurch earthquake”, Bulletin of New Zealand Society of
Earthquake Engineering 44, 205-226.
Green, R.A., Cubrinovski, M., Wotherspoon, L., Allen, J., Bradley, B., Bradshaw, A., Bray, J., DePascale, G., Orense, R., O’Rourke,
T., Pender, M., Rix, G., Wells, D., Wood, C., Henderson, D., Hogan, L., Kailey, P., Robinson, K., Taylor, M., and Winkley,
A. (2012). Geotechnical Aspects of the MW6.2 2011 Christchurch, New Zealand Earthquake, State of the Art and Practice
in Geotechnical Engineering (R. Hryciw, A. Athanasopoulos-Zekkos, and N. Yesiller, eds.), ASCE Geotechnical Special
Publication (GSP) 225, p. 1700-1709.
Ishihara, K. (1985). “Stability of Natural Deposits during Earthquakes” Proceedings of the 11th International Conference on
Soil Mechanics and Foundation Engineering, San Francisco, 1:321-376.
Ministry of Business, Innovation and Employment (MBIE), 2012. “Revised issue of Repairing and Rebuilding Houses affected
by the Canterbury Earthquakes.” December 2012, available at http://www.dbh.govt.nz/guidance-on-repairs-afterearthquake
Russell, J., van Ballegooy, S., Lacrosse, V., Jacka, M. and Rogers, N. (2015). “The Effect of Subsidence on Liquefaction
Vulnerability Following the 2010 – 2011 Canterbury Earthquake Sequence.” In Proceedings. 12th Australia New Zealand
Conference on Geomechanics.
Tonkin & Taylor Ltd., (2013). “Liquefaction Vulnerability Study.” Report to Earthquake Commission, Tonkin & Taylor ref.
52020.0200/v1.0,
prepared
by
S.
van
Ballegooy
and
P.
Malan,
available
at
https://canterburygeotechnicaldatabase.projectorbit.com.
Van Ballegooy S., Cox S. C., Thurlow C., Rutter H. K., Reynolds T., Harrington G., Smith. T., (2014a). “Median water elevation
in Christchurch and surrounding area after the 4 September 2010 Darfield Earthquake. Version 2.” GNS Science Report
2014/18.
Van Ballegooy, S., Malan, P., Lacrosse, V., Jacka, M.E., Cubrinovski, M., Bray, J.D., O’Rourke, T., Crawford, S., Cowan, H.,
(2014b) “Assessment of Liquefaction-Induced Land Damage for Residential Christchurch.” Earthquake Spectra, EERI,
30(1), 31 – 55.
Van Ballegooy, S., Boulanger, R. W., Wentz, R., (2015) “Evaluation of a CPT-based Liquefaction Pocedure at Regional Scale.”
Soil Dynamics and Earthquake Engineering, Special Issue: Liquefaction in New Zealand and Japan. in review
Wotherspoon, L., Bradshaw, A., Green, R.A., Wood, C., Palermo, A., and Cubrinovski, M. (2011). Bridge Performance during
the 2011 Christchurch Earthquake, Seismological Research Letters, SSA, 82(6), p. 950-964.
72
Horizontal Soil Mixed Beam
Ground Improvement as a
Liquefaction Mitigation Method
Beneath Existing Houses
M. Wansbone, S. van Ballegooy
73
6th International Conference on Earthquake Geotechnical Engineering
1-4 November 2015
Christchurch, New Zealand
Horizontal Soil Mixed Beam Ground Improvement as a Liquefaction
Mitigation Method Beneath Existing Houses
M. Wansbone1, S. van Ballegooy2
ABSTRACT
Ishihara (1985) recognised that a thick non-liquefying crust overlying liquefying soils would reduce
the consequences of liquefaction (i.e., sand boils, loss of bearing capacity and differential
settlement). In Christchurch, in the aftermath of the 2010-2011 Canterbury Earthquake Sequence
(CES), detailed engineering assessments of nearly 60,000 single-family houses combined with a
comprehensive regional scale geotechnical investigation, clearly showed that less structural damage
occurred in liquefaction-prone areas containing an intact, relatively stiff non-liquefying crust with
minimum thickness of approximately 3 m. To increase the resilience of the post-CES existing
Christchurch residential housing portfolio (which has been repaired) to future liquefaction damage,
the use of shallow (i.e., 4 m deep) ground improvements to construct stiff, non-liquefying crusts
to mitigate the consequences of liquefaction of the underlying soil layers was evaluated. This paper
presents the results from the in-situ vibroseis dynamic (T-Rex) load testing and dynamic numerical
simulation to examine the liquefaction triggering of Horizontal Soil Mixed (HSM) beam ground
improved soils compared to natural soils. The shake testing of the HSM beam ground improvement
panels as well as the numerical simulations demonstrated that HSM ground improvement resulted
in reduction in the maximum cyclic shear strain ( ) and excess pore water pressure (ru) induced in
the improved soil and hence provide an improvement in liquefaction resistance.
Introduction
The 2010 – 2011 Canterbury Earthquake Sequence (CES) caused widespread liquefaction-related
land and building damage (described in Rogers et al., 2015), affecting 51,000 residential properties
in Christchurch, including 15,000 residential houses damaged beyond economical repair. In
addition, as a result of the ground surface subsidence caused by the CES, the liquefaction
vulnerability has increased in some parts of Christchurch (Russell et al. 2015), increasing the
vulnerability of the repairable houses to liquefaction-related damage in future earthquake events.
In the suburbs most vulnerable to liquefaction damage, the CES revealed the importance of
constructing robust, stiffened foundations capable of resisting the damaging effects of liquefaction
(i.e. angular distortion, lateral stretch and loss of ground support) or instead undertaking ground
improvement to mitigate the damage caused by liquefaction. In response, the New Zealand
Earthquake Commission (EQC) funded an extensive shallow ground improvement trial program
to evaluate the efficacy of various shallow ground improvement methods for reducing the
liquefaction vulnerability of the residential housing portfolio.
There are a number of shallow ground improvement methods suitable for installation on cleared
residential properties (i.e., the houses that were damaged by the CES beyond economic repair and
will be rebuilt). However, options for improving sites where there are existing buildings which are
economic to repair (which do not involve removing or demolishing the building) are limited. A
1
Senior Geotechnical Engineer, Tonkin & Taylor Ltd, Auckland, New Zealand, [email protected]
2
new ground improvement approach, comprising the construction of Horizontal Soil Mixed (HSM)
beams beneath existing houses was developed. The method involves mixing of injected grout into
in-situ soils to construct a series of discreet HSM beams. The development of the construction
methodology is discussed in detail in Hunter et al. (2015).
The main purpose of the HSM ground improvement method is to increase the thickness of the nonliquefying crust beneath existing houses. To achieve this, the HSM beams are installed within the
shallowest layer of soil that would otherwise be vulnerable to liquefaction (the target soil layer).
While it would be possible to construct the HSM beams to form a continuous soil cement raft (i.e.,
with the HSM beams overlapping each other so that there is no unmixed soil between the beams),
this was not considered economic. The HSM beams were therefore designed to prevent
liquefaction, at the design ground motions, between the HSM beams within the target soil layer by
acting to confine the interstitial soil that remains between the beams.
Accordingly, the effectiveness of the beams at confining the interstitial soil and preventing the
triggering of liquefaction within it is a key design question for a given HSM beam layout. In order
to examine this, studies of a typical layout of HSM beams have been undertaken. These studies
comprised in-situ vibroseis dynamic (T-Rex) load testing and dynamic numerical simulation.
T-Rex Shake Testing of the HSM Beam Ground Improvement
Panels of HSM beams were constructed at three sites in eastern Christchurch (Sites 3, 4 and 6,
shown in Wissmann et al., 2015). The T-Rex testing undertaken for the HSM beams was part of a
wider trial assessing a number of different shallow ground improvement techniques. van Ballegooy
et al. (2015a and b) discusses this trial and the T-Rex shake testing methodology in greater detail.
A plan view and cross section schematic of the construction of a double row of the HSM beams,
the instrumentation layout, T-Rex base plate layout and shake testing direction relative to the
beams for one of the panels at Site 4 is shown in Figure 1.
Figure 1. (a) Plan view and (b) cross section of the instrumentation setup, including the Pore
water Pressure Transducers (PPT) and three dimensional (3D) geophones, relative to the T-Rex
base plate and the HSM beams at the third test panel at Site 4.
The vibroseis testing dynamically applied oscillating shear loads from the ground surface at both
the unimproved and the HSM beam improved panels. An array of 3D geophones and pore water
pressure transducers (PPTs) were installed in the ground directly below the T-Rex shaker to
indirectly measure the cyclic shear strain, (from relative displacements between adjacent
geophone sensor locations) and the excess pore water pressure (ru). These measurements were
collected over a range of applied shaking levels and the results for one of the double row of HSM
beam panels constructed at site 4 and the double row of HSM beam panel constructed at Site 6 are
shown in Figure 2 and compared to the adjacent natural soil panels. It is noted that the shaking at
all the HSM beam panels was undertaken in the perpendicular direction apart from the HSM beam
panel at Site 4 which was shaken in both directions (refer to Figure 1).
Figure 2. Natural soil CPT q c and Ic profiles, crosshole VS profiles and T-Rex profiles.
It is noted that the results from the double row of HSM beams at Sites 3 and two panels at Site 4
are not shown because the HSM beams for these panels were not in a constructed correctly in the
regular pattern shown on Figure 1 (van Ballegooy et al., 2015b).
Because the T-Rex shaker applies shear loads at the ground surface to a 2.3 m square plate, the
profiles decay relatively rapidly with depth. The results shown in Figure 2 indicate that for each
of the applied shear stress levels at Site 6 and the higher applied shear stress levels at Site 4 (> 15
kPa), the in the soils between the HSM beams were considerably reduced as a result of the beams
compared to the adjacent natural soil panels, decreasing the potential for development of ru and
hence liquefaction triggering under cyclic loading. However, at the smaller applied shear stress
levels at Site 4 (< 10 kPa), the in the soils between the HSM beams appeared to increase as a
result of the beams compared to the adjacent natural soil panels. Figure 2 also shows the Cone
Penetration Test (CPT) tip resistance (qc) and soil behaviour type index (Ic) as well as the crosshole
shear wave velocity (VS) of the natural soils surrounding the HSM panels (shown as the lighter
blue traces) and within the HSM panel, collected prior to the HSM beam construction (shown as
the darker blue traces). At Site 6 the CPT and VS profile at the HSM panel appears the same as the
surrounding soil, whereas at the Site 4 panel location the VS profile appears to have lower values
compared to the surrounding soils, possibly indicating that the HSM beams were constructed in a
local soft spot. This would increase the values and make the interpretation of the relative
comparisons between the HSM beam improved soil results and adjacent natural soil results
difficult and could possibly explain the increase in values at the lower applied shear stresses.
Dynamic Numerical Modelling of the HSM Beam Ground Improvement
In order to further assess the effectiveness of the HSM beams in mitigating the effects of
liquefaction, numerical modelling of the beams was undertaken. The purpose of this modelling
was to specifically assess how effective the HSM beams were in preventing or limiting liquefaction
triggering in the soil between the beams, and as such supplemented the field testing of the T-Rex
shaker. The purpose of the HSM beams, as discussed above, is to develop a non-liquefying crust
by confining the soil in between the beams to reduce and ru. As such, the numerical modelling
primarily assessed the development of and ru generation of the soil between the beams (within
the non-liquefying crust), although the effects beyond the “crust” itself were also assessed.
The numerical modelling included dynamic analyses with “total stress” (where pore pressure
generation due to shearing was not included in the model, and “effective stress” (pore pressure
generation due to shearing is included) constitutive models. The UBCHyst model (Naesgaard,
2011) was used for the total stress analysis while the PM4Sand model (Boulanger and Ziotopoulou,
2012) was used for the effective stress analysis. The total stress analysis enabled the relative
differences of the for the natural soil and HSM beam soils to be examined while the effective
stress analysis also allowed for the generation of ru, further influencing the values in the latter
stages of the dynamic simulations.
The finite difference program FLAC 7.0 (Itasca Consulting Group, 2011) was used for the 2D
analysis. A soil profile representative of the conditions at Site 4 was used for all numerical analyses
based on borehole, CPT and shear wave velocity (Vs) information (summarised in Figure 2a, b and
c). Two configurations were analysed: an unimproved natural soil model and a model that included
two rows of HSM beams. Figure 3 shows a schematic of the model set up near the ground surface
for the HSM model. A surface surcharge of 10kPa was applied above the beams to approximate
the surcharge from a typical residential building platform with a single storey dwelling. The beams
were modelled as a square cross section with the same equivalent area as the production beams to
simplify the model. A linear elastic constitutive model was used to for the beams.
Figure 3. Schematic of the numerical model and the HSM beams relative to the ground surface
and also relative to the 10 kPa residential building platform load.
For each model configuration, and for both the total and effective stress analyses, a suite of ground
motions were used as dynamic input. These were selected to be representative of the controlling
earthquakes at return periods of 25, 100 and 500 years. These return periods relate to the levels of
ground shaking at the Serviceability Limit State (SLS), Intermediate Limit State (ILS) and
Ultimate Limit State (ULS) that ordinary residential structures must be designed to meet (MBIE,
2012). The SLS, ILS and ULS ground motions were defined as Mw=7.5 equivalent PGAs of 0.13g,
0.20g and 0.35g respectively. However, specific earthquake scenarios were considered in selection
of ground motions which included local (Christchurch) fault events, Canterbury Plains events and
an Alpine fault event. The following earthquake records were selected: Darfield (2010) Canterbury
Aero Club (CACS) recording station (used for the SLS, ILS and ULS events); Darfield (2010)
Riccarton High School (RHSC) recording station (used for the SLS, ILS and ULS events);
Christchurch (2011) CACS (used for the ULS event); Christchurch (2011) RHSC (used for the
ULS event); Landers (1992) Joshua Tree (NGA ID 864) (used for the SLS, ILS and ULS events);
Chi Chi (1999) TCU078 (NGA ID 1512) (used for the SLS and ILS events); and Denali (2002)
Fairbanks – Geophysical Obs., CIGO (2110) (used for the SLS and ILS events). The RHSC and
CACS recordings were de-convolved as required. Recordings were scaled to a target PGA for the
various design motions, accounting for the various magnitudes using the Idriss and Boulanger
(2008) magnitude scaling factors. In total, fifteen different ground motions were analysed, five for
each level of shaking.
Results of the effective stress analyses are presented in Figure 4, in terms of maximum and ru
along the profile labelled Section 3 in Figure 3. The results are presented as the reduction
(percentage change) in max and ru for the HSM beams model when compared to the unimproved
model. Each trace represents the results for a single ground motion. The max profiles clearly show
a variation in behaviour between the soil within the zone of the HSM beams and also below the
HSM beams. A consistent trend of reduced max can be seen at all levels of shaking between the
beams. This result supports the overall improvement concept for HSM beams, which is one of
increasing the thickness of the non-liquefying layer by reducing in the soil between the beams.
The results beneath the beams are somewhat more variable, however there is a trend of increasing
max over the 2m beneath the HSM beams. This result is considered most likely to be due to the
presence of a stiffened layer above the zone of increased , and is perhaps akin to the soft storey
phenomenon in structural engineering. Accordingly, it is considered unlikely that this result is
specific to HSM beams, and is likely to be a phenomenon where any stiffened crust is included,
for example a soil cement raft or gravel raft.
Figure 4. Percentage change in
max
and ru due to HSM beams for SLS, ILS and ULS scenarios.
The max results indirectly indicate the efficacy of the HSM beams in preventing or limiting
liquefaction. The ru results provide a direct measure of this effect. The percentage change in ru
results in Figure 4 are shown vs time since the commencement of strong shaking in each ground
motion record, for the same point in time during the earthquake record for the unimproved and
HSM beam model. The results presented are at three points on Section 3, as shown in Figure 3.
One of the points is at 2 m below the ground surface (bgs) which is between the HSM beams and
the other two points are 2.5 and 3.1 m bgs, below the HSM beams. The results show marked
differences in performance at the various levels of earthquake shaking. At the SLS level of shaking
the results consistently show that the build-up of ru is prevented in the HSM beam model within
and immediately beneath the beams. There is no significant difference in behaviour at the deepest
recorded level. The ILS results show a consistent reduction in r u throughout the earthquake records
between the HSM beams, a delay in ru build up immediately beneath the beams and a small
increase in ru at depth. The ULS results show a delay in the build-up of r u between the HSM beams,
as well as a reduction in ru in two cases. At depth the results are inconsistent but show an increase
in ru build-up for two of the selected ground motions. Overall the results indicate that the HSM
beams tend to reduce r u build-up between the beams, but that this improvement does not extend to
all levels of shaking. At ULS levels of shaking, whilst the tendency to reduce r u is present, it is not
sufficient to prevent liquefaction.
The results in Figure 4 are near the centre of the HSM beams and residential house building
platform. However, residential houses have a limited width, and the behaviour at the lateral extents
of the structures is also of significance, particularly as more highly loaded foundations are typically
found at the edge of the residential building platforms. Figure 5 presents the results of max at the
edge of the residential house building platform and HSM beam extents along Sections 1 and 2 (as
shown on Figure 3), for the SLS ground motions only.
Figure 5.
max
and the percentage change in max near the edge of a house at SLS levels of
earthquake shaking.
The results on the top row of Figure 5 show max for the natural soil. The spike in max across the
profile is a result of the edge of the residential house building platform, indicates that the cyclic
stresses are higher near the edges, resulting in localised liquefaction triggering at lower levels of
shaking compared to the soil further away from the edge effects. The bottom row of Figure 5 shows
the percentage change in max due to the presence of the HSM beams. The results show a consistent
reduction in max between the beams compared to the natural soil case in general. There are some
increases in max immediately adjacent to the beams themselves, however this is not considered
significant due to the relatively thin zone over which this occurs. Overall the results indicate the
beams are effective in reducing max at the edge of building footprints.
Discussion and Conclusions
Ishihara (1985) recognised that a thick non-liquefying crust overlying liquefying soils would
reduce the consequences of liquefaction (i.e., sand boils, loss of bearing capacity and differential
settlement). This was confirmed by the observations following the CES, where less structural
damage occurred in liquefaction-prone areas containing an intact, relatively stiff non-liquefying
crust with minimum thickness of approximately 3 m. The HSM beam ground improvement can be
applied beneath existing structures to increase the non-liquefying crust thickness and hence
decrease the liquefaction vulnerability. In-situ dynamic T-Rex shake testing was undertaken on
HSM beam ground improvement panels to examine the liquefaction triggering and supplemented
by dynamic numerical simulation.
Both the T-Rex shake testing and the dynamic numerical simulations showed that the HSM beams
were effective in confining the interstitial soil that remains between the beams resulting in reduced
and ru at SLS and ILS levels of earthquake shaking. Therefore, at these levels of shaking HSM
beam ground improvement are effective in increasing the non-liquefying crust thickness and also
increasing the crust stiffness (Wentz et al., 2015), decreasing the liquefaction vulnerability of the
existing residential houses. At ULS levels of shaking, whilst the tendency to reduce ru is present,
it is not sufficient to prevent liquefaction and hence at these levels of shaking the HSM beams are
unlikely to have much beneficial effect in increasing the non-liquefying crust thickness. However,
there may still be some reduction in differential settlement at ULS levels of shaking due to the
increase in crust stiffness as a result of the HSM beams (van Ballegooy et al., 2015b).
Acknowledgments
The experimental work described in this paper was funded by the New Zealand Earthquake
Commission and the National Science Foundation. This support is gratefully acknowledged. The
authors would also wish to acknowledge H. Cowan and K. Yamabe from the Earthquake
Commission and K. Stokoe, B. Cox, J. Roberts, S. Hwang, F. Menq, A. Keene, A. Stolte, C.
Hoffpauir, A. Valentine and R. Kent from the University of Texas at Austin who undertook the TRex shake testing, K. Hedberg and E. Tang from Tonkin & Taylor who undertook the numerical
modelling, and Professors J. Bray from the University of California, Berkley and M. Cubrinovski
from the University of Canterbury, who provided support and advice in building the numerical
models.
References
Boulanger, R. W., and Ziotopoulou, K. (2012). PM4Sand (Version 2): A sand plasticity model for earthquake
engineering applications. Rep. UCD/CGM-12/01, Center for Geotechnical Modelling, Dept. of Civil and
Environmental Engineering, Univ. of California, Davis, CA .
Hunter, R., van Ballegooy, S., Leeves, J., & Donnelly, T. (2015). Development of horizontal soil mixed beams as a
shallow ground improvement method beneath existing houses. Proceedings of the 12th Australia New
Zealand Conference on Geomechanics (pp. 650-657). Wellington, New Zealand: NZGS and AGS.
Ishihara, K. (1985). Stability of natural deposits during earthquakes. Proceedings of the 11th International Conference
on Soil Mechanics and Foundation Engineering. 1, pp. 321-376. San Francisco: ISSMGE.
Itasca Consulting Group (2011). FLAC – Fast Lagrangian Analysis of Continua. Version 7.0 [computer program].
Itasca Consulting Group, Minneapolis, Minnesota.
MBIE. (2012). Repairing and rebuilding houses affected by the Canterbury earthquakes. Ministry of Business,
Innovation and Employment. December 2012. Retrieved from http://www.dbh.govt.nz/guidance-on-repairsafter-earthquake.
Naesgaard, E. (2011). A hybrid effective stress–total stress procedure for analyzing soil embankments subjected to
potential liquefaction and flow. (Doctoral dissertation, University of British Columbia).
Rogers, N., van Ballegooy, S., Williams, K., & Johnson, L. (2015). Considering post-disaster damage to residential
building construction - is our modern building construction resilient? Proceedings of the 6th International
Conference on Earthquake Geotechnical Engineering. Christchurch, New Zealand: ISSMGE.
Russell, J., van Ballegooy, S., Lacrosse, V., Jacka, M. and Rogers, N. (2015). The Effect of Subsidence on Liquefaction
Vulnerability Following the 2010 – 2011 Canterbury Earthquake Sequence. Proceedings of the 12th
Australia New Zealand Conference on Geomechanics (pp. 650-657). Wellington, New Zealand: NZGS and
AGS.
van Ballegooy, S., Roberts, J., Stokoe, K., Cox, B., Wentz, F., & Hwang, S. (2015a). Large scale testing of ground
improvements using controlled, dynamic staged loading with T-Rex. Proceedings of the 6th International
van Ballegooy, S., Wentz, F., Stokoe, K., Cox, B., Rollins, K., Ashford, S., & Olsen, M. (2015b). Christchurch Ground
Improvement Trial Report. Report for the New Zealand Earthquake Commission. (In press).
Wentz, R., van Ballegooy, S., Rollins, K., Ashford, S., & Olsen, M. (2015). Large scale testing of shallow ground
improvements using blast-induced liquefaction. Proceedings of the 6th International Conference on
Earthquake Geotechnical Engineering. Christchurch, NZ: ISSMGE.
Wissmann, K., van Ballegooy, S., Metcalfe, B., Dismuke, J., & Anderson, C. (2015). Rammed aggregate pier ground
improvement as a liquefaction mitigation method in sandy and silty soils. Proceedings of the 6th International
Conference on Earthquake Geotechnical Engineering. Christchurch, New Zealand: ISSMGE
74
Rammed Aggregate Pier
Ground Improvement as a
Liquefaction Mitigation Method
in Sandy and Silty Soils
K.J. Wissmann, S. van Ballegooy, B.C. Metcalfe, J.N. Dismuke, C.K. Anderson
75
1-4 November 2015
Rammed Aggregate Pier Ground Improvement as a Liquefaction
Mitigation Method in Sandy and Silty Soils
K.J. Wissmann1, S. van Ballegooy2, B.C. Metcalfe3, J.N. Dismuke4, C.K. Anderson5
ABSTRACT
Ground improvement methods have been used for over 70 years to densify loose sands prone to
liquefaction. Although these methods reduce liquefaction triggering potential and settlement in
densifiable soil, such as loose clean sand, their impacts on soils that are difficult to densify, such as
silty soils, are not well understood. This paper examines the results of full scale testing performed
for Rammed Aggregate PierTM treated soil in Christchurch, New Zealand carried out as part of a
large scale study by the New Zealand Earthquake Commission. The paper describes pier
construction, and outlines test results including pre- and post-installation cone penetration test tip
resistances, crosshole shear wave velocity, and vibroseis shaking tests. The results indicate that soil
densification may be considered to be the primary liquefaction mitigation mechanism in soils with
a soil behavior type index, Ic < 1.8, and that composite dynamic stiffness of the RAP-treated soil
likely dominates the liquefaction resistance mechanism in soils with Ic > 1.8. This paper is of
particular significance because it provides a well-documented link between a widely used ground
improvement method and the mechanisms involved in liquefaction mitigation.
Introduction
Christchurch, New Zealand is founded on a complex, interlayered sequence of alluvial soils
vulnerable to liquefaction-induced land damage from moderate to severe earthquake events.
Widespread ground surface deformation from liquefaction-induced differential and total
settlement and lateral spreading occurred during the 2010 to 2011 Canterbury Earthquake
Sequence (CES). Liquefaction induced damage affected 51,000 residential properties (Figure 1)
with approximately 15,000 residential houses damaged beyond economic repair. The Earthquake
Commission (EQC), a government insurer of private houses in New Zealand, funded a trial
program to evaluate the efficacy of various cost effective ground improvement methods. The
program objective was to investigate the technical viability of using ground improvement to reduce
liquefaction vulnerability for the rebuild or repair of houses. The tested methods include rapid
impact compaction (RIC), Rammed Aggregate Pier™ (RAP) reinforcement, driven timber piles
(DTP), low mobility grout (LMG), resin injection, and shallow gravel and soil cement rafts.
Test panels for each ground improvement method were constructed at three sites in Christchurch
in areas severely affected by liquefaction (Figure 1). The testing phase comprised pre- and postimprovement cone penetration testing (CPT) and crosshole shear wave velocity (V S) testing,
vibroseis T-Rex testing, and blast-induced liquefaction testing. The T-Rex shake test results and
1
Dr. Kord Wissmann, Geopier Foundation Company, Davidson, North Carolina, USA, [email protected]
Dr. Sjoerd van Ballegooy, Tonkin & Taylor Ltd, Auckland, New Zealand, [email protected]
3
Brian Metcalfe, Geopier Foundation Company, Davidson, North Carolina, USA, [email protected]
4
James Dismuke, Golder Associates, Christchurch, New Zealand, [email protected]
5
Clive Anderson, Golder Associates, Christchurch, New Zealand, [email protected]
2
the blast-induced liquefaction test results are presented in van Ballegooy et al. (2015a) and Wentz
et al. (2015) respectively. The blast induced liquefaction tests provided a relative assessment of
the liquefaction susceptibility of unimproved and improved sites, which enabled a measure of
comparison between the improved sites. Pore water pressure measurements made during blasting
for the RAP-improved areas showed that excess pore pressure ratio (ru) values were less than unity
for sensors installed in both silty and clean sand materials. Although the number of measurements
were not sufficient to be conclusive and the ru measurements could be explained by other
mechanisms (e.g., installations in thin sand layers, installations in layers not fully saturated), the
measured site performance and the low ru values resulted in a postulation that the installation of
the RAP elements reduced the liquefaction susceptibility of both the clean sand and silty soil
layers. The purpose of this paper is to describe the results of the measurements for the RAP
treatment and to explore the mechanisms of RAP remediation.
Figure 1. Severity and extent of the mapped liquefaction land damage on the residential land in
Christchurch as a result of the CES. The white areas represent all the non residential land areas.
Rammed Aggregate Pier Ground Improvement Construction
RAP elements were constructed at the test sites using displacement techniques with an excavatormounted mobilram base machine fitted with a high frequency (30 to 40 Hz) vibratory hammer.
The base machine drives a 250 to 300 mm outside diameter open-ended pipe mandrel fitted with
a unique specially-designed 350 to 400 mm diameter tamper foot into the ground. The method
uses hydraulic crowd pressure and vertical vibratory hammer energy to displace and densify the
liquefiable soils. Crushed gravel (typically graded at 20 to 40 mm in particle size) is fed through
the mandrel from a top mounted hopper and compacted in the displaced cavities to create
approximately 600 mm diameter, dense, stiff, aggregate pier elements (Figure 2).
Nine RAP test areas were constructed at three sites in Christchurch. The RAP elements were
spaced 1.5 to 3.0 m on-center in a triangular pattern resulting in an area replacement ratio, Ar.
ranging from 4 to 15%. The piers were installed to depths of 4 m in soil profiles that graded
sequentially from sandy silt and silty sand to clean sand with depth. The siltier deposits were
generally located within the top 1.2 to 2 m of the soil profile and the clean sands deeper than about
2 m. Groundwater was present at approximately 1 m below ground surface (bgs).
Figure 2. RAP ground improvement construction method applied at the three test sites.
At each of the tested areas, a series of in-situ CPT, crosshole V S, and vibroseis investigations were
conducted pre- and post-construction to quantify the improvement from the RAP installations.
Figure 3 shows a schematic of the in-situ tests relative to the RAP locations. The CPTs were
performed equidistant from the three installed RAP elements to conservatively measure the results
at locations furthest from the piers. Crosshole VS tests were performed in the nearby natural soil
and in six of the RAP test areas where the pier spacing was 2.0 m (Ar = 8%). As shown in Figure
3, crosshole VS tests were made both between and across RAPs.
Figure 3. In-situ CPT and crosshole VS test layout relative to the RAP locations.
CPT Investigation Results
Figures 4a and 4b show plots of all of the uncorrected pre- and post-improvement CPT tip
resistance (qc) measurements for three of the RAP test areas where the RAP spacing was varied.
For additional clarity, Figures 4c and 4d illustrate the q c values isolated for the soil layers with a
soil behavior type index, Ic < 1.8 and Ic > 1.8, respectively. The data show that RAP installations
consistently increase qc in soil layers with Ic < 1.8 (i.e. lower fines content, FC) whereas a minimal
improvement in qc occurs in soil layers with Ic > 1.8. These results indicate that soil layers with
higher FC are not appreciably densified by RAP treatment, an observation that is consistent with
those widely reported in the literature for other soil densification methods. Figure 4c and 4d also
show three qc envelope lines representing the computed q c thresholds required to resist liquefaction
for 25, 100 and 500 year return period ground motion cases for Christchurch as specified in the
MBIE., 2014 guidelines computed using the Boulanger and Idriss (2014) liquefaction triggering
method. Comparison of the pre-improvement qc traces against the liquefaction triggering
thresholds shows that the natural soils are generally predicted to liquefy between the 25 and the
100 year return period motions. This prediction is consistent with the observed performance of the
ground through the CES. A comparison of the envelope of post-improvement qc traces for the
clean sand materials (Ic < 1.8) shows that the RAP improved soils are not predicted to liquefy
under the Ultimate Limit State motions when the RAP Ar = 15%, and are not predicted to liquefy
for 85% of cases when the RAP Ar ranged between 5 to 8%.
Figure 4. Pre- and post-improvement CPT traces of RAP improved soil at the three test areas: (a)
q c for all traces, (b) Ic for all traces, (c) q c for all soil layers with Ic < 1.8, and (d) q c for Ic > 1.8.
Figure 5a shows the average computed percentage increases in the Cyclic Resistance Ratio (CRR)
values for various column spacings using the Boulanger and Idriss (2014) liquefaction triggering
methodology. The data presented in Figure 4 indicate that for the silty soil layers (Ic > 1.8) there
is a negligible increase in qc, and hence CRR. Conversely, measured q c and hence computed CRR
(Figure 5a) values increase significantly for soil layers where the Ic < 1.8. As expected the greater
measured and computed percentage increases occurred at higher Ar values.
The computed vertical one dimensional post liquefaction densification settlement, S V1D, and
Liquefaction Severity Number (LSN) vulnerability parameter values (defined in van Ballegooy et
al., 2015b) for all of the pre- and post-improvement CPT traces at all of RAP test areas are shown
in Figure 5b and 5c. The SV1D and LSN values are calculated for the MBIE (2014) specified 500
year return period ground motions over the top 10 m of the soil profile using the Boulanger and
Idriss (2014) liquefaction triggering methodology. The results presented in Figure 5 demonstrate
a significant reduction in liquefaction vulnerability provided by shallow RAP ground improvement
illustrating the benefits produced by RAPs for densifying clean sand and sand with silt. Because
the qc measurements were consistently carried out at the center of the RAP pattern where soil
densification is less than at locations close to the RAP, the results may be considered to be lower
bound conditions. Further, the computation method neglects any reduction in liquefaction potential
that stems from the improved composite stiffness of the reinforced soil.
Figure 5. (a) Average percentage increase in computed CRR at different Ar for the RAP
improved soils at the three test areas. (b & c) Distribution of the SV1D and LSN liquefaction
vulnerability parameters at each CPT location for the natural and RAP improved soils.
Crosshole Shear Wave Velocity Investigation and Vibroseis Shake Testing Results
Crosshole VS measurements were performed at six RAP panels where the pier spacing was 2.0 m
(Ar = 8%). Figure 6 illustrates the measured VS and calculated small strain shear modulus, Gmax,
values for all six test panels. As shown in Figure 6, the crosshole VS traces for the soil matrix
between the RAP columns show a small but discernable increase relative to the natural unimproved
ground for both the upper soil horizon with higher Ic values and the lower soil horizon with lower
Ic values. Because Gmax is proportional to the square of Vs, the increases in the Gmax values shown
in Figure 6b are optically more evident for both soil horizons.
The composite crosshole VS (measured across the RAP elements) values are significantly larger
than those for both unimproved and improved soil because of the presence of the stiff RAP
elements in the measured results. In comparison with the natural soil, the average composite Gmax
values increased by approximately 15 MPa within the upper silty soil horizon and by
approximately 65 MPa within the lower clean sand soil horizon. Unlike the CPT results that
indicated negligible improvement in the siltier soils with Ic > 1.8, the clear improvement in Gmax
suggests the potential for reduced liquefaction potential in the soils with higher FC.
Previous investigators (Baez and Martin, 1993; Priebe, 1995) proposed design methods that
utilized the concept of composite shear strain stiffening for liquefaction mitigation. These methods
considered a reduced shear stress demand on the native soil by considering shear stress attraction
to relatively stiff reinforcing elements. The composite shear stiffness method was later subject to
criticism (Goughnour and Pestana, 1998 and Rayamajhi, et al., 2012) or limitation (Girsang, et al.,
2004; Green et al., 2008) because of the potential for flexural response of the reinforcing elements
at high shear strain levels or because of the potential for gaps to occur between the soil and the
reinforcing element for reinforcing elements constructed from rigid materials such as those used
for piles, deep soil mixing, or jet grout columns. To provide insight into the prevalence of these
mechanisms, the EQC performed high shear strain vibroseis testing at both reinforced and
unreinforced areas at the test sites.
Figure 6. Pre- and post-improvement crosshole VS (a) and Gmax (b) profiles of the RAP improved
soils at the three areas for RAP spacing of 2 m.
The vibroseis testing was performed by a team from the University of Texas, Austin by
implementing the T-Rex mobile shaker to dynamically apply oscillating shear loads from the
ground surface at both the unimproved and the improved soil conditions (van Ballegooy et al.,
2015a). An array of geophones and pore water pressure transducers were installed in the ground
directly below the T-Rex shaker to allow estimation of cyclic shear strain, induced in the ground
and to measure the development of excess pore water pressures. Figure 7 presents both the
previously-discussed Vs profiles and the measured peak shear strain profiles from the T-Rex
shaker testing for natural soil and post-improvement RAP treated soil for two cyclic horizontal
shear stress loading levels. Because the T-Rex shaker applies shear loads at the ground surface to
a 2.3 m square plate, shear strains decay relatively rapidly with depth. The results shown in Figure
7 indicate that for each of the applied shear stress levels, the cyclic shear strains in the RAP
reinforced soil were reduced to approximately 20% to 33% of the cyclic shear strain values
measured in the unimproved soil, which indicates that the composite RAP reinforced ground is
stiffer than the natural unimproved soil by a factor ranging from 3 to 5 at both low and high shear
strain levels. The increase in the high-shear-strain composite stiffness decreases the potential for
development of excess pore water pressure and hence liquefaction triggering under cyclic loading.
Figure 7. Pre- and post-improvement crosshole VS (a), at 5 kPa of horizontal cyclic stress
applied by T-Rex at the ground surface (b & c), and at 15 kPa of cyclic stresses (d & e).
The Christchurch testing program provided a unique opportunity to investigate the efficacy of a
variety of ground improvement methods for mitigating soil liquefaction and provide insight into
the mechanics governing the measured response. CPT qc measurements confirmed that the RAP
displacement method effectively densified clean sand deposits with Ic < 1.8 but did not provide
measureable densification for the upper soil horizon with Ic > 1.8. Small strain crosshole Vs testing
indicated an evident increase in the Vs measurements and corresponding Gmax response for the
improved natural soil and a large increase in Vs and Gmax responses for the composite RAPreinforced ground in both the clean and silty soil horizons. Large strain T-Rex testing showed that
the composite reinforced ground within both the clean sand and silty soil horizons exhibited shear
stiffness values greater than the unimproved soil by a factor of 3 to 5, confirming the effectiveness
of reinforcing non-densifiable soil with RAP elements.
The results of the test program suggest that the improvement in the liquefaction resistance of the
natural soil is related to the increase in the shear stiffness response of the RAP-reinforced ground.
The increase in composite shear stiffness may be explained by a variety of mechanisms. It is likely
that the uncemented RAP materials combined with the vertical ramming inherent in the RAP
construction process results in a well-coupled pier-soil response that transfers shear stresses
effectively across the soil-pier interface whereby the response is a byproduct of the unique
construction process. It is also likely that the response results from the high lateral stresses that are
applied to the natural soil during pier construction. These high lateral stresses serve to increase the
mean stress conditions of the natural soil well above the normally-consolidated stress state (Handy
and White, 2006), creating conditions that have been shown by Harada et al. (2010) to increase
CRR values within the reinforced soils. Regardless of the mechanism, the results presented herein
show that the RAP elements consistently reduce liquefaction susceptibility in both clean and silty
soils through a combination of soil densification and composite ground shear stiffening.
Acknowledgements
Commission and the National Science Foundation. This support is gratefully acknowledged. The
authors would also wish to acknowledge H Cowan and K Yamabe from the Earthquake
Commission and K Stokoe, Brady Cox, J. Roberts, S. Hwang, F. Menq, A Keene, A Stolte, C
Hoffpauir, A Valentine and R Kent from the University of Texas at Austin and Lake Carter of
Geopier Foundation Company.
References
Baez, J.I. and Martin, G.R. (1993). “Advances in the design of vibro systems for the improvement of liquefaction
resistance”, Proc. of the 7th Annual Symposium of Ground Improvement, 1-16.
Boulanger, R.W., and Idriss, I.M., (2014). “CPT and SPT based liquefaction triggering procedures.” Report No.
UCD/CGM-14/01, Center for Geotechnical Modeling, Department of Civil and Environmental Engineering,
University of California, Davis, CA, 134 pp.
Goughnour, R.R. and Pestana, J.M. (1998). “Mechanical behavior of stone columns under seismic loading.” Proc.,
2nd Int. Conf. on Ground Improvement Techniques, Singapore, 157-162
Girsang, C.H., Guitierrez, M.S., and Wissmann, K.J. (2004). “Modelling of the seismic response of the aggregate pier
foundation system.” GeoSupport 2004, pp. 485-496.
Green, R.A., Olgun, S.G., and Wissmann, K.J. (2008). “Shear stress redistribution as a mechanism to mitigate the
risk of liquefaction.” J. Geotech. Earthq. Eng. & Soil Dyn., ASCE IV GSP 181.
Handy, R.L, and D.J. White (2006). “Stress Zones Near Displacement Piers: Plastic and Liquefied Behavior.” ASCE.
Journal of Geotechnical and Geoenvironmental Engineering. Vol. 132. No. 1. January 2006.
Harada, K., Orense, R.P., Ishihara, K., and Mukai, J., (2010). “Lateral Stress Effects on Liquefaction Resistance
Correlations, Bulletin of The New Zealand Society for Earthquake Engineering, Vol. 43, No. 1, March.
Ministry of Business, Innovation and Employment (MBIE), (2014). Q&A for use of the Boulanger and Idriss (2014)
liquefaction triggering assessment method in the Christchurch Rebuild. Available at:
http://www.dbh.govt.nz/guidance-on-repairs-after-quake-issue-07#51
Priebe, H.J., (1995). “The design of vibro replacement,” Ground Engineering, Vol.28, No. 10, pp 31-37.
Rayamajhi, D., Nguyen, T.V., Ashford, S.A., Boulanger, R.W., Lu, J., Elgamal, A., Shao, L. (2012). “Effect of discrete
columns on shear stress distribution in liquefiable soil.” Geo-Congresss 2012: State of the Art and Practice in
Geotechnical Engineering, ASCE Geo-Institute, Oakland, CA, March 25-29.
van Ballegooy, S., Roberts, J., Stokoe, K., Cox, B., Wentz, F., Hwang, S., (2015a). “Large-Scale Testing of Shallow
Ground Improvements using Controlled Staged-Loading with T-Rex” for 6ICEGE.
van Ballegooy, S., Boulanger, R. W., Wentz, F., (2015b). “Evaluation of a CPT-based Liquefaction Procedure at
Regional Scale.” Soil Dynamics and Earthquake Engineering, Special Issue: Liquefaction in New Zealand and
Japan. In review
Wentz, F., van Ballegooy, S., Rollins, K., Ashford, S., Olsen, M., (2015). “Large Scale Testing of Shallow Ground
Improvements using Blast-Induced Liquefaction” for 6ICEGE.
76
Large-Scale Testing of Shallow
Ground Improvements using
Controlled Staged-Loading with T-Rex
S. van Ballegooy, J.N. Roberts, K.H. Stokoe, B.R. Cox, F.J. Wentz, S. Hwang
77
1-4 November 2015
Large-Scale Testing of Shallow Ground Improvements using Controlled
Staged-Loading with T-Rex
S. van Ballegooy1, J.N. Roberts2, K.H. Stokoe3, B.R. Cox4, F.J. Wentz5, S. Hwang6
ABSTRACT
Ishihara (1985) recognised that a thick non-liquefying crust overlying liquefying soils would reduce
the consequences of liquefaction (i.e., sand boils, loss of bearing capacity and differential
settlement). In Christchurch, in the aftermath of the 2010-2011 Canterbury earthquakes, detailed
engineering assessments of nearly 60,000 single-family houses combined with a comprehensive
regional scale geotechnical investigation, clearly showed that less structural damage occurred in
liquefaction-prone areas containing an intact, relatively stiff non-liquefying crust with a minimum
thickness of approximately 3 m. To increase the resilience of the post-earthquake rebuilt/repaired
Christchurch residential housing stock, the use of shallow (i.e., 4 m deep) ground improvements
to construct a stiff, non-liquefying crust and mitigate the consequences of underlying liquefaction
was evaluated. In this paper, the results from the in-situ vibroseis dynamic (T-Rex) load testing are
presented. This testing was able to examine the potential for liquefaction triggering to a depth of
about 3 to 4 m below the ground surface, coinciding with the target depth of the ground improvement
methods investigated as part of this study. The shake testing of the ground improvement panels
demonstrated that, in general, where the shallow ground improvements increased the Cone
Penetration Test (CPT) tip resistance (qc) or the composite crosshole shear wave velocity (V S) of
the improved ground relative to the natural soil, there was a corresponding reduction in the
maximum cyclic shear strain ( ) induced in the improved soil increasing the liquefaction resistance.
Introduction
The 2010 – 2011 Canterbury Earthquake Sequence (CES) caused widespread liquefaction-related
land and building damage (described in Rogers et al., 2015), affecting 51,000 residential properties
in Christchurch, including 15,000 residential houses damaged, beyond economical repair. In the
suburbs most vulnerable to liquefaction damage, the CES revealed the importance of constructing
robust, stiffened foundations capable of resisting the damaging effects of future liquefaction (i.e.
angular distortion, lateral stretch and loss of ground support) or the need to undertake ground
improvements to mitigate the damage caused by future liquefaction in future earthquakes.
Therefore, the New Zealand Earthquake Commission (EQC) funded an extensive shallow ground
improvement trial program to evaluate the efficacy of various shallow ground improvement
methods and determine the cost by undertaking full-scale construction trials on residential
properties. The purpose was to investigate and determine whether there are practical cost effective
shallow ground improvement methods that could be constructed on properties in existing
residential areas to form and/or enhance a non-liquefying crust and reduce liquefaction
1
Graduate Research Asst., Civil, Arc. & Env. Eng., Uni. of Texas at Austin, Austin, USA, [email protected]
3
Professor, Civil, Arc. & Env. Eng., Uni. of Texas at Austin, Austin, USA, [email protected]
4
Professor, Civil, Arc. & Env. Eng., Uni. of Texas at Austin, Austin, USA, [email protected]
5
Principal, Wentz-Pacific Ltd, Napier, New Zealand, [email protected]
6
Graduate Research Asst., Civil, Arc. & Env. Eng., Uni. of Texas at Austin, Austin, USA, [email protected]
2
vulnerability. The methods tested included Rapid Impact Compaction (RIC), Rammed Aggregate
Pier™ (RAP) reinforcement, Driven Timber Poles (DTP), Low Mobility Grout (LMG), Resin
Injection (RES), Gravel Rafts (GR), Soil-Cement Rafts (SCR) and Horizontal Soil-Cement Mixed
(HSM) beams. The construction methodology of each of the tested ground improvement methods
is described in van Ballegooy et al. (2015).
Construction of the various shallow ground improvement methods was undertaken in three
different locations in Christchurch (Sites 3, 4 and 6) in the areas most severely affected by
liquefaction. (The location of the sites is shown in Wissmann et al., 2015.) The testing phase of
the shallow ground improvement trials comprised pre- and post-improvement Cone Penetration
Testing (CPT), seismic crosshole testing, vibroseis T-Rex testing and blast-induced liquefaction
testing. In order to assess the overall effectiveness of the shallow ground improvements to mitigate
the damaging effects of liquefaction, an investigation of two primary mechanisms need to be
investigated: (1) how effective the improvements are in preventing or limiting the triggering of
liquefaction; and (2) how effective they are in reducing the consequences if liquefaction triggering
occurs in the soil beneath the improved zone.
The peak cyclic shear strain ( ) profiles produced by vibroseis T-Rex testing indicated that, at a
depth of approximately 4 m, the maximum at all test panels was consistently < 0.02%. Research
by Dobry et al. (1982) has shown that excess pore water pressures (ru) do not develop until the
peak are greater than 0.01% (the threshold ), and this finding is consistent with the results from
this study. Therefore, it was demonstrated that vibroseis T-Rex testing was only able to examine
the liquefaction triggering to a depth of about 3 to 4 m below the ground surface, which is the
target depth of improvement in this study. Essentially, vibroseis T-Rex testing was used to examine
the effectiveness of the various ground improvement methods to develop a non-liquefying crust
(i.e., the H1 layer described in Ishihara, 1985). However, because vibroseis T-Rex is unable to
induce liquefaction beneath the improvement zone, blast-induced liquefaction trials were also
undertaken (described in Wentz et al., 2015) to examine performance of different shallow ground
improvements in mitigating differential settlement caused by liquefaction of the underlying
unimproved soil layers (i.e., the H2 layer described in Ishihara, 1985).
Vibroseis T-Rex testing was applied to ground improvement test panels composed of Natural Soil
(NS), RIC, RAP, DTP, LMG, RES and HSM beams. Shake testing of the GR and SCR ground
improvement test panels were not undertaken because these materials will not liquefy and hence
there was no need to assess their triggering potential. Due to space constraints in this paper, only
the results of the RIC, RAP and LMG ground improvements compared to the natural soil are
discussed. Wissman et al. (2015) and Wansbone et al. (2015) examine the vibroseis T-Rex testing
of the RAP and HSM beam ground improvements in greater depth. Schematics of the construction
of the RIC, RAP and LMG methods is shown in Figure 1.
Figure 1. Illustration of the RIC, RAP and LMG ground improvement methods.
Vibroseis T-Rex Testing Methodology
The T-Rex truck applies a vertical load of 245 kN to a 2.3-m square baseplate that is set on the
ground, resulting in a pressure of 46 kPa on the ground surface beneath the baseplate. At each test
panel, the T-Rex truck applied a horizontal cyclic load at the ground surface using a 10-Hz
frequency for 100 cycles (10 seconds of shaking, N = 100). The testing typically involved five
loading stages, starting from the lowest level of loading (± 13 kN) to the highest level (± 107 or ±
133 kN). A horizontal cyclic load of +/- 133 kN is equal to +/- 25 kPa cyclic shear stress imparted
on the soil at the ground surface. In the maximum loading stage at the natural (unimproved) soil
test panels, this cyclic loading induces a of approximately 0.3% in the ground beneath the
baseplate at a depth of about 1 m, reducing to around 0.05% at a depth of approximately 3 m .
The soil response was recorded with embedded two dimensional (2D) velocity transducers and
Pore Pressure Transducers (PPTs). The induced at specific locations were evaluated from relative
displacements between adjacent sensor locations. The pore pressure generation was directly
measured with PPTs. Thus, the coupled behaviour between the dynamic response of the soil
skeleton, represented by , and the generated pressure was recorded. These measurements were
collected over a range of applied shaking levels in both natural and improved soils.
A typical test and instrumentation layout for the test panels is shown in Figure 2 (RAP in this case).
Note that the twelve RAP columns shown in Figure 2b are located in the centre of the 7 by 7 m
test panel which contained 22 columns. The location of the T-Rex baseplate during testing and the
direction of shaking are shown in plan view in Figure 2b and cross-sectional perspectives are
shown in Figures 2c and 2d. Before performing the shaking tests, instrumentation was embedded
within the plan footprint (2.3 x 2.3 m) of the baseplate at each test panel. The instrumentation was
installed using a CPT pushing mechanism mounted on the back of the T-Rex truck. The typical
instrumentation array in the test panels, which consisted of four, 2D velocity transducers and five
PPTs, is shown in a plan in Figure 2b and in cross section in Figure 2d.
Reduction of Test Results and Discussion
The raw data collected from the sensor arrays during vibroseis T-Rex testing consists of velocity
and pore pressure time histories at each sensor location. These data were used to compute the ru
(the ratio of generated pore water pressure to the initial vertical effective stress including the static
vertical load imparted by the T-Rex baseplate; ru = u/ v') time history and the induced time
history. An example of these time histories is shown in Figure 2e, corresponding to measurements
at a depth of 2.1 m at one of the natural soil panels at Site 6 for an applied horizontal cyclic loading
of 107 kN (~20 kPa applied horizontal cyclic stress at the ground surface). The velocity time
histories were numerically integrated to obtain displacement time histories, which were then used
to evaluate development. The histories at each PPT location were calculated using the
displacement-based method as described by Cox et al. (2009).
The maximum for each loading stage was linearly adjusted slightly to a nominal level of applied
shear stress at the ground surface so that the for each of the tested ground improvement panels
could be directly compared. For example, the peak shear stress imparted by the T-Rex vibroseis
unit at the ground surface during the second loading stage of the natural soil test panel at Site 6
was recorded as 5.3 kPa; therefore, the estimated peak for this loading stage were multiplied by
a ratio of 5 kPa : 5.3 kPa to linearly adjust the to match a nominal shear stress value for
comparison across test panels. A nominal shear stress level is used because while the input signal
sent to T-Rex vibroseis unit is set at a consistent value for each test panel (e.g. 1.5, 5, 10, 15, 20
and 25 kPa), the true force output depends on the stiffness of the soil as well as various
nonlinearities in the electrical and mechanical systems relating to the operation of the T-Rex
vibroseis unit.
r u = 32 %
Figure 2. (a) Location of the CPT and crosshole VP and VS testing relative to the RAP columns.
(b & d) The relative horizontal and vertical location, respectively, of the sensors beneath the TRex baseplate relative to the RAP columns. (c) Schematic cross section of the T-Rex truck on the
RAP test panel during shaking. (e) Time histories of ru and from data recorded at a depth of 2.1
m at one of the natural soil panels (not the RAP panel) at Site 6.
The last two columns of Figure 3 plot the adjusted values (at the 5 and 15 kPa applied cyclic
stresses at the ground surface) at each PPT location with depth. The blue, yellow, green and red
lines represent the two nominal shear stress profiles for the natural soil panel and the RIC, RAP
and LMG ground improvement panels, respectively.
Pre- and post-improvement CPT and crosshole testing was undertaken at each of the ground
improvement panels as well as the natural soil panels. The location of the crosshole testing relative
to the ground improvement aggregate columns or grout bulbs is shown in Figure 2a. For the RAP
and LMG ground improvements, two set of compression wave velocity (VP) and shear wave
velocity (VS) measurements were made: (1) between the improvement zones and (2) across the
improvement zones. The crosshole testing methodology is described in Stokoe et al. (2014).
The CPT tip resistance (qc) and soil behavior type index (Ic) traces are shown on the first two
columns on the left of Figure 3, respectively. Likewise the crosshole VP and VS traces are shown
on the third and fourth columns from the left, respectively, and the corresponding small-strain
shear modulus (Gmax) profiles calculated from the VS profiles are shown in the fifth column from
the left. Similar to the nominal shear stress profiles, the blue, yellow, green and red lines represent
the profiles for the natural soil, RIC, RAP and LMG test panels, respectively. For the RAP and
LMG ground improvements, the dark green and dark red traces represent the measured VP and VS
across the improvement zone and the light green and light red traces represent the measured VP
and VS between the improvement zones, respectively.
The results in Figure 3 show that, in general, the RAP ground improvement is the most effective
in increasing the CPT q c (which directly correlates with an increase in the Cyclic Resistance Ratio,
CRR) when Ic < 1.8. The RIC ground improvement is also effective in increasing the q c when Ic <
1.8. Little to no increase in qc was observed for the RIC and RAP ground improvements when Ic
> 1.8, nor for the LMG ground improvement at any Ic value.
For both the RAP and the RIC ground improvements, the crosshole-measured VS between the
improvement zones show some improvement on a site-by-site basis relative to the natural ground.
Furthermore, the crosshole VS measured between the LMG bulbs appears to have decreased
relative to the natural ground. Within the RAP ground improvement, however, the composite V S
(measured across the RAP columns) is significantly larger than those for both unimproved and
improved soil because of the presence of the stiff RAP elements. In comparison with the natural
soil, the average composite Gmax values (calculated from the composite VS) increased by
approximately 15 MPa (~40% increase) within the upper silty soil horizon and by approximately
65 MPa (~130% increase) within the lower clean sand soil horizon. Unlike the CPT results that
indicated negligible improvement in the upper siltier soils (with Ic > 1.8), the clear improvement
in Gmax in these soil layers suggests the potential for reduced liquefaction potential. Similar trends
are observed for the crosshole-measured VS across the LMG bulbs, but it is noted that this
increased stiffness across the bulbs is irregular, mainly because the LMG bulbs themselves are
irregular and not continuous.
Figure 3. Natural soil and post-improvement qc and Ic traces (first two columns) and crosshole
VP and VS and corresponding Gmax traces (middle three columns) for the natural soil and RIC,
RAP and LMG ground improvements. Similarly, the last two columns show the vibroseis T-Rex
traces at 5 and 15 kPa of applied cyclic horizontal stress at the ground surface.
The profiles from vibroseis T-Rex testing decay relatively rapidly with depth because the T-Rex
truck applies shear loads at the ground surface. For both load cases shown in Figure 3 (i.e. the 5
and 15 kPa of applied cyclic horizontal stress at the ground surface), the standardised profiles do
not show any noticeable reduction in compared to the natural soil for both the RIC and LMG
ground improvement methods. However, the results for the RAP ground improvement indicate
that for each of the applied shear stress levels, the profiles in the RAP reinforced soil were
reduced by approximately 60% to 80% relative to the natural soil, which indicates that the
composite RAP reinforced ground is stiffer than the natural soil by a factor ranging from 3 to 5.
The increase in composite stiffness (indicated by the crosshole VS across the RAP columns)
decreases the and hence the potential for development of ru (as shown by Stokoe et al., 2014),
increasing liquefaction triggering resistance under cyclic loading. The likely reason that the
reduction in for the same applied loading was not seen in the LMG (even though the measured
crosshole VS across the LMG bulbs is higher) is because the LMG bulbs were not regular
continuous reinforcing elements to stiffen the overall response of the ground, but instead they were
a series of irregular discontinuous bulbs and planes based on visual observation during exhuming
investigations at the end of the ground improvement trial program (van Ballegooy et al., 2015).
It is important to note that Stokoe et al. (2014) make direct comparisons between measured
parameters for the ground improvement panels and the adjacent natural soil measured parameters
at Site 6. Therefore, their conclusions relate specifically to Site 6 and may not apply more
generically across the tested areas. However, the discussion above comparing the ground
improvement results (shown in Figure 3) with the natural soil results relate to how the envelope of
measured parameters for the ground improvements across all the sites have changed compared to
the envelope of measured parameters for the natural soil sites. Therefore, these observations apply
more generically across the tested areas in Christchurch and at some sites the site specific
comparisons may indicate results that vary from the generic observations.
The value at each PPT location for each load stage was converted into an equivalent Cyclic Stress
Ratio (CSR) using the equation:
(1)
=
=
where is the cyclic shear stress (kPa), is the shear modulus (kPa) of the soil which reduces
as the ru increases during cyclic loading. The vertical
with increasing and with decreasing
effective stress (kPa) at each sensor location also includes the additional vertical stress from the
applied vertical load imparted by the T-Rex truck during all dynamic testing. G is calculated from
Gmax x [G/G max] = Vs2 x [G/G max], where is the soil density (kg/m3) and [G/Gmax] is a function
of and
at each instant during loading using the measured pore pressures. The value of G is
calculated using the procedures presented in Stokoe et al. (2016) using site-specific [G/Gmax]-log
empirical relationships published in Menq (2003).
Figure 4 shows the calculated CSR values at each PPT location for all ground improvement panels
for all sites at all loading stages are plotted against the representative composite VS values that
were inferred from the adjacent VS tests (refer to Figure 2a). The VS values were measured without
the weight of the T-Rex truck in place so they were adjusted to account for the influence of the
increased vertical load from the T-Rex truck.
Figure 4. Calculated CSR values at each PPT location for all loading stages at all three trial sites
for the Natural Soil (NS) and RIC, RAP and LMG ground improvements versus composite VS.
The blue, yellow, green and red symbol colours identify the calculated natural soil CSR values for
the Natural Soil (NS), RIC, RAP and LMG ground improvement panels, respectively. The solid
dots represent calculated CSR values where the corresponding residual ru at the end of the vibroseis
T-Rex testing (indicated on Figure 2e) was < 1%. Hollow circles represent calculated CSR values
where the corresponding ru was > 5%. Solid squares represent CSR values where the ru was
between 1 and 5%. Only data points where the soil was close to complete saturation (assumed as
VP > 750 m/s for the purposes of this study) are plotted on Figure 4 because, where the soils were
not fully saturated, the development of ru is likely to have been inhibited.
Figure 4 show that the onset of ru development (ru > 5%) for the nearly saturated soils (VP > 750
m/s) generally occurs at a CSR value of about 0.1 at low values of Vs. The CSR associated with a
ru > 5% increases with increasing Vs (i.e. as the soil becomes stiffer the CSR required to generate
ru > 5% increases). An illustration of this trend in terms of a potential boundary envelope is
presented by the dashed line on Figure 4. This potential boundary does not represent a triggering
boundary curve or a design curve as of now. It is simply presented to show the importance of raw
VS data in liquefaction triggering analyses.
Similar analyses were undertaken by normalizing the Vs data (i.e. VS1) using the Kayen et al.
(2013) procedure. However, when the data were plotted, the VS1 did not separate the data as well
as the raw Vs. Values of ru > 5% are important, because a slightly higher applied load, resulting in
a slightly higher , causes ru to rapidly increase resulting in liquefaction triggering as shown in
Stokoe et al. (2014). Therefore, understanding the value of CSR at which ru begins to rapidly
develop is useful to determine whether liquefaction is likely or unlikely for a given CSR.
Ishihara (1985) recognised that a thick non-liquefying crust overlying liquefying soils would
reduce the consequences of liquefaction (i.e., sand boils, loss of bearing capacity and differential
settlement). This situation was confirmed by the observations following the CES, where less
structural damage occurred in liquefaction-prone areas containing an intact, relatively stiff, nonliquefying crust with a minimum thickness of approximately 3 m. In-situ dynamic vibroseis T-Rex
shake testing was undertaken on natural and RIC, RAP and LMG ground improvement panels to
examine the liquefaction triggering to a depth of about 3 to 4 m below the ground surface,
coinciding with the target depth of the ground improvement methods investigated as part of this
study.
The vibroseis T-Rex testing of the ground improvement panels demonstrated that, in general,
where the shallow ground improvements increased the CPT qc (i.e. for the RAP ground
improvement when Ic < 1.8) or the composite crosshole VS of the improved ground (i.e. for all Ic
values for the RAP ground improvement) relative to the natural soil, there was a corresponding
reduction in the in the improved soils and hence a potential improvement in the liquefaction
resistance.
Conversion of the values into CSR values and plotting them against the corresponding raw Vs
values, separated the cases of potential significant pore pressure generation (ru > 5%) from cases
of minimal pore pressure generation. Raw (i.e. un-normalised) VS appears to separate clearly the
CSR data points with ru values of < 1% and > 5%. In particular, the RAP ground improvement
panels exhibited this relationship, probably because VS captures the stiffness of the composite soilimprovement element system. The CSR results demonstrate that, when the shallow ground
improvements increase the composite crosshole VS of the improved ground relative to the natural
soil, the CRR of the soil increases, reducing the potential for liquefaction triggering.
Acknowledgments
Commission and was partially supported by U.S. National Science Foundation (NSF) grant
CMMI-1343524 and Graduate Research Fellowship Program (DGE-1110007). However, all
findings and recommendations expressed in this paper are those of the authors and do not
necessarily reflect the views of NSF. This support is gratefully acknowledged. The authors would
also wish to acknowledge H Cowan and K Yamabe from the Earthquake Commission, Rick
Fragaszy from the U.S. National Science Foundation and F. Menq, C. Hoffpauir, A. Valentine, R.
Kent, Y. Wang, A. Stolte, and A. Keene, from the University of Texas at Austin.
References
Cox, B., Stokoe, K., & Rathje, E. (2009). An in-situ test method for evaluating the coupled pore pressure generation
and nonlinear shear modulus behavior of liquefiable soils. ASTM Geotechnical Testing Journal, 32(1),11-21.
Dobry, R., Ladd, R., Yokel, F., Chung, R., & Powell, D. (1982). Prediction of Pore Water Pressure Buildup and
Liquefaction of Sands During Earthquakes by the Cyclic Strain Method. Washington, DC.: National Bureau
of Standards.
on Soil Mechanics and Foundation Engineering. 1, pp. 321-376. San Francisco: ISSMGE.
Kayen, R., Moss, R., Thompson, E., Seed, R., Cetin, O., Kiureghian, A. D., . . . Tokimatsu, K. (2013). Shear wave
velocity-based probabilistic and deterministic assessment of seismic soil liquefaction potential. Journal of
Geotechnical and Geoenvironmental Engineering, ASCE, 139, pp. 407-419.
Menq, F.Y. (2003). Dynamic properties of sandy and gravelly soils. Ph.D. dissertation. The University of Texas at
Austin.
Rogers, N., van Ballegooy, S., Williams, K., & Johnson, L. (2015). Considering post-disaster damage to residential
building construction - is our modern building construction resilient? Proceedings of the 6th International
Stokoe, K., Roberts, J., Cox, B., Hwang, S., Menq, F., & van Ballegooy, S. (2016). Effectiness of shallow ground
improvedmtns to inhibit liquefaction triggtering: methodology and analysis of field trials using controlledsource, dynamic loading with T-Rex. Submitted to Soil Dynamics and Earthquake Engineering.
Stokoe, K., Roberts, J., Hwang, S., Cox, B., Menq, F., & van Ballegooy, S. (2014). Effectiveness of inhibiting
liquefaction triggering by shallow ground improvement methods: Initial field shaking trials with T-Rex at
one site in Christchurch, New Zealand. In Orense, Towhata, & Chouw (Eds.), Soil Liquefaction During
Recent Large-Scale Earthquakes (pp. 193-202). London: Taylor & Francis Group.
van Ballegooy, S., Wentz, F., Stokoe, K., Cox, B., Rollins, K., Ashford, S., & Olsen, M. (2015). Christchurch Ground
Wentz, F., van Ballegooy, S., Rollins, K., Ashford, S., & Olsen, M. (2015). Large scale testing of shallow ground
improvements using blast-induced liquefaction. Proceedings of the 6th International Conference on
Earthquake Geotechnical Engineering. Christchurch, NZ: ISSMGE.
Wansbone, M., & van Ballegooy, S., (2015). Horizontal Soil Mixed Beam Ground Improvement as a Liquefaction
Mitigation Method Beneath Existing Houses. Proceedings of the 6th International Conference in Earthquake
Geotechnical Engineering. Christchurch, New Zealand: ISSMGE.
improvement as a liquefaction mitigation method in sandy and silty soils. Proceedings of the 6th
International Conference on Earthquake Geotechnical Engineering. Christchurch, New Zealand: ISSMGE.
78
Large Scale Testing of Shallow
Ground Improvements using
Blast-Induced Liquefaction
F.J. Wentz, S. van Ballegooy, K.M. Rollins, S.A. Ashford, M.J. Olsen
1-4 November 2015
Large Scale Testing of Shallow Ground Improvements using Blast-Induced
Liquefaction
F.J. Wentz1, S. van Ballegooy2, K.M. Rollins 3, S.A. Ashford4, M.J. Olsen5
ABSTRACT
To increase the resilience of new/rebuilt houses in Christchurch, the New Zealand Earthquake
Commission (EQC) funded a study to evaluate the use of shallow (i.e., ≤ 4 m deep) ground
improvements to construct a stiff, non-liquefiable crust. This paper presents the methodology used
for the blast testing, and summarises some of the more significant findings from the overall blast
testing program. This effort included large-scale testing of several ground improvement methods
using controlled blasting to induce liquefaction to depths of 10 to 12 m beneath the zone of
improvement. The blast-induced liquefaction testing demonstrated that ground surface differential
settlement generally increases with increasing total settlement and that improvements that formed
stiffer crusts experienced less structurally damaging differential settlement.
Introduction
The 2010 – 2011 Canterbury Earthquake Sequence (CES) caused widespread liquefaction related
land and building damage, affecting 51,000 residential properties in Christchurch, including
15,000 residential houses damaged beyond economical repair. Most damage occurred from
differential settlement induced by liquefaction. Specifically, differential settlement comprised of
flexural distortion or “curvature” resulted in significantly greater foundation/structural damage
than did rigid tilt. Damage observations and geotechnical assessments of approximately 60,000
residential properties clearly demonstrated that almost no foundation deformation occurred in
areas with liquefaction susceptible soils overlain by an intact, relatively stiff, non-liquefying crust
with a minimum thickness of approximately 3 m. This finding is consistent with those of Ishihara
(1985) who recognised that a thick and/or stiff non-liquefiable crust can reduce the consequences
of liquefaction (i.e., sand boils, loss of bearing capacity and differential settlement). To increase
the resilience of new/rebuilt houses on vulnerable land, the New Zealand Earthquake Commission
(EQC) funded a comprehensive study to evaluate the efficacy and technical viability of using
shallow ground improvement (i.e., < 4 m) to reduce liquefaction vulnerability for the rebuild and
repair of houses. The methods tested included Rapid Impact Compaction (RIC), Rammed
Aggregate Pier™ (RAP) reinforcement, Driven Timber Piles (DTP), Low Mobility Grout (LMG),
Resin Injection (RES), Gravel Rafts (GR), Soil Cement Rafts (SCR) and Horizontal Soil-cement
Mixed (HSM) beams. The construction methodology of each of the tested ground improvement
methods is described in van Ballegooy et al. (2015b). The construction methodology of the HSM
beams is presented in detail in Hunter et al. (2015).
1
Principal, Wentz-Pacific Ltd., Napier, New Zealand, [email protected]
Senior Geotechnical Engineer, Tonkin & Taylor Ltd., Auckland, New Zealand, [email protected]
3
Professor, Dept. of Civil & Env. Eng., Brigham Young Univ., USA, [email protected]
4
Professor, School of Civil & Const. Eng., Oregon State Univ., USA, [email protected]
5
Asst. Prof. Michael Olsen, Sch. of Civil & Const. Eng., Oregon State Univ.,USA, [email protected]
2
Test panels for each ground improvement method were constructed at three sites in Christchurch
in areas severely affected by liquefaction (see Wissmann et al., 2015 for more details). The testing
phase comprised pre- and post-improvement Cone Penetration Testing (CPT) and crosshole
pressure wave velocity (VP) and shear wave velocity (VS) testing, vibroseis T-Rex testing, and
blast-induced liquefaction testing. The CPT, crosshole VP and VS testing and T-Rex shake test
results are presented in van Ballegooy et al. (2015a) and Wissmann et al. (2015). The purpose of
this paper is to describe the methodology and results of the blast-induced liquefaction testing,
which was undertaken at only one of the test sites (Site 4, Wissmann et al. 2015).
The aim of the large scale blast testing was to induce liquefaction below the various ground
improvements to obtain a direct comparison of system performance after treatment. Performance
was evaluated in terms of a method’s ability to control structurally damaging liquefaction-induced
differential settlement (and hence increase post-earthquake resilience). While it is recognised that
blast-induced liquefaction is mechanistically different from liquefaction caused by an earthquake,
the scale and behaviour of the liquefied soil was similar enough to allow assessment of the
performance of the tested ground improvements. Three practice blasts were conducted to optimize
the charge sizes and layouts, followed by seven production blasts to test the ground improvement
methods. This paper focusses on the methodology and findings from production blast 5 (PB5);
however, the analysis and conclusions herein are drawn from the results of the entire EQC blast
testing program.
PB5 included 4 test panels: 1) a 1.2 m thick SCR; 2) a 1.2 m thick geogrid-reinforced compacted
GR; 3) a double-row of HSM beams; and 4) Natural Soil (NS). A specific objective of PB5 was to
produce a significant amount of liquefaction ejecta over the blast area to more closely replicate
(relative to the first 4 production blast tests) the observed liquefaction manifestations caused by
the CES. However, the design of the blasting procedure and layout also needed to achieve a
delicate balance between: 1) generating enough energy to induce sufficient liquefaction for a
sustained period of time to reasonably replicate the liquefaction-related ground surface
manifestations caused by earthquakes; 2) minimising excessive vertical acceleration at the ground
surface, which could cause ground heave / lurch and adversely affect settlement measurements
used to compare the relative performance of the ground improvement methods; and 3) minimising
detrimental vibration effects in neighbouring occupied suburbs.
Subsurface conditions, test panel and charge layout and blasting methodology
The test panel locations, associated charge layouts and subsurface investigations are shown in
Figure 1. Ten CPT and three crosshole VP and VS tests were conducted across the test site to
characterize the subsurface conditions. The subsurface soils at the test site consisted of
approximately 2 m of loose silt/sandy silt overlying generally loose to medium dense silty
sand/sand. Relatively clean, predominantly medium dense sand was consistently present below a
depth of about 4 m across the test site. The CPT and crosshole VS data from the test site are shown
in Figure 2. The two highlighted CPT traces encountered siltier soil layers within the test site based
on the Soil Behaviour Type Index (Ic) values. The depth to groundwater was approximately 1.2 m
and fluctuated daily in the order ± 300 mm, in sync with the tidal influence in the nearby river.
The depth to full saturation, based on the results of the crosshole Vp testing, was approximately 2
to 3 m.
Symmetrical charge layouts were used to ease the interpretation of the results. Double rings of
charges were installed in blast casings around each test panel; an inner 10 m diameter ring and
larger outer 15 m diameter ring. Ten blast casings were evenly spaced around the perimeter of
each ring (at approximately 3.1 m centre-to-centre spacing). The outer ring casings were
horizontally offset from the casings on the inner rings in order to achieve a minimum 3 m
horizontal distance between all charges (to avoid charges in adjacent casings from detonating each
other). The blast casings comprised 12 m long, 80 mm diameter PVC pipe installed through steel
casing using rotary-sonic drilling techniques. The casing was vibrated out of the ground to collapse
the borehole and fill the annulus between the borehole wall and casing. The annulus in the upper
2 m of each hole was filled with cement-bentonite grout seal. Each blast casing contained three
levels or “decks” of gelignite charges as shown in Figure 2. A total of 396.8 kg of explosives were
distributed around the 4 test panels. Angular roading chip gravel (approximately 5 mm dia.) was
used to fill each blast casing up to the depth of the base of the bottom charge, then around the
charges and up to the ground surface. The tops of the casings were covered with blast mats and
sand bags.
Figure 1. Test panel locations and subsurface investigation, instrumentation and charge layouts.
To reduce potential vibration effects on nearby structures, three detonation sequences were
planned in order to spread out the release of energy over a 30 s period. The concept was to detonate
the bottom and middle decks of the inner blast rings first in order to liquefy the soil beneath the
test panels, followed in rapid succession by detonation of the bottom and middle decks of the outer
blast rings to further liquefy the surrounding soil in order to maintain the excess pore water
pressure ratio, ru, and increase the duration of liquefaction. The top decks of charges in the inner
and outer rings were detonated last to liquefy the shallow soil layers and fracture the crust.
Figure 2. CPT and Crosshole VS data from test site and locations of PPTs and gelignite charges.
Test Panel Surcharge and Instrumentation and Lidar Ground Surveys
Instrumentation of the test panels included GE-Druck UNIK 5000 Pore Pressure Transducers
(PPTs), vertical settlement profilometers (Sondex Settlement System by Slope Indicator) and
surface and subsurface geophones. Only the results from the PPTs and profilometers are discussed
in this paper. The locations of the instrumentation in each test panel is shown in Figure 1 and the
depth of the PPTs are shown in Figure 2. Seven PPTs were installed in each test panel around an
approximately 3 m diameter instrumentation ring located at the centre of the panel, at depths of
approximately 2, 3, 4, 5, 7, 9 and 11 m below the ground surface. Two profilometers were installed
on opposite sides of the instrumentation ring to assess where in the soil profile the liquefactioninduced settlement occurred. The profilometers were installed to a depth of 14 m. The minimum
horizontal distance between the profilometers and the PPTs was 1 m.
A surcharge was placed on each test panel to simulate an applied load. The surcharge comprised
four 20 mm x 1.2 m x 2.4 m steel plates set side by side in a rectangular pattern with two 1 m3
concrete blocks placed on top of each plate. The plates were used to evenly distribute the weight
of the concrete blocks, and a 100 mm gap was left between adjacent plates to allow them to move
independently during post-blast settlement. The total weight of the plates/blocks divided by a
2.4 x 5.1 m area equates to about a 16 kPa surcharge. The relative elevations of the four corners
and centre of each steel plate were measured before and after blasting using an optical level
mounted on a tripod located well outside of the test area.
Ground-based Lidar (GBL) and total station surveys were conducted before and after each blast
test to capture the post-blast ground deformations. Each survey consisted of 8 to 15 setups
distributed across the site to minimize occlusions from obstructions. The relative accuracy of the
surveys was estimated as +/-10 mm (3D root mean squared). Digital Elevation Models (DEMs)
with 2 cm resolution were generated from the pre- and post-blast surveys and differenced to
analyse the spatial variability of the settlement. Total station surveys captured stable control points,
lidar reference targets, sand boils, and profilometers elevations, to confirm whether the entire
profilometer had moved, due to liquefaction beneath it, as had happened in earlier tests.
Results of the Blast Testing
During the initial detonation sequence (of the three planned), a detonation wire to one of the
charges in the second sequence was damaged. This tripped a safety switch and prevented firing of
the second and third sequences immediately after the first sequence. The second sequence
detonated 4 minutes after the detonation of the first sequence. By the time the electronics for the
third sequence (i.e., the top deck of charges) were reset and ready for firing, 90 minutes had elapsed
from the time of the first blast. Detonation of the third sequence was deferred to the following day
to allow sufficient time to complete a round of post-blast instrumentation monitoring and lidar
survey to measure the liquefaction related ground surface subsidence caused by the detonation of
the lower and middle decks of the charges. The remainder of this discussion focusses on the results
from the first two detonations. The two blasts produced significant liquefaction ejecta and ground
surface settlement; both within and beyond the blast rings. Figure 3 presents plots of ru with time
for all PPTs located beneath the four test panels. A ru value of 1 (full liquefaction) was reached in
all but the two shallowest PPTs (2.3 m deep) located in the NS and SCR test panels. These PPT
were located within siltier soils and the higher fines content may have inhibited the full build-up
of excess pore pressure. Values of ru were computed assuming a bulk saturated soil density of 18
kN/m3. The excess pore water pressure measured in the shallower PPTs increased during the
dissipation phase (post blasting), indicating an upward flow of water from the underlying liquefied
soils. Figure 3 indicates that ru values at greater depths decreased more rapidly than those at
shallower depths, also indicating an upward flow of water.
Approximately two minutes after the second blast, water and sand (liquefaction ejecta) flowed out
of the ground at discrete locations for up to 30 minutes. The mapped locations of sand ejecta are
shown in Figure 4. The duration of surface flow was consistent with the duration of sustained
excess pore pressure measured by the PPTs. Notably larger amounts of water ejecta and some sand
ejecta occurred on the southern end of the blast area (near the HSM beam test panel). In
comparison, there was notably less water ejecta and very little sand ejecta across the northern
portion of the blast area (around the SCR test panel). The ejected fine grey sand was typical of the
soil material encountered below a depth of 2 to 3 m across the entire test site, and was the same
(in terms of colour, texture and grain size) as the soil ejected during the CES.
Figure 4 shows the change in ground surface elevation after the blast test, along with summary
tables of the 15th, 50th and 85th percentiles of settlement values for each test panel (7 x 7 m area)
and corresponding central surcharge zone (4 x 4 m area). The settlement patterns are relatively
symmetrical for 3 of the 4 test panels; ranging from little settlement at a horizontal distance of 5
m from the outer perimeter of blast casings, to about 130 to 250 mm in the centre of the test panels.
The total settlement of the SCR test panel was less than that of the other three test panels (Figure
4). The highlighted red trace of Ic values for CPT 36410 in Figure 2 indicates that a layer of silty
to very silty soil (approximately 1 m thick) was present at a depth of about 6.5 m beneath the SCR
test panel. In general, the soils beneath the SCR test panel also contain more silt than those beneath
the three adjacent test panels. A reduction in liquefaction-induced settlement beneath the SCR as
a result of the higher silt content helps explain the pattern of settlements shown in Figure 4. The
profilometer settlement data (Figure 3) indicates that for 3 of the 4 test panels, the majority of the
settlement occurred over the depth interval between approximately 1.5 and 7.5 m.
Figure 3. Excess pore pressure ratio ru vs time and vertical settlement data for each test panel.
Analysis of the Results, Discussion, and Conclusions
The following discussion considers test data from all of the production blast tests. The blastinduced liquefaction settlement of the 17 test panels included in the study was compared to the
predicted earthquake-induced one-dimensional post-liquefaction reconsolidation settlement (SV1D)
for the 25, 100 and 500 year return period design ground motions as specified in the MBIE (2012)
guidance, which represent a M 7.5 earthquake and peak ground accelerations (PGAs) of 0.13g,
0.20g and 0.34g, respectively. This comparison provides insight into the equivalent level of
earthquake shaking required to generate a liquefaction demand similar to that created by the blasts.
The pre- and post-improvement liquefaction triggering/settlement analyses (Figure 5d) were
performed utilising the CPT-based procedures by Idriss and Boulanger (2008) and Zhang et al.
(2002). Also shown are the measured 15th, 50th and 85th percentiles of the blast-induced total
settlement values for each panel (sampled from the 7 x 7 m area in the centre of each blast ring).
It can be seen that the measured settlements resulting from PB5 are, in general, larger than the
settlement (SV1D) predicted to result from the 500 year return period earthquake scenario.
Some of the ground improvement test panels had greater total settlement than others due to the
variability of soil conditions across the test site and the resulting spatial variation of the
development of excess pore water pressures induced during blast testing. Some test panels
performed well in limiting the differential settlement of the supported surcharge even though they
underwent larger total settlements. Likewise, some of the test panels did not perform well in
limiting the distortion of the supported surcharge even though smaller total settlement occurred.
Figure 4. Summary of measured ground surface settlement at each test panel.
Therefore, in order to appropriately evaluate the performance of the ground improvements under
blast-induced liquefaction, it was necessary to express the differential settlement of the test panels
in terms of flexural distortion; specifically, the deviation from planar tilt. To do this, a best fit plane
was fitted to the surface settlement profile over a 4 x 4 m area in the centre of each test panel with
the tilt component removed from the differential settlement. The tilt-removed differential
settlement of the test panels was calculated as the 85th percentile of the measured settlement with
the tilt removed minus the 15th percentile of the measured settlement with the tilt removed.
The calculated differential settlement and differential settlement with tilt removed for each test
panel is plotted against the total measured blast-induced settlement in Figures 5a and 5b (yellow
dots represent results from PB5). Also plotted (Figure 5c) is the differential settlement with tilt
removed vs differential settlement with tilt. Figure 5a shows that as the amount of measured blastinduced total settlement increased, the differential settlement also increased. This is an expected
outcome; however, the ratio of differential settlement to total settlement ranged between 0.2 and
0.5.
Figure 5. (a) Differential settlement vs total settlement, (b & c) differential settlement with tilt
removed vs total settlement and differential settlement, respectively, and (c) liquefaction
settlement compared with predicted CPT-based volumetric strain calculated settlement (SV1D).
Figures 5b and 5c show that the test panels with relatively stiffer shallow ground improvements
(i.e., stiffer non-liquefiable crusts measured by the crosshole VS testing), the amount of differential
settlement with tilt removed was notably less than for those with relatively less stiff improvements.
These results indicate that the stiffness of the surface crust should be considered in addition to the
thickness as originally proposed by Ishihara (1985).
In summary, the controlled use of large-scale blast testing was found to be a useful tool to assess
the performance of various shallow ground improvements in mitigating the effects of liquefaction.
The blasting was shown to generally replicate the liquefaction demand predicted to occur as a
result of the target levels of earthquake ground shaking. This was demonstrated through the use of
a robust instrumentation program to measure excess pore pressure development, ground surface
settlement and layer specific settlement within the soil profile.
Acknowledgments
Commission. This support is gratefully acknowledged. The authors would also wish to
acknowledge H. Cowan and K. Yamabe from the Earthquake Commission. The authors appreciate
the assistance of the numerous individuals involved in the testing.
References
Hunter, R., van Ballegooy, S., Leeves, J., & Donnelly, T. (2015). Development of horizontal soil mixed beams as a
shallow ground improvement method beneath existing houses. Proceedings of the 12th Australia New
Zealand Conference on Geomechanics (pp. 650-657). Wellington, New Zealand: NZGS and AGS.
Idriss, I. M., & Boulanger, R. W. (2008). Soil liquefaction during earthquakes. Earthquake Engineering Research
Institute. Oakland, CA: Monograph MNO-12, pp. 261.
on Soil Mechanics and Foundation Engineering, 1, pp. 321-376. San Francisco.
MBIE. (2012). Repairing and rebuilding houses affected by the Canterbury earthquakes. Ministry of Business,
Innovation and Employment. December 2012. Retrieved from http://www.dbh.govt.nz/guidance-on-repairsafter-earthquake
van Ballegooy, S., Roberts, J., Stokoe, K., Cox, B., Wentz, F., & Hwang, S. (2015a). Large scale testing of ground
improvements using controlled, dynamic staged loading with T-Rex. Proceedings of the 6th International
van Ballegooy, S., Wentz, F., Stokoe, K., Cox, B., Rollins, K., Ashford, S., & Olsen, M. (2015b). Christchurch Ground
improvement as a liquefaction mitigation method in sandy and silty soils. Proceedings of the 6th
International Conference on Earthquake Geotechnical Engineering. Christchurch, New Zealand: ISSMGE.
Zhang, G., Robertson, P., & Brachman, R. (2002). Estimating liquefaction-induced ground settlements from CPT for
level ground. Canadian Geotechnical Journal, 39(5), 1168-1180.
80
Utilizing direct-push crosshole testing
to assess the effectiveness of soil
stiffening caused by installation of stone
columns and Rammed Aggregate Piers
L. M. Wotherspoon, B.R. Cox, K.R. Stokoe II, D.J. Ashfield, R.A. Phillips
81
1-4 November 2015
Utilizing direct-push crosshole testing to assess the effectiveness of soil
stiffening caused by installation of stone columns and Rammed
Aggregate Piers
L. M. Wotherspoon1, B.R. Cox2, K.H. Stokoe II3, D.J. Ashfield4, R.A. Phillips5
ABSTRACT
This paper outlines a methodology for the assessment of the effectiveness of soil stiffening caused
by installation of shallow vibro-replacement stone columns (SC) and Rammed Aggregate PiersTM
(RAP) in Christchurch, New Zealand for liquefaction mitigation. Direct-push crosshole tests were
performed before and after ground improvement using custom-built cone sensors to measure
compression wave (Vp) and shear wave velocity (Vs). Full scale case study results presented here
and in other studies highlight the usefulness of the crosshole technique, as it can reliably demonstrate
the stiffening effect of the composite soil-improvement mass. Crosshole testing showed that the SC
and RAP ground improvement techniques seem to develop composite stiffness characteristics of the
improved zone through two quite different mechanisms. Vibro-replacement SCs at typically-used
area replacement ratios (ARRs) appear to provide greater stiffening of the native clean sand soils
within the improved zone compared to the RAP at their typically-used ARRs. However, despite less
stiffening of the native soil within the improved zone, RAPs appear to be stiffer discrete inclusions
than the SCs, as evidenced by their high composite Vs values, which compensates for the reduced
stiffening of the native clean sand soils.
Introduction
A full-scale field testing program was designed to assess the effectiveness of soil stiffening caused
by installation of shallow vibro-replacement stone columns (SC) and Rammed Aggregate PiersTM
(RAP) for liquefaction mitigation in Christchurch, New Zealand (Tonkin & Taylor Ltd 2014).
These ground improvement techniques were installed in soils ranging from clean sands to sandy
silts at a number of sites across the city. To assess the stiffening effect, compression wave (Vp) and
shear wave (Vs) velocity measurements were made throughout the depth of the improved zone
using direct-push crosshole tests with custom-built cone sensors containing a three-dimensional
geophone array. Where possible, tests were performed before ground improvement to characterise
the properties of the virgin soil. After ground improvement installation, testing was carried out to
characterise the properties of the soil within the improved zone (i.e. between SC/RAPs), and the
1
Research Fellow, Civil & Environmental Engineering, University of Auckland, Auckland, NZ,
[email protected]
2
Assistant Professor, Civil, Architectural & Environmental Engineering , University of Texas, Austin, USA,
[email protected]
3
Professor, Civil, Architectural & Environmental Engineering , University of Texas, Austin, USA,
[email protected]
4
Tonkin & Taylor Ltd, Christchurch, NZ, [email protected]
5
Tonkin & Taylor Ltd, Christchurch, NZ, [email protected]
composite properties of the soil and SC/RAP by testing across representative individual SC/RAPs.
This paper provides a detailed outline of the direct-push crosshole testing methodology and data
interpretation approach used during this study. Results from two case study sites are then presented
for stone column and RAP installation in clean sand sites. We focus on the stiffening effect of
these ground improvement methods, with small strain shear modulus GMAX = Vs2ρ (where ρ =
density). Other potential beneficial effects of these methods for liquefaction mitigation, such as
drainage or increased lateral stress, are not discussed herein.
Ground Improvement Methods
All stone column ground improvements tested as part of this research project were constructed
using a dry bottom feed method and river gravel-derived aggregate with at least two broken faces.
A vibroflot is first vibrated down to the desired depth of improvement, and then compressed air
feeds the stone though the vibratory probe out through its base. As the vibroflot is partially
withdrawn, aggregate is delivered into the resulting void and the weight of the vibroflot is applied
onto the aggregate in combination with a vibratory load. This process is performed in stages as the
probe is progressively withdrawn towards the ground surface to form a column. RAP are also
constructed by initially vibrating a mandrel into the soil to the desired depth. The mandrel is
partially withdrawn and aggregate is then fed down through the mandrel into the resulting void.
Thin layers of aggregate are then compacted using a patented process involving repeated impacts.
This process is performed in stages as layers are compacted and the mandrel is withdrawn towards
the ground surface to form the RAP. Both methods result in densification of both the aggregate
and the surrounding soil.
Direct Push Crosshole Testing
Direct-push crosshole tests were performed to determine Vp and Vs as a function of depth using
custom-built cone sensors containing a three-dimensional geophone array designed and
constructed at the University of Texas. The geophones are housed in a stainless steel cone chassis
with dimensions similar to a typical cone penetrometer test (CPT) tip. A source and a receiver
sensor were advanced separately into the ground to the same depth using standard CPT rods and
two small-scale cone penetrometer rigs. Data was acquired using a Data Physics ACE Quattro
dynamic signal analyser connected to a laptop.
The test setup is shown schematically in Figure 1a, with the horizontal spacing between the source
and receiver rods ranging from 1.5 – 1.9 m. This distance was defined based on the diameter of
the column being tested, using a 500 mm offset on each side of the SC/RAP diameter in an effort
to avoid areas of bulging on the side of the SC/RAP that would hinder advancement of the rods.
At each test location the sensors were initially pushed to a depth of 0.4 m below ground level and
the first test performed. The sensors were then both advanced at 200 mm intervals to determine Vp
and Vs down to at least 1 m below the depth of SC/RAP installation. CPT operators continually
monitored the rods to maintain their verticality during testing.
At each depth testing was performed using a hammer impact source applied to the top of the source
rod, with three separate tests performed at each depth and stacked to increase signal-to-noise ratio.
This impact develops compression waves (P-waves) that travel down the length of the source rod
to its cone tip. The vertically oriented sensor (SV) at the bottom of the source rod (Figure 1a) was
used to trigger the data acquisition system, eliminating the need to determine the P-wave travel
time from the source impact at the top of the rod to the wave arrival at the cone tip. The P-wave
reaches the end of the rod and creates both radially propagating P-waves and horizontally
propagating, vertically polarised shear waves (Shv-waves) at the base of the source that follow the
path shown in Figure 1a. Shv-waves, referred to as simply S-waves hereafter, were detected by the
vertically oriented sensor in the receiver rod (RV). In a similar fashion, P-waves were detected by
the horizontal geophone in the receiver rod that was oriented in line with the travel path (RH1).
The other horizontally aligned geophone (RH2) is not used in these analyses.
Tests were performed across each SC/RAP to characterise the composite properties of the soil and
the SC/RAP, and in-between SC/RAPs to characterise the properties of the soil within the
improved zone (Figure 1b). In-between tests were positioned in order to characterise the soil that
was furthest away from the SC/RAP elements, representing the soil in the improved zone that was
likely to be least affected by the installation of these elements (i.e. the least improved soil). Where
possible, tests were performed prior to ground improvement or outside the improved zone to
characterise the virgin (or unimproved) soil properties. Similar spacing between source and
receiver rods were used for the composite, improved soil zone and virgin measurements at each
test location.
Source rod (S)
Receiver rod (R)
Between
Impact
1.5-1.9 m
Composite
S
S
P-wave
R
R
RV
SV
SH1
P- & S-wave
RH1
RH2
SH2
Stone column/RAP
Stone column/RAP
(a)
(b)
Figure 1. Schematic of the direct push crosshole testing method a) elevation view, b) plan view
Calibration Method
Throughout the field testing programme, calibration testing was carried out to correct for the
trigger calibration time between the source and receiver rods. The source and receiver rods were
tightly clamped together above ground using hose clamps at the top and the bottom of the rod. The
RH1 and SH1 geophones in the source and receiver rods were aligned, representing the alignment
during field testing with a zero separation distance. A hammer impact was applied to the top of the
source rod and the waveform recorded with the same data acquisition parameters used in field
testing. Compressive wave (RH1) and shear wave (RV) geophone readings were used to define
the trigger calibration time (tT) for each sensor orientation. Typically, the calibration time was
negative, meaning that by the time the data acquisition was triggered the vibration had already
been detected by the receiver geophones. This type of calibration is important when utilizing
source-to-receiver crosshole rather than receiver-to-receiver crosshole.
Data Interpretation
The stacked/averaged waveform was used to identify the travel time (tRAW) for first arrivals of
direct compressive and shear waves at each depth. Waterfall plots that combine traces from each
depth into a single plot were used to more easily identify trends with depth, as indicated for the
example data in Figure 2. The picked/interpreted direct P-wave first arrivals are shown in Figure
2a, while Figure 2b shows both the picked first arrivals of the direct P-wave from Figure 2a
(crosses) and the picked direct S-waves (circles). Once the first arrivals of the direct waves were
picked, the velocity (V) of each wave type was simply defined using:
V = SR / tC
tC = tRAW + tT
(1)
where tC is the corrected travel time and SR is the spacing between source and receiver rods.
Picking of the first P-wave arrivals were fairly straightforward, with the first clear departure/break
used at each depth. Once saturated soils were encountered, the P-wave arrivals were characterised
by a high frequency signal, which is evident in Figure 2a below a depth of 3 m. The main issue to
account for in the choice of first arrivals was the effect of P-wave refraction along the saturation
depth or other stiff layers. However, refraction effects were minor due to the combination of the
short travel distance and small depth increments.
Picking of S-wave first arrivals was complicated by a number of factors, and the first pulse in each
trace of the waterfall plots was not always representative of the direct horizontal S-wave travel
path. Typically the highest amplitude pulse with the correct polarity (in this case, a positive
break/departure) in the initial stages of each trace was assessed as the most likely direct wave
arrival. The main sources of waves arriving prior to the direct horizontal S-wave were the result
of: (1) A strong P-wave signal arriving near the time of the S-wave may be picked up by the Swave sensor (RV), which usually only occurred in unsaturated soils where the P-wave velocity
was rather slow. This effect is clearly shown in the upper 1.6 m in Figure 2b, with small amplitude
signals that start at the point of the P-wave pick at the corresponding depth; (2) If the test depth
was near the boundary of a layer with a higher velocity, refracted waves may travel along the layer
boundary and arrive at the receiver prior to the direct wave. Usually in this case the amplitude of
the refracted wave is less than the direct wave. This effect was often evident below the base of the
improved zone, with the amplitude of this refracted wave reducing with depth below this zone. An
example of this is shown in Figure 2b below a depth of 5.8 m. Here the stronger amplitude pulse
is chosen rather than the smaller amplitude earlier arrivals.
One of the main shortcomings of the receivers used in these trial tests was the inability to measure
the subsurface deviation of the source and receiver rods during testing. As stated above,
precautions were taken to assure the CPT rods were initially vertical and deviated as little as
possible in the soft soils during testing. Furthermore, the total pushing depths were typically less
than 8 m, limiting the distance the rods could drift/deflect over. However, on occasion trends in
the Vp and Vs with depth were used to identify and account for any deviation issues. Newer versions
of the sensors, developed after the conclusion of this testing, have inclinometers installed adjacent
to the geophones to allow tracking of the location of the tip of each rod.
(a)
Travel Time (ms)
2
4
6
8
0
-1
-1
-2
-2
-3
-3
-4
(b)
Travel Time (ms)
10
Depth (m)
Depth (m)
0
0
0
5
10
15
20
-4
-5
-5
-6
-6
-7
-7
-8
-8
Figure 2. Waterfall plot a) In-line horizontal geophone records used to define compression wave
velocity with P-wave picks (crosses); b) Vertical geophone records used to define shear wave
velocity with S-wave picks (circles) and P-wave picks from a) (crosses). This waterfall plot is
not related to the results presented in Figure 3 or 4.
Case Studies
The details of two case studies that are presented in this paper are outlined in Table 1. A stone
column and a RAP installation are presented, both in predominantly clean sands. The nominal
diameter of the SC/RAP, the design spacing, pattern and installation depth, and the average area
replacement ratio (ARR, equal to the ratio of the SC/RAP cross sectional area and the improved
area) is summarised for each site. Note that the SCs were installed with ARRs approximately twice
that used at the RAP sites, with these ARR values fairly representative of recent installations in
clean sands in Christchurch. For each case study composite tests are denoted as C, between tests
as B, and virgin soil tests as V, with the number of each test also indicated (i.e. C1 for composite
test 1).
Table 1. Summary of case study ground improvement details
Site Method
A
SC
B
RAP
Soils
Clean sand
Clean sand
Diameter
0.68 m
0.60 m
Spacing
1.70 m c/c
2.00 m c/c
Depth
4.0 m
4.0 m
ARR
15.1%
8.5%
Pattern
Triangle
Triangle
Stone Columns
Figure 3 summarises the soil profile data from crosshole (Vp and Vs) and pre-improvement CPT
testing (tip resistance qc and soil behaviour type index Ic (Robertson & Wride 1998)) at Site A,
with fines content (FC, % passing 0.075 μm sieve) from laboratory testing of 1-4% in the relatively
clean sand profile. The horizontal spacing between source and receiver was approximately 1.7 m
at this location, therefore for composite tests, approximately 40% of the travel path was through
the SCs.
There is a clear increase in Vs within and below the column installation depth for all tests (both
across/composite C and between B) compared to the virgin soil measurements. Surprisingly, there
is little difference in the C (composite) and B (between) Vs measurements, suggesting that the
stiffness of the stone columns elements are not too dissimilar from the stiffness of the surrounding
improved soil. Overall, the SC installation increased Vs by approximately 50 m/s throughout the
improved zone, a 70-100% increase in GMAX. The Vp measurements are dominated by the degree
of soil saturation, and indicate a slight reduction in post-improvement saturation over most of the
improved zone. This reduction in the degree of saturation was a characteristic evident at a number
of other sites where SCs were installed (Tonkin & Taylor 2014). At Site A testing was carried out
18 days after SC installation.
Pre- and post-improvement CPT soundings were also able to assess the level of improvement of
the soil between the SCs across a range of soil types (Tonkin & Taylor Ltd 2014). However, CPT
soundings cannot be used to infer the composite soil-column stiffness.
(a)
0
Depth (m )
Silt Mix
Sand Mix
-4
WT
V1
C1
B1
C2
B2
(d)
Sand
-3
WT
(c)
Gravelly
-2
Improved Zone
-1
(b)
-5
CPT1
-6
-7
0
600
1200
1800
V (m /s)
P
0
100
200
V (m /s)
S
300
0
5
10 15 20
q (MPa)
C
1
2
3
I
C
Figure 3. Site A characteristics: a) P-wave velocity; b) S-wave velocity; c) CPT tip resistance; d)
soil behaviour type index
Rammed Aggregate Piers
Figure 4 summarises the soil profile data from crosshole and pre-improvement CPT testing at Site
B, with fines content from laboratory testing of <5% in the relatively clean sand profile below 1
m depth. The horizontal spacing between source and receiver was approximately 1.6 m at this
location, therefore for composite tests approximately 38% of the travel path was through the RAPs
(comparable to the 40% travel path for the SC test summarised here).
Below 1 m depth, the profile is relatively clean sand, and there is a clear difference in the C and B
Vs measurements. Soil between the RAP units showed Vs increases of approximately 20 m/s
compared to the virgin soil measurements throughout the clean sand, an increase in GMAX of 2535%. Across the RAP there was a much larger increase in Vs of between 40-115 m/s in the clean
sand, a 180-220% increase in GMAX. This suggests very effective stiffening of the aggregate in the
RAP, and modest stiffening of the soil between RAPs. Consistent with the SC results, there was
also an increase in Vs beneath the improved zone, with this improvement shown to decrease with
depth. In this case, the Vp measurements do not indicate much of a change in the degree of soil
saturation in the improved zone before and after installation, a characteristic that was evident at
other sites where RAP were installed (Wissman et al. 2015).
(a)
0
(b)
(c)
Depth (m )
-5
0
Silt Mix
-4
Sand Mix
V1
C1
B1
Sand
-3
Improved Zone
-2
Gravelly
WT
WT
-1
(d)
CPT1
600
1200
V (m /s)
P
1800
0
100
200
V (m /s)
S
300
0
5
10 15 20
q (MPa)
C
1
2
3
I
C
Figure 4. Site B characteristics: a) P-wave velocity; b) S-wave velocity; c) CPT tip resistance; d)
soil behaviour type index
Verification Testing and Quality Assurance
This and other studies have shown that generally, both vibro-replacement stone columns and RAPs
are effective at increasing the Vs of cleaner sand sites, with the stiffening effect of stone columns
and RAPs reducing as the silt content of the soil profile increased (Tonkin & Taylor Ltd 2014, van
Ballegooy et al. 2015, Wissman et al. 2015). RAP have been found to be effective in silty sand
deposits, however at these sites a high degree of quality assurance (QA) and verification testing
was undertaken. In order to ensure that the desired ground improvement characteristics are
achieved, verification testing is important when using SC/RAPs, especially in silty soils, so that
modifications to the methodology can be made (such as increasing the area replacement ratio)
when the desired results are not achieved. While the crosshole method is useful for SC assessment,
the improvement of a SC site may be largely demonstrated by CPT soundings in the soil between
the columns, as this method seems to rely on an increase in the stiffness of the soil surrounding
the SC to provide the desired composite stiffness of the improved zone. Given there appears to be
smaller changes in the properties of the soil surrounding the RAP, crosshole testing may be very
useful, as it can reliably demonstrate the stiffening effect of the composite soil-RAP mass.
Conclusions
This paper outlines the use of direct push crosshole testing to assess the effectiveness of soil
stiffening caused by installation of shallow vibro-replacement stone columns (SC) and Rammed
Aggregate PiersTM (RAP) in Christchurch for liquefaction mitigation. Case study results
highlighted the usefulness of the crosshole technique, as it can reliably demonstrate the stiffening
effect of the composite soil-improvement mass. More traditional post-construction verification
testing methods cannot be used to infer this composite stiffness. Crosshole testing showed that the
SC and RAP ground improvement techniques seem to develop composite stiffness characteristics
of the improved zone through two quite different mechanisms. Very similar composite (i.e., across
columns/RAPs) and between columns/RAPs Vs measurements were evident at the SC sites, while
there was a clear distinction between these measurements at the RAP sites. Thus, the vibroreplacement stone columns at typically installed ARRs appear to provide greater stiffening of the
native clean sand soil within the improved zone compared to the RAP. However, it should be noted
that the SCs were installed with ARRs approximately twice that used at the RAP sites. Despite
showing less stiffening of the native soil within the improved zone, RAPs with typically installed
ARRs appear to be stiffer discrete inclusions than the SCs due to their high composite Vs values,
which may compensate for the reduced effect of the RAP on stiffening of the native clean sand
soils. A lowering of Vp was also identified at some SC sites after installation, indicating reductions
to the degree of saturation, an effect not evident at RAP installation locations. This change in the
degree of saturation over time will be of interest for future study, as it may have an impact on the
triggering of liquefaction.
Acknowledgments
We acknowledge the Canterbury Geotechnical Database for some of the site investigation data
used in this study. EQC, CERA, their data suppliers and their engineers, Tonkin & Taylor, have
no liability to any user of this data or for the consequences of any person relying on them in any
way. This work was partially supported by U.S. National Science Foundation (NSF) grant CMMI1343524. However, all findings and recommendations expressed in this paper are those of the
authors and do not necessarily reflect the views of NSF.
References
Robertson PK & Wride CE (1998). Evaluating cyclic liquefaction potential using the cone penetration test, Canadian
Geotechnical Journal 35: 442-459.
Tonkin & Taylor Ltd (2014). Crosshole seismic testing of stone column ground improvement works, Factual Report,
Tonkin & Taylor Ltd.
Wissman KJ, van Ballegooy, S, Metcalfe, BC, Dismuke JN & Anderson CK (2015). Rammed aggregate pier ground
improvement as a liquefaction mitigation method in sandy and silty soils, 6th International Conference on
Earthquake Geotechnical Engineering, 1-4 November, Christchurch, NZ.
van Ballegooy S, Roberts JN, Stokoe II KH, Cox BR, Wentz FJ & Hwang S (2015). Large-scale testing of shallow
ground improvements using controlled staged-loading with T-Rex, 6th International Conference on Earthquake
Geotechnical Engineering, 1-4 November, Christchurch, NZ.
82
83
84

Appendix B - EQC Earthquake Commission

Transcription

Similar documents

Angiosperm Latin Name: Cyperus involucratus Common Name

Angiosperm Latin Name: Clivia miniata Common Name: Kaffir Lily

Technical Sheet - Microbes Biosciences

HSM T 15007 ROTARY TRIMMER

Angiosperm Latin Name: Euphorbia ingens Common Name

Project Profile: Horasan Kurtozu Sarıkamış Highway

PDF view - Woody Plants Database

Revel Casino News 2013

Angiosperm Latin Name: Callisia fragrans Common Name: Basket