Presented at the 25th USSD Annual Meeting and Conference, 2005

DEEP SOIL MIXING FOR SEISMIC REMEDIATION

OF THE
CLEMSON UPPER AND LOWER DIVERSION DAMS

Lee Wooten, P.E.1 Ben Foreman2

ABSTRACT

The Clemson Upper and Lower Diversion Dams are similar, random earthfill dams
constructed in 1960-1961 to protect lands and facilities at Clemson University. Because
of a loose silty sand / sandy silt alluvial foundation layer, the dams were susceptible to
downstream liquefaction slope failures during or following the design earthquake (amax =
0.2g). The remedial design to prevent seismic instability consisted of deep-soil mixing
along the downstream berm of each dam. The soil-mix elements were designed to form
transverse shear walls oriented perpendicular to the dam axis which would carry the
seismic loads in lieu of the weak soil layer and a longitudinal wall oriented parallel to the
dam axis at the upstream end of the transverse walls which would assist in resisting the
movement of soils between the transverse walls.

Design of the soil-mix remediation addressed the feasibility of soil-mix, slope stability,
dynamic behavior of the dam, anchorage of the soil mix walls, alternate failure surfaces,
continuity of drainage, and quality assurance. Construction demonstrated that the use of
soil mix shear walls to provide reinforcing of the loose soil against seismic deformations
and liquefaction shear slides can be feasible and cost effective. Construction also
provided lessons for future soil mix projects.

INTRODUCTION

Dams

The Clemson Upper and Lower Diversion Dams (referred to in this paper as the Upper
and Lower Dams or as the Clemson Dams) were constructed in 1960 and 1961 as part of
the Hartwell Reservoir project to protect lands and facilities at Clemson University. The
Upper and Lower Dams are located on the perimeter of Hartwell Lake in Pickens County,
South Carolina. The dams restrict the Hartwell Reservoir impoundment from one section
of the Seneca River basin adjacent to Clemson University and the Town of Clemson.

The Lower and Upper Dams are very similar, random earthfill dams with an inclined
chimney drain and horizontal drainage blanket to collect and filter seepage. The Lower
and Upper Dams have lengths of about 3,000 and 2,100 feet, respectively, and a
maximum height above the valley of about 80 feet. Both dams have had a history of
seepage-related problems. Several remedial measures were constructed to mitigate these
1
Design Division Manager, GEI Consultants, Inc., 1021 Main Street, Winchester, MA 01890
2
Chief, Soils Section and Project Engineer, U.S. Army Corps of Engineers, Savannah District, 100 West
Oglethorpe Avenue, Savannah, GA 31401
 -2-

seepage problems, culminating in the installation of a concrete cutoff wall across the
upstream side of the crest of each dam in the early 1980s, which essentially eliminated
seepage concerns. Figures 1 and 2 show the plan of the Upper Dam and a simplified
cross section characteristic of both dams.

Figure 1. Plan - Upper Diversion Dam

Figure 2. Simplified Cross Section of Clemson Diversion Dams

Susceptibility To Liquefaction Slope Failures

The USACE performed numerous subsurface investigations in the early 1980’s at the
Clemson Dams because of seepage problems. These investigations documented the
presence of a loose silty sand / sandy silt alluvial foundation layer. This alluvial deposit
varied in thickness from 7 to 28 feet. Most N-values in the alluvium ranged between 3
and 30 blows per foot.
 -3-

Due to the low range of the N-values in the alluvium, the USACE initiated investigations
into the potential for liquefaction slope failures of the Clemson Dams. The USACE and
GEI undertook a series of investigations and analyses that 1) estimated undrained steady-
state strengths (Sus) for zones of the alluvium, 2) estimated stability factors of safety less
than 1.0 for the 0.2g design seismic event for both upstream and downstream slopes using
these steady-state strengths, 3) used Newmark-type sliding analyses to determine that a
seismic event would cause enough strain to reach these steady-state strengths and create
instability in the downstream direction only, and 4) used a two-dimensional finite
element model to predict locations where accumulated strains exceeded triggering strains,
confirming instability for the downstream slopes and suitable stability for the upstream
slopes for the post-seismic condition. Prof. Liam Finn of the University of British
Columbia performed the two-dimensional finite element modeling using the computer
model TARA. These analyses are described in Wooten, et al. (2003 and 2004). As
background for these analyses, we refer the reader to Poulos (1981) for the steady-state
model of soil behavior, Castro et al. (1989) for application of the steady-state model to an
embankment dam analysis (Lower San Fernando), Castro (1994) for the triggering
analysis methodology, and Finn et al. (1986) for the TARA dynamic finite element
model.

All the different types of analyses used to estimate the potential seismic behavior of the
Clemson Dams confirmed that, upon the occurrence of the design seismic event, a failure
of the downstream sections of the dams would be likely. Consequently, the dams would
no longer be able to retain the normal reservoir. Thus, remediation of the downstream
sections of the dams was required for the dams to withstand the design earthquake, but it
was concluded there was no need to remediate the upstream sections of the dams.
Because of these conclusions of seismic instability, the USACE proceeded with remedial
design.

Conceptual Design

Based on the results of the previous evaluations, the USACE engaged GEI to develop
conceptual designs for remedial measures to prevent a liquefaction failure and excessive
deformations of the downstream sections of the dams. Conceptual designs were
developed for four remedial techniques: 1) jet grouting, 2) deep soil mixing, 3) stone
columns, and 4) excavation and replacement. The jet grouting, deep soil mixing, and
stone columns all involved methods to increase the shear strength in the alluvial layer
under the downstream berm. The excavation and replacement option would replace the
loose alluvial material under the downstream berm with compacted fill. Costs estimated
for these options are listed below.
 -4-

Option Estimated Cost

Jet Grouting $10,700,000 to $14,100,000
Deep Soil Mixing $8,200,000 to $9,800,000
Stone Columns $14,200,000 to $14,700,000
Excavation and Replacement $9,800,000 to $10,900,000

The remediation alternatives listed above were all technically feasible. The two most
attractive alternatives were deep soil mixing, because it had the lowest estimated costs,
and excavation and replacement because it would be the simplest to confirm that the
desired results had been achieved and because modeling its performance would be more
straightforward. Other factors that affected the attractiveness of each alternative included
the risk of having a stability failure or damage to the dams during construction,
construction traffic impacts on the roads and Clemson University, impacts on existing
structures due to increasing the size of the stability berm, impacts of weather on the
different operations, and the aesthetics of each alternative. The deep soil-mixing
alternative was selected based on an evaluation of all criteria.

DESIGN

Soil Mix Reinforcement Layout

The soil mix reinforcement design consisted of a series of soil-mix shear walls installed
through the downstream berm of the dams. Soil mix columns would be overlapped to
create nominal 3-foot-thick, 50-foot-long transverse shear walls oriented perpendicular to
the axis of the dam. The walls would be spaced at 15.5-foot centers and would begin
approximately 130 to 140 feet downstream of the centerline of each dam. The shear
walls provide the major reinforcement of the downstream slope to resist both seismic
deformation of the alluvium and a downstream slope failure during and following a
seismic event. The transverse walls would be embedded into the overlying berms and
into the underlying sand and gravel layer or weathered rock. A 3-foot-wide longitudinal
wall parallel to the dam axis would also be constructed at the upstream end of the
transverse walls. The longitudinal wall would prevent softened alluvium from squeezing
or flowing between the transverse walls. Figure 3 shows the plan layout for the soil mix
walls at the Upper Dam. We describe the major design analyses in the following
sections.
 -5-

Figure 3. Layout of Soil Mix Shear Walls, Upper Diversion Dam

The design concept for installation called for the existing berm to be excavated in
sections to create a work platform about 8 feet above the elevation of the downstream toe
of the dam. Deep soil mixing elements would be installed from the excavated working
surface through the loose alluvium layer. The target 28-day unconfined compressive
strength for the soil-cement mix would be about 400 psi to provide sufficient strength if
the strength in the alluvium were to drop to steady-state values. Figure 4 shows a
schematic cross section of the remediated downstream slope design.

Figure 4. Schematic Cross Section of Remediated Downstream Slope

Dynamic Analysis

After selection of the deep soil-mixing remedial alternative, Dr. Finn performed TARA
finite element analyses to evaluate the performance of the modified section. These
follow-up finite element analyses addressed the dynamic behavior of the remediated dam.
Our major concern was that the remediated stiffer downstream soils might cause
increased strains under the upstream slope to the point of triggering liquefaction of the
upstream slope. We also wanted to estimate the actual strain of the remediated
downstream soils.

The results of the finite element analyses of the modified section of the dam using the
same strength parameters as in previous work indicated that the modified downstream
 -6-

section of the dam developed negligible deformations as a result of the design seismic
event and that the response of the upstream section of the dam was essentially unaffected
by the downstream modifications. The dynamic analyses showed that the soil mix
reinforcement decreased deformation and strain to the point that we did not expect the
alluvium to exceed triggering strains.

Soil Cement Strength Selection and Criteria

We estimated the following shear stresses within a 3-foot-thick soil-mix wall for the
noted conditions:

Condition Soil-Mix Shear

Wall Stress
Static - alluvium strength drops to Sus 82 psi
Static - alluvium does not lose strength, but soil- 60 psi
mix walls take all stress within remediated zone
Dynamic - alluvium does not lose strength and 55 psi
load is shared with soil

Assuming a shear strength of about 166 psi, between ½ to ѿ the unconfined compressive
strength of 400 psi, gave us a factor of safety of about 2 for the conservative Sus case (82
psi) and about 3 for the most likely peak dynamic stresses (55 psi).

We recognized that contractors have a variety of soil-mix equipment and methodologies,

and we wanted to allow contractors the flexibility to optimize soil-mix shear wall
installation for their methods. To do so, we structured the specifications for soil cement
strength (f’sc) as a function of the average wall width (Wa) and wall spacing (S) using
the following equation:

f'sc • 77.4 psi x (S/Wa)

For 3-foot-thick shear walls, spaced 15.5 feet on center, this equation requires a 400-psi
soil cement unconfined compressive strength. We restricted the wall spacing to no more
than 12.5 feet plus the wall width. This formula also provided a mechanism for
estimating a revised wall spacing if soil cement strengths in the field were lower than
required.

Soil Cement Mix Feasibility Evaluation

We collected additional field and laboratory data for final design to 1) establish the extent
of the alluvium layer for installation of the soil-cement shear walls and 2) determine the
feasibility of achieving the design soil-cement strengths in the alluvial soils. The USACE
performed 48 borings at the Clemson Dams, which included SPT sampling to delineate
the alluvium and large diameter spoon and tube sampling to collect bulk samples for the
soil cement mix testing. The laboratory testing included index testing (water content,
 -7-

Atterberg limits, organic content, and grain size analyses) to characterize the alluvium
and a soil-cement mix testing program to evaluate whether or not we could achieve the
compressive strength required in the soil-cement.

Based on observation and index testing, we identified five predominant soil types present
in the alluvial foundation soils of the two dams: 1) silty sand, 2) silty sand/sandy silt, 3)
low to medium plasticity silt, 4) clay, and 5) silty sand/sandy silt with organics. Soil for
batches 2 and 3 were taken from the alluvium at each dam. Soil for batches 1, 4, and 5
were taken from the Upper Dam alluvium only. Each soil type was tested using varying
cement contents (300, 450, and 600 pounds per cubic yard (lb/cy)).

The laboratory soil mix tests demonstrated that soil cement unconfined compressive
strengths in excess of 400 psi could be achieved with the alluvial soils. Figure 5 is an
example of the test results on the low to medium plasticity silt batch samples.
800

700
Peak Compressive Strength (psi)

cement content = 600 lb/cy, w/c = 0.7

600

500

400

300
cement content = 450 lb/cy, w/c = 0.9
200
cement content = 450 lb/cy, w/c = 0.7
100

0
0 7 14 21 28 35 42 49 56 63

Days Cured

Figure 5. Soil-Cement Mix Test Results - Batch 3 - Low-Medium Plasticity Silt

We typically prepared specimens in each batch at two cement contents. Nine tests were
typically performed on each cement content batch, three each at 7 days, 28 days, and 56
days. In general, for a given cement content, the batch mixes resulted in unconfined
strengths that ranged from highest to lowest in batches 2 (silty sand/sandy silt), 5 (silty
sand/sandy silt with organics), 1(silty sand), 3(low to medium plasticity silt), and 4
(clay).

On the basis of the soil mix tests, we concluded that soil cement of suitable strength,
generally in excess of 400 psi, could be produced using the alluvium materials if
sufficient cement was added. The soil mix test results were included in the contract
 -8-

documents; however, we specified that the contractor was responsible for the final mix
design.

Anchorage Design

Anchoring the reinforced soil mix zone into the underlying and overlying layers is
necessary to keep deformations below triggering strains and thus provide a higher factor
of safety (1.6) for the post-earthquake embankment. If slippage occurs at these
interfaces, the alluvium could drop in strength to Sus and the factor of safety will
decrease to about 1.2. Slippage could also result in larger dynamic movements and
potentially greater damage to the embankment.

The anchorage was achieved by extending the soil-mix shear walls into the underlying
sand and gravel/weathered rock about 2 to 4 feet and into the overlying embankment
about 8 feet. We used the TARA models of peak dynamic stresses to estimate loads on
the shear wall anchorage. We estimated that the shear walls would carry about 70% of
the shear stress in the remediated zone based on the element stresses for soil zones
immediately adjacent to the remediated zone. We calculated anchorage resistances using
simple frictional shear models for the sides, top, and bottom of the shear walls and using
a simple pile resistance model from Broms (1964) using three times the passive
resistance at the ends of the walls.

The design anchorage embedments provided a factor of safety of 1.1 against peak
dynamic stresses. A lower factor of safety for the anchorage was selected because we
expected that even if interface slippage did occur, the resulting movements would not be
great enough to cause a loss of soil strength. Furthermore, embankment failure should
not occur even if the soil were to lose strength.

Alternative Failure Surface Analyses

We selected the position of the remediated zone to 1) reinforce the dam and 2) provide
for construction access. We located the wall as far downstream as feasible, away from
the steepest part of the embankment, to reduce excavation difficulties. We checked
embankment steady-state stability for failure surfaces that extended under, over, and
upstream of the final shear wall positions.

Drainage Design

A longitudinal wall was added at the upstream end of the shear walls (see Figures 3 and
4) because of concerns of potential slippage of soil between the shear walls. The
longitudinal wall however would block drainage through the drainage layer on top of the
alluvium. Consequently, we added a filtered seepage collection drain upstream of the
wall. Six-inch-diameter pipes routed through windows in the longitudinal wall were
added to carry seepage flow from the collection drain to downstream outlets.
 -9-

Quality Assurance Design

Soil-cement strengths in soil-mix construction will vary significantly, mostly because of

variations in the base soil (gradation, cohesion, water content, and chemistry), but also
because of variations in mixing methods (speed, time, equipment), slurry water/cement
ratio, and cement or water chemistry. We expected that typical soil-cement cylinder tests
(3- to 6-inch diameter) would be highly variable because of the small scale of the
specimens, especially where lumps of unmixed soil could remain after mixing.

We designed the soil-cement quality assurance program to allow for this expected
variability by:

(1) Requiring twelve 28-day f’sc tests on wet-grab samples for each wall,

(2) Requiring that the average 28-day f’sc for each wall exceed the strength equation,

(3) Allowing only one of the twelve 28-day f’sc tests per wall to fall below Ҁ’s of the
strength equation,

(4) Requiring two cores and twelve strength tests on core samples for walls where the
wet-grab samples did not meet strength criteria, and

(5) Specifying that only the Contracting Officer would determine acceptance of soil-
cement shear walls based on interpretation of the quality assurance data.

We expected that wet-grab sampling would be the most expedient and most economical
way to retrieve samples of the soil-cement. We specified that wet-grab samples be taken
with a device that could take discrete samples at pre-determined depths, could accept up
to 6-inch-size soil lumps, and could provide sufficient volume to produce four 6-inch-
diameter cylinders. We wanted the sampler to be large so that we sampled significant
soil lumps and not just the more fluid slurry matrix. We further specified that soil lumps
be reduced in size to 1 inch and be included in the 6-inch-diameter cylinders.

We expected that coring of the soil-cement would be slower and more expensive and
would produce a smaller diameter test specimen with greater variability than the wet-grab
samples. Consequently, we specified minimum 3-inch-diameter cores only where wet-
grab sample strengths fell below criteria.

To ensure that the contractor’s soil-mix equipment and design mix would meet design
criteria, we required that two test walls be constructed at each dam and that the test wall
soil cement meet specifications prior to construction at a dam.
 -10-

CONSTRUCTION OF SOIL MIX WALLS

Bids

Six contractors bid on the Clemson project. The following two tables summarize the
project bids and bids on soil-cement items.

Clemson Total Bid Summary Production Soil Cement Bid Summary

Government Estimate $7,768,587 Winning Bid $107/yard3
Winning & Low Bid $7,744,657 Low Bid $70/yard3
High Bid $12,592,400 High Bid $199/yard3
Average $9,248,526 Average $126/yard3

The production soil cement bid prices were estimated from the contractors’ bids for soil
cement installation as defined by the volume encompassed by the soil mixing
improvements, including unmixed soil between walls. We multiplied the actual bids by
15.5/3.0 to obtain these estimates. The contractors also provided bids for additional soil-
mix walls on a straight soil cement volume basis as summarized below.

Additional Soil Cement Bid Summary

Winning & Low Bid $220/yard3
High Bid $819/yard3
Average $556/yard3

The USACE selected RAITO Inc. (RAITO) of Crofton, Maryland on the basis of their
bid package. RAITO started construction of the Clemson seismic remediation project in
December 2003 and reached substantial completion of the project in March 2005.

Construction Methods

RAITO proposed and used a different working elevation for installation of the soil mix
walls. The original design called for excavation of the berm down to the top of the soil
mix walls. RAITO constructed the soil mix walls from a level about 8 to 12 feet above
the top of the soil mix walls, thus reducing the cost for site work before and after wall
installation because most of the berm would remain in place. Working from the higher
level did increase the efforts required to construct the walls and to install the upstream
collection drain.

RAITO constructed the soil mix walls using several significant pieces of equipment and a
refined computerized control system. The primary installation equipment consisted of a
crawler base machine equipped with a leader to support and guide the electric top drive
and the six-axis soil mixing shafts. The cutting heads and mixing blades on the shafts
were 3 feet in diameter and could be rotated at either 20 or 40 rpms. The mixing blades
extended over the bottom 10 feet of the shafts and overlapped adjacent mixing columns
by 1 foot resulting in an average wall width of 2.76 feet. Adjacent augers were vertically
offset by 1 foot. The leader was equipped with electronic inclinometers to control the
 -11-

verticality of the soil mixing shafts. Figures 6 and 7 are photos of the deep soil-mix rig
and the mixing shafts.

Figures 6 & 7 – Deep Soil-Mix Rig and Mixing Shafts

Slurry was injected through the hollow stems and the tips of the mixing shafts into the
soil-cement mix. A computer controlled batch plant and six variable speed, positive
displacement pumps with flow monitors supplied cement slurry to the mixing shafts.
Figure 8 shows the batch plant and pumps.

Figure 8 – Soil-Mix Slurry Batch Plant & Pumps

 -12-

RAITO monitored in real time the parameters of depth, shaft rotation speed, penetration
and withdrawal rates, and the slurry injection rate for every vertical segment of each
column. These four parameters together with the average quantity of slurry injected per
each three-foot segment of each column were recorded for submittal. RAITO also
monitored “cable load”, which was the actual weight that the cables supported during the
mixing process, and the electrical load on the motors to evaluate drilling resistance and to
prevent overloading the motors.

Soil Cement Quality Control and Quality Assurance

Prior to the start of work, RAITO took additional soil samples and conducted their own
soil mix test program. Based on their tests and the GEI soil mix test results, RAITO
proposed an initial soil mix of 500 kg/m3 (843 lb/cy). The cement was subsequently
reduced during production to 450 kg/m3 (758 lb/cy) and then to 400 kg/m3 (674 lb/cy)
based on the high strengths obtained in the production soil cement. RAITO used
water/cement ratios that ranged from 0.6 to 0.7.

RAITO’s electronic control and data collection of depth, penetration and withdrawal
rates, and slurry injection rate provided the primary quality control to ensure that enough
cement slurry was distributed suitably and efficiently over the full depth of all soil
elements. Figure 9 shows a portion of a typical installation record that was generated by
RAITO’s automated system.

Figure 9 – Soil-Mix Installation Record

Specific gravity tests were used to ensure that batch plant mixing produced suitable
cement slurry.
 -13-

Testing of cured, as-mixed soil cement samples was the primary quality assurance for the
soil cement. However, the wet-sampling method that we expected to use did not work.
During the installation of the first set of test walls, RAITO tried numerous methods to
obtain wet grab samples for quality assurance with no success. All attempts to advance
and retrieve a sampler within soil mix columns resulted either in loss of or damage to
samplers, refusal in embankment soils, or clogging of the samplers with embankment
soils. The ~16-foot-thick overlying denser embankment soils, even when mixed with
slurry to the surface, either clogged the samplers or provided resistance to advancement
attempts to the point that samplers were damaged or lost.

Due to the inability to obtain wet grab samples, all quality assurance samples were
obtained using full-depth, continuous cores of the shear walls taken at approximately 28
days. RAITO used a drill rig designed especially for soil cement mix sampling to obtain
~2.56-inch diameter core samples. Based on the consistently high strengths of the
samples, the USACE agreed to sample on a frequency of at least one core for each day of
production, typically about two shear walls. Core samples offered the advantages over
wet grab samples of 1) continuity through the entire depth of the walls, thus allowing test
specimens to be selected at various depths, and 2) sampling the in situ, hardened soil
cement rather than a remixed specimen.

The USACE selected six specimens to be tested at 28 days from each core. Based on the
average wall width of 2.76 feet and a 15.5-foot spacing, the average soil cement 28-day
f’sc was required to be at least 435 psi. In addition, no specimen was allowed to have a
28-day f’sc less than 290 psi. All samples were stored and tested on site. Additional
cores were taken if these strength criteria were not met. The USACE accepted soil-mix
walls only if test results and the cores demonstrated that there were no repetitive weak
zones at the same elevations in the walls.

The soil-mix walls installed by RAITO generally met strength criteria. The USACE
accepted the soil cement in all of the 100 shear walls constructed in the Lower Dam and
in all but 27 of the 105 shear walls in the Upper Dam. In the 27 walls for the Upper Dam
where soil cement strengths did not meet criteria, the lower strengths appear to have been
the result of either organic content and/or low pH soils within the alluvium layer. Where
shear walls did not meet soil cement strength requirements, we required RAITO to install
additional adjacent shear wall elements to provide a total shearing resistance equivalent
to that of original design with full strength shear walls.

Penetration/Anchorage Of Soil Mix Shear Walls

Both achieving and judging the penetration of the soil-mix shear walls into the
underlying dense sand and gravel or weathered bedrock proved to be difficult tasks
during production. The ability of the soil-mix equipment to penetrate dense soils was
limited and irregular. The specifications required a 4-foot penetration into the dense sand
and gravel or a 2-foot penetration into the weathered bedrock.
 -14-

We expected that we would be able to judge the increased resistance of the top of the
denser layers using parameters such as cable load and electrical current load on the
motors. However the correlation of low cable load and high electrical current load with
the top of denser layers as determined by borings was inconclusive.

Consequently, the following termination/refusal criterion was adopted in order to

determine the bottom of a shear wall element: penetration rate remains at 0.2 foot/min or
less, with a cable load of 1 ton or less for a period of 1 minute. This criterion was
selected because 1) it generally resulted in adequate penetrations, and 2) a lower
penetration rate and/or longer hold period caused significant wear on the mixing shafts
and motors with very minimal additional penetration.

About 20 percent of the shear walls did not achieve the required penetrations using the
termination/refusal criterion. For shear walls where required penetrations were not
achieved or were not being achieved, we either 1) added supplemental shear wall
elements, 2) determined that the penetration and anchorage was suitable because of
localized dense deposits in the lower alluvium, or 3) performed water stroking in advance
of the shear wall installation. Water stroking consisted of predrilling with water injection
through the dense embankment materials. The soil mix wall was then formed by drilling
with injection of the grout slurry. The water stroking typically produced increased
penetrations of the shear walls.

The large number (six) of mixing shafts on the RAITO rig might have affected its
capacity to achieve greater penetrations. One or two of the augers could encounter
resistance and prevent penetration of the other augers. However, we would not
recommend restricting the number of mixing shafts because of the efficiency and general
good performance of RAITO’s six-axis rig.

LESSONS

As with most projects, we would have done a few things differently if given the benefit
of hindsight; however, we believe that most of the design elements and specifications did
suit the project purposes quite well. Some of the more significant lessons that we learned
or that the project reinforced were:

x Expect zones of significant deviation from the design soil mix strength. These
deviations should be anticipated in both the selection of factors of safety and in
preparing the QA specifications. Set up specifications to allow for the occasional
random low strength sample. Patterns of low strength samples, such as at a
consistent depth, should not be allowed. Also, specifications should consider the
skewing effect of occasional very high strength samples on the calculated average
strength.
 -15-

x Cover as large a range of soils as practical in the design phase soil cement mix
test program. Use the borings for the test program to better define remediation
layer boundaries.

x Test soil samples for pH. Perform soil mix test program on any low pH zones of
soil encountered.

x Similarly, test any soil samples that appear to have organic content for both
organic content and pH. Perform soil mix batch tests on any organic soils.

x Use coring in lieu of wet grab sampling for QA on soil mix zones. Use percent
recovery and RQD as part of the QA criteria. Allow for additional follow-up
cores to evaluate uncertainties in soil mix strengths.

x Include a number of additional soil borings and core borings as part of the
construction program. Chances are good that you will use a significant number of
each during construction to better define areas of uncertainty.

x Define the required remedial actions if soil mix walls have either low strengths or
limited penetration.

x Require the use of a computerized data collection system that tracks the quantity
of cement grout injected within each soil-mix column over every 1 to 3 feet. We
were able to focus QA testing on any low cement content zones because RAITO’s
controls and data collection provided these data.

x Realize that soil-mix contractors can advance their equipment through overlying
soils and that excavation to the top of the soil mix zone may not be necessary.

SUMMARY

The use of soil mix shear walls to provide reinforcing of the loose soil against seismic
deformations and liquefaction shear slides at the Clemson Dams was feasible and cost
effective.

REFERENCES

Broms, Bengt B. (1964), “Lateral Resistance of Piles in Cohesionless Soils,” ASCE

Journal of Soil Mechanics and Foundations Division, May.

Castro, G.; Keller, T.O.; and Boynton, S.S. (1989). "Re-Evaluation of the Lower San
Fernando Dam. Report 1: An Investigation of the February 9, 1971 Slide," Report No.
GO-89-2, U.S. Army Corps of Engineers, WES.
 -16-

Castro, G. (1994). "Seismically Induced Triggering of Liquefaction Failures," 13th

ICSMFE Special Volume on Earthquake Geotechnical Engineering, New Delhi.

Finn, W.D. Liam; Yogendrakumar, M.; Yoshida, N.; and Yoshida, H. (1986). “TARA-3,
A program to compute the response of 2-D embankments and soil-structure interaction
systems to seismic loadings,” Department of Civil Engineering, University of British
Columbia, Canada.

Poulos, S.J. (1981). "The Steady State of Deformation," Journal of Geotechnical

Engineering Division, ASCE, 107(GT5), May, pp.5513-5562.

Wooten, R.L., Castro, G., Gregory, G., and Foreman, B. (2003). “Evaluation and Design
of Seismic Remediation for the Clemson Upper and Lower Diversion Dams,”
Proceedings of the 23rd Annual United States Society on Dams Conference, April.

Wooten, R.L., Castro, G., Finn, W.D.L., and Foreman, B. (2004). “Seismic Modeling
And Triggering Analyses For The Clemson Upper And Lower Diversion Dams,”
Proceedings of the 2004 Annual Conference, Association of State Dam Safety
Officials, September.

GEI Consultants, Inc. contact for questions about the Clemson Diversion Dams
remediation project:

Lee Wooten, P.E.

GEI Consultants, Inc.
1021 Main Street
Winchester, MA 01890
Phone: (781)721-4034
e-mail: [email protected]
Company web site: www.geiconsultants.com