Partial Retrofit of Bridges Using

Micropiles
Project 4ST-20A of the Consequence Minimization
Program
Partial Retrofit of Bridges

Rory Philip Alfred Ball

URA/Home Institution: University of Illinois at Urbana-Champaign
URA Advisor: Professor James H. Long

21
Project 4ST-20A - Partial Retrofit of Bridges

Research performed by Rory Ball, under the supervision of Professor James H. Long at the
University of Illinois at Urbana-Champaign.
January-August, 2002

Title of Project: Partial Retrofit of Bridges Using Micropiles

Abstract:

Working in conjunction with seismologists and geotechnical engineers, investigators in this

project have developed typical ground motions for different recurrence intervals for various
bridge sites. The bridges were evaluated to determine the expected level of damage for the
different earthquakes. Estimated costs for retrofit and partial retrofit vs. no action were made
and the selected retrofits were installed.

In this report, a highway bridge that was deemed unprepared for earthquake will be discussed. It
is located at mile-marker four of Interstate 57, in Cairo, Illinois north of the New Madrid fault
zone. This paper will specifically focus on the Micropile retrofit that was installed. These
Micropiles were used to seismically strengthen the existing foundation and they will be
explained in detail about what they are, why they are used, and how they are installed and tested.
To aid in this bridge investigation, computer modeling was performed and was subsequently
compared with in-situ tests. The result of this comparison was the conclusion that results from
each test lined up well with one another.

22
1.1 Background and Details of Micropiles

Traditional pile foundations are constructed by hammering relatively long piles made of steel,
precast concrete, or wood into the ground until the desired strength is achieved. Unfortunately,
these piles require a large crane that supports the pile driver as it hammers the pile into the
ground. This process can be quite noisy and the ground vibration caused by pile driving may
cause damage to nearby structures.

Originally invented in Italy, Micropiles were mainly used as a quick application in underpinning
historical landmarks. Today, Micropiles are used for both the original foundation and structural
retrofitting of bridges, dams, buildings, and many other structures. Micropiles are installed using
rotary drilling and grouting techniques rather than a conventional pile driver. They are also quite
versatile. They can be drilled through almost every ground condition with minimal vibration,
disturbance, and noise, at any angle below horizontal, and in places with minimal headroom.
Because Micropiles have all of these advantages, they are increasingly being used in urban
settings where access restrictions prevent pile driving.

Micropiles are usually under 10-inches in diameter (100-250 mm), 25 or more feet in length (7.5
or more meters), and 70-225 kips in capacity (300-1000 kN). By using innovative grouting
methods, Micropiles can achieve high grout-to-ground bond strengths to be generated around the
pile perimeter; therefore the Micropile can carry load in the same manner as a friction pile.
Micropiles are filled with concrete, but it is primarily the steel elements (occupying up to 50% of
the hole volume) that transmit the load from the structure to the soil.

There are several different types of Micropiles. They are classified (but not limited to) as
follows:

Type A: Gravity head Micropiles. These are non-pressurized and use sand-cement "mortars" or
neat cement grouts [a fluid mixture of hydraulic cement (a cement that is capable of setting and
hardening under water) and water].
Type B: Grout is injected as temporary drill casing or auger is withdrawn. Pressurized once at
low pressure (0.3-1 MPa; 44-145 psi)
Type C: Grout is gravity placed, allowed to set for 15-25 min, and then a second batch of grout is
injected at moderate pressure through a sleeved grout pipe.
Type D: Grout is gravity placed and allowed to harden. When primary grout has hardened, more
grout is injected through a sleeved port grout pipe. A movable packer is used so that specific
horizons may be treated several times if necessary. High pressure is used (2-8 MPa; 290 -1,160
psi).
[source: ISSMFE website]

1.2 This is why I Studied It

The New Madrid Fault System has the potential to devastate a vast area of the continental United
States if a large earthquake is unleashed by this fault. This fault zone lies within the central
Mississippi valley and crosses five state lines; cutting across the Mississippi River in three places
and the Ohio River in two places. This fault system has been labeled the most earthquake prone

23
region of the United States east of the Rocky Mountains and averages more than 200 measured
events per year (1.0 or more on the Richter scale), or about 20 per month. The underlying
geology of this region has the potential to cause destruction of more than 20 times that of
California. The USGS estimates that a major earthquake in this area occurs every 200-300 years
(the last one in 1812). Even with so much possibility of a devastating earthquake rocking this
region, the New Madrid Fault Zone remains one of the most ill-prepared earthquake areas with
many buildings being made out of masonry. USGS seismologists anticipate an earthquake in
this fault zone to shake a large area sometime within the next century. Structures that are not
properly mitigated against earthquake loads may be severely damaged and cause billions of
dollars in damage to the Central United States and the U.S. Economy. [Paragraph source: Center
for Earthquake Studies and the USGS]

1.3 Project 4ST-20A

1.3.1 The Bridge Studied

The bridge location that the author specifically worked on is located at the Southern end of
Illinois, close to where the Ohio and Mississippi Rivers join. There, at mile marker four on
Interstate 57, the two north-bound lanes and the two south-bound lanes are separated into two
1770 foot long spans (shown in Figure 1.) Each span has 24 hammerhead piers that consist of
timber piles driven into clay and dense sand. These parallel spans were built by Illinois
Department of Transportation (IDOT) in the late 1960s and are located only about 70 miles north
of the New Madrid fault line, where three great earthquakes took place in the early 1800s. The
author should mention that this bridge is over land and only over water when there is flooding of
the Mississippi River. From analysis, engineers determined that this bridge could be prone to
earthquake damage and it should therefore be retrofitted to strengthen it against excessive
deflections. Because this bridge is about a third of a mile long, IDOT concluded that it would
only be financially feasible to retrofit just one of the spans. If the case were that an earthquake
knocked down the south-bound lanes, the two north-bound lanes could be switched to
accommodate both directions of traffic and assure a thoroughfare.

Figure I: Wide-angle view of the bridge section studied on Interstate 57.

24
1.3.2 Bridge Piers and Existing Foundation

Enclosed in each hammerhead pier pile cap are three rows of timber piles. There are 24 piles in
each pier foundation, each 12 inches in diameter, tapered to 8 in. at the bottom, and all about 36
inches in center-to-center spacing. The outer rows of piles are battered from the vertical axis one
inch per foot of length to help aid in lateral stiffness. Specific pile details can be found in Table
1. The pile cap formed over the timber piles creates a fixed-head connection and is usually about
two feet thick.

Table I. Pile properties

Timber pile Micropile

Area (in2) 113.1 15.7
Moment of Inertia (in4) 1018 160
GJ Torsional Stiffness (lbs-in2) 1.16E+09 3.59E+09
Young's Modulus (psi) 1250000 34640000
EI (lbs-in2) 1272500000 5542400000
Length (ft.) 25 34
Pile Diameter (in.) Top = 12, Bottom = 8 9.625
Pile Numbers (See Figure 4) 4-9,12-23,26-31 1-3,10,11,24,25,32,33,34

1.3.3 Soil Profile

There were a total of 29 original soil borings from 1968 that were drilled along the alignment of
the north and south bound lanes of Interstate 57 and a total of seven new borings from 1999 that
were drilled along the centerline of the bridge. Boring depths ranged from 41.5 ft to 66.5 ft for
the existing borings and 66.5 ft to 81.5 ft for the new borings. The subsurface investigation
consisted of soil samplings, visual classification, strength tests, and Standard Penetration Tests
(SPTs). The soil profile that was used on this project consists of four soil layers contributing to
an overall depth of 50 ft. All soil information can be found in Figure 2. The soil profile that
was modeled on computer consisted of a top layer of 60 in. of submerged soft clay followed by
another 60 in. of submerged soft clay with differing properties of the overlaying layer of clay.
The third layer was 150 in. of submerged soft clay which was comprised of properties shown in
Figure 2. The unconfined clay strengths vary from about 0.7 tsf near the ground surface to
about 1.1 tsf at about 11 ft below the ground surface, and then decrease in strength to about 0.4
tsf at the bottom of the clay layer. These top three layers of clay are therefore considered soft to
medium from these shear strengths. Finally, the bottom layer modeled was 330 in. of sand. The
bottom layer is medium dense sand from analyzing blow counts - which ranged from 8 to 35
blows per foot. This information is all summarized in Figure 3. Note: The profile described is
an average of the whole area; therefore, each layer can vary by several feet.

25
Figure II: Soil profile that was used in the LPILE and GROUP5 computer programs.

Figure III: Summary of soil profile from SPT tests.

26
1.3.4 Micropile Retrofit

Micropiles can have installation bearing strength of up to 200 tons. For the southern Illinois
bridge considered, they were designed for a capacity of 160 kips and were chosen because they
could be installed underneath the low overhead of the overlaying bridge decks. The pile retrofit
is comprised of ten Micropiles, each made with a steel casing and filled with cement grout.
These vertical piles are 9.625 in. in diameter and are 34 ft. long, with a fixed-head connection
that is formed into the pile-cap of the timber piles. The exact specifications of the Micropiles are
shown in Table 1.

These piles consist of an upper section with a heavy walled casing, and a lower, uncased section
with a centered high-strength bar. This casing is a 0.545 in. steel sleeve that is filled with high
strength grout and an inner # 11 or # 20 dooeydag bar in the center. Retrofit specifications called
for a 0.4 in. or less deflection is the axial deflection. The plans also called for piles to be
installed between 25 to 30 feet deep, but on-site testing showed that the 240 Micropiles could be
installed at 20 feet and still achieve the desired strength. These piles were combined in five foot
sections and twisted together until this depth was achieved. Even with the savings from the
decreased bond length, the project still cost about 2.5 million dollars.

To create the retrofit, the pile cap around the timber piles was chemically cracked so that these
piles would be exposed. Micropiles were then drilled down around the existing foundation, and
a new pile cap was poured to tie together all the piles.

2.1 Computer Modeling

2.1.1 Computer Programs

LPILE is a program used to determine the response of a single pile to lateral and moment
loading. The program computes the response of the pile using either internally generated p-y
curves based on the given soil properties, or user input p-y curves. A large part of my
contribution to the project was setting up and running several computer modeling-programs to
predict displacements of single and groups of piles subjected to lateral loads. LPILE uses the p-y
curves to determine lateral displacement for defined loads. P-y curves give a relationship
between soil displacement versus pile displacement along the length of the pile and help the
program iterate to generate several deflection and moment graphs. (Brown and Shie, 1991) The
program GROUP5 is an extension of the program LPILE (also calculates axial loads) and
calculates the reaction of a group of piles to a specified load. The effects of pile-soil-pile can be
accounted for in several different ways: GROUP5 allows the softening of response to lateral
loads from these nearby piles from p-y modifiers the user can specify, or the user may ask the
program to internally generate the modifiers. GROUP5 can therefore determine one pile’s
influence on another pile that is near it. (Reese and Wang, 2000) When a pile is deflected, it
creates an area of stress in the soil it is pushing against, and the movement of that stressed soil
affects nearby piles. (Poulos and Randolph, 1983) In his individual research, the author created a
complete model of the underlying foundation of these bridge piers both pre- and post-retrofit.
The results of that analysis using different lateral loads and interaction factors was compared to

27
other computer programs that our team of researchers in the Geotechnical Engineering
Department had been experimenting with. To have the ability to model pile group deflection
under lateral loads is very important in Earthquake Engineering. With results from computer
modeling programs such as GROUP5, some of the uncertainty that goes along with foundation
engineering can be eliminated.

GROUP5 parametric studies were conducted on the retrofitted pile group by varying pile
diameter and stiffness to determine the effect on overall group displacement. All loads were
applied in the longitudinal direction to identify loads required for specific deflections of the
group. One load was defined by the occurrence of pile pull-out to identify the ultimate load for
the group. A 1.5 in. deflection was considered a large displacement and a 0.75 in. deflection was
defined as a more reasonable displacement that might occur when the group was loaded.
Parametric studies were conducted with loads that would create the 0.75 in. and 1.5 in.
deflections. For each set of deflections, there were three main cases investigated in these studies.
First, the group was simulated under specified interaction effects with the p-y multipliers shown
in Figure 4 (Specified Case). Next, soil-pile-soil interactions were considered with the p-y
modifiers that were determined with the automatically generated interaction effects that
GROUP5 created. Finally, the pile group was simulated by ignoring pile-soil-pile interaction
effects.

Figure IV: Plan view of the existing pile foundation (inside 24 piles) and the Micropile
retrofit (outside 10 piles); each pile is labeled with its own p-y modifier.

In the first runs, the lateral load behavior of the existing group of timber piles was evaluated.
Then the retrofit was added and the group was analyzed under different loading. With the

28
retrofit in place, additional studies were conducted by altering the diameters and stiffness of the
piles. The p-modifiers that are typed next to each pile shows how if the group is loaded from the
left, the front piles have a reduced factor because they are not as stiff as the back piles that are
pushing against soil with no other piles behind them.

2.1.2 Case Studies

The following pile group scenarios were constructed on GROUP5 so that deflection could be
estimated from user defined loads:

Case 1: Bridge foundation without the retrofit.

Case 2: Bridge foundation with the proposed retrofit.
Case 3: Case 2, diameters of all piles are changed to that of the timber piles- 12 in.
Case 4: Case 2, diameters of all piles are changed to that of the Micropiles- 9.625 in.
Case 5: Case 2, but the E of the Micropiles is changed to 795315 psi
Case 6: Case 2, but the E of the timber piles is changed to 5444400.8 psi
Case 7: Case 5, but all of the pile diameters are now changed to 12 in.
Case 8: Case 5, but all of the pile diameters are now changed to 9.625 in.
Case 9: Case 6, but all of the pile diameters are now changed to 9.625 in.
Case 10: Case 6, but all of the pile diameters are now changed to 12 in.

2.1.3 Computer Modeling Results

Of the three cases (Automatically Generated, Specified, and No p-y Modifiers), each graph has
three groups of lines which are based on the EI assigned to each case as follows:

Case 1 - EI (all piles) = 1272500000 lb-in2

Cases 9,10,6 - EI (all piles) = 5542400000 lb- in2
Cases 2,4,3 - EI (timber) = 1272500000 lb- in2, EI (micro) = 5542400000 lb- in2
Cases 5,7,8 - EI (all piles) = 1272500000 lb- in2

In all three deflections for the specified p-y modifier case (.75 in., 1.5 in., and ultimate), the cases
remain in the same order of stiffness. Organized from the most to the least stiff would be Case
10-6-9-3-2-4-7-5-8-1. This order is consistent with the EI and the pile diameters of the cases
considered. Comparing the graphs in Figures 9, 10, and 11 of the Group Load vs. Group
Displacement for the three cases, there is an obvious difference directly related to the p-y
modifiers used. The group is behavior is stiff when interaction between the piles is ignored
because each pile is acting independently in the group, and the only interaction occurs through
the pile cap. In the automatically generated p-y case, GROUP5 estimates the pile-soil-pile
interaction effects to be lower than the no p-y modifier case. The graph titled, “Specified p-y
Modifiers,” is notably more sensitive to loading compared with the other two cases.

29
3.1 Lateral Load Tests

3.1.1 Pile Layout

My research team at the University of Illinois traveled down to the bridge site in Southern
Illinois to conduct several lateral load tests aside the southeast corner of pier 17 of the north-
bound span. There, 12 Micropiles were installed in a zigzag pattern each with a 28-day
compressive strength of the cement grout required to exceed 4000 psi. #11 rebar were installed
the full length of all the test piles except for pile no.s 5, 6, and 7, which had #20 bars.

The area around these 12 piles was excavated to a depth of six feet so that test equipment could
be mounted on the pile heads. A 2 in layer of gravel was place on the floor of the test pit to keep
the research being conducted above the water table. Figure 5 (a) and (b) are illustrations of the
test pit and give pile geometry and numbering.

Figure V (a): Photo of pit excavation with test piles.

30
 Figure V (b): Plan view showing test pile positions.

3.1.2 Testing System

Using a loading frame that was mounted on the piles, two pile were tests at the same time by
simultaneously applying a lateral load to pull them together. The field setup is illustrated in
Figure 6. These loads were applied by using hand-pumped, 30-kip hydraulic jacks that were
mounted within the loading frame. The applied load was monitored by a pressure gage mounted
on each hydraulic jack that had an accuracy of 100 psi. Deflections were carefully measured
using four Linear Variable Differential Transformers (LVDTs) and loads were monitored by two
load cells that were incorporated into the testing structure. All load cells were calibrated prior to
testing by personnel from Newmark Structural Engineering Laboratory (NSEL) Facility.

31
 Figure VI: Photo of loading system.

3.1.3 Predictions and Testing

The author’s role in this testing was to help with the structural setup, to record the deflections
that were achieved, and to compile the test data back at the lab to prepare for the formal report.
The computer program LPILE was used to predict lateral load behavior. From estimated soil
properties, pile properties, and applied loads, predicted pile behavior was recorded so we would
be able to compare these results with the actual field test data. The displacement predictions that
the author had developed using LPILE modeling lined up fairly well with the experimental
deflections.

3.1.4 Conclusions

To ensure only a negligible amount of interaction between the two piles being pulled together,
we tested piles that were far enough apart. Results show that the measured displacement lines up
well with the predicted displacements from LPILE. Any differences in test data could be
attributed to varying soil strength conditions, different bending stiffness of the piles, or a need
for changes to the p-y curves that were defined in the LPILE program. From further analyses it
was concluded that there were two piles that failed structurally. One of the failures can be seen
in the photograph in Figure 7. These two piles exhibited a larger deflection than the other piles
tested. To inspect them, we excavated to a depth of 8 ft and examined the upper portion of the
pile. We found that these failures occurred in the Micropiles at the threaded-connections close to
the surface. At these connections, the bending moment was very high and the joints may have
been under or over tightened or even cross-threaded when the machines drilled them into the
ground. Because of this, the threaded connection joining the casing had ruptured – slipping
between the threaded connections. We concluded that this could be a possible problem with this
retrofit, a presence of a structurally weak spot in the Micropile. The actual cause of this weak
spot is unknown, but in his formal report to IDOT, Professor Long believes the following could
be explanations for the ruptured connection:

32
 1) It is possible that the threaded connections were structurally sound, but were ruptured
during the process of excavating and leveling the test pit area. The bucket of the
excavating equipment may have caught on the Micropile, subjecting the Micropile to
excessive load and causing structural damage to it.
2) It is possible there was a defect in the structural steel at these threaded connections
that resulted in very weak structural properties. In this case, either the process of
excavation or the lateral load tests could have ruptured the section.
3) It is possible that the threaded connections were cross-threaded during installation
and therefore were inherently weak. It is also possible that the axial load test program
served to weaken the threaded connections in these piles.

It is important to note that because this was a free-headed connection test, the moment near the
surface would decrease when it would be tied into a fixed-head group pile-cap (as the actual case
is).

Figure VII: Pile #2 significantly deflecting after a structural failure.

4.1 Pile Bending Test

Upon request, a 10 ft section of Micropile was filled with grout and allowed to cure at the site.
Our group at the University of Illinois transported this pile to the testing bay at NSEL where the
pile was instrumented with strain and dial gauges and then placed in bending. The photo in
Figure 8 displays what the bending test setup looked like.

33
 Figure VIII: Bending test setup.

The setup included nine half-inch strain gauges that were used to measure the strain at various
points on top and bottom of the pile. The vertical deflections were measured by installing four
DCDTs at various positions on top of the pile and a deflection gage at the bottom of the pile.
The important results are that the 147 ksi yield strength was confirmed and also that about 95%
of the Micropile’s stiffness comes from the EI of the steel. The design only called for 80 ksi, so
the 147 definitely exceeded the contract requirements. If this 10 foot section were to not be
continuous, but rather two five foot sections with a threaded connection, the bending strength of
such a joint could have been tested. Unfortunately, this was not available to our research team at
the time.

In conclusion, the engineering practice should pay more attention to the strength of joints. If
these threaded joints were to fail under seismic loading, the retrofitted foundation may not
perform to expectations and the consequences could be disastrous. The lateral load tests that we
performed were the first that Illinois DOT has conducted, and more should be done to help test
and confirm lateral deflections.

5.1 Future Testing

The in-situ lateral load tests that were performed were important to not only verify the integrity
of the Micropiles, but they were also valuable to compare to computer modeling output from the
program LPILE. The Micropiles were found to displace about 10% greater than the predicted
displacements from LPILE, but overall there was good agreement between the analyses. The

34
next step in Micropile testing would be to test groups of piles subjected to lateral loads. In
section two of this report, the computer program GROUP5 was discussed along with the research
the author completed on it. To compare in-situ field results to the GROUP5 (and similar
program) output, much in the same way the single pile tests were compared with LPILE output,
pile group tests could be performed in the field or in the lab to accomplish this. With
conclusions that could be made from research such as these pile group tests, the engineering
community could place more confidence in the computer modeling programs they use to predict
foundation motion under seismic loading.

In addition, further testing should be performed on grouted threaded connections to help to give
consideration when the specifications and designs are drawn. Threaded connections currently do
not have much research behind their moment capacity, and testing for this should take place.
Furthermore, because the 1968 soil borings did not align with the more updated borings, better
subsurface investigation techniques should be used if more funding were available.

Works Cited

Brown, D.A., and Shie, C-F. “Modification of p-y Curves to Account for Group Effects on
Laterally Loaded Piles.” Proc., Geot. Eng. Congress. Boulder: June 10-12 1991. Vol.1, pp.
479-490.

Center for Earthquake Studies, Southeast Missouri State University

http://www2.semo.edu/ces/ces2.html

International Society for Soil Mechanics and Geotechnical Engineering (ISSMFE).

http://tc17.poly.edu/mp.html

Long, James H. “Results of Lateral Pile Load Tests.” Formal Report. July 4, 2002.

Poulos, H.G., and Randolph, M.F. “Pile Group Analysis: A Study of Two Methods.” J. of Geot.
Eng. March, 1983. Vol. 109 (3).

Reese, L.C., and Wang, S-T. Group Version 5.0 User’s Manual. Ensoft Inc. Austin: 2000.

USGS. “Uncovering Hidden Hazards in the Mississippi Valley.”

http://quake.wr.usgs.gov/prepare/factsheets/HiddenHazs/

35
Figure IX:
Automatically Generated p-y Modifiers by GROUP5

3000

2500

2000
Group Load (kips)

1500

1000

500

0
0 5 10 15 20 25

Group Displacement (inches)

Retrofit
Retrofit-All D =12 in
Retrofit-All D = 9.625 in
Retrofit-EI of Timber
Retrofit-EI of Micropile
Retrofit-EI & D of Timber
Retrofit-EI of Timber, D of Micro
Retrofit-EI & D of Micropile
Retrofit-EI of Micropile, D of Timber
Existing Pile Group

36
 Figure X: No p-y Modifiers

3000

2500

2000
Group Load (kips)

1500

1000

500

0
0 5 10 15 20 25

Group Displacement (inches)

Retrofit
Retrofit-All D =12 in
Retrofit-All D = 9.625 in
Retrofit-EI of Timber
Retrofit-EI of Micropile
Retrofit-EI & D of Timber
Retrofit-EI of Timber, D of Micro
Retrofit-EI & D of Micropile
Retrofit-EI of Micropile, D of Timber
Existing Pile Group

37
Figure XI: Specified p-y Modifiers

3000

2500

2000
Group Load (kips)

1500

1000

500

0
0 5 10 15 20 25

Group Displacement (inches)

Retrofit (Case 2)
Retrofit-All D =12 in (Case 3)
Retrofit-All D = 9.625 in (Case 4)
Retrofit-EI of Timber (Case 5)
Retrofit-EI of Micropile (Case 6)
Retrofit-EI & D of Timber (Case 7)
Retrofit-EI of Timber, D of Micro (Case 8)
Retrofit-EI & D of Micropile (Case 9)
Retrofit-EI of Micropile, D of Timber (Case 10)
Existing Pile Group (Case 1)

38