Professional Documents
Culture Documents
Abstract: Departments of transportation are increasingly embracing pipe ramming for culvert installation under roadways due to its
cost effectiveness and ability to mitigate problems associated with open-cut trenching. Despite the increase in use, little technical guidance
Downloaded from ascelibrary.org by New York University on 05/15/15. Copyright ASCE. For personal use only; all rights reserved.
is available for the engineering of pipe-ramming installations. This study presents the analysis of the performance of an instrumented
610-mm-diameter steel pipe installed using pipe ramming. Measurements include ground surface movement and dynamic force and velocity
waveforms to obtain driving stresses, hammer-pipe energy transfer, and static and dynamic soil resistance during the installation. Ground
movements are compared to existing settlement prediction models. Inverted normal probability distribution models commonly used in tunnel
engineering were evaluated and were observed to capture the observed settlement close to the center of the pipe but did not accurately predict
the observed transverse settlement profiles. The transfer of energy was observed to range from as low as 17–39% of the estimated hammer
energy. Compressive stresses were observed to remain relatively constant over the penetration length observed and were well below the yield
stress of the pipe. Soil resistance derived from wave equation analyses were compared to four pipe-jacking models to evaluate their accuracy
and applicability for planning pipe-ramming installations. The jacking models bracketed the static soil resistance components of the wave
analysis, indicating that the models may be adopted for pipe-ramming applications pending empirical modification. DOI: 10.1061/(ASCE)
CF.1943-5509.0000463. © 2014 American Society of Civil Engineers.
Based on cost, pipe ramming was selected by the contractor as ond stage and segment 2 for third stage) and driven, respectively, to
the preferred trenchless pipe installation method for the crossing cross the embankment. The project utilized a 400-mm pneumatic
to avoid the loss of income resulting from the temporary closure HammerHead Mole hammer, produced by HammerHead Trench-
of the road and disruption of the conveying operations. less Equipment of Oconomowoc, Wisconsin, with an estimated
Project requirements called for an open-ended steel pipe of rated energy of 6.4 kN-m and rate of blow of 231 stroke= min.
610-mm-diameter and 13-mm wall thickness, 33.5 m in length. The hammer was fitted to the rear end of the pipe with collets that
A cutting shoe with an external diameter of 630 mm and internal facilitate the connection and distribute the impact force of the ham-
mer to the edge of the pipe casing. The pipe-collet-hammer system
diameter of 550 mm was welded to the lead edge of the first seg-
was held in place by tensioned chains hooked to eye pads welded
ment of the pipe to reinforce the pipe section at the face and facili-
on each segment of the 610-mm-diameter pipe (Fig. 1). The tension
tate face penetration. The cutting shoe creates an overcut in the
in the chains is an important factor in successful pipe-ramming ap-
ground and is thought to reduce the frictional drag on the pipe
plications, as loose chains can result in significantly reduced trans-
due to the excavation of a cavity slightly larger than the diameter
fer of energy, and if left unattended, can lead to a loss of seating of
of the pipe (Simicevic and Sterling 2001). The steel pipe installa-
the hammer on the pipe.
tion commenced with the placement of a leveled 0.3-m-thick
compacted crushed rock bedding layer in the launching area on
the west side of the embankment. Due to the topography adjacent Field Instrumentation
to the embankment, excavation of insertion or receiving pits was
not required. Steel tracks approximately 20-m long and typically Monitoring of ground movements due to trenchless culvert instal-
used with auger boring operations were placed on the leveled lation is a relatively common practice because intolerable magni-
ground to help guide the hammer-pipe system and maintain the tudes of heave and settlement could cause loss of service to
required grade and alignment during the installation (Fig. 1). adjacent buried utilities, structure foundations, or roadway pave-
The placement of the first segment of the pipe-hammer system ments. Despite the need for field performance data, very few case
on the desired grade and alignment is critical, because correcting histories of pipe-ramming installations with ground movement
the alignment after approximately 1.2 m of pipe insertion is measurements have been reported. Additionally, no measurements
extremely difficult as pipe ramming is a nonsteerable technique of soil resistance to ramming are documented other than those re-
(Najafi 2008). Following installation of pipe segments, the tracks ported by Stuedlein and Meskele (2012). In order to broaden the
performance data of pipe-ramming installations, field instrumenta-
tion was used to monitor the performance of the pipe installation.
The instrumentation included ground deformation monitoring
points (GDMPs) and dynamic strain gages and accelerometers to
observe the pipe response to impact driving.
1 2 3 4
1.8 m
the advancing pipe follows an inverted normal probability distribu-
tion curve and is symmetrical about the vertical axis normal to the
0 longitudinal centerline through the pipe (Fig. 4). The ground dis-
1.8 m
placement can be represented by
2
−y2
1 2 3 4
Sz ðyÞ ¼ Smax e 2i ð1Þ
Downloaded from ascelibrary.org by New York University on 05/15/15. Copyright ASCE. For personal use only; all rights reserved.
0 8m 16 m 24 m 32 m
Distance from West toe of embankment where Sz ðyÞ = vertical settlement at a distance y from the centerline;
Smax = maximum centerline settlement; and i = horizontal distance
Fig. 2. Profile and plan view of the access road and conveyer belt from the pipe axis to the inflection of the settlement trough (Fig. 4).
embankment and settlement monitoring points The maximum settlement Smax can be obtained by equating the
total volume of settlement trough and the ground loss due to the
over-cut, V s
Observation of Impact Forces and Velocity Z ∞ pffiffiffiffiffiffi
π
The pile-driving analyzer (PDA) is a widely used computerized V s ¼ ½D2c − D2 ¼ Sz ðyÞdy ¼ 2π · i · Smax ð2Þ
4 −∞
system for dynamic load testing of piles during impact pile driving.
It is composed of high-frequency strain gauges and accelerometers
(Fig. 3) that are bolted on diametrically opposed sides of a pile, or Vs
Smax ¼ pffiffiffiffiffi
ffi ð3Þ
in this case the 610-mm-diameter pipe, to acquire the signals of i 2π
dynamic strain and acceleration due to hammer impact (Hannigan
et al. 1996). The strain and acceleration measurements are con- where V s = volume of soil loss per unit of length; Dc = external
verted by the PDA to force (i.e., F ¼ εEA) and velocity (i.e., V ¼ diameter of the cutting shoe or excavated bore; and D = external
∫ adt) time histories, respectively, where ε = strain, E = modulus of diameter of the pipe. Originally, Schmidt (1969) and Peck (1969)
the pipe, A = cross-sectional area of the pipe, and a = acceleration suggested that the width of the settlement trough (i) can be de-
of the propagating wave. The force and velocity records, in con- scribed as a function of the depth of the pipe below the ground
junction with wave equation analysis, are commonly used to evalu- surface (z0 ) and diameter of the pipe (D). Cording and Hansmire
ate the load-bearing capacity of piling, driving stresses, and the (1975) approximated the area under an inverted normal probability
energy transferred from hammer to a pile during driving. Details distribution curve with a triangular area and derived the half-width
on high-strain dynamic testing and measurement are given in of the settlement trough (w), given by
ASTM D-4945 (ASTM 2008). PDA measurements were made
Dc π ϕ
for the second stage of the pipe-ramming installation (i.e., the sec- w ≈ 2.5i ¼ þ zo tan − ð4Þ
ond, 9.1-m-long pipe segment). The strain gauges and piezoelectric 2 4 2
accelerometers were mounted approximately 2D (1.22 m) from the
rear end of the pipe segment (Fig. 3). where z0 = depth of cover to the center of the pipe in meters.
O’Reilly and New (1982) generated another relationship for
the width of the settlement trough as a function of the depth of
Analysis of Ground Deformation the pipe. They performed multiple linear regression on field data
the access road, where Section 1 was located, was 1.3 m, whereas (i.e., set) and therefore penetrate the ground. Analysis of the dy-
the height of the main portion of the embankment where the other namic response of the pipe can be performed through the use of
three sections were located was 4.1 m. Because Section 1 had such stress wave theory. The force and velocity measurements near
a small depth of cover, the empirical approaches described above the rear end of the pipe are used to evaluate the hammer perfor-
do not adequately capture the settlement observed, which ranged in mance, driving stresses, and soil resistances during pipe-ramming
magnitude from 1 to 8 mm. Because Section 2 lies along the crest installation. The PDA applies the simple-closed form case method
of the embankment, settlements were anticipated to be greater than solution of the wave equation (Goble et al. 1975; Rausche et al.
for level ground conditions due to amplification of ramming- 1985) and provides the energy transferred, driving stresses, and soil
induced vibrations associated with slope crests. Fig. 5 presents resistances associated with driving in real time from the dynamic
the monitoring results for Sections 2, 3, and 4, and indicates that measurements of force and velocity. The energy transfer from the
the largest settlements were those at the crest of the embankment. hammer for each impact can be computed by integrating the prod-
Generally the three empirical models appear to capture the ob- uct of force and velocity signals with respect to time, which is given
served settlement close to the center of the pipe; however, measure- by Goble et al. (1980)
ments at distances greater than 1 m on either side from the center of
Z
the pipe suggest a greater extent of settlement than that predicted. t
The maximum observed settlement and heave were approximately Etransfer ðtÞ ¼ FðtÞ · vðtÞdt ð5Þ
0
14 and 6 mm, respectively, indicating that total magnitude of set-
tlement is small to negligible. The observations did not agree with
where Etransfer = energy transferred; F = force trace (i.e., time
history); and v = velocity trace at head of the pipe. The maximum
energy transferred to the pipe for each hammer strike can be
Distance from the Center of the Pipe (m)
obtained by taking the maximum value of Etransfer . This allows an
-5 -4 -3 -2 -1 0 1 2 3 4 5
-10 evaluation of the performance of the hammer utilized in the pipe-
ramming installation.
The stress developed in the pipe during impact driving can
0 be attributed to compressive, tensile, and flexural forces. However,
Settlement (mm)
multiplied by the pipe face velocity, Rd ¼ J c · Z · V face , with the E ¼ 210 GPa; material density, γ ¼ 7881 kg=m3 ; wave speed,
assumption that the damping effects are concentrated at the pipe c ¼ 5155 m=s; and nominal pile impedance, Z ¼ 964 kNs=m.
face (Goble et al. 1975). The Case damping factor has a significant In general, the energy transfer depends on the type of hammer,
effect on the magnitude of the soil resistance predicted by the Case the hammer-pipe connection (e.g., collets, ram cones), the hammer
method, and the selection of its value is an important task. For rec- alignment, and the degree of tension on restraining chains (Meskele
ommendations on typical ranges, see Goble et al. (1975). The static and Stuedlein 2013). The actual energy transfer profile decreases
soil resistance component can be computed by (Rausche et al. with penetration length, from 39 to 17%, of the rated hammer
1985) energy. The reduction in efficiency is believed to result from a com-
bination of infrequent tensioning of the restraining chains and the
Fðt1 Þ þ Z · vðt1 Þ Fðt2 Þ − Z · vðt2 Þ cumulative increase in inertial forces of the spoils, which had com-
Rs ¼ ð1 − J c Þ þ ð1 þ J c Þ
2 2 pletely filled the pipe at the end of the drive. The induced compres-
ð8Þ sive stress profile indicates a somewhat constant average stress
of 60 MPa, which is significantly smaller than the allowable com-
where F = measured force; v = measured velocity; t1 = time of pressive stress in the pipe. Eq. (6) was used to compute the com-
initial impact; t2 = time of reflection of initial impact from pipe pressive stress based on an assumed piston diameter of 330 mm,
face (t1 þ 2L=c); and L = length of pipe. The Case method offers because manufacturers do not report their piston geometries, and
an approximate estimate of the total soil resistance and does not resulted in a slight overestimate of the actual maximum compres-
distinguish between the casing and face resistance components. sive stress. It appears that Eq. (6) can be adopted for the prediction
Thus, a more reliable estimate of the soil resistance requires a of compressive stress in the absence of wave propagation analysis,
rigorous analysis of the force and velocity traces by performing provided the piston diameter can be accurately estimated.
a signal matching process. The Case Pile Wave Analysis Program
(CAPWAP) (Rausche et al. 1972) is probably the most widely 8
Max. Energy transferred (kN-m)
used signal matching program that can compute the total static
7 Rated Energy of Hammer
soil resistance (i.e., the casing and face components) following
installation. 6
Fig. 6 depicts the force, the product of velocity and impedance, 5
and the energy transfer time histories for a hammer blow at a pen- 4
etration length of 11.9 m. A qualitative interpretation of the soil
resistance can be made by observing the form of the force and 3
Actual Energy Transfer
velocity curves between the time of peak impact loading (t ¼ 0) 2
and first reflected stress wave from the pipe face (t ¼ 2L=c). 1
Initially, there exists proportionality between the force and velocity
0
traces loading up to the peak via F ¼ Z · v. Thereafter, separation 10 12 14 16 18 20 22
of the two curves is observed after t ¼ 6 m=s, indicating the (a) Length of penetration (m)
250
Max. Compressive stress (Mpa)
Energy (kN-m)
1500 2.5
1000 2.0
100 Predicted with Eqn. 6
500 1.5
0 1.0
50
-500 0.5 Induced max. Compressive stress
-1000 0.0 0
2 L/C
0 2 4 6 8 10 12 10 12 14 16 18 20 22
Time (ms) (b) Length of penetration (m)
Fig. 6. Force, velocity, and energy time histories for a hammer blow at Fig. 7. Case method results of pipe performance: (a) maximum trans-
a penetration length of 11.9 m ferred energy; (b) maximum induced compressive stress in the pipe
pipe ramming in detail. These soil resistance models were applied vide good estimates of casing resistance at a penetration length of
to the installation of the 610-mm-diameter pipe installation using 11.9 m, but reduce in accuracy at a penetration length of 20.2 m.
an angle of internal friction ϕ ¼ 42°, unit weight of the soil Additionally, the total static soil resistance (i.e., face and casing
γ ¼ 20.5 kN=m3 , and soil-pipe interface angle δ ¼ 0.5ϕ. resistance) estimates produced using the ATV and PJA methods
Fig. 8 presents the observed total static soil resistance (i.e., com- overestimate the CAPWAP-based total resistance, largely because
prising face and casing) and total soil resistance (i.e., sum of static of the inaccurate estimate of face resistance, as shown in Fig. 8. The
and dynamic) to ramming and static soil resistances computed us- Terzaghi (1943) approach appears to provide a good estimate
ing the various pipe-jacking and microtunneling methods presented of total static soil resistance as a result of compensating overpre-
in Stuedlein and Meskele (2012). Static soil resistances were diction and underprediction of face and casing resistance, respec-
observed in real time using the case method solution to the wave tively. Despite the wide range in prediction performance, empirical
equation and CAPWAP analysis for two hammer impact blows modification of these methods for application to pipe ramming
corresponding to penetration lengths of 11.9 and 20.2 m. The per- could provide suitable soil resistance estimates as additional data
formance of the pipe-jacking models for prediction of static soil become available.
resistance can be evaluated by comparing to the Case method static As shown in Fig. 8, the total soil resistance (i.e., static and dy-
soil resistance obtained in the field and the CAPWAP-based static namic) is larger than the static resistance alone. Thus, the selection
soil resistance, both of which are presented in Fig. 8. The selection of a pipe-ramming hammer must consider the dynamic component
of the appropriate Case damping constant for comparison to the of resistance. Research is presently underway to address the esti-
pipe-jacking methods must be informed; for this case, a Case mation of dynamic soil resistance to pipe ramming.
damping constant of J c ¼ 0.8 was selected based on the results
of the more accurate CAPWAP analysis. The static soil resistances
were observed to be relatively high initially at the resumption of Summary and Conclusions
driving following the welding of the second pipe segment, and de-
creased thereafter as ramming continued. The initial high resistan- Development of technical guidance and implementation in practice
ces were likely due to a loss of soil arching over the interval of requires the assessment of appropriate engineering calculations and
welding while the decrease in soil resistance in the later ramming validation against full-scale performance data. This paper presents
stage can be associated with reestablishment of the soil arch above the performance of a pipe-ramming installation derived from mon-
itoring a 610-mm-diameter instrumented steel pipe. Vertical ground
the pipe. Because the pipe-ramming installation was performed
movements, dynamic strains, and accelerations were observed dur-
“in the dry,” that is, above the water table, the reduction in driving
ing the production installation. Methods for the estimation of driving
resistance cannot be attributed to positive excess pore water pres-
stresses, energy transfer, soil resistance, and settlement have been
sures. The effect of piston velocity and hammer energy on the total
evaluated to examine the applicability of existing techniques devel-
(static and dynamic) soil resistance can be observed over penetra-
oped for pile-driving, pipe-jacking, and microtunneling techniques.
tion lengths of 18–21 m. The sharp reduction in total soil resistance
Ground deformations observed during the installation indicated
is partially attributed to the smaller dynamic soil resistances
that settlements and heave were relatively small, less than 15 and
10 mm, respectively. The settlement profiles observed were evalu-
700 ated using the inverted normal probability distribution approach for
PJA Total soil
600 ATVA 161 resistance tunneling and pipe jacking, including methods by O’Reilly and
Soil Resistance (kN)
Terzaghi (1943) New (1982), Cording and Hansmire (1975), and Mair and Taylor
500
Staheli (2006) Static soil (1997). Generally, these methods seem to predict the maximum
400 resistance observed settlements close to the center of the pipe; however,
CAPWAP