See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/268474427

Theoretical Model to Predict Pulling Forces for Horizontal Directional Drilling

Installations

Conference Paper · November 2012

DOI: 10.1061/9780784412619.193

CITATIONS READS

2 673

3 authors, including:

Liangxue Cai Guangli Xu

Southwest Petroleum University Southwest Petroleum University
15 PUBLICATIONS 82 CITATIONS 12 PUBLICATIONS 68 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Liangxue Cai on 16 November 2018.

The user has requested enhancement of the downloaded file.

ICPTT 2012 © ASCE 2013 1917

Theoretical Model to Predict Pulling Forces for Horizontal Directional Drilling

Installations

Liangxue Cai1, Guangli Xu2 and Limin He3

1 Doctor, School of Petroleum Engineering, Southwest Petroleum University,

Chengdu, 610500, China. Email: [email protected]
2 Doctor, School of Petroleum Engineering, Southwest Petroleum University,
Chengdu, 610500, China. Email: [email protected]
3 Professor, College of Pipeline and Civil Engineering, China Univ. of Petroleum,
Qingdao, 266555, China. Email: [email protected]

ABSTRACT

The paper presents a model for predicting pulling force during pulling back phase in
horizontal directional drilling. The analytical method develops the method given by
Polak et al. in three aspects: (1) the soil is described by Winkler model and it
constitutes an elastic support for pipe during pulling back phase, (2) considering the
nonlinear rheological properties of slurry, power law model is used in calculation of
fluidic drag friction, and (3) the resistance exerted on drill string is considered and
pulling force at carriage is calculated. Except for the weight and weight friction of
pipe outside pilotbore and fluidic drag friction, components of pulling force are
calculated considering the effect of pipe-soil interaction by two parameters, namely
pipe displacement and coefficient of chock effect which represent the displacement
of pipe-soil contacting point and the clamping effect of soil on pipe in cross section
during pulling back phase, respectively. A numerical program is compiled according
to the given method. Comparison of HDD field data and predictions of pulling force
indicates that the proposed method is rational.

KEYWORDS

Horizontal directional drilling; Pulling force; Drill string

INTRODUCTION

Horizontal directional drilling (HDD) technology has made great progress since the
first HDD rig was built in 1964 (Mohammad and Sanjiv, 2004). At present, the
maximum push-pull capacity of drill rig reaches 600 t and the applicable formations
extend from clay and sand to gravel, permafrost (Hair, 2011) and part of rock. HDD
becomes more and more popular for underground constructions of
telecommunication cables, natural gas transmission and distribution network, oil and
ICPTT 2012 © ASCE 2013 1918

gas pipelines, water pipes and power cables (Yan Chunwen, 2010). In the application
of HDD technology, the prediction of pulling forces is an important research area
which provides the basis for the design of crossing project, selection of drill rig,
stability evaluation of pipeline during constructing process, and development of drag
reduction technology. Supposing that pulling force is a linear function of the
installation distance, an empiric coefficient is regressed to predict pulling forces
according to field test data (Baumert, 2003). However, it is difficult to obtain
resistance coefficients which can be used in wide scope since too many factors affect
the pulling force and the cost of HDD experiments is high. Most scholars
(Driscopipe, 1993; Drillpath, 1996; Huey et al., 1996; ASTM, 2005; Cheng and
Polak, 2007) study the problem using analytic methods. Components of the pulling
force, including weight and weight friction of pipe, friction due to directional
changes of pilotbore, fluidic drag friction, are calculated separately and summed to
get the total pulling force.
Based on the theoretical model of pulling force prediction given by Cheng and Polak
(2007), the paper presents a new theoretical method to predict forces which develops
in three aspects, including: (1) soil described by Winkler model is treated as an
elastic body, (2) considering the nonlinear rheological properties of slurry, power law
model is used, and (3) the resistance exerted on drill string is considered. A
numerical program is coded using language C based on the new method and the
accuracy of the model is evaluated by comparison of the pulling force predictions
and field data.

PHYSICAL MODEL FOR PREDICTING PULLING FORCE

The pulling force derived from existing prediction methods is the force imposed on
pulling head (named as pulling force at pulling head). The drag resistance is
composed of the above three components and is a result of interaction between pipe,
slurry and soil. However, it is the pulling force imposed on drill string at the carriage
(named as pulling force at carriage) that is used to choose drill rig. Resistance
exerted on drill string, which is a result of interaction between drill string, slurry and
soil, is considered compared with pulling force at pulling head. The contribution of
resistance exerted on drill string to pulling force at carriage is about 30% (Baumert,
2003).
Physical model for predicting pulling force at carriage is established. Pilotbore
profile is in vertical plane for most HDD projects. Deformations of pipe and drill
string only occur in this plane. Based on the assumption, formulations referring to
mechanical analysis in two-dimensional plane are given. Pilotbore profile has to
change direction in horizontal plane to bypass obstacles in some HDD projects. In
such cases, horizontal angle changes can be used in formulations for friction due to
directional changes of pilotbore directly and the projected angle in vertical plane is
ICPTT 2012 © ASCE 2013 1919

used in formulations for weight and weight friction of pipe. Discussions and
examples in the paper is done only in vertical plane for simplicity. Physical model
and initial parameters used in calculations are given as follows.

Pilotbore profile

y
Exit
Entrance
o Ground Surface

1 x
S1

βs-s-i

2 S2 Si+1
Si
Si-1 i βs-h-i βs-h-(i+1)
βs-h-(i-1) (+ve)
3 i+1
i-1 βs-s-(i+1)
βs-s-(i-1)
Horizontal Line

Fig. 1. Example of the pilotbore profile

Pilotbore profile, i.e. the centerline of pilotbore, is assumed to be a broken line

formed by straight line segments. The discrete points where the coordinates are
determined during pilotbore drilling phase are the key points of the broken line. Let
pipe entrance be selected as the origin of the x and y coordinates, as shown in Fig. 1.
Key points from pipe entrance to exit are denoted by K1、K2、…、Kn, successively. A
pilobore profile including n key points is formed of n-1 straight segments and the
two end points of segment Si are points Ki and Ki+1. Initial parameters required in
calculations for pilotbore profile are coordinates of key points which can be obtained
in two steps, i.e. coordinate calculation and coordinate transformation according to
field records.

Layer
Soil described by Winkler model is assumed as elastic body. It deforms due to the
pipe or drill string forces during pulling-back process. The displacement w(x, y) of a
point at wall of pilotbore is in direct proportion to the stress q(x, y) at the point and is
independent of the stress of other points. The constitutive equation of Winkler model
is:
q( x, y)  kw( x, y) (1)

where: k is the modulus of foundation support in N/m3. Initial parameter required in

calculations for layer is the modulus of foundation support at key points of pilotbore
profile which can be acquired according to geological investigation data of HDD
project, structure of pilotbore and relative soil mechanics knowledge.
It is assumed that the layer type of segment Si is same as that of point Ki.
ICPTT 2012 © ASCE 2013 1920

Pipe and drill string

l
MA
A x
qB ymax
P
B
y F=P/cos(qB)
Fig. 2. Analysis of large deflection cantilever

Pipe is treated as isodiametric cylindrical shell and two physical models are used in
analysis: ① infinite pipe with outer radius rp, wall thickness tp, pipe density p and
elastic modulus Ep is located in unlimited pilotbore, and ② cantilever with length L,
flexural rigidity EI which is equal to pipe (Fig. 2). Two models are used to calculate
different parameters and detailed analysis is present later. In large scale HDD project,
ballast is often put in pipe to reduce pulling force during pulling back phase, in
which case the weight of ballast can be treated as the weight of pipe and the
equivalent density p-eq of pipe is used in calculations. Initial parameters required in
calculations for pipe and drill string are rp、tp、Ep、p(or p-eq).
Drill string is non-equant cylindrical shell since the outer diameter of drill rod joint is
larger than that of drill rod. Drill string is assumed to be isodiametric cylindrical shell
for simplicity and geometry of drill rod is used in calculations. Similar to pipe, two
models are built to analyze drill rod. Initial parameters required in calculations for
drill string are outer radius rd, wall thickness td, density d, and elastic modulus Ed of
drill rod.

Slurry flow
Ariaratnam et al. (2007) investigates rheological characteristics of slurry through
experiment. It is indicated that power law model and H-B model had good
consistency with experimental data. Power law model is most feasible for theoretical
investigation of slurry since H-B model is too complex to be used. The constitutive
equation of power law model is as follows:
  K n (2)
where: K is the consistency index in Pa sn; n is the flow index.
ICPTT 2012 © ASCE 2013 1921

RI
Slurry
r
rP
Z
Direction of Pipe Pulled
vP Pipe
RB
v(r)
Wall of Pilotbore

Fig. 3. Model for slurry flow in annulus

Model for laminar flow of power law fluid in concentric annulus is used (Fig. 3) to
describe the slurry flow. The pipe and drill string is pulled with the velocity vp in the
opposite direction of slurry flow (in the same direction after crossover point).
Cylindrical coordinate is built as shown in Fig. 3. Initial parameters required in
calculations for slurry flow are vp, K, n and density of slurry s.

MATHEMATICAL MODEL FOR PREDICTING PULLING FORCE

μAPC
Direction of Pipe Pulled F
PC B
TB
PB
μAPA MB
MA Td μBPB

A PA W
TA

Si

Fig. 4. Interaction analysis between pipe (or drill string) and soil

Components of pulling force are calculated separately and summed to gain the
pulling force at carriage. The relevant resistance calculations for pipe and drill string
in pilotbore are done piecemeal. Fig. 4 shows the interaction diagram between pipe
(or drill string) and soil. Forces exerted on pipe segment in interaction analysis
between pipe and soil include the tensions TA, TB and moments MA, MB on both ends
of pipe segment, the normal reaction forces PA, PB, PC, the weight of pipe W, the
buoyancy F and fluidic drag friction Td.
ICPTT 2012 © ASCE 2013 1922

Kci
① γs-s-i
Extension Line of Pipe
② γs-h-i
Horizontal Line
Ki ③ fp-t1-i
Line Tangent to Pipe Pa-i
①
④ ④ ft-h-i
Z ②
③ ⑤ fp-t2-i
Kci-1 Sp-(i-1) Kbi
Pa-(i-1) γAZ-h-(i-1) Sp-i
Kci+1
Ki-1 Si-1
⑤
Pa-(i+1)
Si
Ki+1
Kbi-1 (+ve)

Kbi+1

Fig. 5. Geometric details of interaction analysis between pipe and soil

A large number of geometric parameters are involved in calculations, including: the

coordinates of pipe axis, crown and bottom (denoted by Pa-i, Pc-i, Pb-i, i=1, 2, …, n) at
key point Ki, the angle βs-h-i (i=1, 2, …, n-1) between pilotbore segment Si and
horizontal line, the angle βs-s-i (i=1, 2, …, n) between pilotbore segment Si-1 and Si,
where S0 and Sn denote the ground surfaces at pipe entrance and exit separately, as
shown in Fig. 1. The two ends of pipe segment Sp-i (i=1, 2, …, n-1) are key points Pa-i
and Pa-(i+1), the angle γs-h-i (i=1, 2, …, n-1) between pipe segment Sp-i and horizontal
line, the angle γs-s-i (i=1, 2, …, n) between pipe segment Sp-(i-1) and Sp-i, where Sp-0
and S p-n denote the pipes on ground surfaces at entrance and exit separately, the
angle ft-h-i (i=1, 2, …, n) between the line tangent to the pipe at key point Pa-i and
horizontal line, fp-t1-i and fp-t2-i (i=1, 2, …, n-1) denote the angles between pipe
segment Sp-i and the lines tangent to the pipe at key points Pa-i and Pa-(i+1) separately,
as shown in Fig. 5. All angles involved in the paper are positive anticlockwise. The
detailed formulations for calculating components of pulling force at carriage when
the pipe is pulled back to key point Ki are present as follows.

Weight and weight friction of pipe

In pulling back process, the pipe can be divided into two parts, i.e. pipes outside or
inside pilotbore according to its location. The pulling force component (Tg)i to
overcome the frictional resistance resulting from the pipe weight outside pilotbore
and the pipe’s weight outside pilotbore is:
 i 1

(Tg )i   wP g cos 0  wP sin 0   L   Lk  (3)
 k 1 

where: wP is the weight of pipe per unit length, g is the friction coefficient between
ICPTT 2012 © ASCE 2013 1923

pipe and ground, L is the total length of pipe, Lk is the length of pilotbore segment Sk
representing portions of pipe that has been pulled into pilotbore and  0 is the angle
between horizontal line and the ground surface in front of pipe entrance.
The pulling force component (Tb)i to overcome the frictional resistance resulting
from the submerged weight of pipe inside pilotbore and the pipe’s submerged weight
itself is calculated by:
i 1
(Tb )i   ( Kc-k Lk wb-k cos s-h-k  Lk w sin s-h-k ) (4)
k 1

where: Kc-k is the coefficient of chock effect relevant to pilotbore segment Sk, which
indicates the clamping effect of soil on the pipe, ｗ is the submerged weight of pipe
per unit length, b-k is the friction coefficient between soil and pipe at pilotbore

segment Sk.

Frictions due to directional changes of pilotbore

The pulling force increases when pipe negotiates the corner of pilotbore because the
directional changes of pulling force and bending stiffness of pipe can increase the
interaction force between pipe and soil, which can be named by capstan effect and
bending effect separately. The calculating method of two resistances when pipe
negotiates the corner Ki is present.

Capstan effect The increase of pulling force resulting from capstan effect, named by
capstan force, can be calculated by means of static equilibrium analysis. Chehab
(2008) derived the formulation for calculating capstan force. However, the clamping
effect of soil on the pipe is not considered in the analysis. Here the formulation to
calculate capstan force considering the clamping effect is derived. It is assumed that
the contact segment as pipe negotiates the corner is circular arc with radius R and
central angle β, as shown in Fig. 6. Tensile forces T1 and T2 act on the arc pipe
segment along tangent direction at two ends. Since the change in the tensile force
over infinitesimal element dT is insignificant compared to T, the normal force N is
assumed to act along the line bisecting dβ. The increase of pulling force F1 due to
capstan force when pipe negotiates the corner at Ki is given by:
 
F1  TKi eb-i Kc-i s-s-i  1 (5)

where, TKi is the total resistance force acting on pipe before it negotiates the corner
Ki, b-i is the friction coefficient between soil and pipe at Ki, Kc-i is the

corresponding coefficient of chock effect at point Ki.

ICPTT 2012 © ASCE 2013 1924

Δβ
dβ

Direction of Pipe Pulled

T1

T2
N T

T+dT Fr

Fig. 6. Analysis of Capstan effect

Bending Effect As pipe negotiates the corner of pilotbore, interaction with pilotbore
walls results in frictional force against the direction of movement because of the
bending stiffness of pipe. The shape of pipe inside pilotbore needs to be estimated to
calculate the increase of pulling force due to bending effect, since there are
corresponding formulations for different scenarios. Two assumptions are used in
analysis: ① the line tangent to pipe at the contact point is perpendicular to the biset
line of the angle when pipe bends around the corner of pilotbore, ② the displacement
of contact point is along normal direction when pipe and soil interact.
The shape of pipe inside pilotbore is determined by the structure of pilotbore profile.
Soil is assumed to constitute a stiff support for pipe in the beginning of calculations.
Initial geometric parameters for interaction analysis between pipe and soil are
estimated since the shape of pipe can be determined by the geometry of pilotbore.
There are two primary scenarios (Fig. 7) to model the geometry of pipe inside
pilotbore and the criteria to distinguish the scenarios are:
s-s-i  s-s-i 1  0 Scenario 1

s-s-i  s-s-i 1  0 Scenario 2

The scenarios of pipe need to be further subdivided for calculating frictional force
resulting from bending effect. Cheng and Polak (2007) have presented the criteria to
distinguish the detailed scenarios: scenario 1 is subdivided into three scenarios
according to γAZ-h-i, ft-h-i and the maximum deflection of pipe, and scenario 2 is
subdivided into two scenarios according to the maximum deflection of pipe. The
details are not repeated in this work.
ICPTT 2012 © ASCE 2013 1925

Scenario 1 Scenario 2

Ki+1
B
A B
Z Ki Ki+1
A
Ki Si Si

Scenario 1a Scenario 2a
Ki+1
A B
Ki A B
Ki Ki+1
Si
C1 C2
C2 Si
Scenario 1b
Ki+1
B

A
Ki Si
C1

Fig. 7. Geometric configurations for pipe in a segment AB

Pipe model ② is used to analyze the interaction between pipe and soil and the large
deflection theory is utilized in order to accommodate cases of large deformations
happening due to abrupt changes in the pilotbore profile. The basic equations for
calculating contact force in vertical plane are present as follows (Ge and Chu, 1997).
The ratio of the maximum deflection ymax (occurs at the free end B) and the
projection length l of deformational cantilever in the x-axis is calculated by:
ymax Fy (q B )
 (6)
l Fx (q B )

qB cos q
where, Fx (q B )   dq (7)
0
sin(q B  q )

qB sin q
Fy (q B )   dq (8)
0
sin(q B  q )

Integral functions Fx (qB ) and Fy (q B ) are dimensionless transcendental functions

of rotary angle θ of cantilever and can be calculated using numerical integration.

The normal force P acting on pipe is given as:
EI cos q B 2
P  Fx (q B ) (9)
2l 2
There are two cantilevers for each pipe segment. The normal forces are calculated
separately and are added to that of relevant contact point. The increases of normal
force acting on pipe for two contact points Ki and Ki+1 are zero if the pipe segment Si
is in scenario 1a, i.e. Pi(Si)=Pi+1(Si)=0. Based on the large deflection theory, there are
two ways to calculate the normal force of cantilever for other four scenarios: (1)θB
ICPTT 2012 © ASCE 2013 1926

can be determined using Eqs. (6)—(8) when ymax and l are given, then the normal
force P is calculated using Eq. (9). Scenario 1 and 2 adopt this method. And (2)
combined with the geometry of pipe inside pilotbore, l can be calculated when θB is
given using the following equation:
2 RB cosq B  rp  rp cosqB
l (10)
tan q B  Fy q B  Fx q B 
Then the normal force P is calculated using Eq. (9). Scenario 1b and 2a adopt this
method. The formula for calculating the given parameters are listed in Table 1.
The normal force Pi acting on pipe at point Ki is determined based on the assumption
that soil is rigid body. As soil described by Winkler model constitutes an elastic
support for pipe, the displacement sp of contact point (named by pipe displacement)
occurs due to the contact force between pipe and soil. The detailed calculation of sp is
present in following paper. Based on the assumption ② in this section, the
coordinates of contact points are rearranged according to the obtained displacements.
Then the normal forces at each point are re-calculated assuming stiff soil. Repeat the
above steps and iterative computation does not end until the normal forces are no
longer changed. It should be noted that there are two other contact points (denoted by
C1 and C2, as shown in Fig. 7) except Ki and Ki+1 when pipe segment Sp-i is in
Scenario 1b or 2a. The normal forces (denoted by Ps-i-1 and Ps-i-2) at C1 and C2 are
also involved in the criteria to end the iterative computation. Let Pi and Kc-i denote
the normal force and coefficient of chock effect at point Ki when iterative
computation ends and the increase of pulling force F2 due to bending effect at point
Ki is:
F2  Kc-i b-i Pi  Kc-i b-i Ps-i -1  Kc-i b-i Ps-i -2 (11)

Table 1. Formula to calculate the given parameters for different scenarios

scenarios of Sp-i ymax l θB
Ki lAZ·sin(|γAZ-h-i-fp-t1-i|) lAZ·cos(|γAZ-h-i-fp-t1-i|) -
Scenario 1*
Ki+1 lBZ·sin(|γBZ-h-i-fp-t2-i|) lBZ·cos(|γBZ-h-i-fp-t2-i|) -
Ki lAB·sin(|fp-t1-i|) lAB·cos(|fp-t1-i|) -
Scenario 2
Ki+1 lAB·sin(|fp-t2-i|) lAB·cos(|fp-t2-i|) -
Ki - - |βs-h-i-ft-h-i|
Scenario 1b
Ki+1 - - |βs-h-i-ft-h-(i+1)|
Ki - - |βs-h-i-ft-h-i|
Scenario 2a
Ki+1 - - |βs-h-i-ft-h-(i+1)|
*Inflection point Z is determined using the small deflection theory (Cheng and Polak,
2007). lAZ and lBZ denote the lengths of pipe segment AZ and BZ separately, and
γAZ-h-i, γBZ-h-i denote the angles between AZ, BZ and horizontal line.
The increase of pulling force due to directional changes of pilotbore as pipe
ICPTT 2012 © ASCE 2013 1927

negotiates point Ki is given by:

Tf-i  F1  F2 (12)

Fluidic drag friction

During pulling back phase, slurry flows to ground surface in the opposite direction of
pipe moves and it changes the direction of flow 1/2 to 2/3 into pipe installation in the
pilotbore (Knight et al., 2001). Shear stress exerts on the outer surface of pipe from
slurry when relative movement occurs between pipe and slurry. It is kind of
resistance for pipe pulled and is named by fluidic drag friction. Contribution of
fluidic drag friction to pulling force at pulling head is less than 10% (Cai et al., 2010)
and the prediction of pulling force is more accurate when fluidic drag friction is
considered (Duyvestyn, 2009).
Based on the physical model of slurry flow, the velocity components of slurry in the
radial and circumferential directions are zero assuming that the effect of gravity is
ignored. The velocity v of slurry distributes symmetrically with respect to z-axis and
depends on r alone. The momentum conservation equation for incompressible fluid
of z-axis in the cylindrical co-ordinates is simplified as the following equation:
1 dr  dp
 0 (13)
r dr dz
where: τ is the shear stress of slurry, dp/dz is the pressure gradient through z-axis.
The boundary conditions are  =0 at r= RI, v=vP at r=rp and v=0 at r= RB , where RI,

rp and RB are radii of the interface of maximum velocity, the pipe and the borehole,
respectively. By applying the boundary conditions and Eq. (2) the following
equations are obtained for the velocity of the slurry:
 1
n  R2
1

v  r  1  dp    n

 p  
rp 2 K

     I  r  dr rp  r  RI
 dz    r 
v r    1 1
(14)
 RB  1  dp   n  RI2  n
       r   dr RI  r  RB
 r  2 K  dz    r 
The total discharge of slurry is given by:
RB
Q   2 r  v  r   dr (15)
rp

RI can be calculated by Eq. (14) since the results of two equations are same at r= RI.
Then Eq. (14) is substituted into Eq. (15) to find the pressure gradient dp/dz. Both
calculations can be solved through method of function approximation. The pulling
force component to overcome fluidic drag friction when pipe head negotiates point
Ki is:
ICPTT 2012 © ASCE 2013 1928

n
 dv  r   i 1
Td i K    Dp  Lk (16)
 dr r r  k 1
 p 

In calculation of fluidic drag friction, the crossover point is assumed to be located at

7/12 into pipe installation according to excavations completed by Knight et al. (2001).
The total discharge Q of slurry is equal to the sum of pump discharge and discharge
of slurry extruded from pilotbore when pipe is pulled in while Q decreases to zero
after crossover point occurs. It should be noted that fluidic drag friction is still
greater than zero when Q decreases to zero.

Resistance exerted on drill string

Similar to pipe inside pilotbore, resistance forces acting on drill string include weight
and weight friction of drill string, frictions due to directional changes of pilotbore
and fluidic drag friction. It should be noted that the discharge of slurry in annulus
between drill string and pilotbore is zero before crossover point occurs and fluidic
drag friction acting on drill string is resistance. The direction of slurry flow is same
as that of drill string pulled after crossover point occurs and fluidic drag friction
acting on drill string can becomes motive force in certain situation. The pulling force
component to overcome resistance exerted on drill string when pipe is pulled at point
Ki is given by:
n
(Ts )i  (Tb' )i  (Td' )i   Tf-' k (17)
k i

where: (Tb' )i is weight and weight friction of drill string, (Td' )i is fluidic drag friction

exerted on drill string, Tf-k' is the increase of pulling force when drill string

negotiates the corner at point Kk.

Pulling force at carriage

After evaluating all the components of pulling force, the pulling force at carriage
when pipe is pulled at point Ki can be calculated from the equilibrium:
i 1
Ti  (Tg )i  (Tb )i  (Td )i   Tf-k  (Ts )i (18)
k 1

ANALYSIS OF HDD INSTALLATION TESTS

Table 2. Parameters of HDD installations in University of Waterloo

parameter Dp tp p Ep Dr tr r Er
-3
unit mm mm kg·m MPa mm mm kg·m-3 GPa
ICPTT 2012 © ASCE 2013 1929

HD3-2 168 15.3 920 700 71 10.9 7800 210

HD3-3 219 12.9 920 700 71 10.9 7800 210
HD3-2 250 120 1200 6.4366 0.3063 0.05 0.59 0.244
HD3-3 300 120 1200 6.4366 0.3063 0.05 0.63 0.256

The University of Waterloo initiated a field research program to study the behavior
of pipes during installations using HDD in 1996 and the third phase of the program
was executed in 2001. Three pipes (HD3-1, HD3-2, HD3-3) were installed and
pulling forces at pulling head were collected in tests (Duyvestyn, 2004). Data of
HD3-2 and HD3-3 is chosen to evaluate the accuracy of predicting model proposed
in the paper. All parameters used in calculations are listed in Table 2. The values of
friction coefficients between pipe and soil are determined using method given by Cai
(2011) and values for rheological parameters of slurry are calculated according to
experimental data measured by Ariaratnam et al. (2003).

30 30

Predicted pulling force at pulling head

25 Predicted pulling force at carriage 25
Measured pulling force at pulling head

20 20
Pulling Force /kN
Pulling Force /kN

15 15

10 10
Predicted pulling force at pulling head
Predicted pulling force at carriage
5 5 Measured pulling force at pulling head

0 0
-20 0 20 40 60 80 100 120 140 160 180 -20 0 20 40 60 80 100 120 140 160 180

Distance /m Distance /m

HD3-2 HD3-3
Fig. 8. Comparison of field tests results and pulling force predictions

Fig. 8 shows the measured pulling force at pulling head, predictions of pulling force
at pulling head and pulling force at carriage for HD3-2 and HD3-3. It is indicated
that predicting results are very close to the measured values in the range of
installation distance and theoretical model presented in the paper is rational. During
the whole phase of pulling back, pulling force at carriage is always higher than
pulling force at pulling head except for the termination at which two values are
equal.

CONCLUSIONS

A theoretical model for predicting pulling forces during pulling back phase in
horizontal directional drilling installations is built by analyzing four components of
pulling force and creating physical models to calculate the corresponding component
ICPTT 2012 © ASCE 2013 1930

in the paper. The analytical method develops the method given by Polak et al. in
three aspects: (1) the soil is described by Winkler model and it constitutes an elastic
support for pipe during pullback, (2) considering the nonlinear rheological properties
of slurry, power law model is used in calculations, and (3) the resistance exerted on
drill string is considered and pulling force at carriage is calculated. Except for the
weight and weight friction of pipe outside pilotbore and fluidic drag friction,
components of pulling force are calculated considering the effect of pipe-soil
interaction by two parameters, namely pipe displacement and coefficient of chock
effect which represent the displacement of pipe-soil contacting point and the
clamping effect of soil on pipe in cross section during pulling back phase,
respectively. Comparison of field test data and prediction results shows that the
accuracy and reliability of the model is high.

ACKNOWLEDGEMENTS

This work was financially supported by the fundamental research funds (natural
science) (No.2012XJZ002) and the special fund of China’s central government for
the development of local colleges and universities —— the project of national
first-level discipline in Oil and Gas Engineering (No. 030009001).

REFERENCES

Ariaratnam, S.T., Harbin, B.C., Stauber, R.L., 2007. Modeling of annular fluid
pressures in horizontal boring. Tunnelling and Underground Space Technology
22, 610-619.
Ariaratnam S. T., Stauber R. M., Bell J., et. al.. Evaluation of rheologic properties of
fluids returns from horizontal directional drilling. NASTT/ISTT No Dig
Conference. Arlington, 2003.
ASTM F1962-05, 2005. Standard guide for the use of maxi-horizontal directional
drilling for placement of polyethylene pipe or conduit under obstacles, including
river crossings. ASTM International, West Conshohocken.
Baumert, M.E., 2003. Experimental investigation of pulling loads and mud pressures
during horizontal directional drilling installations. Ph.D. Thesis, University of
Western Ontario, London, Ontario.
Cai, L.X., 2011. Investigation on dynamics of pipe during pullback in horizontal
directional drilling installations. Ph.D. Thesis, China University of Petroleum,
Qingdao, China.
Cai, L.X., He, L.M., Wang, X., et al., 2010. Prediction model of pulling loads in
horizontal directional drilling installations and its characteristic parameter
sensitivity analysis. Journal of China University of Petroleum 34(6), 114-119.
 ICPTT 2012 © ASCE 2013 1931

Chehab, A.G., 2008. Time dependent response of pulled-in-place HDPE pipes. Ph.D.
Thesis, Queen’s University, Kingston, Ontario.
Cheng, E., Polak, M.A., 2007. Theoretical model for calculating pulling loads for
pipes in horizontal directional drilling. Tunnelling and Underground Space
Technology 22, 633-643.
DrillpathTM, 1996. Drillpath theory and user’s manual. Infrasoft L L C, Houston,
Texas, USA.
Driscopipe, 1993. Technical expertise application of driscopipe in directional drilling
and river crossings. Tech. Note No. 41.
Duyvestyn, G., 2009. Comparison of predicted and observed HDD installation loads
for various calculation methods. ISTT No-Dig, Toronto, Ontario.
Duyvestyn, G., 2004. Field and numerical investigation into pipe bursting and
horizontal directional drilling pipeline installation ground movements. Ph.D.
Thesis, University of Waterloo, Waterloo, Ontario.
Duyvestyn, G., Knight, M., Polak, M.A., 2001. Excavation and analysis of two
HDPE pipes installed using HDD construction techniques. No-Dig 2001.
Nashville.
Ge, R.H., Chu, ZH.J., 1997. A way to calculate the deformation of great flexibility of
slender long overhanging beam under concentrated load. Chinese Journal of
Applied Mechanics 14(4), 71-77.
Hair, J.D., 2011. Considerations in the application of horizontal directional drilling to
pipeline construction in the arctic. OTC Arctic Technology Conference,
Houston, Texas, USA. 22086-MS.
Hu, B., 2005. Underground pipelines protection in directional drilling crossing
construction. Shanghai Gas 4, 5-7.
Huey, D.P., Hair, J.D., McLeod, K.B., 1996. Installation loading and stress analysis
involved with pipelines installed in horizontal directional drilling. No-Dig, New
Orleans.
Knight, M., Duyvestyn, G., Gelinas, M., 2001. Excavation of surface installed
pipeline. Journal of Infrastructure Systems 7(3), 123-129.
Mohammad, N., Sanjiv, G., 2004. Trenchless technology: pipeline and utility design,
construction, and renewal. McGraw-Hill Professional.
Shu, H.M., Tong, X.H., 2002. Engineering finite element method. China University
of Petroleum Press, Dongying, China, pp. 93-97.
Yan, CH.W., 2010. Trenchless technology review and survey in China and USA.
Trenchless Technology 1, 1-9.

View publication stats