Effect of Pipe Stiffness on Maxi-HDD Pull Loads

Lawrence M. Slavin, Ph.D., A.M.ASCE1 and Mohammad Najafi, Ph.D., P.E., M.ASCE2

Abstract: The ASTM standard F1962, Standard Guide for Use of Maxi-Horizontal Directional Drilling for Placement of Polyethylene Pipe
or Conduit Under Obstacles, Including River Crossings, provides a procedure for estimating pull loads and stresses on polyethylene (PE) pipe
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/04/15. Copyright ASCE. For personal use only; all rights reserved.

as primarily a function of the drilled path and buoyant weight of the pipe in the borehole. Although the (capstan) effect of tension at route
bends is reflected in the analysis, it is assumed that pipe bending stiffness effects may be ignored for the typically very flexible PE material.
However, it is of interest to gain a quantitative understanding of the potential impact of such additional factors on the required pull force in
somewhat stiffer pipe materials, such as polyvinyl chloride (PVC) or steel. The present investigation therefore focuses on an analytical means
of estimating the associated frictional drag forces, allowing an appreciation of the possible importance of such effects and, as appropriate, an
extension of the general ASTM F1962 procedure to a broader range of pipe materials. The magnitude of the additional installation forces
depends on the bending stiffness of the pipe, as well as the nature of the bend(s) or path curvature and the pipe and borehole dimensions. The
results help provide a better understanding of the factors influencing the pull loads in the maxi-horizontal directional drilling (maxi-HDD)
installation of pipelines. DOI: 10.1061/(ASCE)PS.1949-1204.0000092. © 2012 American Society of Civil Engineers.
CE Database subject headings: Drilling; Pipes; Stiffness; Polyethylene; Load factors.
Author keywords: Directional drilling; HDD; Pipes; Stiffness; Pull load; ASTM F1962.

Introduction buoyancy) and tension at route bends (capstan effect), and any
contribution caused by fluidic drag. The objective of the present
The ASTM F1962, Standard Guide for Use of Maxi-Horizontal paper is to provide an analytical basis for estimating the additional
Directional Drilling for Placement of Polyethylene Pipe or Conduit drag forces caused by finite pipe stiffness, thereby determining
Under Obstacles, Including River Crossings, provides overall their relative importance and, if appropriate, allowing a means
guidelines for a maxi-horizontal directional drilling (maxi-HDD) of extending the original formulas in ASTM F1962 to a broader
operation. This document addresses preliminary site investigations, range of pipe materials.
safety and environmental considerations, regulations, damage
prevention, bore path design, project implementation, inspection, Flexible (Polyethylene) Pipe
and site cleanup. One of the more significant contributions of
ASTM F1962 is the provision of a rational, analytical method Fig. 1 illustrates a typical geometry for a maxi-HDD operation cor-
for selecting the polyethylene pipe strength requirements based responding to a river crossing, similar to that shown in ASTM
on the estimated installation loads on the polyethylene (PE) pipe. F1962. The horizontal projection of the pipe path comprises four
Thus, ASTM F1962 provides a means of determining project fea- segments, including those spanning the pipe entry to exit point (L2 ,
sibility and initial design details. L3 , and L4 ) and the additional length L1 . The quantity L1 allows for
The original equations in the ASTM standard for determining handling at both ends and other possible effects (e.g., path curva-
required pull loads were developed assuming negligible pipe bend- ture, thermal contraction, stretching). The projected bore length
ing stiffness, which is characteristic of PE material. Other pipe ma- (Lbore ) is given by Eq. (1) as follows:
terials, including relatively rigid plastics [e.g., polyvinyl chloride Lbore ¼ L2 þ L3 þ L4 ð1Þ
(PVC)] and more obviously, iron or steel, may lead to appreciable
reaction forces at route bends and path curvature as the pipe is The entry segment (A–B) and exit segment (C–D) may each be
pulled along the borehole during the horizontal directional drilling of uniform curvature, but not necessarily. The intermediate hori-
operation. These reaction forces correspond to additional sources of zontal segment (L3 ) may be of zero length depending on project
frictional drag on the pipe, impeding its placement, requiring geometry. The term H represents the depth of the installation rel-
greater installation tension than may otherwise be required consid- ative to the elevation at the pipe entry and exit points.
ering only the frictional drag due to the pipe weight (including Using the terminology presented in Fig. 1 and Eq. (1), ASTM
F1962 provides a set of relations to predict the required pull force
1
President, Outside Plant Consulting Services, Inc., 15 Lenape Ave., (T A , T B , T C , and T D ) corresponding to the leading end of the poly-
Rockaway, NJ07866-1019 (corresponding author). E-mail: [email protected] ethylene pipe reaching point A, B, C, and D, as presented in
2
Director, Center for Underground Infrastructure Research and Eqs. (2a)–(2d) as follows:
Education (CUIRE), Univ. of Texas at Arlington, 428 Nedderman Hall,
Arlington, TX 76019-0308. E-mail: [email protected] T A ¼ eν a α · ν a · wa · ðL1 þ L2 þ L3 þ L4 Þ ð2a Þ
Note. This manuscript was submitted on April 14, 2011; approved on
July 6, 2011; published online on January 17, 2012. Discussion period open T B ¼ eν b α · ðT A þ ν b · jwb j · L2 þ wb · H
ν a · wa · L2 · eν a α Þ
until July 1, 2012; separate discussions must be submitted for individual
papers. This paper is part of the Journal of Pipeline Systems Engineering ð2b Þ
and Practice, Vol. 3, No. 1, February 1, 2012. ©ASCE, ISSN 1949-1190/
2012/1-22–31/$25.00. T C ¼ T B þ ν b · jwb j · L3
eν b α · ðν a · wa · L3 · eν a α Þ ð2c Þ

22 / JOURNAL OF PIPELINE SYSTEMS ENGINEERING AND PRACTICE © ASCE / FEBRUARY 2012

J. Pipeline Syst. Eng. Pract. 2012.3:22-31.

 bore / pipe path pipe entry

β A α
D

H
C B
pipe exit

L4 L3 L2 L1

Lbore

Fig. 1. Geometry for maxi-HDD operation (reprinted with permission of Outside Plant Consulting Services, Inc.)
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/04/15. Copyright ASCE. For personal use only; all rights reserved.

T D ¼ eν b β · ðT C þ ν b · jwb j · L4
wb · H 3. Direct resistance or assistance to the movement of the sub-

merged pipe to a point of greater or lesser depth, due to the

eν b α · ½ν a · wa · L4 · eν a α Þ ð2d Þ net buoyant forces (positive upward); and
4. Distributed fluidic drag due to the flow of the slurry along the
where wa = empty aboveground weight (downward positive) of the
length of the pipe.
pipe; and wb = net buoyant weight (upward positive) of the pipe as
The various terms such as “ν b · jwb j · L2 ” or “ν a · wa · L2 ” in
submerged in slurry below ground; ν a and ν b = corresponding
Eqs. (2a)–(2d) reflect the primary contribution to the required pull
aboveground and belowground Coulomb “coefficients of friction.”
force, due to frictional drag resulting from the effective weights of
Coulomb friction assumes that the surface drag (force per unit area)
the pipe. The exponential terms (eν a α , eν b α , and eν b β ) represent am-
is proportional to the local normal pressure between two surfaces.
plification factors due to the tension at route bends (capstan effect).
The buoyant weight may reflect the use of antibuoyancy tech-
The wb · H terms correspond to the direct impact of the buoyant
niques, including the use of liquid ballast (e.g., water) inside the
pipe. The pipe entry angle (α) and exit angle (β), relative to the forces in resisting or assisting the pipe placement along the seg-
horizontal, are expressed in radians (1 rad equals 180° ÷ π). For ments A–B or C–D, respectively. The contribution of the fluidic
a bore path of approximately uniform curvature from the entry drag is estimated by the incremental tensile force term of Eq. (4).
and exit points to the horizontal segment (B–C), the lengths L2 Additional details and discussions regarding the development of
and L4 may be estimated as the following: this model are provided by Slavin and Petroff (2010).
An additional (fifth) component that would typically contribute
L2 ¼ 2H=α ð3a Þ to frictional drag on the surface of the pipe, further increasing the
required pull force, is that which is due to any finite pipe bending
L4 ¼ 2H=β ð3b Þ stiffness. Such effects are presently not considered in the ASTM
F1962 model, due to the minimal impact for polyethylene pipe.
In addition to the calculated loads as given by Eqs. (2a)–(2d), an The objective of the present paper is to estimate the magnitude
incremental tensile force, ΔT, must be added to account for the of these effects, as appropriate for other materials, which would
drag effect of the drilling fluid/slurry (fluidic drag), which is deter- presumably be a function of the particular pipe and installation
mined from the magnitude of the hydrokinetic pressure (ΔP) as parameters (e.g., pipe stiffness, diameter, borehole clearance, and
follows: radius of curvature).
Fig. 2 illustrates the general manner in which the bending stiff-
ΔT ¼ ΔP · ðπ=8Þ · ðd 2h
d 2p Þ ð4Þ ness of a pipe results in additional normal (reaction) forces and in-
creased frictional drag force, where the pipe comes into contact
where d h = diameter of borehole; d p = outer diameter of PE pipe; with the inner surface of the borehole at a change in curvature along
and ΔP = incremental drilling fluid pressure in the borehole at the the path. The illustrated curvature represents one type of path
leading end of the pipe during the pullback operation, which is in geometry, corresponding to a single discrete bend, to be sub-
addition to the hydrostatic pressure corresponding to the head sequently defined. Other types of borehole curvature may include
(depth) of relatively dense slurry. The incremental tension (ΔT)
is properly added to the local tension (T A , T B , T C , or T D ) as speci-
fied in Eqs. (2a)–(2d), for each of the four points.

ASTM F1962 Model and Pipe Stiffness

The development of the engineering model in ASTM F1962 for
estimating the required pull forces on the pipe during the maxi-
HDD operation considers the following four effects:
1. Distributed normal pressure, and associated frictional drag, due
to the effective (buoyant) weight of the pipe, both within and
outside the borehole;
2. Normal pressure, and associated frictional drag (capstan Fig. 2. Additional forces as a result of bending stiffness of pipe
effect), due to the tension at path bends (e.g., points A, B, (reprinted with permission of Outside Plant Consulting Services, Inc.)
and C);

JOURNAL OF PIPELINE SYSTEMS ENGINEERING AND PRACTICE © ASCE / FEBRUARY 2012 / 23

J. Pipeline Syst. Eng. Pract. 2012.3:22-31.

 a sequence of relatively small magnitude bends, representing path the center of the hole, with straight borehole sections immediately
undulations, not considered in the present analysis. preceding and following the bend, beyond which other discrete
The reaction forces required to deflect a “stiff” pipe to conform bends, of possible different radii, may exist. Although other, more
to a vertical route bend, as shown in Fig. 2, would be lower than general, route bends, such as non-uniform radius of curvature,
otherwise required due to the beneficial effect of the distributed may occur in practical applications, the present model is useful
buoyant forces also acting laterally on the pipe that help it conform for understanding the potential magnitude and significance of the
to the local curvature. Similarly, any significant tension that may effect of bending stiffness in HDD installations.
have previously developed at the bend, will exert a local bending The angle θb at the point of closest contact of the pipe with the
moment that also helps reduce the otherwise-required reaction borehole wall at the bend, is less than or equal to half of the total
forces. As a result, without direct consideration of these effects, bend of angle γ. For the case shown in Fig. 3 (complete bend), the
the additional frictional drag attributed to the bending stiffness finite difference between twice the angle θb and the full angle γ
may be significantly overestimated. Therefore, although a reason- corresponds to a pipe segment of uniform curvature, which is es-
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/04/15. Copyright ASCE. For personal use only; all rights reserved.

able degree of conservatism is desirable and consistent with the sentially equal to the curvature of the borehole itself. The pipe also
original intent of ASTM F1962 (Slavin and Petroff 2010), the experiences uniform (zero) curvature along its straight sections
present investigation will briefly consider or discuss these effects, beyond the distances f
as appropriate. Since some pipe materials may not display conventional linear
elastic properties, engineering judgment may be required to select
an appropriate value for the modulus E. For example, for a visco-
Discrete Bend
elastic plastic material, the effective modulus depends on the du-
The magnitude of the reaction forces indicated in Fig. 2 depends on ration of the load or deflection corresponding to the time required
the geometric details of the bore path and the pipe characteristics, for a portion of the pipe to enter and deform to a route bend, remain
including the outer diameter and bending stiffness. The significant at the given curvature, and then exit from such a bend. The appro-
parameters required for determining the forces are shown in Fig. 3, priate moduli may therefore have a tendency to change somewhat
and are defined as: as the pipe traverses a bend. In general, however, the time frames of
• d h = borehole diameter interest would not be large (but rather, hours or minutes) and any
• d p = pipe diameter change in the effective modulus would not be expected to be major
• f = horizontal (projected) distance to point (1) of closest contact (Plastics Pipe Institute 2008). Thus, for the present purposes, the
of pipe with borehole straight segment pipe segment between the reaction forces N 1 and N 2 (or between
• E = pipe effective elastic modulus N 3 and N 4 ) may reasonably be considered to act as an elastic beam.
• I = pipe area moment of inertia For pipes joined at discrete lengths, it is assumed that the cou-
• l = horizontal (projected) distance between points (1 and 2) pling flexes similar to the native pipe (e.g., fused PE) or that the
of contact of pipe with borehole straight and curved segments coupling is a local rigid element with insignificant effect on the
• N i = local reaction force at point i (i ¼ 1, 2, 3, or 4) overall flexibility or size of the joined pipeline. The analysis, there-
• γ = total bend angle (radians) fore, does not directly consider the effect of restrained joints that
• θb = partial angle, at point 2, of closest contact of pipe with bore- may allow a finite deflection angle at each coupling while travers-
hole bend (radians), for “complete” bend (defined subsequently) ing a route bend.
• ρh = borehole radius of curvature at centerline
Regarding Fig. 3, the geometry is clearly not drawn to scale, as
the borehole radius of curvature would typically be at least two Theoretical Analysis
orders of magnitude greater than the borehole (or pipe) diameter.
In general, the physical phenomena associated with pulling a pipe
It may also be assumed, and may be subsequently verified, that the through a borehole are similar to those encountered when placing
angle θb (“complete bend”, as subsequently discussed) is relatively a cable within a duct or conduit. For example, the capstan effect
small; i.e., θb ≪ 1 radian. The X–Y coordinates (not shown) are incorporated in the ASTM F1962 model is analogous to that an-
defined with their origin at the center of the pipe cross-section, alyzed by Buller (1949) and Rifenburg (1953) for pulling power
at the point of closest contact with the straight segment of the bore- cables through conduits. Similarly, the method for estimating the
hole (point 1). reaction forces due to pipe stiffness at bends along the bore path
Furthermore, it is assumed that the local discrete bend along may follow that employed by Griffioen (1993) for determining the
the borehole is characterized by a uniform radius, measured to effects of cable stiffness when placing cables within ducts.
Several observations may be made that facilitate the analysis.
Assuming that θb is relatively small, and ignoring tension and
weight or buoyancy forces, the reactions forces N i are approxi-
mately equal on the basis of equilibrium considerations and sym-
metry. Thus
N ¼ N1 ¼ N2 ¼ N3 ¼ N4 ð5Þ

where N = local reaction force, which appears four times at

each bend.
In addition, the pipe and borehole dimensions (d p and d h ,
respectively) which are on the order of inches (possibly several
feet for a very large pipe installation), are negligible compared
the borehole radius (ρh ), which is typically on the order of several
Fig. 3. Terminology at discrete (complete) bend (reprinted with
hundred feet (1 inch ¼ 2:54 cm; 1 foot ¼ 30:48 cm). Thus, the ra-
permission of Outside Plant Consulting Services, Inc.)
dius of curvature of the pipe, at the portion where it conforms to the

24 / JOURNAL OF PIPELINE SYSTEMS ENGINEERING AND PRACTICE © ASCE / FEBRUARY 2012

J. Pipeline Syst. Eng. Pract. 2012.3:22-31.

 inner surface of the borehole at the bend (Fig. 3), is essentially
equal to that of the bend itself, ρh .
To estimate the magnitude of the reaction forces, the pipe seg-
ment between points 1 and 2 is assumed to deflect as an elastic
beam, fixed at point 2, with characteristic stiffness (EI). The three
unknowns N, l, and θb , may then be determined by three indepen-
dent conditions relating the pipe deflection, slope, and curvature
with the geometry of the borehole. For example, on the basis
of the first two (deflection and slope) compatibility conditions
(Young 1989), and assuming θb ≪ 1, then

N · l3 =ð3EIÞ ≈ l tan θb
½ρh · ð1
cos θb Þ þ ðd h
d p Þ

Fig. 4. Magnitude of “complete” bend angle as function of installation
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/04/15. Copyright ASCE. For personal use only; all rights reserved.

≈ l · θb
½ρh · ðθ2b Þ=2 þ ðd h
d p Þ ð6Þ geometry

and
relatively large 300-mm (12-in.) pipe for such an installation would
N · l2 =ð2EIÞ ¼ tan θb yield a somewhat greater angle of 0.17 rad, which is still considered
≈ θb ð7Þ much less than 1; more typical, larger radii, such as 60 m (200 ft) or
greater, correspond to smaller values. Fig. 4 shows the functional
Dividing Eq. (6) by Eq. (7) yields dependence of the bend angle θb on the ratio of the clearance to
the borehole radius, for the typical range of values of the latter
l ¼ 3½ρh · ðθb Þ=2 þ ðd h
d p Þ=θb  ð8Þ parameter.
Eq. (14) may now be combined with Eq. (11) to obtain a direct
Furthermore, based on Fig. 3
expression for the reaction force N, redesignated as N b to empha-
l ≈ f þ ρh · sin θb size its applicability for the case of a complete bend
≈ f þ ρh · θb ð9Þ N b ¼ EI=f2 · ½6ðd h
d p Þ · ρ3h 1=2 g
for which it may be shown that ¼ EI=f2ρ2h θb g ð15Þ
f ¼ ρh · ðθb Þ=2 þ 3ðd h
d p Þ=θb ð10Þ
Eq. (15) indicates that the reaction force N b is inversely propor-
Using Eqs. (7) and (9), the normal force N may also be tional to the magnitude of the (complete) bend angle θb , which rap-
expressed as a function of the angle θb idly decreases with diminishing clearance, as shown in Fig. 4.
Fig. 5 directly illustrates the strong (inverse) dependence of the
N ¼ 2EI · θ3b =9½ρh · ðθ2b Þ=2 þ ðd h
d p Þ2 ð11Þ reaction force N b on the clearance, for a given path geometry
(borehole radius ρh ). For convenience, the reaction force N b has
The value of θb , corresponding to a complete bend, may then been normalized relative to the normalized clearance ðd h
d p Þ=ρh ,
be determined by the third compatibility condition corresponding such that
to matching the curvature of the pipe to that of the borehole at
point 2 (Young 1989) N b · ρ2h =EI ¼ 1=f2 · ½6ðd h
d p Þ=ρh 1=2 g ð16Þ
N · l=EI ¼ 1=ρh ð12Þ The non-colinear nature of the reaction forces N i , implies that
these loads are only approximately equal, and that there is also a
Dividing Eq. (7) by Eq. (12)
compressive force induced within the pipe in the vicinity of the
l ¼ 2θb · ρh ð13Þ bend (Griffioen 1993).

which, combined with Eq. (8), yields the angle θb at the point (1) of
closest contact of the pipe with the bend of the borehole

θb ¼ ½6ðd h
d p Þ=ρh 1=2 ð14Þ

The general magnitude of θb may now be verified to be much

less than 1. For example, for a maxi-HDD installation of a 600-mm
(24-in.) pipe in a borehole with 50% clearance (300 mm or 12 in.),
using a minimum radius of curvature of 185 m (600 ft) [consistent
with the guidelines in ASTM F1962 for a 150-mm (6-in.) drill rod],
a value of 0.1 rad is calculated. (More typical, larger radii corre-
spond to lower values.) This small angle would also correspond
to larger pipe sizes, based on industry recommendations of a maxi-
mum clearance of 300 mm (12 in.) (HDD Consortium 2008).
Smaller diameter pipes, with smaller magnitude clearances of 50%
of the pipe’s outer diameter, correspond to lower values of θb for the
same bore path radius. As another example, a mini-HDD installa-
tion of a 100-mm (4-in.) pipe [115-mm (4.5 in.) outer diameter],
Fig. 5. Normalized reaction force (complete bend) as function of
in a borehole with 50% clearance and a radius of curvature of only
normalized clearance
30 m (100 ft) also results of approximately 0.1 rad. However, a

JOURNAL OF PIPELINE SYSTEMS ENGINEERING AND PRACTICE © ASCE / FEBRUARY 2012 / 25

J. Pipeline Syst. Eng. Pract. 2012.3:22-31.

 Incomplete Bend

The above formulas [Eqs. (15) and (16)] for the reaction force N
follows from Eq. (12), which assumes the pipe conforms to a finite
portion of the borehole bend between points 2 and 3 of Fig. 3. This
situation is designated as a complete bend for which the total bend
angle γ satisfies the following inequality
γ ≥ 2θb ð17Þ
where the portion θb is determined by Eq. (14). For more shallow
bends, such that
γ < 2θb ð18Þ
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/04/15. Copyright ASCE. For personal use only; all rights reserved.

the bend is of insufficient extent for the pipe to conform to the cur-
Fig. 7. Normalized reaction force as function of normalized bend angle
vature, and is designated as an incomplete bend, as shown in Fig. 6,
for which θ is defined as half the total bend angle, or
θ ¼ γ=2 ð19Þ
which is less than the complete bend angle (i.e., θ < θb ).
In this case, points 2 and 3 coincide and the net reaction force
at point 2=3 is interpreted as N 2 þ N 3 , or twice that at point 1 or 4.
N again refers to N i (I ¼ 1, 2, 3, 4), for which Eq. (5) is still appli-
cable, as well as Eqs. (6)–(11), with θb replaced by θ. Combining
these relations yields

NðθÞ ¼ 8EI · ðθ=θb Þ3 =fρ2h · θb · ½3ðθ=θb Þ2 þ 12 g for θ=θb ≤ 1

¼ Nb for θ=θb > 1
ð20Þ

where NðθÞ = explicit function of the incomplete bend angle θ

(θ < θb ) and the radius of curvature ρh . The relationship between
the reaction forces corresponding to a complete versus incomplete
bend is shown in Fig. 7. For this purpose, the reaction force NðθÞ Fig. 8. Normalized reaction force (incomplete bend) as function of
has been normalized relative to the reaction force N b for a complete normalized clearance
bend [Eq. (15)], and rewritten as a function of the normalized bend
angle term θ=θb , such that
Fig. 7. Thus, for a given (incomplete) bend angle θ, a smaller
NðθÞ=N b ¼ 16ðθ=θb Þ3 =½3ðθ=θb Þ2 þ 12 for θ=θb ≤ 1 relative clearance corresponds to a lower value of θb and corre-
¼1 for θ=θb > 1 ð21Þ spondingly higher values of θ=θb and NðθÞ=N b , as well as N b .
The net effect is reflected in Eq. (22) and shown in Fig. 8, for sev-
The magnitude of the reaction force increases with the value of eral discrete bend angles (5, 10, and 15°) as follows:
the incomplete bend angle, reaching a maximum value when the
bend is complete, (i.e., equal to θb ). Beyond this angle, the reaction NðθÞ · ρ2h =EI ¼ 2θ3 =9½θ2 =2 þ ðd h
d p Þ=ρh 2 for θ ≤ θb
force N is equal to N b , the value corresponding to a complete bend. ¼ 1=f2 · 6ðd h
d p Þ=ρh 
1=2
g for θ > θb ð22Þ
The importance of the clearance (relative to the bend radius of
the borehole) is emphasized by Figs. 4 and 5, in combination with The results for the various angles coincide at low values of clear-
ance, since the presumably incomplete bend angles θ approach (and
exceed) the magnitude of a complete bend θb , which diminishes
with decreasing clearance, as shown in Fig. 4. For such values of
θ > θb , the reaction force reaches its peak, as indicated in Fig. 7,
independent of the magnitude of the actual bend angle θ. A rapidly
increasing trend of the reaction force with decreasing clearance is
evident by the sharp rise in the former for relatively low values of
the latter, which may be on the order of 0.002 or less for practical
maxi-HDD installations.

End of Pipe Traversing Bend

The configuration shown in Fig. 2 assumes that the leading end of
the pipe has traversed the extent of the discrete bend, and extends
into the straight portion of the bore path, beyond the bend itself.
Fig. 6. Incomplete bend (reprinted with permission of Outside Plant
However, for the special case when the pipe is still passing through
Consulting Services, Inc.)
the extent of the bend, the reaction forces will be shown to be larger

26 / JOURNAL OF PIPELINE SYSTEMS ENGINEERING AND PRACTICE © ASCE / FEBRUARY 2012

J. Pipeline Syst. Eng. Pract. 2012.3:22-31.

Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/04/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. Pipe end traversing (complete) bend Fig. 10. Pipe end traversing general bend

than the previous estimates. This physical situation may also apply As the pipe end begins to enter a complete bend, or possibly
at the pipe entry or exit location if the borehole has finite curvature traverses an incomplete bend, the subtended angles θ0 or θ″ will
at these points. For example, if the bore path segment A–B or C–D be less than the minimum value of Eq. (27), as shown in Fig. 10,
of Fig. 3 is of uniform curvature, the reaction forces at point A or D in which θ0 is not initially assumed equal to θ″ . During this tran-
will be greater than that at point 1 or point 4 of Fig. 3, for which the sition, the instantaneous total angle traversed is designated as γ0 , for
end of the pipe extends beyond the curved portion. Fig. 9 depicts which
such a situation, which is initially considered to be analogous to
that of the complete bend shown in Fig. 3. θ0 þ θ″ ¼ γ 0 ð28Þ
The present analysis again follows Griffioen’s procedure for
this configuration (Griffioen 1993). In this case, because of the In this case, points 2 and 3 coincide and the net reaction force at
asymmetry on opposite sides of the bend, the pair of equal reaction point 2=3 is interpreted as N 2 þ N 3 , and individual component
forces N 1 , N 2 (N 1 ¼ N 2 ) may not be assumed to be equal to the pair forces are not necessarily equal. However, as for the complete bend
N 3 , N 4 (N 3 ¼ N 4 ). However, since the geometry corresponding to considered previously, N 1 ¼ N 2 and N 3 ¼ N 4 . Eqs. (20)–(22) may
the section of the bend containing the reaction forces N 3 , N 4 is be verified to be applicable to the latter pair of reactions (N 3 , N 4 ),
identical to that shown in Fig. 3, these forces may each be consid- with θ replaced by θ″. Thus
ered to be equal to N b as given by Eq. (15) or (16), and the angle θb
N 4 ðθ″ Þ · ρ2h =EI ¼ 2ðθ″ Þ3 =9½ðθ″ Þ2 =2 þ ðd h
d p Þ=ρh 2 for θ″ ≤ θb
given by Eq. (14). Considering the reactions (N 1 or N 2 ) on the
opposite (leading) side of the bend, the beam bending relation ð29Þ
for lateral deflection may be shown to yield
In addition, the beam deflection equation reflected in Eq. (7)
N 1 · l3 =ð3EIÞ ≈ l · θ0b
½ρh · ðθ0b Þ2 =2 þ ðd h
d p Þ ð23Þ applies, such that

Based on Fig. 9 N 4 ðθ″ Þ · ðl0 Þ2 =ð2EIÞ ¼ θ″ ð30Þ

l ≈ ρh · θ0b ð24Þ Similarly, it may be verified that Eqs. (23) and (24) are still
applicable for the variables at the leading end of the pipe, such that
Furthermore, the beam bending relation for imposed curvature
requires the following: N 1 ðθ0 Þ · l3 =ð3EIÞ ≈ l · θ0 − ½ρh · ðθ0 Þ2 =2 þ ðd h
d p Þ ð31Þ

N 1 · l=EI ¼ 1=ρh ð25Þ and

l ≈ ρh · θ 0 ð32Þ
Manipulating these three equations containing three unknowns
(N 1 , l, and θ0b ) then yields Considering the additional condition that the bending moments
must be equal on both sides of the contact point 2=3, then
ðN 1 Þb · ρ2h =EI ¼ 1=½6ðd h
d p Þ=ρh 1=2 ð26Þ
N 1 ðθ0 Þ · l ¼ N 4 ðθ″ Þ · l0 ð33Þ
where ðN 1 Þb is introduced to emphasize that the value applies to the
case of a complete bend, and for which the reaction force is exactly The Eqs. (28)–(33) consist of six equations for six unknowns
twice that of Eq. (16). Eqs. (24)–(26) may then be used to verify (θ0 , θ″ , l, l0 , N 1 , and N 4 ) mathematically determining a unique sol-
that the angle θ0b corresponding to a complete bend is again given ution, which would be a function of the pipe and borehole param-
by Eq. (14): eters and the instantaneous angle γ0 . However, since this set of
nonlinear algebraic equations does not appear to be tractable with
θ0b ¼ ½6ðd h
d p Þ=ρh 1=2 respect to conveniently obtaining the theoretical solution, a simple
¼ θb ð27Þ expedient is considered for the present purpose of estimating the
effect of the local reaction forces. On the basis of the preceding
Thus, for the case of the pipe end within the route bend, the total results for which the leading end of the pipe has passed an incom-
of the related reaction forces acting at the bend is equal to two times plete bend, as shown in Fig. 6, in combination with that of the
that indicated in Eq. (26) plus two times that indicated in Eq. (16), previous case for the pipe end traversing a complete bend, as shown
for a total of six times that of Eq. (16), or 6N b. in Fig. 9, the following approximate solution is assumed:

JOURNAL OF PIPELINE SYSTEMS ENGINEERING AND PRACTICE © ASCE / FEBRUARY 2012 / 27

J. Pipeline Syst. Eng. Pract. 2012.3:22-31.

 where Ro and Ri = outer and inner radii of the pipe, respectively;
and A = cross-sectional area of the pipe. The cross-sectional area of
the pipe (A) is as follows:

A ¼ πðR2o
R2i Þ
¼ πd 2p · ½ð1
ð1
2=DRÞ2 =4 ð36Þ

where DR = pipe dimension ratio, defined as the pipe outer diam-

eter (d p ) divided by the wall thickness. The moment of inertia may
be calculated as follows:

I ¼ A · d 2p ½ð1 þ ð1
2=DRÞ2 =16 ð37Þ

Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/04/15. Copyright ASCE. For personal use only; all rights reserved.

Previous relationships (shown in Figs. 5, 7, and 8), describe the

dependence of the bending reaction forces on the bore path char-
Fig. 11. Normalized reaction force of leading end of pipe when
acteristics, including radius of curvature and borehole clearance. It
traversing a bend
is also useful to understand the dependence of the reaction forces
on the pipe diameter d p. For example, it has previously been noted
that, in the absence of appreciable bending forces (such as PE pipe),
for a specified route geometry (e.g., bore path length, depth, entry
N 4 ðθ″ Þ · ρ2h =EI ≈ 2ðθ″ Þ3 =9½ðθ″ Þ2 =2 þ ðd h
d p Þ=ρh 2 ð34a Þ or exit angles) and fixed diameter-to-wall thickness dimension ratio
(DR), the predicted pulling loads based on the procedure in ASTM
N 1 ðθ0 Þ ≈ 2N 4 ðθ″ Þ ð34b Þ F1962 are directly proportional to the square of the pipe diameter,
d 2p (Slavin 2010). This is due to the buoyant weight relationships, as
and well as the relationship associated with the relatively relatively mi-
nor contribution of the fluidic drag of Eq. (4), assuming a consistent
θ0 ≈ θ″ ≈ γ0 =2 ð34c Þ percentage clearance is maintained. (A 50% minimum borehole
clearance guideline applies to pipes of 610-m (24-in.) diameter
This approximate solution is considered applicable as the pipe or less. For larger pipes, a 305-mm (12-in.) minimum clearance
enters a complete bend, prior to subtending an angle equal to 2θb , is recommended, which effectively reduces the relatively small
and for the traversal of an incomplete bend. It is recognized that contribution of the fluidic drag component.)
these results do not represent an exact solution, but are considered Thus, since the net cross-sectional tensile strength of the pipe is
reasonable with respect to estimating the potential impact of this also proportional to the square of the diameter for a fixed DR value,
condition. Furthermore, in many cases, such as in the subsequent the ability of the pipe to withstand the predicted pull loads for a
sample results, the angle θb corresponding to a complete bend is given pipe installation distance remains essentially unchanged,
relatively low and is often exceeded, such that the relevant reaction considering the average cross-sectional stresses in relation to the
forces are given by Eqs. (16) and (26), which are greater than the pipe pull strength (i.e., “safe pull stress”). However, as suggested
assumed transitional forces and are therefore conservative. by the previous equations for the inertia term I, the dependence of
Fig. 11 qualitatively illustrates the reaction force N 1 at the lead- the reaction forces may increase at a rate above and beyond that of
ing end of the pipe as it traverses a bend, for which the normalized the square of the pipe diameter d 2p depending on the normalization
term θ0 =θ represents the progress along the bend (and beyond). factors incorporated in Figs. 5 or 8. The net effect is subsequently
(The portions of the trends following the peak values are not based discussed for either plastic or metallic pipes, with respect to the
on actual calculations, and are only intended to indicate anticipated reaction force terms N b or NðθÞ, which represent characteristic val-
trends based on the present analytical model.) For a complete bend, ues and trends for various conditions (e.g., pipe spans complete or
as defined by Eq. (27) (γ0 ¼ 2θb ), the force initially increases, as the incomplete bend and pipe end traversing bend).
pipe enters the curve, achieving a peak value given by Eq. (26),
which is equal to twice the characteristic force level N b given Plastic Pipes
by Eq. (16). The force on the pipe end remains constant, at the peak
For plastic pipes, with a characteristically low elastic modulus, the
value, until the pipe begins to transition to a subsequent straight
bore path curvature may be assumed to be dictated by the stiffness
section of the borehole, during which the reaction force N 1 gradu-
and corresponding bending capability of the drill rods, to which the
ally decreases to the N b value. Beyond this point, the force on the
plastic pipe would readily conform. Thus, the bending radius term
very end of the pipe vanishes as the pipe negotiates the extended
ρh may be considered to be relatively constant over a wide range
straight section of the borehole. Fig. 11 also qualitatively illustrates
of pipe sizes (diameters) for a typical maxi-HDD installation.
the estimated reaction forces for an incomplete bend.
Furthermore, for pipe sizes of 600-mm (24-in.) diameter or greater,
a clearance of 300 mm (12 in.) is considered to be a “good practice”
(HDD Consortium 2008). Thus, the local borehole radius ρh and a
Pipe Stiffness normalized clearance term ðd h
d p Þ=ρh may be assumed to be es-
The pipe bending stiffness (EI) is the product of the material effec- sentially constant. This indicates that the normalized reaction force
tive elastic modulus (E) and the area moment of inertia (I). For a NðθÞ · ρ2h =EI shown in Fig. 8 remains constant over a range of
hollow pipe, the value of I is given as follows (Young 1989): pipe diameters and corresponding moments of inertia value I.
However, because Eqs. (22)–(24) indicate that the characteristic I
I ¼ ðπ=4ÞðR4o
R4i Þ effectively has quartic (power of four) dependence on the pipe
diameter for fixed DR value, it is evident that the reaction force
¼ A · ðR2o þ R2i Þ=4 ð35Þ NðθÞ increases at the rate of d 4p , considerably greater than the

28 / JOURNAL OF PIPELINE SYSTEMS ENGINEERING AND PRACTICE © ASCE / FEBRUARY 2012

J. Pipeline Syst. Eng. Pract. 2012.3:22-31.

 quadratic dependence (d 2p ) of the otherwise predicted tensile load outer diameter, the normalized clearance term ðd h
d p Þ=ρh is equal
and pipe strength, as described previously. Thus, the stiffness to a constant value of d p =2ρh , resulting in an essentially constant
effect, as reflected in the bending reaction forces, will become in- value of the normalized reaction force NðθÞ · ρ2h =EI, for fixed bend
creasingly significant for larger pipe sizes, increasing the total drag angle θ. Considering the preceding discussion on plastic pipes, rec-
on the pipe and the required tensile load and corresponding average ognizing that the characteristic I effectively has quartic dependence
cross-sectional stress. Furthermore, in addition to the average on the pipe diameter, for fixed DR value, the reaction force NðθÞ may
cross-sectional stress, the local bending stresses increase in direct be seen to increase at the rate of only d 2p —consistent with the quad-
proportion to the pipe diameter for an assumed fixed local borehole ratic dependence of the estimated tensile load and pipe strength. This
curvature. Both types of stresses are added at several locations again suggests that, for pipes of less than 600-mm (24-in.) diameter,
along the route to ensure the peak stress condition is considered. the bending stiffness is less likely to significantly impact the required
For somewhat smaller diameter pipes [i.e., less than 600-mm installation forces of the pipe. However, for mini-HDD applications,
(24-in.) diameter], the borehole clearance is not fixed at the the presence of path undulations resulting from frequent path cor-
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/04/15. Copyright ASCE. For personal use only; all rights reserved.

300-mm (12-in.) level, but is assumed to be 50% of the pipe’s outer rections may have a significant cumulative effect on the required pull
diameter. Thus, the normalized clearance term, ðd h
d p Þ=ρh is load, similar to the installation of plastic pipes, and is the subject of a
equal to d p =2ρh , which, for a fixed local borehole radius, is directly future investigation.
proportional to the pipe diameter. For the range of normalized For steel or iron pipes of 600-mm (24-in.) diameter or greater,
clearances such that the angle θ remains greater than θb for which a fixed clearance of 12 in. applies, the normalized clear-
[Eq. (14)], the reaction force NðθÞ is equal to N b with the normal- ance ðd h
d p Þ=ρh is inversely proportional to the borehole radius,
ized term N b ρ2h =EI given1 by Eq. (16). In this case, the reaction force and therefore also inversely proportional to the pipe diameter. As
3
N b is proportional to d p2 , increasing at a slightly lower rate than the indicated in Eq. (22), and shown in Fig. 8, the normalized reaction
quartic dependence for pipes of larger than 600-mm (24-in.) diam- force NðθÞ · ρ2h =EI, for fixed bend angle θ, increases
1
for larger pipe
eter. For smaller bend angles, such that θ < θb , Eq. (22) indicates sizes. The rate of increase is proportional to ρ2h as the magnitude of
a limiting trend (for low angles θ), of reaction force NðθÞ that is a complete bend θb decreases with larger bend radius [Eq. (14)],
proportional to d 2p , similar to that of the otherwise predicted tension such that θ > θb . The net effect is that the reaction force NðθÞ in-
21
and corresponding pipe strength. This suggests that the pipe diam- creases at the rate of d p2 in comparison to the quadratic dependence
eter for this size range (d p < 600 mm) plays a somewhat less of the predicted tensile load and pipe strength, corresponding to only
significant role with respect to the impact of the bending reaction a slightly greater impact of bending stiffness for large-diameter steel
forces on the installation of the pipe along a specified route. The or iron pipe products. Thus, in contrast to plastic pipes, it appears that
subsequent examples illustrate the overall trend and dependence of the large-sized metallic pipes (> 600 mm) may be less dramatically
the reaction forces on pipe size. impacted by the bending stiffness effects essentially because of the
The pipe stiffness characterization for mini-HDD installations is proportionally larger bend radii considered for the bore path.
similar to that of the maxi-HDD installations. Because the pipes
placed by mini-HDD are generally of less than 300-mm (12-in.)
diameter, the 50% borehole clearance guideline applies and the pre-
Results
vious discussion regarding pipes of less than 600-mm (24-in.) The theoretical model will be applied to representative examples
diameter, placed by maxi-HDD, is relevant. For mini-HDD, the as shown in Table 1 in order to better understand the potential im-
smaller-sized drill rods will produce smaller borehole bend radii, plications of the bending stiffness in practical applications, and to
and the normalized clearance term, ðd h
d p Þ=ρh , will be of similar verify the anticipated trends for both plastic and metallic pipes. The
order of magnitude as that for the larger pipelines installed by maxi- results apply to the entry and exit segments A–B and C–D, repre-
HDD. As a result of these considerations, and in combination with senting route bends in a vertical plane, as shown in Fig. 1. Nominal
its relatively short placement distances, it would generally be an- pipe sizes ranging from 300 to 1,200 mm (12–48 in.) in diameter
ticipated that the impact of the bending reaction forces during mini- are considered; Table 1 also indicates the actual pipe diameter, d p .
HDD placement of the small-diameter pipes is not major. However, For the present purposes, representative values for the pipe dimen-
mini-HDD operations are characterized by an additional detrimen- sions (i.e., wall thickness) have been selected, but are not neces-
tal effect in comparison to the typically well controlled maxi-HDD sarily equal to that of presently manufactured or marketed
installations because of the presence of path undulations resulting products. In particular, a single artificial DR value is selected
from frequent path corrections (Slavin and Petroff 2010). Similar to for the ductile iron pipe, on the basis of an average dimension ratio
the greater importance of the capstan effect in such mini-HDD in- over the sizes of interest, as calculated from those available from
stallations, the corresponding sequence of bends may have a signifi- one manufacturer. In general, however, the physical properties and
cant cumulative effect on the required pull load. This phenomenon characteristics are reasonably representative of manufactured prod-
may be investigated in a future analysis focusing on mini-HDD in- ucts and published data and specifications. For example, a short-
stallations of pipes characterized by significant bending stiffness. term elastic modulus of 689:4 N=mm2 (100;000 lb=sq in:) and
2;758 N=mm2 (400;000 lb=sq in:) is considered appropriate for
Metallic Pipes
high-density polyethylene (HDPE) and PVC pipe, respectively,
In contrast to plastic pipes, metal (e.g., steel or iron) pipes have an and a value of 165;470 N=mm2 (24 million lb=sq in:) is assumed
extremely high elastic modulus, which will often dictate the bore for ductile iron pipe. Corresponding material specific gravities, γa ,
path curvature to maintain the bending stresses within allowable are equal to 0.955, 1.40, and 7.10, respectively.
limits. Because the bending strains and stresses are proportional to The drilling fluid density in combination with the material den-
the ratio of the pipe diameter to bending radius, the term d p =ρh may sity determines the buoyant weight of the pipe wb , possibly reduced
be considered to be relatively constant over a wide range of pipe by internal liquid (e.g., water) ballast, representing the primary
sizes (diameters) for a typical maxi-HDD installation; that is to source of drag and the major influence on the required pull force.
say, the bore path would be designed with proportionally larger bend The buoyant weights, for the case of internal water ballast, are in-
radii for larger pipes. Thus, for pipes of less than 600-mm (24-in.) cluded in Table 1 for an assumed drilling fluid specific gravity
diameter and for which the borehole clearance is 50% of the pipe’s equal to 1.2. Because of the greater density of ductile iron material,

JOURNAL OF PIPELINE SYSTEMS ENGINEERING AND PRACTICE © ASCE / FEBRUARY 2012 / 29

J. Pipeline Syst. Eng. Pract. 2012.3:22-31.

 Table 1. Example Pipe and Borehole Characteristics (Vertical Bends)
dh
dp
Nominal size=dp wb (N=m) (mm)
(mm) (in.) EI (N·mm2 (lb·in:2 ) (lb=ft) ρh (m) (ft) (in.)
HDPE, DR 11, E ¼ 689:4 N=mm2 (100;000 lb=sq in), γa ¼ 0:955
300=325 2:06 × 1011 174 365 162
(12=12:75) (7:16 × 107 ) (11.9) (1,200) (6.375)
600=610 2:58 × 1012 614 365 305
(24=24:0) (8:99 × 108 ) (42.1) (1,200) (12.0)
900=915 1:31 × 1013 1,383 365 305
(36=36:0) (4:55 × 109 ) (94.8) (1,200) (12.0) Fig. 12. Effective incremental route length due to bending reaction
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/04/15. Copyright ASCE. For personal use only; all rights reserved.

1;200=1;220 4:13 × 1013 2,459 365 305 forces for typical entry and exit route bends
(48=48:0) (1:44 × 1010 ) (168.5) (1,200) (12.0)
PVC, DR 25, E ¼ 2;758 N=mm2 (400;000 lb=sq in:), γa ¼ 1:40 bend at the entry and exit segments A–B and C–D. Furthermore,
it is also conservatively assumed that the leading end of the pipe
300=335 4:85 × 1011 120 365 168 at point D, and the trailing portion of the pipe in the vicinity of
(12=13:2) (1:69 × 108 ) (8.2) (1,200) (6.6) point A, are in the process of traversing their respective bends.
600=655 7:09 × 1012 458 365 305 As a result, the local reaction forces at these points are estimated
(24=25:8) (2:47 × 109 ) (31.4) (1,200) (12.0) as twice that of the characteristic reaction force N b acting on the
900=975 2:87 × 1013 1,010 365 305 portion of the pipe that precedes or is beyond the bend—i.e., Point
(36=38:3) (1:20 × 1010 ) (69.2) (1,200) (12.0) C or B, respectively. With the pipe fully installed in the borehole,
1200=1290 1:28 × 1014 1,776 365 305 the sum of the reaction forces, considering both segments, is there-
(48=50:8) (3:71 × 1010 ) (121.7) (1,200) (12.0) fore equal to 2 × 6N b , or 12N b. In practice, the bore path would
typically include an initially straight segment at the drill rig/pipe
Ductile iron, DR 56, 165;470 N=mm2 (24 million lb=sq in:), γa ¼ 7:10 exit end (e.g., corresponding to the first drill rod placed) such
300=335 1:39 × 1013
197 402 168 that the leading end of the pipe would be somewhat beyond the
(12=13:2) (4:84 × 109 ) (
13:5) (1,320) (6.6) actual extent of the bend, reducing the reaction force, consistent
600=655 2:03 × 1014
753 786 305 with the decreasing trend in Fig. 11. Similarly, the trailing portion
(24=25:8) (7:07 × 1010 ) (
51:6) (2,580) (12.0)
of the pipe may be supported external (e.g., on rollers) to the pipe
entrance such as to create a relatively straight entry path, which
900=975 9:85 × 1014
1;661 1,167 305
may also tend to reduce the local reaction (and possible
(36=38:3) (3:43 × 1011 ) (
113:8) (3,830) (12.0)
associated frictional) force.
1; 200=1; 290 3:04 × 1015
2;922 1,548 305 To help interpret the significance of the corresponding drag
(48=50:8) (1:06 × 1012 ) (
200:2) (5,080) (12.0) effect, the reaction force 12N b is normalized with respect to the
buoyant weight, wbuoyant , as shown in Fig. 12. This ratio may be
considered to represent an incremental route length and associated
the net buoyant weight for this type of pipe is negative, indicating larger required pull force or, alternatively, a possible required re-
that the pipe will rest on the bottom of the borehole, in comparison duction in route length for a limited pull load.
to the plastic pipes that will be pushed upward against the crown.
The effective drag phenomena are similar for both cases, as re-
flected in Eqs. (2a)–(2d). Discussion of Results
Reasonable borehole radii of curvature for a maxi-HDD instal-
The increasing trends of normalized reaction forces with pipe diam-
lation have also been designated, consistent with the previous dis-
eter for the three pipe materials and products described in Table 1,
cussions, and the borehole clearance selected at the maximum of
and as observed in Fig. 12, are consistent with the previous predic-
300 mm (12 in.) for pipes greater than 600 mm (24 in.), and at 50%
tions based on the increasing stiffness characteristics relative to the
of the pipe’s outer diameter for smaller sizes. The constant 370-m
square of the pipe diameter, d 2p . The higher rate of increase for the
(1,200-ft) borehole radius of curvature selected for the plastic pipes ductile iron pipe relative to the plastic products is also as antici-
is well within the range of application of the HDPE products but it pated. In general, however, the magnitudes of the impact of the
approaches the recommended limit for the largest PVC pipe shown. reaction forces are quite low in comparison to typical placement
For the ductile iron pipe, the ratio of the pipe’s outer diameter to capabilities, assuming pipe joints that maintain most or all of
bending radius d p =ρh is maintained at 1 to 1,200, consistent with an the original tensile strength of the pipe, such as fused or welded
industry standard for steel products (e.g., ASTM F1962 for steel joints, especially for the assumed use of ballast. For example, the
drill rods). equivalent incremental lengths for HDPE are two orders of mag-
The calculated pipe bending stiffness (EI) values shown in the nitude lower than the practical placement distances for such prod-
table are based on the minimum pipe wall thickness corresponding ucts (Slavin 2010). This apparent minimal impact of the bending
to the indicated dimension ratio (DR). forces is in spite of the various conservative assumptions incorpo-
In combination with the borehole characteristics in Table 1, rated into the model and/or its application.
Eq. (14) indicates the magnitude of a complete bend θb to be in Furthermore, for the present case of route bends in a vertical
the range of 2 to 4°, corresponding to a minimum total bend angle plane, the reaction forces required to deflect a pipe to conform
γ of 4 –8°. Referring to Fig. 1, because the typical pipe entry and to the route bend would be lower than estimated above, due to
exit angles (α and β) are in this range or are greater, it may be the beneficial bending effect of the distributed net buoyant forces
conservatively assumed that the pipe will experience a complete acting laterally on the pipe, at both sides of the inner contact points

30 / JOURNAL OF PIPELINE SYSTEMS ENGINEERING AND PRACTICE © ASCE / FEBRUARY 2012

J. Pipeline Syst. Eng. Pract. 2012.3:22-31.

 (points 2 and 3). Thus, considering the buoyant weight acting along The results of direct application of the equations indicate that the
the unsupported length l between points 1 and 2 (and between quantitative impact of the bending stiffness increases with pipe size,
points 3 and 4) in Fig. 3, as given by Eq. (13), it may be verified but is nonetheless low for the cases considered, especially for the
that the corresponding total lateral bending forces are an order of relatively flexible HDPE products. Furthermore, it is demonstrated
magnitude greater than that of the calculated characteristic discrete that the analytical model tends to be overly conservative with re-
reaction force N b . In addition, any significant tension that may have spect to ignoring the mitigating effects of the lateral buoyant forces
previously developed at the bend will exert a local beneficial bend- in tending to deflect the pipe to conform to the bend, essentially
ing moment that also helps reduce the magnitude of any reaction eliminating the otherwise required local reaction forces. Although
forces. These considerations indicate that the calculated forces are not considered in the present analysis, any tensile load that has
significantly overestimated and, recognizing their already minor developed during the pulling operation, due to drag and other pos-
impact, may essentially be ignored for the representative cases con- sible effects acting on the portion of the pipe preceding the bend
sidered in Table 1. in question, would also have a mitigating effect in reducing the
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/04/15. Copyright ASCE. For personal use only; all rights reserved.

Nonetheless, there may be cases in which these effects may be reaction forces. Thus, except for a combination of circumstances,
more significant and should be considered or addressed, as appro- such as a large-diameter pipe, small borehole bend radius, relatively
priate. The examples presented in Table 1 and Fig. 12 consider thick pipe, low pipeline tensile capability (e.g., weak joints) and
pipes with a uniform outer diameter, such as representative of a correspondingly limited placement distance, reduced borehole
continuous-length HDPE pipe, and possibly including fused joints. clearance (e.g., large couplings), and very low buoyant weight
Fused joints or possibly rigid couplings for other products, how- and/or horizontal bends, or metal pipe with relatively large wall
ever, would typically be characterized by a locally larger diameter thickness for which the increased stiffness may result in a signifi-
reducing the local clearance. Assuming that industry-recommended cant increase in required drill-rig pull force capability, the pipe
practices for minimum borehole clearance [i.e., 50% the product’s bending stiffness does not appear to be an important factor in af-
outer diameter, but 300 mm (12 in.) maximum] are based on the fecting the net pull force. In such exceptional cases, the formulas
pipe diameter and not the coupling, the effective clearance will be provided in this paper may be used to provide an estimate of the
somewhat reduced by the incremental diameter due to the coupling. reaction forces to determine the associated increased drag force and
Some such joints or connections comprise an integral bell which pull load. The resulting tensile forces correspond to (increased)
effectively increases the local diameter by twice the wall thickness, average cross-sectional stress to be added to the bending stresses
corresponding to a locally decreased borehole clearance by an because of the imposed curvature, following usual engineering
amount equal to 2d p =DR. An analysis similar to the development practices (e.g., ASTM F1962).
of the present model, would be relatively complex, but it is reason- Future investigations will attempt to quantify the effects of vari-
able to consider an upper bound of the reaction forces as deter- ous departures from the somewhat idealized model considered in
mined by reducing the effective clearance term (d h
d p ) in the the present analysis. Such studies may address the mitigating effect
various equations [e.g., Eq. (15)] by the indicated amount. The re- of tension and/or net buoyant weight on the bending reaction
sulting increase in the reaction forces is not major, and would not forces, and the significance of discrete route bends of nonuniform
in itself alter the conclusion regarding the essentially negligible radii of curvature. In addition, while the presence of nonrigid cou-
impact of the bend stiffness on pull load, especially due to the plings, allowing a finite deflection angle at each joint, would intui-
mitigating effect of the buoyant weight for bends in a vertical plane. tively be expected to significantly reduce the effects of pipe
However, larger-diameter joints, in combination with a signifi- stiffness and associated overall reaction forces, it is conceivable that
cantly reduced borehole radius, relatively thick pipe, low tensile the cumulative effect of a long string of such sections—depending
capability (e.g., weak joints) and correspondingly limited placement on the specific route geometry, and borehole, pipe, and joint
distance, and very low buoyant weight and/or horizontal bends, the dimensions—may have a greater impact than that corresponding to
relative effect for large diameter pipes may be considerably more an equivalent continuous pipe, therefore deserving further analysis.
important. For metal pipes with relatively large wall thickness,
the increased stiffness may result in a significant increase in required References
drill-rig pull force capability, although not necessarily exceeding the
increased pipe strength. In such cases, the formulas provided herein Buller, F. H. (1949). “Pulling tension during cable installation in ducts or
may be used to provide a conservative estimate of the reaction forces pipes.” Gen. Electr. Rev., 52(8), 21–23.
to determine the associated increased drag force and pull load. Griffioen, W. (1993). Installation of optical cables in ducts, Plumettaz,
Bex, Switzerland.
Horizontal Directional Drilling (HDD) Consortium. (2008). Horizontal di-
Summary rectional drilling—Good practices guidelines, 3rd Ed., North American
Society of Trenchless Technology (NASTT), Liverpool, NY.
The quantitative effect of pipe bending stiffness on required pull Plastics Pipe Institute (PPI). (2008). The Plastics Pipe Institute handbook of
loads during a typical maxi-HDD operation has been analyzed us- polyethylene pipe 2nd Ed., PPI, Irving, TX.
ing a mathematic model that considers the pipe as a beam, resulting Rifenburg, R. C. (1953). “Pipeline design for pipe-type feeders.” Trans.
in lateral reaction forces at discrete bends along the borehole path, Am. Inst. Electr. Eng., Part 3, 72(2), 1275–1288.
contributing to increased frictional drag. The resulting equations Slavin, L. M. (2010). “Parametric dependency and trends of HDD pull
incorporate generally conservative assumptions, and provide a loads.” J. Pipeline Sys. Eng. Pract., 1(2), 69–76.
convenient means of estimating the potential impact as a function Slavin, L. M., and Najafi, M. (2011). “Maxi-HDD pull loads for entry and
of the installation parameters, including bore path radius of curva- exit points at different elevations.” Proc., Int. Conf. on Pipelines and
Trenchless Technology, ASCE, Reston, VA.
ture, borehole clearance, and pipe stiffness, which is a function of
Slavin, L. M., and Petroff, L. (2010). “Discussion of ASTM F1962, or
pipe geometry (diameter and wall thickness) and effective elastic ‘How are the pulling load formulas derived and how are they used?’.”
modulus. The equations have been applied to representative prod- Proc., NO-DIG 2010, North American Society of Trenchless Technol-
ucts ranging from 300 mm (12 in.) to 1,200 mm (48 in.) in nominal ogy (NASTT), Liverpool, NY.
size, including HDPE, PVC, and ductile iron, installed using Young, W. C. (1989). Roark’s formulas for stress and strain, 6th Ed.,
maxi-HDD. McGraw-Hill, New York.

JOURNAL OF PIPELINE SYSTEMS ENGINEERING AND PRACTICE © ASCE / FEBRUARY 2012 / 31

J. Pipeline Syst. Eng. Pract. 2012.3:22-31.