SPE 84246
A Compendium of Directional Calculations Based on the Minimum Curvature Method
S.J. Sawaryn, SPE, J.L. Thorogood, SPE, BP plc.
Copyright 2003, Society of Petroleum Engineers Inc.
This paper was prepared for presentation at the SPE Annual Technical Conference and
Exhibition held in Denver, Colorado, U.S.A., 5 – 8 October 2003.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
Abstract
The minimum curvature method has emerged as the accepted
industry standard for the calculation of 3D directional surveys.
Using this model, the well’s trajectory is represented by a
series of circular arcs and straight lines. Collections of other
points, lines and planes can be used to represent features such
as adjacent wells, lease lines, geological targets and faults. The
relationships between these objects have simple geometrical
interpretations, making them amenable to mathematical
treatment. The calculations are now used extensively in 3D
imaging and directional collision scans, making them both
business and safety critical. However, references for the
calculations are incomplete, scattered in the literature and have
no systematic mathematical treatment. These features make
programming a consistent and reliable set of algorithms more
difficult. Increased standardisation is needed.
Investigation shows that iterative schemes have been used
where explicit solutions are possible. Explicit calculations are
preferred because they confer numerical predictability and
stability. Though vector methods were frequently adopted in
the early stages of the published derivations, opportunities for
simplification were missed because of premature translation to
Cartesian coordinates.
This paper contains a compendium of algorithms based on the
minimum curvature method (includes co-ordinate reference
frames, toolface, interpolation, intersection with a target plane,
minimum and maximum TVD in a horizontal section, point
closest to a circular arc, survey station to a target position with
and without the direction defined, nudges and steering runs).
Consistent, vector methods have been used throughout with
improvements in mathematical efficiency, stability and
predictability of behaviour. The resulting algorithms are also
simpler and more cost effective to code and test. This paper
describes the practical context in which each of the algorithms
is applied and enumerates some key tests that need to
be performed.
Introduction
The first reference to the minimum curvature directional
survey calculation method is credited to Mason and Taylor1 in
1971. In the same year, Zaremba2 submitted an identical
algorithm that he termed the circular arc method. In the
minimum curvature method, two adjacent survey points are
assumed to lie on a circular arc. The arc is located in a plane
which orientation is defined by the known inclination and
direction angles at the ends. By 1985 the minimum curvature
method was recognised by the industry as one of the most
accurate methods but it was regarded as cumbersome for hand
calculation3,4. The emergence of well trajectory planning
packages to help manage directional work in dense well
clusters increased its popularity. It was natural to use the same
model for both the surveys and the segments of the well plan
trajectories. Today, with the wide spread use of computers,
computational power is no longer an issue and the method has
emerged as the accepted industry standard.
Industry Requirements
Over the years, various algorithms based on the minimum
curvature method have been published for the construction of
increasingly complex trajectories and tasks such as
interpolation. Since these algorithms have emerged piecemeal,
they have tended to use different nomenclatures and
mathematical techniques for their solution. The result of this
piecemeal development is duplicated and inefficient computer
code and a poor understanding of the engineering integrity of
the systems.
Safety and Business Criticality
An undetected fault in the coding or use of directional
surveying and collision scanning software has been classified
as having the potential to cause property damage,
environmental damage, personal injury or loss of reputation5.
The integrity of these business and safety critical drilling
systems is therefore a concern. Modern 3D imaging and
directional scanning packages execute thousands of
calculations for each task. Increased automation of the
workflows associated with these tasks means that most
calculations must be taken for granted and will pass
unchecked. Sawaryn6 et. al., have described a process for
managing these systems and identified a number of
requirements related to the equations they use. Specifically,
equations must be traceable back to the source documentation
2 SPE 84246
that must clearly explain their purpose, limitations and use.
The general characteristics of the published algorithms can be
assessed against these requirements.
Angular Change Like other survey calculation methods,
the minimum curvature algorithm was originally developed to
calculate a well’s position from directional surveys. The
spacing between the survey stations was normally 30 to 500ft.
At that time with typical build rates, the total angle change
over a 100ft course length would rarely be allowed to exceed 5
deg and the final inclination of most of these early wells was
below 90 deg. When creating directional well plans the total
angle change between adjacent stations in the plan may be
considerably larger. These days, in designer wells the angular
change between two adjacent points on a well plan may
exceed 90 deg and the final inclination often exceeds 90 deg.
One well7 is recorded as having reached 164.7 deg inclination.
Many of the published algorithms do not contain an explicit
definition of the maximum permitted angle change. The
multiple solutions arising from periodicity of the trigonometric
equations involved makes this a serious concern.
Mathematical Behaviour The possibility of multiple
solutions means the results of the calculations may not always
be as intended, unless great care has been paid to their
implementation. Some algorithms employ iterative schemes so
that even if the scheme coverges, there is no guarantee that it
converges to the correct solution. Ideally, iterative schemes
should be accompanied by proof of convergence. At the very
least they should be thoroughly tested over some specified
range of variables. Additionally, there are cases where no
solution exists and extra code is needed to trap this condition.
Explicit expressions are more predictable and usually confer
advantages in speed and maintainability of the computer code.
For certain values, for example in geometrically straight hole,
expressions may be indeterminate. One solution adopted by
Zaremba is to define a suitably small number at which the
expression jumps abruptly to the asymptotic value2. However,
this can give rise to random differences between software
packages6. A better method is to develop series expressions
that enable a smooth transition to be maintained.
Common Constructs The absence of consistent
mathematical methods and nomenclature may hide common
constructs and potential simplifications in the coding of the
algorithms. For example, an equation of the form C = A sin(α)
+ B cos(α) appears in many of the geometric constructions
associated with the minimum curvature method. This equation
can be solved explicitly for α and several mathematically
equivalent forms exist. Zaremba proposed the form (1) that is
used throughout this paper.
(
 A ± A2 + B 2 − C 2 ) 
1
2
α = 2 tan 
−1

 B+C  …(1)
When presented in this way, it can be seen that no real
solutions exist if C2 > A2 + B2. This inequality has simple
geometric interpretations and several examples of its use are
highlighted in this paper.
Inconsistent nomenclature also leads to implemenational
difficulties. Review shows the nomenclatures used in the
literature are neither consistent with each other, nor consistent
with accepted mathematical practice. An example is the
definition of the normal vector that mathematical convention
has pointing towards the centre of curvature. This is opposite
to the convention used in the earlier drilling literature. Because
of the expansion in directional drilling applications, symbols
inevitably conflict. The Industry Steering Committee on
Wellbore Survey Accuracy (ISCWSA) has proposed some
standards8, but with limited scope. We conclude the SPE
documentation standards9 associated with this subject area are
no longer adequate and need revising.
Directional Calculations
A consistent vector notation is used throughout this paper. This
simplifies the development of the 3D equations and improves
the clarity and presentation of the results. For convenience, the
main vector operations are summarized in the appendix. In
some cases, series expansions have been used to ensure the
smooth transition of an expression to what would otherwise be
an indeterminate form. The thresholds at which the series
approximations should be used depend on the machine
precision. The constants used in this paper assume calculations
are good to at least 9 significant digits. The angle α subtending
the arc may assume values such that 0 ≤ α < π . Throughout, it
is also assumed the start and end points of the arc are not
coincident. Until such time as the standards are officially
revised, we have chosen to maintain commonality with earlier
papers on this subject and use a vector b pointing away from
the centre of the arc. For comparison with mathematical texts,
the normal vector is therefore –b.
Reference Frames
Co-ordinate Reference Frame The traditional reference
frame for directional work uses North, East and Vertical
coordinates that comprise a right handed set, Fig. 1. A point
N,E,V can be represented by the vector p, equation (2).
N
φ
E
θ
t
V
Fig. 1
SPE 84246 3

N  cosθ cos φ 
p =  E  …(2) h =  cosθ sin φ  …(6)

V   − sin θ 

A unit direction vector t can be represented in terms of the

− sin φ 
local inclination θ and azimuth φ, equation (3). The inclination
and azimuth values can be calculated from the vector’s r =  cos φ  …(7)
components using the expressions (4) and (5).  0 
∆N  sin θ cos φ  0 
t  
=  ∆E  =  sin θ sin φ  …(3)
v = 0 …(8)
 ∆V   cos θ 
1

Dogleg Severity
(
 ∆N 2 + ∆E 2 ) 
1
2

θ = tan −1
 …(4) Dogleg severity is a measure of the change in inclination
 ∆V  and/or direction of a borehole, Fig. 3. The change is usually
expressed in degrees per 100 ft of course length in oilfield
units3 and degrees per 30m in metric units. Dogleg severity is
 ∆E  used to determine stress fatigue in drill pipe, casing wear and
φ = tan −1   …(5)
casing design loads. It can also be a limiting factor in casing
 ∆N 
running and directional drilling operations. For the minimum
By using this reference frame an implicit assumption is made curvature method, the expression for the dogleg severity takes
that the earth is flat. For moderate distances from the origin the form (18000*α /π)/(D2−D1) in oilfield units. The
this assumption holds. For larger distances, such as those difference in measured depths D2 – D1 between the points is
encountered in extended reach wells Earth’s curvature is referred to as the course length S12.
important and corrections to the coordinates must be made.
Williamson and Wilson discuss the matter in detail10. N

Borehole Reference Frames Two reference frames are

associated with the borehole, Fig. 2. The first frame is formed φ1
by the highside, rightside and tangent unit vectors h, r and t
respectively. These form a right-handed, mutually orthogonal
E
set. In curved hole, the second frame comprises the normal, t1 φ2
binormal and tangent unit vectors –b, n and t respectively. θ1
These also form a right-handed, mutually orthogonal set. The α t2
angle between the highside vector h and normal vector –b is θ2
the toolface angle τ.
h

τ -b V
Fig. 3

Most expressions found in the literature involve the

r t
v assumption about the actual path of the wellbore, yet it is
Fig. 2
The highside, rightside and vertical unit vectors are
represented in equations (6), (7) and (8). Expressions for the
calculation of an arc cosine. These have been used even
though it is recognised that cosines of small angles are more
difficult to handle accurately than sines of small angles3.
Equation (9) developed by Lubinski does not make any
mathematically equivalent to the expression traditionally used
in the minimum curvature method. An expression for tan(α/2)
is readily developed from it.
  φ − φ1  2 
1

normal and binormal vectors –b and n can be found in the   θ − θ1 

α = 2 sin sin 2  2
−1
 + sin θ 1 sin θ 2 sin 2  2  
appendix (A-8 to A-12).   2   2  

…(9)
4 SPE 84246

Since the trigonometrical identity and computational
advantages of (9) were recognised it is surprising it has not
been adopted earlier. The dogleg severity can be related to
both the radius of curvature and the curvature κ of the arc
using the relationships (10).
situation is of no practical concern. This may not be the case
for adjacent points in a well plan trajectory, which may be
separated by considerable distances.

1 α α*
κ= = = …(10)
R S12 S *

Survey Calculation
Accurate determination of wellbore position is critical to well
placement, collision avoidance, reservoir modeling and equity p1
determination. Though the accuracy of the minimum curvature p2
method is acknowledged, Stockhausen and Lesso11 showed
that modern drilling practices could introduce systematic Fig. 5
errors even with survey intervals as frequent as 100ft.
Interpolation
n12 It is often required to identify the coordinates of a particular
point, say p* on a trajectory Fig. 6. In all cases, the problem
reduces to one of interpolation or extrapolation on an arc
b1 po defined by the positions p1 and p2 and directions t1 and t2 of its
R α end points. The algorithms presented here may be used for
p1 both functions. The interpolation may be driven by one of
b2 several parameters such as measured depth, subtended angle,
inclination, azimuth, northing, easting or vertical ordinate.
n12
S12
t1

R
p2 α∗
t2 p1 α
Fig. 4

The position of the next survey point p2 is calculated from p1

using (11), Fig. 4. The shape factor f(α) equals tan(α/2)/(α/2).
Details are summarised in the appendix (A-6 to A-18). t1
p*
sin θ 1 cos φ1 + sin θ 2 cos φ 2 
S12 f (α )  
p 2 = p1 +  sin θ 1 sin φ1 + sin θ 2 sin φ 2 
t* p2
2 t2
 cosθ 1 + cosθ 2 
…(11) Fig. 6

Straight Hole Conditions When α equals zero the shape
factor is mathematically indeterminate, so for α < 0.02 radians
the series expansion (12) should used instead12. The series is
presented in Horner13 form to minimise both the number of
arithmetic operations and the propagation of errors.
α 2  α 2  α 2  31α 2    …(12)
f (α ) ≈ 1 + 1+ 1 + 1 +  circle in the reference frame must be defined.
12  10  168  18   
There is a second possible solution of (9), equal to (2π – α).
The measured depth between the survey stations must be the
same in both cases, implying the second solution has a greater
curvature, Fig. 5. When calculating directional surveys, the
density of survey stations, behaviour of the bottom hole
assemblies and knowledge of the toolface settings means this
Before discussing each of these cases it is worth reviewing the
properties of a circular arc. The subtended angle, inclination
and azimuth of a point on a circular arc can be determined
solely by the attitude of the circle in the coordinate reference
frame. Knowledge of the size of the circle enables course
lengths to be determined. Finally, it is only if a north, east or
vertical ordinate is needed that the absolute position of the
Interpolation on Measured Depth The association of
events or observations such as formation tops and overpulls
with points on the wellbore is a common requirement.
Interpolations on measured depth determined from the pipe
tally are therefore the most common interpolation mode. If S*
is the course length along the arc at which the properties are
required, the relationships α* = (S*/S12)α and (α - α*) = (1 -
SPE 84246 5

S*/S12)α can be used to reduce the interpolation on measured (17) and (18). Choose the smallest root unless it is less than or
depth to the interpolation on subtended angle (13). equal to both θ1 and θ2 when the largest root should be chosen.

Interpolation on Subtended Angle The expression (13) (

A = sin θ1 cos φ * − φ1 ) …(16)
enables the direction vector t* at a point p* on an arc to be
determined solely from the direction vectors of its start and B = cosθ 1 …(17)
end points and the angle subtended from the first point p1 to
the point of interest. Refer to the appendix (A-19 to A-27). C = cosα * …(18)

*
t =
(
sin α − α *
t1 +
)
sin α *
t2 …(13)
As the subtended angle was originally determined from an
azimuth on the arc a solution must exist. Finally, the position
sin α sin α p* is determined using the minimum curvature equation (11).
Many of the algorithms presented in this paper involve the Generally, if C2 > A2 + B2 the orientation of the arc is too
determination of the subtended angle α* as a first step. The shallow for the inclination at any point on it to reach the
relationship (13) provides a convenient means of determining desired value. For details, see the appendix (A-28 to A-30).
the other parameters at the point once the subtended angle has
been found. For example, once t* is known, the corresponding Straight Hole Conditions Azimuth varies linearly with
point p* can be calculated using the minimum curvature measured depth in near straight hole conditions. For angles α
equation (11). < 10-4 radians, expression (19) should be used to determine the
corresponding measured depth. If the azimuth values in either
Straight Hole Conditions When the subtended angle α the numerator or denominator straddle north, the minimum
equals zero, both terms in (13) are indeterminate. For α < 0.02 angular difference should be used. The remaining properties
radians these terms which contain factors of the form can be determined by interpolating on this depth.
sin(cα)/sinα may be expanded14 in another Horner series (14).
 φ * − φ1 
sin (cα )  1 c    7 c 
S * ≈ S12   …(19)
2 −1
2 2
≈ c + α 2  c  −  + α 2  c 

 
 360 + c  36 + 120    φ 2 − φ1 
sin α  6 6     
Interpolation on Inclination The determination of
measured depths corresponding to some inclination range
  31  −7  1 c    2
+ α 2  c + c 2  + c 2  −  

appears in the automation of calculations for hole cleaning and
  15120  2160  720 5040    rock mechanics. Firstly check that both points θ1 and θ2 are
 
not equal to π/2. In this condition the arc would lie in the

 127  − 31  7  −1 c 2        horizontal plane and no solutions are possible. In the appendix
+ cα 2  + c2  + c 2  + c 2  +   (A-31) it is shown that the subtended angle α* can be
 90720  
 604800
   43200  30240 362880       
 calculated using (1). In this case the constants A, B and C are
…(14) given by the expressions (20), (21) and (22).

It should be noted this interpolation mode is impossible when A = cosθ 2 − cos α cosθ 1 …(20)
the subtended angle is identically zero, unless the constant c
can be defined in some other way, such as by using the ratio of B = sinα cosθ1 …(21)
the measured depths.
C = sin α cos θ * …(22)
Interpolation on Azimuth Occasionally, it is necessary to
truncate a well plan trajectory at a depth so that it is lined up If C2 > A2 + B2 then the orientation of the plane containing the
with some specified direction. Firstly check the condition arc is too close to the horizontal to enable the desired
inclination to be reached and no solutions exist. The
sin θ 1 sin θ 2 sin (φ 2 − φ1 ) ≠ 0 to determine that the arc does corresponding azimuth value can be determined from (23).
not lie in the vertical plane and that a solution exists. The
subtended angle is then determined using (15).
φ * = tan −1 
( )
 sin θ 1 sin φ1 sin α − α * + sin θ 2 sin φ 2 sin α * 
 ( )
sin φ1 − φ * sin α sin θ 1  ( )
 sin θ 1 cos φ1 sin α − α + sin θ 2 cos φ 2 sin α 
* *

α * = tan −1 
( *
) (
*
) 
 sin φ − φ 2 sin θ 2 + sin φ1 − φ sin θ 1 cos α  …(23)
…(15) Straight Hole Conditions Inclination varies linearly with
measured depth in near straight hole conditions. For the
In the appendix it is shown that the expression (A-29) used to subtended angle α < 10-4 radians, expression (24) should be
calculate the inclination θ* from φ* and α* is of the form C = used to determine the course length. The remaining properties
A sin(θ*)+ B cos(θ*). The solution for θ* is determined using can be determined by interpolating on this depth.
(1). In this case the constants A, B and C are given by (16),
6 SPE 84246

 θ * − θ1  If C2 > A2 + B2 the curvature of the arc is too large to intersect

S ≈ S12 
*
 …(24) the plane and no solutions exist. If C2 = A2 + B2 then the arc
 θ 2 − θ1  just touches the plane and there is only one solution. For C2 <
A2 + B2 there are two intersections and in this case the
Interpolation at a Plane A plane can be used to represent subtended angles to both of them may be less than π. The two
many geological features such as formation horizons and solutions correspond to the plus and minus signs in (1). This
faults. In 3D visualisation tools, collections of interlocking situation may be encountered when landing in a pay zone and
planes are used to represent complex geological features. the assembly is not building at the desired rate. Using this
Imaginary planes can be used to represent lease boundaries or interpolation mode it is possible to determine if and where the
the north, east and vertical coordinate limits on directional bottom of the zone will be breached and the point at which the
plots. The point p* at which the well meets or crosses these well is expected to re-enter. The lost production interval can
features is of great practical interest, Fig. 7. The plane is then be calculated directly.
uniquely defined by its normal vector m and any point px on it.
Straight Hole Conditions When the subtended angle α
R equals zero the above solution is indeterminate. For α < 10-4
α radians a series expansion (29) is used to determine the
p1
α∗ measured depth. See the appendix (A-37 to A-44). No solution
is possible if the line is parallel to the plane, when (m • t 1 )
equals zero.
px
t1
p2 ζ =
(
m • p x − p1 ) [(m • t 2 ) − (m • t 1 )]
…(28)
S12 (m • t 1 )
2

t2 p* m

S* ≈
(
m • p x − p1  ζ ) 
1 − (1 − ζ ) …(29)

(m • t1 )  2 

Fig. 7 Orientation of the Target Plane The normal vector m

There are five possible relationships between the arc and
the plane.
• An infinite number of intersections, when the plane
containing the arc and the target plane are parallel. This
condition should be tested first by establishing if both
(m • t 1 ) and (m • t 2 ) are equal to zero.
• Two intersections, when the arc completely cuts
the plane.
• One intersection, when the arc just touches the plane.
• No intersections, when the curvature of the arc is
too large.
• Lastly, if the curvature is so small then for practical
purposes the problem is reduced to the intersection of a
straight line with a plane.
defining the orientation of the target plane can be constructed
in terms of the dip angle Θ and dip azimuth Φ of the plane
(30).
− sin Θ cos Φ 
m =  − sin Θ sin Φ  …(30)
 cos Θ 
North, East and Vertical Interpolation The interpolation on
north, east or vertical ordinates are particular cases of the more
general expression. The values of the vectors px and m to use
in each of these are listed in Table 1. For example, if the well
is building and approaching an eastern lease line use the
vectors listed under the heading East, with E* set equal to the
numeric value of the eastern boundary limit.

In the appendix (A-32 to A-36) it is shown that the subtended North East Vertical
angle α* to the plane can be calculated using equation (1). In
this case the constants A, B and C are given by the expressions N *  0 0
  E *  0
(25), (26) and (27). px 0    
A = (m • t 1 )sin α …(25) 0  0  V * 
 
B = (m • t 1 )cosα − (m • t 2 ) …(26) 1 0  0 
0  1 0 
C=
( )
m • p x − p 1 α sin α
+ (m • t 1 ) cos α − (m • t 2 )
m  
0
 
0
 
1
S12
Table 1. Choice of vectors px and m for interpolating on north,
…(27) east and vertical ordinates.
SPE 84246 7

Turning Point In horizontal wells, the trajectory is steered Their algorithms were iterative and depended on the explicit
to remain in the reservoir section, successively building and determination of the centre of curvature of the arc, which
dropping inclination to avoid breaching the bottom or top of causes problems for small angle changes. In the appendix (A-
the pay zone. The calculation of the turning points at which 48 to A-65) it is shown how the minimum curvature and
the well becomes horizontal is of interest from a reservoir minimum distance can be determined explicitly, without
perspective Fig. 8. Since these points represent the maximum reference to the centre of curvature. For a fixed radius of
and minimum vertical depths, they are also needed to curvature there are two possible trajectories from p1 to p3
determine the numerical range of axes in well plots. shown by the solid and dashed lines in Fig. 10. All the
trajectories lie in a plane. The solid line shows the minimum
distance to the target. The other trajectory shown by the
t2 dashed line requires an angle change greater than π. Note that
there are four mathematical solutions but that two of these are
physically unrealisable because they would require the well to
p2 turn back on itself.
α
t* α∗ γ
n12
p1 p3
p* p1
t1 v
Fig. 8

This case is equivalent to interpolating on an inclination. In

this special case, direct determination of the azimuth φ* at the
turning point is possible (31). Fig. 10

 cos θ 1 sin θ 2 sin φ 2 − cos θ 2 sin θ 1 sin φ1 

φ * = tan −1 
 cos θ 1 sin θ 2 cos φ 2 − cos θ 2 sin θ 1 cos φ1 
…(31)
In deriving (31), it is assumed that θ 1 < θ 2 and so represents a
minimum inclination. Should θ 1 > θ 2 then the inclination is a
maximum. In this case the direction of t* will change by π and
the azimuth becomes (φ* + π). The course length S* to the
turning point can then be calculated from the relationship S* =
S12α*/α using (13). Finally, its position vector p* can be
calculated using the t1 and t* vectors with the minimum
curvature equation (11). See the appendix (A-45 to A-47).
Position at Target Defined
Hogg and Thorogood15 described expressions for the
minimum curvature and minimum distance from a point p1 on
a wellbore with direction t1 to a target p3, Fig. 9.
t1
p1
p2
R t2
α
β
p3
t3
Fig. 9
 The changes in angle can be categorised according to the
position of the target point p3 relative to the starting point p1
and the direction t1. These categories are represented by the
regions A to F in Fig. 11. Targets falling in region A can be hit
with angle changes less than π/2 (90 deg.), B with angle
changes less than π (180 deg.), C with angle changes less than
3π/2 (270 deg.) and D with angles changes less than 2π (360
deg.). Areas E and F cannot be hit at the prescribed build
up rate.
C B
D
F
p1
E
t1
A
Fig. 11
The appropriate region is determined by calculating the
distance ψ between the points p1 and p3, the perpendicular
distance η from p1 in the direction t1 and the distance ξ normal
to t1, equations (32), (33) and (34) respectively. Refer to
Fig. 12.
8 SPE 84246

t1 critical radius of curvature Rc given by (38). The

corresponding angle change αc is given by equation (39).
ξ
ψ2
p3 Rc = …(38)
2(ψ 2 − η 2 )2
1

η ψ
 ψ 2 −η 2 ( ) 
1
2
p1 α c = 2 tan  −1
 …(39)
 η 
The course length can be calculated as S12 = Rcαc. Finally, t2 is
calculated from (37) with the tangent section length β equal
to zero.
Fig. 12
2 Position and Direction at Target Defined
ψ 2 = p 3 − p1 …(32) Advances in surveying and geosteering techniques have
enabled multiple targets to be penetrated as a matter of course.
(
η = p 3 − p1 • t 1 ) …(33) The targets may be at different geological horizons or different
fault blocks in the same horizon. In these cases, the well’s
trajectory must be lined up and its direction on entry to or exit
ξ = (ψ 2 − η 2 )
1
2
…(34) from the target must be defined as well as its position. The
trajectory can no longer be achieved with a simple build and
From Fig. 11 it can be seen that if η ≤ 0 and ξ ≤ 2 R then hold profile. An additional curved section must be added and
in the general case the trajectory is three-dimensional, Fig. 13.
α ≥ π which is outside the imposed limit on the angle However, since the target can be specified so that the
change. With the restriction 0 ≤ α < π only targets in areas trajectory lies completely in a plane this calculation can also
A, B and E apply. be used to design nudge profiles to increase well separation
directly beneath a well cluster. Liu and Shi16,17 described an
Minimum Distance To Target If R < ψ + η then the
2 2 2 iterative scheme using coordinate transforms for the solution
of the equations. Different calculations were used for each of
tangent section length β is greater than zero. Because of the
the two arcs.
imposed restriction 0 ≤ α < π , the target must be in areas A or
B, Fig. 11. The values of the tangent section length and angle
We offer an alternative iterative solution based on the
change corresponding to the minimum distance to the target
geometrical symmetry of the problem and the minimum
are calculated using (35) and (36).
distance to target scheme described earlier in this paper. The
1
advantage of this scheme is that only the subtended angles α1
(
β = ψ 2 − 2 R ψ 2 − η 2 2  )
2 1
…(35) and α2 of the arcs need to be determined each iteration.
 
Inspection of Fig. 13 shows that in the general case, each of
the two sets of points p1, p2, p3 and p2, p3, p4 are geometrically
 η−β  similar. This suggests applying the minimum distance to target
α = 2 tan −1  1 
…(36) algorithm alternately for each of the sets of points.
(
 2 R − ψ 2 − η 2 2  ) t1
The course length is then calculated as S12 = Rα. In straight p1
hole ψ equals η and as long as R is finite (35) reduces to the p2
distance between the points p3 and p1. The straight-hole case R1 t2
therefore degenerates to a straight line rather than an arc with α1
an infinitely large radius, conferring stability to the β
calculation. Finally, t2 is calculated using (37).

S12 f (α ) α2
p 3 − p1 − t1 R2
t2 = 2 …(37) p3
S12 f (α ) t4
+β t3 p4
2
Fig. 13

Minimum Curvature To Target If R ≥ ψ + η the 2 2 2

Let p1, j and p4, j be the targets and α1, j and α2, j be the
target must lie in region E and the build rate is insufficient to subtended angles at the jth iteration. To start the scheme, the
hit it. To hit the target, the build rate must be increased to the
SPE 84246 9

subtended angle α1,0 is calculated using the minimum distance

to target algorithm between the points p1 and p4. This
corresponds to the trajectory 1 in Fig. 14. t2

po
p2
p1 α
p1,1
p1,2 t* α∗
n12
2
px
3 p1
1
p*
p4,1
t1
b* b1
Fig. 15
p4
p4,0
The closest approach is reached when the vector (px – p*) is
Fig. 14 normal to the curve. The subtended angle α* to this point is
calculated using expressions (43), (44) and (45).
A new target p1,1 is calculated using expression (40).

 α 1, j  (
η1 = p x − p1 • t 1 ) …(43)
p1, j +1 = p1 + R1 tan  t 1 …(40)
 2  η2 = (p x
− p )• t
1 2 …(44)
The subtended angle α2,0 is now calculated for the second arc
using the points p4 and p1,1. When performing the calculation,  η cosα − η 2 S12  
α * = tan −1 η1  1 +  …(45)
α  
the direction of the borehole at p4 must be set to –t4 in order to
calculate the correct angle. This corresponds to the trajectory 2   sin α
in Fig. 14. A new target p4,1 is calculated using expression (41)
which completes the first iteration. The corresponding position p* on the arc is determined by
interpolating on the subtended angle α* and then using the
 α 2, j  minimum curvature equation. Lastly, the distance is calculated
p 4, j +1 = p 4 − R2 tan  t 4 …(41) from the magnitude of the vector (px – p*). Mathematical and
 2  practical difficulties arise when the distance to the point is of
The next iteration is started by calculating the subtended angle the same magnitude or exceeds the radius of curvature of the
α1,1 using the points p1 and p4,1 and so on. Iteration continues arc. Restrictions on space preclude further discussion.
until the desired precision is achieved. A suitable criterion for Straight Hole Conditions Expression (45) is indeterminate
convergence is given by (42). in straight hole, when α equals zero. For α < 10-4 radians,

[
ε > (α 1, j +1 − α 1, j )2 + (α 2, j +1 − α 2, j )2 ] small angle approximations are used and expression (46) is
1
2
…(42) used to calculate the course length S* directly.

Four iterations are usually sufficient to reduce the error ε S12 η1

below 10-5 radians. Note that on completing the calculation,
the direction vectors of the second build and hold trajectory
must be reversed before use.
From a safety critical systems perspective, neither the above
scheme, nor that proposed by Liu and Shi are completely
satisfactory. In neither case is convergence proved and no
definitive statement is made regarding the conditions under
which no solutions exist, for example the radii of curvature are
too large to hit the target. Further work is required.
Closest Approach
The calculation of the closest distance of a point px to a
circular arc, Fig. 15 is the key to the construction of the
normal plane collision avoidance diagram described by
Thorogood and Sawaryn18.
S* ≈ …(46)
η1 − η 2 + S12
The point p* is calculated by interpolating on measured
depth assuming straight hole conditions. For these very small
subtended angles the value of the shape factor f(α*) in the
minimum curvature equation equals unity. Again, the distance
is calculated from the magnitude of the vector (px – p*).
Toolface Angle
Liu and Shi16 provided useful expressions for the toolface at
the start and end of the arc in terms of the inclination and
azimuth values at its ends. In the appendix (A-66 to A-79) it
is shown how the general vector equation provided by
Thorogood and Sawaryn18 may be expanded in terms of the
inclination and azimuth values at any intermediate point on the
arc as well as at its ends (47).
10 SPE 84246

tan τ * =
[ ( ) ( ) (
sin θ * sin θ 2 cos α * sin φ 2 − φ * + sin θ 1 cos α − α * sin φ * − φ1 )] The expressions (51) and (52) can be used to derive the
( )
cos α − α * cos θ 1 − cos α * cos θ 2 equivalents (54) and (55) of Liu and Shi’s expressions. The
…(47) representation of the azimuthal component is simplified.

At the ends, this expression reduces to those presented by Liu

and Shi. The forms (48) and (49) have a small advantage κ θ* = κ
( )
cos α − α * cos θ 1 − cos α * cos θ 2 …(54)
because they are not singular when the inclination at either of sin α sin θ *
the ends is zero. Dividing both numerator and denominator by
a factor that can be zero causes the singularity.
κ φ* = κ
( ) ( ) (
sin θ 2 cos α * sin φ 2 − φ * + sin θ 1 cos α − α * sin φ * − φ1 )
 sin θ 2 sin (φ 2 − φ1 )  …(48) sin α sin θ *
τ 1 = tan −1  
 sin θ 2 cosθ1 cos(φ 2 − φ1 ) − sin θ1 cosθ 2  …(55)
Straight Hole Conditions Again, for α < 10-4 radians,
 sin θ 1 sin (φ 2 − φ1 )  …(49) small angle approximations to (54) and (55) are used and
τ 2 = tan −1  
 sin θ 2 cosθ 1 − sin θ 1 cosθ 2 cos(φ 2 − φ1 )  expressions (56) and (57) are used to calculate the curvatures.
If the azimuth values in (57) straddle north, the minimum
Straight Hole Conditions For α < 10-4 radians, small angle angular difference should be used.
approximation to (47) is used giving (50). The toolface is
constant over the arc and is undefined when α equals zero. If θ * − θ1 θ 2 − θ1
the azimuth values in (50) straddle north, the minimum κ θ* ≈ *
≈ …(56)
S S12
angular difference should be used.

 φ − φ1  φ * − φ1 φ 2 − φ1
tan τ ≈ sin θ 1  2
*
 …(50) κφ ≈
*
≈ …(57)
 θ 2 − θ1 
*
S S12
Curvature Implementation and Testing
The total curvature κ is a constant 1/R on a circular arc, Fig. The algorithms in this paper are presented in logical order with
16. Liu and Shi presented expressions for the inclination and the later, more complex cases using results of earlier ones. It is
azimuthal components of the curvature at any point on an arc recommended that the routines are coded and tested in
in terms of the toolface angle at its start. this order.

Once coded, a good test procedure is to calculate values in two

∆α ∆θ h* different ways. For example, a point can be interpolated using

each of measured depth, inclination and azimuth determined
p* from the previous calculation in cyclic order. The results
R should be identical. The trajectory presented in Table 2 is
b* constructed with both the wellbore position and direction at
∆S θ the target defined. This example can be used to test most of the
∆φ r* t* algorithms presented here. Station numbers with alphabetical
suffixes indicate interpolated points. Williamson8 presents
other trajectories that may be used as tests in both oilfield and
metric units.
∆S sinθ
Fig. 16 The simple representation afforded by the circular arc
construct allows for a consistent treatment of all the other
Using vector methods the inclination and azimuthal curvatures associated mathematical operations. Representation of the
can be shown to be (51) and (52). For further details see the wellbore path in a different form, for example a spline or
appendix (A-80 to A-84). These two components satisfy polynomial would necessitate rederivation of all the constructs
Wilson’s19 expression (53) for the total curvature κ. in this paper. This may not be a simple task.

κ θ* = κ cos τ * …(51) Conclusions

1. Previously unpublished algorithms have been presented
for small angle approximations associated with a circular
sin τ * …(52) arc, the determination of a turning point, a general
κ φ* = κ
sin θ * expression for toolface angle and minimum distance to a
target with and without the direction at the target defined.
κ = (κ θ*2 + κ φ*2 sin 2 θ * )
1 2. Vector methods are a useful tool for 3D directional
2
…(53) calculations and often result in simpler expressions
compared with other means. Their use is recommended.
SPE 84246
3. Because of trigonometric identities, many forms exist for
any one expression. This can cause confusion. Some
forms are computationally more efficient and reliable
than others.
4. A standard nomenclature is required for all directional
work that is compatible with other, related subject areas.
Consistency with accepted mathematical conventions
should also be reviewed.
5. Further work is needed on the point to target algorithm
with both position and direction at the target defined. In
particular, the conditions under which no solutions are
possible should be identified.
6. Multiple solutions to the trigonometric equations exist.
The range of the variables must be carefully stated and
tested for.
7. Increasingly complex trajectory plans emerging from
work on designer wells means segments of the well plan
may exceed 180 deg and the alternative solutions may
need to be considered.
8. Care should be taken to design expressions so that both
the numerator and denominator are not divided by a term
that would make it indeterminate at some point.
9. Representation of the wellbore path in a form other than
a circular arc would necessitate rederivation of all the
constructs in this paper.
Nomenclature
b = Negative unit normal vector
c = Ratio of the course lengths S*/S12 on an arc
f = Geometrical shape factor
h = Unit highside vector
m = Unit vector normal to a plane
n = Unit binormal vector
p = Position vector in N,E,V coordinates, L, ft
r = Unit rightside vector
t = Unit direction vector
v = Unit vertical vector in N,E,V coordinates, L, ft
A = Constant
B = Constant
C = Constant
D = Measured depth, L, ft
E = Easting, L, ft
N = Northing, L, ft
R = Radius of curvature, L, ft
S = Course length, L, ft
V = Vertical, L, ft
Greek Symbols
α = Subtended angle, radians
β = Length of the tangent section, L, ft
γ = Angle between the binormal and vertical vectors,
radians
ε = Angular error tolerance, radians
ζ = Substituted variable
η = Substitution for a dot product, L, ft
θ = Inclination angle, radians
κ = Curvature, radians/L, radians/ft
ξ = Substitution for a dot product, L, ft
τ = Toolface angle, radians
11
φ = Azimuth angle, radians
ψ = Substitution for a dot product, L, ft
∆ = A difference in a parameter
Θ = Dip angle, radians
Φ = Dip azimuth, radians
Subcripts and Superscripts
c = Critical value
j = Iteration counter
o = Centre of curvature of the arc
x = Position of a defining point
1,2,3,4 = First, second etc. point, arc or property
θ = Inclination component
φ = Azimuthal component
* = Component to be determined
Acknowledgments
The authors thank BP plc. for permission to publish this paper.
References
1. Mason, C.M., Taylor, H.L.: “Systematic Approach to Well
Surveying Calculations”, SPE3362, June 1971
2. Zaremba, W.A.: “Directional Survey by the Circular Arc
Method, SPEJ February 1973;Trans., AIME, 255.
3. API Bull. D20, Directional Drilling Survey Calculation
Methods and Terminology, first edition, December 31, 1985
4. Walstrom, J.E., Harvey R.P., Eddy H.D.: A Comparison of
Various Directional Survey Models and an Approach To
Model Error Analysis”, JPT August 1972
5. Sawaryn, S.J., Sanstrom, W., McColpin, G.: “The
Management of Drilling Engineering and Well Services
Software as Safety Critical Systems”, SPE 73893 presented
at the International Conference on Health, Safety and
Environment in Oil and Gas Exploration and Production,
Kuala Lumpur, Malaysia, 20-22 March 2002
6. Sawaryn, S.J. et. al.: ”Safety Critical Systems Principles
Applied to Drilling Engineering and Well Services
Software”, SPE84152 presented at the Annual Technical
Conference and Exhibition, Denver, Colorado, 5-8
October 2003.
7. “Delivery Update - Autotrack Milestones”, Baker Hughes
INTEQ (2002) 18
8. Williamson, H.S.: “Accuracy Prediction for Directional
Measurement While Drilling”, SPEDC December 2000,
Vol 15 No. 4
9. SPE Letter and Computer Symbols Standard, 1993
10. Williamson, H.S., Wilson, H.F.: “Directional Drilling and
Earth Curvature”, SPEDC March 2000, Vol 15 No. 1
11. Stockhausen, E.J., Lesso, W.G.: “Continuous Direction and
Inclination Measurements Lead to an Improvement in
Wellbore Positioning”, SPE/IADC 79917, Amsterdam
February 2003.
12. Thorogood, J.L.: “Well Surveying Data”, World Oil, April
1986, 100
13. Knuth, D.E.: “The Art of Computer Programming Volume 2:
Seminumerical Algorithm”, Third Edition, Addison Wesley,
1997, 485-488
14. Wolfram Research, “Mathematica 4 Standard Add-On
Packages”, Wolfram Media Inc.,1999, 34
15. Hogg T.W., Thorogood J.L.: “Performance Optimisation of
Steerable Systems”, ASME Energy Resources Technology
Conference, New Orleans, January 1990
16. Liu, X., Shi, Z.: “Improved Method Makes a Soft Landing of
Well Path”, OGJ October 22nd 2001
12 SPE 84246

17. Xiushan, L., Jun, G., "Description and Calculation of the t1 × t 2

Well Path with Spatial Arc Model," Natural Gas Industry, n12 = …(A-8)
Beijing, Vol. 20, No. 5, 2000, 44-47. sin α
18. Thorogood J.L., Sawaryn S.J.: “The Travelling Cylinder: A
Practical Tool for Collision Avoidance”, SPE 19989 b1 = t 1 × n12 …(A-9)
19. Wilson, G.J.: "An Improved Method for Computing
Directional Surveys," J. Pet. Tech., Vol. 20, Aug. 1968, 871- b 2 = t 2 × n12 …(A-10)
76.
20. Weatherburn, C.E.: Elementary Vector Analysis, G. Bell and Substituting n12 from A-8 into A-9 and A-10 and using the
Sons, London (1967) expression A-4 for the vector triple products gives A-11 and
21. Abramowitz and Stegun: Handbook of Mathematical A-12.
Functions, Dover, New York (1972), Chap. 4, page 75
t 1 cosα − t 2
SI Metric Conversion Factors
b1 = …(A-11)
sin α
ft x 3.048* E-01 = m
deg/100 ft x 0.984252 E-00 = deg/30 m t 1 − t 2 cosα
*Conversion factor is exact b2 = …(A-12)
sin α
Appendix The position p2 is calculated from p1 using A-13. Substituting
Summary of Vector Methods b1 and b2 from A-11 and A-12 into A-13 gives A-14.
Weatherburn20 presents details of all the vector methods used
in this paper. The key constructions are the dot (or scalar) p 2 = p1 + R(b 2 − b1 ) …(A-13)
product • and vector cross product × . Let a1, a2, a3 and b1, b2,
b3 and c1, c2, c3 be the N,E,V components of the vectors a, b
p 2 = p1 + R
(1 − cosα ) (t +t2) …(A-14)
and c respectively. Let α be the angle between the a and b sin α
1

vectors and n be a unit vector normal to both a and b. The dot

and cross products are then described by expressions A-1, A-2 Finally, recalling the trigonometric identity tan(α/2) = (1-
and A-3. cosα)/sinα and that S12 = Rα we obtain the now familiar
expression for the minimum curvature equation A-15.
a • b = a1b1 + a2 b2 + a3b3 …(A-1)
S12 α 
p 2 = p1 + tan  (t 1 + t 2 ) …(A-15)
= a b cosα …(A-2) α 2
Straight Hole Conditions For small angles, tan(z) can be
a × b = a b sin α n …(A-3) expanded in a Taylor series21, A-16. Using the first five terms
of the expansion, the factor tan(α/2)/(α/2) can be put in
Horner12,13 form, A-17.
a × (b × c ) = (a • c ) b − (a • b ) c …(A-4)
z 3 2 z 5 17 z 7 62 z 9
a • (b × c ) = a1 (b2 c3 − b3 c 2 ) + a 2 (b3 c1 − b1c3 ) tan ( z ) = z + + + + + ... …(A-16)
3 15 315 2835

+ a3 (b1c 2 − b2 c1 ) …(A-5) tan

α
2 ≈ 1+ α
2  α2  α 2  31α 2    …(A-17)
1 + 1 +  
Minimum Curvature α 12  10  168 1 + 18   
   
Zaremba2 presented the following derivation. Refering to Fig.
2
4, the direction vectors t1 and t2 at the arc’s ends are given by
A-6 and A-7. For small angles the shape factor can be treated as unity and
A-15 reduces to A-18, which is recognised as the balanced
sin θ 1 cos φ1  tangential3 survey calculation method.
t 1 =  sin θ 1 sin φ1  …(A-6)
S12
 cosθ 1 
p 2 ≈ p1 + (t 1 + t 2 ) …(A-18)
2
Interpolation
sin θ 2 cos φ 2 
Refering to Fig. 6, interpolation involves determining the
t 2 =  sin θ 2 sin φ 2  …(A-7) position p* at some point on the arc given a criterion. The
corresponding direction vector at the point is t*, A-19.
 cosθ 2 
The binormal vector n12 and vectors b1 and b2 at the arc’s ends sin θ * cos φ * 
  …(A-19)
t =  sin θ * sin φ * 
*
are given by the expressions A-8, A-9 and A-10.
 cosθ * 
 
SPE 84246 13

The binormal vector n12 can also be written in terms of the Interpolation on Azimuth Dividing the easting and
direction vectors at the start point and point at which the northing components of A-24 eliminates sinθ* giving the
interpolation is to take place, A-20. expression A-28.

n12 =
t1 × t
*
…(A-20) =
( )
sin φ * sin θ 1 sin φ1 sin α − α * + sin θ 2 sin φ 2 sin α *
sin α * ( )
cos φ * sin θ 1 cos φ1 sin α − α * + sin θ 2 cos φ 2 sin α *
Equating A-8 and A-20 and taking the cross product of both …(A-28)
sides of the equality with t1 gives A-21. Next, expand the terms sin(α-α*) in both numerator and
(
t1 × t1 × t
*
)= t 1 × (t 1 × t 2 ) …(A-21)
denominator of A-26 using the trigonometric identity sin(α-
α*) = sinα cosα* - sinα* cosα. The terms involving sinα* are
sin α * sin α then collected on the left hand side of the equals sign and
Using A-4 to expand the triple cross products and rearranging terms involving cosα* are collected on the right hand side.
for t* gives A-22. The azimuth terms are then combined using the same
trigonometric identity to give the expression (15) for α* used
sin α * (t 1 cosα − t 2 ) to interpolate on azimuth in the body of the paper. The
t = t 1 cosα * −
*
…(A-22)
sin α traditional form of the dogleg severity equation3 A-29 can be
Multiply the numerator and denominator of the first term in A- used to determine θ*. This is of the form C = A sin(θ*)+ B
22 by sinα and collect terms in t1 and t2. After simplification, cos(θ*).
this gives the important relationship A-23 that is the
foundation for all the interpolation formulae.
( )
cos α * = sin θ 1 cos φ * − φ1 sin θ * + cosθ 1 cosθ *

*
t =
(
sin α − α *
t1 +
sin α * )
t2 …(A-23)
…(A-29)
sin α sin α If the arc lies in the vertical plane then (v • n12) equals zero and
the solution is single valued. The vector n12 is given by A-8.
Substituting A-6, A-7 and A-18 in A-23 gives A-24. Expanding the scalar triple product using A-5 gives the
sin θ * cos φ *  expression A-30.
sin θ 1 cos φ1 

θ φ * sin α − α * ( )  sin θ  sin θ1 sin θ 2 sin (φ 2 − φ1 )
= 1 sin φ1  +
*
 sin sin   =0 …(A-30)
sin α sin α
 cosθ *   cosθ 1 
 
Interpolation on Inclination Extracting and rearranging
the vertical component of A-24 gives A-31 which is of the
sin θ 2 cos φ 2  form C = A sin(α*) + B cos(α*). Once α* has been found, the
sin α * 
sin θ 2 sin φ 2  …(A-24) azimuth component is determined from A-28.
sin α 
 cosθ 2  sin α * cosθ * = (cosθ 2 − cos α cosθ 1 )sin α * +

Interpolation on Measured Depth Since the radius of the sin α cosθ 1 cosα * …(A-31)
arc is fixed, the ratio of the subtended angles is identical to the
Interpolation at a Plane Refering to Fig. 7, A-14 can be
ratio of the course lengths to the same points, A-25.
used to calculate the point p* from p1. The radius R can be
α* S* expressed as S12/α to give A-32.
= …(A-25)
α S12 *
p = p1 +
(
S12 1 − cosα * )(
t1 + t
*
) …(A-32)
Substituting α* from A-25 into A-23 gives the expression A- α sin α *
26 for t* in terms of the course lengths.
Now use A-22 to substitute for t* in A-32 to give A-33.

sin 1 −
S *  
α
 sin
 S* 
 α *
p = p1 +
(
S12 1 − cos α * ) ( )
1 + cos α * t 1 −
*   S12    S12  α sin α * 
t = t1 + t2 …(A-26)
sin α sin α
For small values of α the expression A-26 can be expanded in sin α * (t 1 cosα − t 2 )  …(A-33)

a Taylor series. Evaluation of the terms is tedious and sin α 
computer assistance14 was used to establish the expression (14)
presented in the body of the paper. For small angles, the Now take the dot product of A-33 with the normal vector m of
simple expression A-27 can be used. the plane and rearrange slightly.

 S*   S* 
t ≈ 1 −  t 1 +   t 2
*
…(A-27)
 S12   S12 
14 SPE 84246

( *
m • p − p1 = ) (
S12 1 − cos α *  ) ( )
1 + cos α * (m • t 1 ) − ζ =
(
m • p x − p1 ) [(m • t 2 ) − (m • t 1 )]
…(A-41)
α sin α *  S12 (m • t 1 )
2

sin α * ((m • t 1 ) cos α − (m • t 2 ))  …(A-34) − (m • t 1 ) ± (m • t 1 )(1 + 2ς ) 2

1

sin α  S = * …(A-42)
 (m • t 2 ) − (m • t 1 )
The equation of the plane20 is given by A-35, showing (m • px)  
equals (m • p*).  S12 
(
m• px − p = 0
*
) …(A-35) Since 2ζ is much smaller than unity, the square root may be
expanded in a series using the first four terms of the binomial
expansion21, A-43 to give the expression A-44.
Equation A-35 can be used to eliminate p* from A-34. After
rearranging, the resulting expression A-36 is of the form C = A z z2 z3 z4
(1 + z )
1

sin(α*) + B cos(α*).
2
= 1+ − + − +L …(A-43)
2 8 16 128
(
m • p x − p1 α sin α ) + (m • t 1 ) cosα − (m • t 2 ) =  ς2 ς3 
S12 − (m • t 1 ) ± (m • t 1 )1 + ς − + 
 2 2 …(A-44)
S ≈
*

(m • t 1 )sin α sin α * + [(m • t 1 )cosα − (m • t 2 )]cosα *  (m • t 2 ) − (m • t 1 )

 
…(A-36)  S12 
From A-44, it can be seen that the positive root must be
Straight Hole Conditions For small angles, equation A-36 chosen so that the expression degenerates to the straight-line
is badly behaved. Small angle approximations must be used case when ζ equals zero. Factoring A-44 gives the expression
and the interpolation must be conducted with respect to (29) in the body of the paper.
measured depth. The expression A-18 can be used to calculate
the position of p* from p1, A-37. Turning Point From Fig. 8 the vector t* can be written as

( )
S * A-45. The angle between the n12 and v vectors is γ.
* *
p = p1 + t1 + t …(A-37)
2 * n12 × v
t = …(A-45)
Sustituting A-27 into A-37 gives A-38. sin γ
Using A-8 for n12 and expanding the resulting vector triple
S *2
2S12
(
(t 2 − t 1 ) + S * t 1 − p * − p1 = 0 ) …(A-38) product gives A-46.

*
− (v • t 2 ) t 1 + (v • t 1 )t 2
Taking the dot product of A-38 with the normal vector m of t = …(A-46)
the plane and substituting (m • px) for (m • p*) as before, gives sin γ sin α
a quadratic equation A-39 in the course length S*. A-39 has Note that (v • t2) equals cosθ2 and (v • t1) equals cosθ1 and that
the solution A-40.
θ* equals π/2 at the turning point. Using these values and
(m • t 2 ) − (m • t 1 ) S *2 + (m • t )S * − m • ( p − p ) = 0 expressions A-19, A-6 and A-7 for t*, t1 and t2 in A-46
2 S12
1 x 1 gives A-47.

…(A-39) cos φ *  sin θ 2 cos φ 2 

 * cosθ 1  sin θ sin φ  −
 sin φ  = sin γ sin α  2 2


− (m • t 1 ) ± (m • t 1 ) +
2
(
2m • (t 2 − t 1 ) m • p x − p1  2 ) 1  0 
   cosθ 2 

 S12 
S* = sin θ 1 cos φ1 
 m • (t 2 − t 1 ) cosθ 2 
  sin θ 1 sin φ1  …(A-47)
 S12  sin γ sin α 
 cosθ 1 
…(A-40)
Dividing the easting by the northing components gives
To simplify the manipulation, define a variable ζ according to expression (31) in the body of the paper for the azimuth φ* of
A-41. Expression A-40 can then be written as A-42. the turning point.
SPE 84246 15

Position at Target Defined η 2 − 2 Rη sin α + R 2 sin 2 α

In the most general case, the target p3 can be hit with one β2 = …(A-60)
cos 2 α
curved section of radius R and one straight section of length β,
Fig. 9. Using A-15 the position p3 of the target can be written The substitution of β from A-53 and β 2 from A-60 into A-54
as A-48. results in A-61.
α  2 R sin α
p 3 = p1 + R tan  (t 1 + t 2 ) + β t 2 …(A-48) ψ 2 = 2 R 2 (1 − cos α ) + (η − R sin α ) +
2 cos α
Taking the dot product of A-48 with t1 and also with itself η 2 − 2 Rη sin α + R 2 sin 2 α
gives A-49 and A-50 respectively. + …(A-61)
cos 2 α
(p 3
)
− p1 • t 1 = R sin α + β cosα …(A-49)
Multiplying through by cos2α and rearranging terms gives A-
62. Completing the square on the right hand side of A-62
p 3 − p 1 = 2 R 2 (1 − cosα ) + 2 βR sin α + β 2 …(A-50)
2
gives A-63.
ψ 2 cos 2 α = 2 Rη sin α (cosα − 1) + η 2 +
Let

ψ 2 = p 3 − p1
2
…(A-51)
( )
R 2 1 + cos 2 α − 2 R 2 cosα …(A-62)

(
η = p 3 − p1 • t 1 ) …(A-52)
(ψ 2
)
− η 2 cos 2 α = [η sin α − R(1 − cosα )]
2
…(A-63)
Taking the square root of A-63 and rearranging for R results in
Making the substitution A-51 in A-50 and A-52 in A-49 gives equation A-64 of the form C = A sin(α) + B cos(α). The
A-53 and A-54. The variable ξ is determined using solution of such an equation is given by (1).
Pythagoras’s theorem, Fig. 12.
(
R = η sin α +  R ± ψ 2 − η 2 2  cosα )
1
…(A-64)
η = R sin α + β cosα …(A-53)  
After substituting the values of the constants A, B and C and
ψ 2 = 2 R 2 (1 − cosα ) + 2 βR sin α + β 2 …(A-54)
comparing the result with A-58 we obtain the important result
Multiplying A-53 by 2R gives A-55. Rearranging A-54 gives A2 + B2 – C2 = β 2. Setting β equal to zero provides a useful
A-56. test to determine if the target can be hit at the specified radius
of curvature. Finally, the expression for α/2 is given by A-65.
2 R 2 sin α + 2 βR cosα = 2 Rη …(A-55)
α   η−β 
tan  =   …(A-65)
2 βR sin α − 2 R cosα = ψ − β − 2 R
2 2 2 2
…(A-56) 2
   2 R − ψ 2 − η 2 ( )
1
2

Squaring both A-55 and A-56 and adding the results
Setting β equal to zero and R equal to Rc in A-65 gives the
eliminates α and results in a quadratic in β 2, A-57. Solving
expression (39) in the body of the paper for the subtended
the quadratic gives an explicit expression for β, A-58
angle αc for the minimum curvature to target.

β 4 − 2 β 2ψ 2 + (ψ 4 − 4 R 2ψ 2 + 4 R 2η 2 ) = 0 …(A-57) Toolface Angle

The toolface angle is determined by the dot products between
1 the -b, h and r vectors, Fig. 2.
β = ψ 2 ± 2 R (ψ 2 − η 2 ) 
1

(− b )
2
2
…(A-58)
• h = cosτ *
* *
  …(A-66)

Inspecting A-58, the minimum curvature to target will occur

when β equals zero and the radius equals Rc, A-59. Since the
(− b *
• r ) = sin τ
* *
…(A-67)
radius cannot be negative, the negative root of A-58 must be Expressions A-66 and A-67 lead directly to the vector
the correct one. equation A-68 for toolface angle presented by Thorogood
and Sawaryn18.
ψ2 …(A-59)
Rc =
( )  − b* • r * 
1

2 ψ −η 2 2 2
tan τ * =  * *  …(A-68)
To find the angle α, rearrange A-53 for β and square the result −b •h 
to give A-60. The rightside unit vector lies in the horizontal plane normal to
both the v and t* vectors, A-69. Evaluating the expression
gives (7) in the body of the paper.
16 SPE 84246

* v×t
* Inserting the expressions for the scalar triple products A-77
r = …(A.69) and A-78 into A-76 gives the final form for (-b* • r*), A-79.
sin θ *
The highside unit vector lies in the vertical plane normal to
(− b *
•r
*
) = sin θ 2 ( ) ( ) (
cos α * sin φ 2 − φ * + sin θ 1 cos α − α * sin φ * − φ1 )
both the r* and t* vectors, A-70. Combining A-69 with A-70 sin α
and expanding the triple vector product gives A-71. Evaluating …(A-79)
the expression gives (6) in the body of the paper. The general expression (47) given in the body of the paper for
* * * toolface angle τ* is obtained by substituting A-75 and A-79
h = r ×t …(A-70)
into A-68. For straight hole conditions, small angle
approximations must be used.
t cosθ * − v
*
*
h = …(A-71)
sin θ * Curvature
Refering to Fig. 16, Frenets20 formula for total curvature κ
The negative normal unit vector b* is normal to both the t* gives A-80.
and n12 vectors. Combining A-8 and A-19 with A-72 gives A-
*
73. dt
= −κ b
*
*
…(A-80)
*
b = t × n12 *
…(A-72) dS
The definitions for κ θ* and κ φ* as the inclination and azimuthal
*
b =
(
cos α − α * t 1 − cos α * t 2 ) …(A-73) components of the curvature give expressions A-81 and A-82.
sin α
*
Taking the dot product of A-73 and A-71 gives an expression dt
= κ θ* h
*
…(A-81)
for (-b* • h*), A-74. Taking the dot products gives A-75. dS *

( ) ( )
*
 t * • t 2 t 1 − t * • t 1 t 2   t * cos θ * − v  1 dt
( ) = κ φ* r
*
* * …(A-82)
− b • h = − •  sin θ dS
* *
 sin α   sin θ
*

Eliminating the differential between A-80 and A-81 and taking
…(A-74) the dot product of both sides with h* gives A-83 for κ θ* .
(− b *
•h =
*
) (
− cos α * cos θ 2 + cos α − α * cos θ 1 …(A-75) ) Using A-66 and the expression A-75 for (-b* • h*) gives the
sin α sin θ * expression (54) in the body of the paper.
Taking the dot product of A-73 and A-69 gives an expression
for (-b* • r*), A-76. κ θ* = κ cosτ * …(A-83)

(− b *
•r =
*
) [ (
cos α * t 2 • v × t
*
)] − cos(α − α ) [t • (v × t )]
*
1
*
Eliminating the differential between A-80 and A-82 and taking
sin α sin θ *
the dot products of both sides with r* gives A-84 for κ φ* .
…(A-76) Using A-67 and the expression A-79 for (-b* • r*) gives
expression (55) in the body of the paper.
Using A-5, the scalar triple products become A-77 and A-78.
sin τ *
( )
t 1 • v × t = − sinθ1 sin θ sin φ − φ1
* *
( *
) …(A-77) κ φ* = κ
sin θ *
…(A-84)

• (v × t ) = sin θ (
sin θ * sin φ 2 − φ * )
*
t2 2 …(A-78) The signed values of A-83 and A-84 are commonly referred to
as the build rate and walk rate respectively.

No. MD Inc. Azi. N E V DLS τ κθ κφ κ R

(ft) (deg) (deg) (ft) (ft) (ft) (deg/100ft) (deg) (deg/100ft) (deg/100ft) (deg/100ft) (ft)
1 702.55 5.50 45.00 40.00 40.00 700.00 - 38.14 1.57 12.89 2.00 2864.78
2 1964.57 29.75 77.05 154.78 393.64 1895.35 2.00 6.85 1.99 0.48 2.00 2864.78
2a 4250.00 29.75 77.05 408.84 1498.82 3879.60 0.00 - - - - -
3 5086.35 29.75 77.05 501.82 1903.25 4605.73 0.00 -135.72 2.15 4.22 3.00 1909.85
3a 8504.11 80.89 300.71 1967.04 1033.30 7050.00 3.00 -20.54 2.81 1.07 3.00 1909.85
3b 8828.04 90.00 297.31 2123.40 751.22 7075.71 3.00 -20.27 2.81 1.04 3.00 1909.85
3c 9151.97 99.11 293.92 2262.88 460.41 7050.00 3.00 -20.54 2.81 1.07 3.00 1909.85
4 9901.68 120.00 285.00 2500.00 -200.00 6800.00 3.00 -23.58 2.75 1.39 3.00 1909.85
Table 2. Test case based on a trajectory with both its position and direction defined at the target