IADC/SPE 151448

Improving Casing Wear Prediction and Mitigation

Using a Statistically Based Model
Sarah Mitchell, WWT International
Yanghua Xiang, Pegasus Vertex International

Copyright 2012, IADC/SPE Drilling Conference and Exhibition

This paper was prepared for presentation at the 2012 IADC/SPE Drilling Conference and Exhibition held in San Diego, California, USA, 68 March 2012.
This paper was selected for presentation by an IADC/SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not
been reviewed by the International Association of Drilling Contractors or the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily
reflect any position of the International Association of Drilling Contractors or the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any
part of this paper without the written consent of the International Association of Drilling Contractors or 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 IADC/SPE copyright.

Abstract
Wells are now routinely drilled both in deepwater and on land to depths that were previously considered impossible. In these
environments, casing design is critical to safely and successfully drilling and producing wells, and unexpected casing wear
can result in significant costs or even the loss of a well. As part of a successful casing design strategy, the engineer must
assess the maximum permissible casing wear required to maintain casing integrity. Then, steps must be taken to ensure that
casing wear thresholds are not exceeded.
Casing wear models use the number of drill string revolutions and contact force between the drill pipe and casing to calculate
wear. The contact force is calculated using the dog-leg severity within the well, with the maximum dog-leg severity often
determining the location and extent of the most severe casing wear. There is often a large discrepancy between predicted and
actual casing wear because of survey quality and inaccurate estimates of dog-leg severity and total revolutions. These
discrepancies result in predictions of contact force and drill string revolutions that are in error by 50% or more.
To improve the accuracy of casing wear models, an extensive database was created from a wide variety of wells with
measured depths greater than 13,000ft. The database results in a statistically based model for determining dog-leg severity
within vertical, build, and tangent sections, as well as total drill string revolutions at various levels of confidence to bound
average and maximum expected contact force and casing wear.
Case histories compare measured wear with predictions of casing wear based on original well data and the statistically based
model. The case histories also demonstrate the effect of various drilling parameters on casing wear, and evaluate the
effectiveness of non-rotating protectors in preventing casing wear.
Project Goal
The goal of this project was to more accurately quantify casing wear risk by improving casing wear analysis accuracy. To do
this, data from a large number of wells was analyzed to generate probabilities for dog-leg severity in common well types and
also correlate those to actual backmodeled casing wear factors. The results will allow an engineer to analyze what the
expected casing wear might be for an average (P50) horizontal well, and then evaluate the maximum expected wear for a 1 in
10 case (P90), 1 in 20 case (P95), or 1 in 100 (P99) case.
Surveys
All casing wear software, and torque and drag software as well, use a directional survey to determine the side force or contact
force between the drill string and wellbore. These points within a directional survey can be a representation of a planned well
path, or it can be taken from actual downhole measurements. The survey points are then connected into a single line
representing a best approximation of the wellpath with the information given.

IADC/SPE 151448

For many reservoir engineers, the positional accuracy of the wellpath is important only in terms of determining if the well
reached the targets. But for casing wear modeling, the side force or contact force between the drill pipe and casing has a
direct effect on wear rate. Lubinski (1961, 1984) have shown that the severity of curvature in the wellbore is related to the
contact force. The severity of curvature in a well is represented as dog-leg severity DLS, normally expressed in /100ft or
/30m.
Tortuosity
In modeling a well using a planned wellpath, an amount of tortuosity, or artificial curvature, is added to an otherwise
perfectly smooth planned wellpath to more closely represent the quantity and magnitude of curvature that will be inherent in
the real wellbore. Tortuosity can take many forms, including sinusoidal, helical, or randomly generated deviation as shown
by Samuel (2005) and others. In modeling a well using actual surveys, the distance between survey points, also called the
survey station frequency, and the accuracy of the tool taking the measurements, can greatly influence the accuracy of the
calculated wellpath to the actual path of the wellbore.
The way that the tortuosity is used is important. Because the effects of torque and drag are cumulative over the entire drill
string, the effect of dog-leg severity and resulting contact force at any one point is not as important as the overall average
level of contact force over the entire well. In contrast, when modeling casing wear, the goal is to calculate the minimum
burst and collapse pressures for the casing string resulting from the maximum amount of casing wear. In this case, the
maximum dog-leg severity value, and not the average dog-leg severity, is important because an acute dog-leg over a small
distance can cause severe contact loads and result in failure of the casing at that location. As a result, the average dog-leg
severity over an area is important in running torque and drag simulations, but an accurate representation of the maximum
expected dog-leg severity is important in modeling casing wear.
Casing Wear Calculation
Earlier studies of casing wear have concluded that most wear is caused by drill string rotation rather than reciprocation.
Rotating tool joint against casing interior wears crescent shaped grooves in casing.
To predict the rate of casing wear by rotating drill string tool joint, it is necessary to express the rate of wear in terms of fieldmeasured parameters, which include the mud system, tool joint material, casing material, dog-leg severity, rotary speed and
ROP, and the tension along the drill string (torque and drag). Bradley(1975),Bol(1986),Williamson (1985) and Hall(1993,
1994, 2005), and others have all helped to develop methods of calculating casing wear.
The volume of casing wall removed per foot in time t hours is mathematically expressed in the equation below.

WV WF SFdp Dtj 60 N t
Where:
WV
WF
SFdp
Dtj
N
t

Casing wear volume per foot, in3/ft

Casing wear factor, defined as the ratio of friction factor to specific energy, E-10psi-1
Side force on drill string per foot, lbf/ft
Tool joint OD, inch
Rotary speed, RPM
Rotating time, hour

The definition of wear factor is the ratio of friction factor to specific energy, which is the amount of energy required to
remove a unit of steel. An example of some experimentally determined wear factors are shown below in Table 1.

Mud type
Tool joint material
Rotating
Drill Pipe protector

Table 1. List of Experimentally Determined Wear Factors

Selections
Water or Water based, Steel tool joint
Oil based mud, Steel tool joint
Smooth tungsten carbide
Very rough tungsten carbide
Other proprietary casing-friendly hardbanding
Pipe protector started with rusted casing
Pipe protector with average casing interior
Pipe protector after polishing casing

-1

Wear Factor (E-10 psi )

0.5 to.40
0.3 to 5
8.5
1625
1 to 6
4.1
2.1
0.06

Additional information on the methods for calculating casing wear for the purposes of this paper can be found in the
appendix.

IADC/SPE 151448

Well Survey Database

To provide a statistical understanding of the average and maximum expected dog-leg severity within a variety of wells,
surveys from more than a hundred recent directional wells from around the world were collected and entered into a database
for analysis. The wells were segregated by category. These categories include horizontal, ERD (extended reach drilling),
deepwater, and other directional wells. Horizontal wells are wells that are primarily vertical then horizontal. ERD wells
have a reach ratio of 2:1 or greater. Deepwater wells are drilled in approximately 1500ft of water or greater. The directional
well category includes wells such as build-and-hold and S-shaped wells. The surveys of each of these wells were then further
segregated into vertical, build, and tangent or horizontal sections where possible. Also included in the horizontal well
category was a low angle build section for pad-drilled wells that made short build to a 5 to 20 degree inclination angle.
The methodology for this study is to first analyze each section of each well regarding its average dog-leg severity and
maximum dog-leg severity. Next, we examined similar wells of the same category and section. This produces the average
dog-leg severity for the data group and the average maximum dog-leg severity within the data group, with its associated
standard deviations on both parameters. Next, we evaluate the average dog-leg severity and average maximum dog-leg
severity of the several data groups in the database using standard statistical methods (Lapin 1990) to determine the
cumulative probability at 50%, 90%, and 95% levels (P50, P90, P95).
As an example, all of the wells within the Directional Well category are shown in Figure 1. The average and maximum
dog-leg severities for the build sections are represented in the histogram chart in Figure 2. As can be seen, the average and
maximum dog-leg severity generally fit into a normal distribution patterns. So, the average or mean was calculated for each
distribution. For example, when building angle, the average directional well has an average dog-leg severity of 1.5 /100ft,
and a maximum dog-leg severity of 2.94 /100ft. In this same example, 90% of these wells will have an average and
maximum dog-leg severity that is less than 1.28 standard deviations from mean. So, a P90 case will have an average dog-leg
severity of 2.50 /100ft or less, and a maximum dog-leg severity of 5.04 /100ft or less in the build section.

Figures 1 and 2. Chart showing all wells in the directional category, with a historgram of average and maximum dog-leg severity in
the build section for each of these wells.

To calculate the distribution of dog-leg severities within each case, we use the average and maximum expected dog-leg
severity to calculate a distribution pattern of dog-leg severities within each section, using the formulas shown in Table 2.
These results are shown in Tables 3, 4, 5, and 6. An example of the probability distribution is shown below in Figure 2 for a
P90 case. In the P90 case, there is a 90% probability that on a 95ft frequency survey, the average and maximum dog-leg
severity will be less than the values listed for each section of the well.
Probability
P50

Table 2. Explanation of the Source of the Values Given in Well Statistics Tables.
Average DLS
Maximum DLS
Expected Standard Deviation of DLS
P50 =
MP50 = max
P50=(MP50 - P50) / s

P90

P90 = + 1.28

MP90 = max + 1.28max

P90=(MP90 - P90) / s

P95

P95 = + 1.64

MP95 = max + 1.64max

P95=(MP95 - P95) / s

IADC/SPE 151448

Where:

max
max
s

Average (arithmetic mean) of the Average DLS values for all wells within a category
Standard deviation of mean DLS values for all wells within a category
Mean of the maximum DLS values for all wells within a category
Standard deviation of the maximum DLS values for all wells within a category
Standard deviations based on the average number of survey points within each section. This represents the number
of standard deviations based on a probability that the maximum dog-leg severity is taken from 1 in x, where x is the
average number of survey measurements for a section.

Using the example above, for the directional well category, the average build section has 30 survey points, with a P50=1.50
/100ft and MP50=2.94 /100ft. So, a dog-leg severity of 2.94 /100ft is a 1 in 30 chance in the average well, given that 30
survey points have been taken. So, this equates to this maximum value occurring 1.83 standard deviations away from the
mean value. So in this case, the standard deviation, P50, would be:
(2.94-1.50)/1.83 = 0.78 /100ft.
There is a considerable difference in dog-leg severity between the various types of wells. In many of the ERD and deepwater
wells, rotary steerable drilling systems are used routinely. In the horizontal and directional well categories, some wells use a
rotary steerable BHA, and some wells use a conventional directional BHA. As shown in Figure 3, there is a positive
correlation between the average DLS and standard deviation of the DLS in each section of each well, with the standard
deviation of the dog-leg severity ranging from about 0.3 to 0.7 times the average dog-leg severity.

Figure 3. Graph showing the average vs. standard deviation of dog-leg severity for wells used in this database to illustrate that the
standard deviation increases with increasing build rate.

Confidence
Levels
P50
P90
P95

Table 3. Horizontal Well Data Group Showing Dog-Leg Severity (/100 ft) in Various Sections
of the Wells at Various Confidence Levels
Vertical Section
Build Section
Horizontal Section
Low Angle Build
Average
Std. Dev.
Average
Std. Dev.
Average
Std. Dev.
Average
Std. Dev.
0.32
0.45
9.96
2.65
1.53
1.87
1.37
1.29
0.49
0.91
14.53
4.63
2.26
3.38
2.01
1.73
0.53
1.03
15.81
5.19
2.47
3.81
2.20
1.86
Table 4. Deepwater Well Data Group Showing Dog-Leg Severity (/100 ft) in Various Sections
of the Wells at Various Confidence Levels
Vertical Section
Build Section
Tangent Section
Confidence
Levels
Average
Std. Dev.
Average
Std. Dev.
Average
Std. Dev.
P50
0.24
0.35
1.57
1.02
0.42
0.81
P90
0.37
0.64
2.58
1.78
0.59
1.36
P95
0.41
0.73
2.87
1.99
0.63
1.52

IADC/SPE 151448

Table 5. ERD Well Data Group Showing Dog-Leg Severity (/100 ft) in Various Sections of the Wells
at Various Confidence Levels
Vertical Section
Build Section
Tangent Section
Confidence
Levels
Average
Std. Dev.
Average
Std. Dev.
Average
Std. Dev.
P50
0.51
0.41
2.18
1.47
0.75
0.75
P90
1.29
1.03
3.33
2.44
1.37
1.21
P95
1.50
1.21
3.65
2.71
1.55
1.34
Table 6. Directional Well Data Group Showing Dog-Leg Severity(/100 ft) in Various Sections of the
Wells at Various Confidence Levels
Vertical Section
Build Section
Tangent Section
Confidence
Level
Average
Std. Dev.
Average
Std. Dev.
Average
Std. Dev.
P50
0.39
0.43
1.50
0.78
0.61
0.75
P90
0.81
0.81
2.50
1.38
0.91
1.20
P95
0.92
0.91
2.78
1.55
1.00
1.33

Effect of Rotary Steerable vs. Conventional Directional BHA on Dog-Leg Severity

Difficulty was encountered in differentiating wells with various BHA types within the database. For example, horizontal
wells are drilled with different BHA types in different sections of the well. This is in contrast to deepwater wells that are
most often entirely drilled with some type of Rotary Steerable System (RSS). To examine the effect of RSS versus
conventional directional methods, two wells drilled in the Gulf of Mexico were examined. The wells were similar in shape,
with a build and hold shape at 60 to 70 inclination. Measured depth was adjusted to allow an overlay for comparison. As
seen in Figure 4 and Table 7, the average build rate is similar, but overall the standard deviation in dog-leg severity was much
greater using the conventional directional BHA. Therefore, in the vertical and horizontal or tangent section of most
directional wells, there will be an increase in the standard deviation and maximum dog-leg severity when using a
conventional directional BHA.
These results agree with those of Weijermans et al (2001). They found that the standard deviation of dog-leg severity in
wells drilled with RSS was approximately half that of a conventional directional BHA for a given average build rate.

Figure 4. Graph comparing the dog-leg severity of a conventional directional BHA vs. rotary steerable BHA in similar Gulf of Mexico
wells.

Average

Table 7. Comparison of Dog-Leg Severity Statistics of

Conventional vs. Rotary Steerable BHA (/100 ft)
Build
Tangent
Conventional
Conventional
Rotary Steerable
Rotary Steerable
BHA
BHA
2.10
2.52
0.88
0.20

Standard Dev.

0.85

0.72

0.61

0.16

Maximum

3.96

3.83

4.01

0.77

IADC/SPE 151448

Effect of Survey Accuracy and Survey Station Frequency on DLS

When using actual surveys to model casing wear, the effect of the frequency of the survey readings on the accuracy of the
mode was considered. Although it would seem that more measurements in a well interval would improve the accuracy of
the modeling of the dog-leg severity, these casing wear and torque drag models rely on the rate of change in direction, or
dog-leg severity of the wellbore. A change in direction over a very small distance can result in very large values for dog-leg
severity. Therefore, if survey points are too close together, the noise coming from sources, such as the resolution and
accuracy of the survey device, makes it difficult to discern actual dog-legs within the data noise. Gabris and Hansen (1988)
showed that there can be a variation in the accuracy of survey measurements taken in the same wellbore with difference
devices.
As an example, if surveys are taken at 5ft intervals using a typical MWD or Gyro device with an accuracy of +/- 0.1 for both
inclination and azimuth, the resulting error band would be 1.7 /100ft, completely masking most real dog-legs within the
well. But, if that same tool were making measurements every 31ft, then the error band is only 0.3 rees/100ft. The table
below shows the error band created when taking surveys at various intervals. As a general rule, surveys taken at intervals
less than 15ft result in too much noise to accurately represent the actual wellpath. Table 8 shows the correlation between the
survey device and the possible error band in the dog-leg severity.
Table 8. Error in Dog-Leg with Varying Survey Station Frequency. (/100ft)
Survey Station Frequency
Survey Accuracy

5 ft

15 ft

31 ft

+/- 0.1

1.7

0.6

0.3

95 ft
0.1

+/- 0.25

4.2

1.4

0.7

0.2

+/- 0.5

8.5

2.8

1.4

0.5

Conversely, if the surveys are taken too far apart then the survey will ignore or underestimate dog-legs that are normally
present in the well. As an example, Figures 5 and 6, along with Table 9 compares dog-leg severities taken from the same
survey sampled at varying intervals. In Figure 5 shown on the left, the survey was taken using a gyro through the drill pipe,
so the sag in the pipe is visible as a large variability in dog-leg severity at 5ft intervals. In Figure 6, the survey on the right is
a gyro survey run in the casing. For the 5ft survey, too much noise is generated by the survey tool and the periodic sag in the
pipe. When viewed at 15ft intervals, some of the same noise is still present, but short period dog-legs are better defined. In
both cases, when the surveys are viewed at 95ft intervals, short interval dog legs are not clearly defined.
In most cases, surveys taken every 31ft provide a good balance in producing a reasonable picture of actual dog-legs within
the wellbore without too much noise from survey tool accuracy. With surveys taken every 95ft, the average dog-leg appears
to be similar to the build section of surveys taken at a frequency of 15ft to 31ft. In the tangent section, the higher resolution
surveys appear to do a better job of finding the smaller period dog-legs, so the standard deviation increases.

Figures 5 and 6. These graphs show the dog-leg severity of a high resolution survey of 2 wells sampled at varying survey intervals.

IADC/SPE 151448

Table 9. Average and Standard Deviation of the Example Wells Shown Above (/100 ft)
Example Well 1, RSS
Example Well 2, RSS
Build Section
Tangent Section
Build Section
Survey
Frequency
5ft
15ft
31ft
95ft

Average
5.15
3.40
3.26
3.29

Standard
Deviation
2.72
1.91
1.02
0.96

Average
3.96
1.99
0.80
0.50

Standard
Deviation
2.40
0.95
0.41
0.30

Average
N/A
1.47
1.39
1.35

Standard
Deviation
N/A
0.8
0.67
0.62

Back-model of Casing Wear Logs

Casing wear logs were available for several of the wells within the database. Back-models were run on several of these wells
using actual surveys taken every 95-100ft, unless otherwise specified. All the modeling parameters were set to match the
actual parameters for each well so that the drill pipe, casing, RPM, and ROP parameters were as close as possible to the
recorded values. Then, the wear factor was adjusted to achieve the best possible fit to the average wear as well as maximum
measured wear. Examples are shown below in Figures 7 and 8. A summary of the results of these back-models is shown in
Table 10. Some of these back-models were taken from USIT (ultrasonic imager tool) logs which give a precise measurement
of wall thickness, while others were taken from multi-finger caliper logs, which give a reading of the inside geometry of the
casing. In the second set of wear data, the percent (%) wear is inferred from nominal wall thickness values for the type of
casing used.

Figures 7 and 8. Casing wear back-model results for a S-shape land well in the western US, and a horizontal well in North Dakota.
Notice that in both of these wells, the maximum wear is far greater than predicted when the average wear is matched up.

Well Type
Horizontal
Horizontal
Horizontal
Horizontal
Deepwater
Deepwater
Deepwater
S-Shape Directional
S-Shape Directional
S-Shape Directional
S-Shape Directional
S-Shape Directional

Table 10. Comparison of Back-Modeled Wear Factors

Wear Factor
Mud
BHA Type
Location
to Match Average
Type
-1
Wear, E-10psi
Conventional
North Dakota
OBM
60
Conventional
North Dakota
OBM
30
RSS
Texas
OBM
20
Conventional
Oklahoma
OBM
10
RSS
Gulf of Mexico
SBM
2-5
RSS
Gulf of Mexico
SBM
7
RSS
Gulf of Mexico
SBM
3
RSS
Colombia
OBM
1.5
RSS
Colombia
OBM
1.5
RSS
Colombia
OBM
1.5
RSS
Texas
WBM
5
Conventional
Rockies
WBM
10

Wear Factor
to Match Maximum
-1
Wear, E-10psi
240
60
20
20
10
7
10
2
2.5
3
10
18

The horizontal wells in this study required relatively high wear factors from 20 to 100 E-10psi-1. This wear factor is much
greater than other wells. The high wear rate could be due to several factors that are particular to horizontal drilling.
First, in horizontal wells, drill pipe wear is often a serious issue in the horizontal sections, so hardbanding is re-applied
frequently, and case is not taken to grind the hardbanding flush with the pipe. Raised hardbanding decreases the contact area
and significantly increases the contact pressure between the drill pipe and casing. This hardbanding can also be fairly
aggressive as the drill pipe suppliers attempt to protect their investment at the cost of wear to the casing. Another issue is that

IADC/SPE 151448

in horizontal wells, devices are often used that use rotation or vibration to break static friction while slide drilling, but also
cause relative motion that could increase casing wear when slide drilling. Horizontal wells also have a relatively steady
contact force throughout the horizontal section because True Vertical DepthTVD is being held constant. And finally,
buckling plays an important role in increasing wear lower in the well due to increased side force from compression in the
drill pipe.
The deepwater wells examined in this project show a wear factor of 2 to 10 E-10psi-1. This relatively low wear factor is likely
be due to a combination of factors, including: high strength casing, premium casing-friendly hardbanding, a lower degree of
dog-leg severity because of extensive use of rotary steerable BHA, and also the use of premium synthetic oil-based mud. All
of these factors generally result in lower friction and wear on the casing. In the deepwater wells analyzed, there were often
one or two small areas that were 10 to 50ft in length that required a wear factor of 7 to 10x10-10 psi to match the modeled
wear to actual wear values. These were usually located at the mudline, or near and within the formation salt strata. All of the
deepwater wells universally used rotary steerable systems, so while dog-leg severity is kept lower than some of the other well
types, because of the use of rotary steerable systems, they also drill at high RPM which can increase the total amount of
revolutions.
The directional well category used a variety of casing, drilling systems, and ROP, so the wear factor ranged from 1.5 to a high
of 18 E-10psi-1. In general, the lower wear factors were seen in wells with casing-friendly hardbanding, high strength casing,
oil based mud, and rotary steerable systems.
Due to the relative scarcity of casing wear logs, the back-modeled wear factors should be used as a guideline. The sample
size of casing wear back-models was not sufficient for the purposes for determining a statistical distribution of casing wear
factors. Future work will focus on increasing the size of the casing wear back-model database.
It is important to note that in most of these cases, the wear factor could be adjusted so that the predicted and actual wear were
approximately coincident, however there often existed short-distance spikes in wear greater than predicted. As shown
below in Figures 9 and 10, this is a result of wellbore curvature that is shorter than the survey interval. This results in dog-leg
severity that is underestimated by an amount related to the ratio of the length of the dog-leg severity divided by the survey
station interval. Examples of acute casing wear over short intervals are shown in the casing wear logs in Figures 9 and 10.
Note that the casing wear log shown in Figure 8 is from a deepwater Gulf of Mexico well drilled using synthetic oil-based
mud and a rotary steerable BHA; hence, even when drilling with the latest in technology, short distance casing wear is
possible. This could be caused by by conditions such as a result of drilling between two different formations, such as when
drilling near salt, or at sidetrack points where a whipstock can cause an abrupt change in direction, or could be caused by
buckled casing that would not otherwise show up in a survey taken prior to running the casing.

Length of wear = 20ft

Length of wear = 30ft

Figures 9 and 10. Examples of casing wear logs showing acute, short distance wear.

IADC/SPE 151448

New Statistically-Based Tortuosity Model

Using the statistical data gathered from the well surveys, a new tortuosity model was created so that the distribution of dogleg severity in the tortured survey matches expected values based on the probability level, section, and well type being
analyzed.
In this new model, tortuosity is applied so that dog-leg severity is varied in a distribution pattern based on the statistical data
gathered from the well database. Tables 3 to 6 are used as a basis for the average () and standard deviation () in the
equations below, with the exception that in a build or drop section, the planned build or drop rate is used as the average value.
For build or drop values far outside the statistical values, a standard deviation that is 0.3 to 0.7 times the average expected
build rate is permissible, as shown in Figure 3.
The tortuosity model must have zero as a lower bound, because negative values for dog-leg severity are not allowed, so a
normal distribution could not be used. As shown in the example in Figure 11, a gamma distribution provided good
approximation to distributions of actual dog-leg severity data.

Figure 11. Comparison of the distribution of dog-leg severity of an actual vertical deepwater well and the results from a tortuosity
simulation using the gamma distribution model described in this paper.

A mathematical description of a gamma probability distribution is given in the appendix, but in general, gamma distributions
are represented by two parameters, and , that define the shape of the distribution curve. In general, larger values of
produce a distribution that is less skewed, whereas larger values of produce a flatter curve. The values that were found to
provide a good fit for the gamma distribution function over the range of expected dog-leg severities are listed below:

Examples of the gamma distribution results for dog-leg severity are shown for P50 case wells in Figures 12-15. In the
tortuosity model, a random value is generated, ranging from 0 to 1, and the computer calculates the inverse gamma function
to return the dog-leg severity based on the average () and standard deviation () values provided. This dog-leg severity is
then translated into a tortured survey that best fits the planned well trajectory.

10

IADC/SPE 151448

Figures 12-15. Graph showing distribution of dog-leg severity vs. the randomly generated value for the P50 case for both the
deepwater and directional well category.

One important addition to the model is to include corrections for analyzing the vertical or low inclination tangent sections of
wells drilled with a conventional directional BHA. In these cases, it is necessary to account for some of the dog-legs created
by slide drilling corrections that would otherwise be unaccounted for in the directional survey. To do this, zones are inserted
with dog-legs represented by the Low Angle Build column in the horizontal well statistical Table 3, or the Build column
in the deepwater, ERD or directional well found in Tables 4, 5, and 6. Generally these dog-leg severities represent the build
rate created by the directional BHA used when correcting back to vertical. This magnitude of dog-leg severity, in the range
of 3 to 6 deg/100ft, are added to as-drilled surveys taken at intervals of more than 30ft to account for short period dog-legs
that do not show up in an as-drilled survey. As an example, when drilling a horizontal well, often the surface casing is drilled
with a vertical packed hole BHA, so the hole below the surface casing would include some of the corrections mentioned
above, perhaps 2 survey points each 1500ft.
An average well can be analyzed using the values in the P50 case. The P90 and P95 represent more severe cases with a
decreasing probability of occurrence, but should be considered in planning and risk assessment.
Two examples are shown below in Figures 16 and 17 compare the Statistically-Based Tortuosity with the actual as-drilled
surveys. Figure 16 shows an S-shaped well drilled in Texas. Figure 17 is an S-shaped deepwater well drilled in the Gulf of
Mexico. The graphs compare the planned, tortured P50 case, and actual dog-leg severity. As you can see, the actual dog-leg
severity matches closely to the dog-leg severity predicted by the Statistically-Base Tortuosity Model.

IADC/SPE 151448

11

Figures 16 and 17. Comparison between planned, actual, and tortured surveys using the Statically-Bsed Tortuosity Model.

When running casing wear analyses, wear factors listed below in Table 10 shows a range of possible wear factors based on
back-model average casing wear patterns. To predict the maximum wear within the casing, these values should be multiplied
by a factor of 2. In general, wells drilled using flush, casing-friendly hardbanding and synthetic oil-based mud will result in
values on the lower side of the range shown in Table 10.
Table 10. Suggested Wear Factors Based on Casing Wear Backmodels
Wear Factor Range to Model Average Wear
Wear Factor Range to Model Maximum Wear
Well Type
-1
-1
E-10psi
E-10psi
Horizontal
10 - 60
Use a value 2 to 4(where is justification for 4)
Deepwater (using RSS)
1.5 - 7
times the average wear factor to determine
ERD
1.5 - 10
maximum wear.
Other Directional Wells
1.5 - 10

As an example of the use of this Tortuosity model, a typical horizontal well drilled in the Bakken Formation of North Dakota
was modeled. The example well used 7 inch, 26ppf casing to 7000ft, with 7 inch, 32ppf casing to 11000ft. A conventional
directional assembly was used with 4 inch drill pipe. The well is drilled with a 10000ft TVD and 10000ft VS. The P50
average well case was examined, using a moderate wear factor of 30 x10-10/psi to model average wear, and 60 x10-10/psi to
model maximum wear. The dog-leg severity of the proposed and tortured survey is shown in Figure 18 and the resulting
wear is shown in Figure 19.
In the P50 case for the Bakken well, using a wear factor of 60 x10-10/psi, the maximum expected wear is 35-40% of the
casing thickness. In the P90 case, with the same wear factor, the maximum predicted wear is 60-70% of the casing thickness.
These results were compared with a casing wear log from another North Dakota well not included in the database. The
results from the actual well are shown in Figures 18 and 19. The results compare well with the results predicted by the
Statistically-Based Tortuosity Model, shown as a dotted line in the figures below. In general, the model is accurate in the
magnitude and general location of dog-legs within the well, but cannot predict the precise location of severe dog-legs.

Figures 18 and 19. Example of a model of a typical Bakken North Dakota well, showing the tortured survey dog-leg severity, and
resulting casing wear.

12

IADC/SPE 151448

Casing Wear Mitigation Using Non-Rotating Protectors

Moore et al (1996) demonstrated that Non-Rotating Protectors (NRP) help to reduce casing wear. Several wells in the
database used NRP protect the casing in the areas of highest calculated contact force. These wells were analyzed to
determine the effect of NRP on casing wear.
In one example, the well was re-drilled through a window cut lower in the well. The original wellbore was drilled without
protectors, and a sidetrack was drilled with 6 million revolution using NRP to cover most of the casing from 8000-12000ft
MD. Casing wear logs were run before and after drilling the sidetrack. These logs are shown below in Figure 20 below,
indicating that there was no additional wear.

Figure 20. Comparison of casing wear after initial wellbore without NRP use, then after a sidetrack requiring over 6 million revs.
With the use of NRP, there is no discernible increase in wear even after 6 million revs.

In other wells in the database, NRP were installed in high side force areas of s-shape wells shown in Figures 21 and 22. In
both of these wells, the entire string was not covered. The wells were back-modeled to calculate the average wear factor in
the area without protectors. Then the measured wear was compared with the actual wear with the use of NRP. As seen in
Figures 16 and 17 below, the protectors were effective at preventing wear to the casing. The wear factor in the areas with
NRP installed was reduced from 1.5 to 3 E-10psi-1down to 0.5 to 1.0 E-10psi-1.

Figures 21 and 22. Casing wear logs from two wells drilled with Non-Rotating Protectors to help protect a portion of the casing.

IADC/SPE 151448

13

Conclusions
The statistical examination of a database of over 100 wells provides insights into assessing casing wear risk. The conclusions
from this examination are the following:

The well category type, including horizontal wells, deepwater wells, ERD wells, and directional wells, affects the
dog-leg severity and thus affects casing wear.

The variance in dog-leg severity increases with build rate.

Rotary steerable systems generally produce less variance in dog-leg severity than conventionally drilled wells.

Survey frequency affects the error of the measured dog-leg severity. Generally, measurements at less than 15 ft
intervals produce excessive variance in dog-leg severity measurements and intervals of about 95 ft or more
significantly underestimate dog leg severity.

A statistically-based tortuosity model is presented that appears to more accurately predict actual dog-leg severity for
the purposes of modeling casing wear and assessing casing wear risk.

Most wells are not surveyed at a frequency required to accurately depict acute, short distance dog-legs within the
well, and surveys taken while drilling would not measure deviations caused by buckled casing or inadequate support
from cement. As a result, wells often require a wear factor twice the average value to predict areas of maximum
casing wear.

Often, back-modeled wear factors are greater than laboratory test results would suggest, particularly in horizontal
wells.

References
Bol, G.M., October 1986, Effect of Mud Composition on Wear and Friction of Casing and Tool Joints, SPE Drilling Engineering
Bradley, W., and Fontenot, J., February 1975, The Prediction and Control of Casing Wear, SPE 6399
Chan, A. et al, August 2008, Evolution of Innovative Test Methodology for Evaluation of Hardfacing Materials in both Cased and Open Hole Environments,
IADC/SPE Asia Pacific Drilling Technology Conference in Jakarta, SPE/IADC 115206
White, J., and Dawson, R., March 1987, Casing Wear: Laboratory Measurements and Field Predictions, SPE Drilling Engineering
Gaynor et al, March 2001, Tortuosity versus Micro-Tortuosity - Why Little Things Mean a Lot, SPE/IADC 67818
Gabris, P.M., and Hansen, R.R., Martch 1988, A Field Comparison of Directional Accuracy of MWD in Comparison with Six Other Survey Tools,
SPE/IADC 17214
Gaynor et al, October 2002, Quantifying Tortuosities by Friction Factors in Torque and Drag Model, SPE Annual Technical Conference in San Antonio,
SPE 77617
Lapin, L. ed. 1990, Probability and Statistics for Modern Engineering, Second Edition, PWS-Kent Publishing Company
Lubinski, .A, Maximum Permissible Dog-Legs in Rotary Boreholes, Journal of Petroleum Technology, Feb 1961
Lubinski, A., and Williamson, J. Usefulness of Steel or Rubber Drillpipe Protectors, Journal of Petroleum Technology, April 1984
Moore et al, May 1996, Reduction of Drill String Torque and Casing Wear in Extended Reach Wells Using Non-Rotating Drill Pipe Protectors, SPE
Western Regional Meeting in Anchorage, SPE 35666
Hall, R.W., and Malloy, K.P., April 2005, Contact Pressure Threshold: An Imporant New Aspect of Casing Wear, SPE Productions and Operations
Symposium in Oklahoma City, SPE 94300
Hall. R.W. et al, February 1994, Recent Advances in Casing Wear Technology, IADC/SPE Drilling Conference in Dallas, SPE 27532
Hall R.W. et al, February 1993, Laboratory Casing Wear Tests, Paper No. 93-PET-3 Energy Sources Technology Conference
Samuel, G.R., et al, February 2005, Tortuosity Factors for Highly Tortuous Wells: A Practical Approach, SPE/IADC Drilling Conference in Amsterdam,
SPE/IADC 92565
P.Weijermans, et al, March 2001, Drilling with Rotary Steerable System Reduces Wellbore Tortuosity SPE/IADC 67715
Williamson, J.S. September, 1985 Casing Wear: The Effect of Contact Pressure, Journal of Pet. Tech., SPE 10236

14

IADC/SPE 151448

Appendix: Casing Wear Calculation

Earlier studies of casing wear have concluded that most wear is caused by drill string rotation rather than reciprocation.
Rotating tool joint against casing interior wears crescent or moon-shaped grooves in casing.
To predict the rate of casing wear by rotating drill string tool joint, it is necessary to express the rate of wear in terms of fieldmeasure parameters, which include:
Mud system (abrasiveness)
Tool joint material
Casing wear resistance
Dog-leg severity
Rotary speed and ROP
Tension along the drill string (torque and drag)
The volume of casing wall removed per foot in time t hours is mathematically expressed in the equation:

WV WF SFtj SD

(1)

The equaion to calculate the sliding distance traveled by the tool joint is:

SD Dtj 60 N t

Ltj
12 Ldp

(2)

Calculating the side force per tool joint is given by the equation:

SFtj

SFdp Ldp 12
Ltj

(3)

Combining Eq.(1), (2) and (3) will result in the following casing wear volume equation:

WV WF SFdp Dtj 60 N t

(4)

Wear Geometry
A typical wear groove is shown in the following figure:

The relationship between wear depth and casing wear volume is:

WV 12( r 2 2 P( P R)( P r )( P S ) R 2 )
S = R - (r - h), inch
P = (R + r + S) / 2, inch

cos ( R 2 S 2 r 2 ) / 2 RS
R sin
arctg (
)
R cos S

(5)

IADC/SPE 151448

15

Wear Factor
Many researchers, including Bradley(1975),Bol(1986),Williamson (1985) and Hall(1993, 1994, 2005),have concluded that
the volume of steel removed from casing is proportional to frictional work done by the tool joint rotating against the casing.

The definition of wear factor is the ratio of friction factor to specific energy, which is the amount of energy required to
remove a unit of steel. The unit for wear factor is E-10 psi-1. Therefore, when a wear factor is reported as 8, the actual value
used in casing wear calculation is 8E-10 psi-1.
Many experiments were conducted to find the casing wear factors under different mud system, tool joint material, casing
interior, drill string protectors. Among them, Maurer Engineering Inc. conducted a joint-industry project DEA-42. It was
reported that more than 300 laboratory tests were performed under DEA-42 to determine the wear factors for various drilling
conditions. The casing wear testing machine consisted of a device for rotating a tool joint sample, approximately 5 inches
long, against a casing sample at a set RPM and side load. The wear was recorded at pre-determined intervals and used to
derive a wear factor.
In the SPE Paper Recent Advances in Casing Wear Technology, Dr. Russell Hall, et al, illustrates some of the variation of
wear factor vs. mud additive and/or lubricants, pipe protectors, etc.
Gamma Distribution
The gamma distribution is an effective distribution for cases where the values are non-negative and skewed. The gamma
probability density function (Lapin, 1990) is given by the equations:

Spreadsheet and statistical software packages have the ability to output gamma distributions given the variables and .
Abbreviations
Tool joint OD, inch
Dtj
h
Wear depth, inch
Total length of tool joint, inch
Ltj
Length of each joint of pipe, feet
Ldp
N
Rotary speed, RPM
r
Tool joint outer radius, inch
Side force on the tool joint, lbf/in
SFtj
Side force on each joint of pipe, lbf/ft
SFdp
SD
Sliding distance, inch
t
Rotating time, hour
WF
Casing wear factor, defined as the ratio of friction factor to specific energy, E-10psi-1
WV
Casing wear volume per foot, in3/ft
R
Casing inner radius, inch