IADC/SPE 87100

New Method of Representing Rock Properties Over Entire Bit Run Improves Computer
Generated Bit Recommendations
D.A. Curry, Baker Hughes OASIS; P.B. Perry, Baker Hughes OASIS; and J.M.Evans, Singleway Computing Ltd.

Copyright 2004, IADC/SPE Drilling Conference

of drilling problems, and of the selection of bits to overcome
This paper was prepared for presentation at the IADC/SPE Drilling Conference held in Dallas, these problems.
Texas, U.S.A., 24 March 2004.
During development, we identified a weakness in the
This paper was selected for presentation by an IADC/SPE Program Committee following
review of information contained in a proposal submitted by the author(s). Contents of the
conventional approach to computer supported bit selection
paper, as presented, have not been reviewed by the International Association of Drilling (i.e. making recommendations for each foot or sample in the
Contractors or 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 International interval to be drilled). This approach is adequate if the bit
Association of Drilling Contractors or Society of Petroleum Engineers, their officers, or
members. Papers presented at IADC/SPE meetings are subject to publication review by
selection is for a section of the well where the formations /
Editorial Committees of the International Association of Drilling Contractors and Society of rock properties are homogeneous. However, bit selection is
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the International Association of Drilling more difficult if the interval to be drilled contains formations
Contractors and Society of Petroleum Engineers is prohibited. Permission to reproduce in print that possess a wide variation in their drilling properties, as is
is restricted to a proposal of not more than 300 words; illustrations may not be copied. The
proposal must contain conspicuous acknowledgment of where and by whom the paper was often the case in the real world.
presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A.,
fax 01-972-952-9435. The authors set out to devise methods that would allow
algorithms to make bit selection decisions based on the
formation properties over the entire length of a bit run.
Abstract
This paper describes methods of calculating measures that
In this paper, we demonstrate how the calculation of properties
describe both the variations in the drilling properties over the
for an interval of the wellbore as a whole can support reasoned
length of the bit run, and the challenges posed for bit selection
selection of drill bit features for that application by a rule-
by the variations in those drilling properties. These measures
based expert system. We show how the use of measures that
capture the characteristics of the interval of rocks that affect
represent the rocks drilling properties over the entire bit run
the choice of mechanical design features for drill bit selection
leads to significantly better bit design recommendations than
in that interval.
can be made using just instantaneous foot-by-foot
This approach provides an analysis of drill bit features
representations of drilling properties.
suitable for drilling heterogeneous applications, which cannot
be achieved using instantaneous indices alone. Our approach
Introduction
allows us to implement algorithms to automate this analysis
Several computer-based methods1,2 exist to determine from
and to replicate the complex decision making undertaken by
wireline logs indices that represent a rocks drilling properties,
drill bit and drilling engineers.
and to use those indices to make automatic drill bit
recommendations. Commonly these methods represent values
Rule-Based Bit Selection
of rock property indices at individual depth stations through a
In order to make computer-generated bit selections, we first
well bore.
identified, for each class of bit, all the design features that
Like many other oilfield tasks, drill bit selection is a
affect significantly the life and performance of the bit.
complex problem, which requires a considerable degree of
Some of these features are explicitly mechanical e.g. PDC
expert knowledge. Many research teams have attempted to
cutter diameter and blade count; others borrow existing
develop computer systems to tackle this3 and similar
formalisms for characterising the bit e.g. IADC code for roller
problems, including drilling operations analysis and formation
cone bit cutting structure. Each feature has a range of possible
damage analysis / stimulation4,5, using techniques from the
numeric or categoric values, as appropriate.
artificial intelligence community, including rule-based and
Rules, triggered by characteristics of the drilling
case-based reasoning.
environment ('attributes'), are used to constrain the
The authors have been developing an expert system for
recommended range for these features of each class of bit.
drill bit selection6,7 for over ten years. This development has
Together these constrained ranges provide a blueprint for a bit
been an extended interactive process involving in-depth
design able to drill the interval successfully.
discussions with a range of bit design and application experts
In this paper we use the cutting structure feature of TCI
from the developer's various research and application
(Tungsten Carbide Insert) bits as an example. The categoric
departments over a number of years. It has resulted in a very
values for this feature encode the first two digits of an IADC
analytical approach to bit selection, and has produced a
code in order to characterise the bits cutting structure, which
significant improvement in the company's understanding both
2 IADC/SPE 87100

gives a range of values 4-1 to 8-4. Recognising that the 4-1-X constrained ranges for bit features for each zone; the ranges
code covers a wide range of cutting structures we have chosen were then intersected to produce the recommendation for the
to add an additional value of 4-0 that is not present in the bit run overall. This approach had some advantages over that
current IADC bit coding. The lightest set TCI bits are then which recommends a value for a bit feature for each foot of a
assigned this code to represent their cutting structure. We file of foot based data; in particular, we could reason about the
recognise that the IADC code for a bits cutting structure is length of the zone of homogenous conditions. (Our rule for
not assigned according to any physically measurable selecting cutting structure code on the basis of rock hardness
parameters such as insert count or projection. It is however a included consideration of the thickness of the rock section; a
widely accepted, and indeed the only readily accessible 20ft stringer would be expected to have a more damaging
method for characterising and comparing roller cone bits effect than a 2ft stringer). Nevertheless this approach was
cutting structures. fundamentally similar to the foot by foot approach, and was
Our initial rules for cutting structure code considered the subject to much the same problems.
hardness of the rock coupled with the planned rotary speed
and the thickness of the interval of rock to be drilled, the The Problem
expected level of chert or pyrite inclusions, and the balling The main problem with both the zone based approach and the
tendency of the rock coupled with the drilling fluid to be used. foot by foot approach is combining the recommendations from
Subsequently a rule has been added to eliminate the lightest each foot or from each zone:
set bits when high levels of abrasion are expected. if loose constraints are used which only eliminate bit
feature values which could cause bit failure, these
Analysing a Bit Run constraints can be intersected easily, but in many cases the
Our initial system required the user to break up the interval to resultant constrained range can be too wide to be useful
be drilled into a number of segments or zones, each of which if on the other hand, an attempt is made to recommend
was basically homogenous. The user had to identify these features which will result in good performance for the
zones, and characterise the drilling environment within foot/zone, it is likely that there will be no intersection
each zone. between the individual recommendations.
As time went on, and foot based data from offset wells
became more readily available, we developed tools to allow If the individual depth station or zone recommendations
the user to begin his analysis directly from foot-based data are not treated as constraints to be intersected, then
detailing the physical properties of the interval to be drilled. considerable interpretative skills are necessary to examine this
These may be in the form of actual logs acquired from output and select an effective compromise recommendation
wireline logging on offset wells, or a composite log for the bit run. This selection is not straightforward viewing
constructed to describe the environment expected in the the foot by foot recommendations for cutting structure code
planned well. Data quality in these logs is paramount; and choosing a cutting structure code to drill the whole
typically, they will have analysed to identify errors, and often interval can easily lead to wrong selection. What value(s)
will have a small moving average window applied to reduce should be selected from the set of foot by foot or zone by zone
erratic outliers. recommendations to arrive at an overall recommendation for
Mason8 pioneered the use of wireline sonic data as an aid the run? Should it be the lightest cutting structure
to bit selection. Sonic travel time and lithology were used to recommendation for any point in the run, the mean, the mode
estimate the unconfined compressive strength or UCS, and (most frequently occurring value), or the heaviest cutting
UCS estimates were then used to select an appropriate roller structure? All have their associated problems.
cone bit cutting structure. This approach uses data for an The example data shown in Figure 1 gives offset
individual depth station to select the most appropriate cutting interpreted lithology, sonic and UCS curves for a planned
structure for that particular depth. While undoubtedly a great 400m bit run, which contains a wide variation of rock
step forward in quantifying the bit selection process, this characteristics. The bit run consists of 10-20ksi UCS
results in a range of cutting structure recommendations over sandstones and conglomerates interspersed with thin stringers
the bit run, which the user has to reconcile manually. of soft shale and coal.
Unconfined Compressive Strength (UCS) figures and The right hand track in Figure 1 shows the results of
interpreted lithology are central to our approach, augmented running a simple rule based on rock type and UCS to produce
with whatever other variables are available. If UCS is not a cutting structure code recommendation for each depth
available directly it can be calculated from wireline sonic station. This rule recommends a 4-1 cutting structure for the
transit time and lithology according to previously published soft shale and coal, and 4-4 to 5-2 for the harder sands, up to
algorithms9. Proprietary algorithms are used to calculate 5-4 for the hardest sandstone stringers. These
indices for abrasivity10, balling tendency11, impact damage recommendations would be fine if the planned bit run were
potential and bit whirl potential. homogenous, but bit selection in the example is difficult since
Our tools assist the user in identifying homogenous zones the interval to be drilled is a mixture of different conditions.
within the interval to be drilled, and characterise automatically Foot by foot recommendations do not tell us which bit to use
the drilling environment for each zone. for such an interval.
In this early version of the system, recommendations were Which is the best bit to use in our example? In this case,
then made for each zone, by eliminating bit design features do we select a 4-1 cutting structure, as recommended for the
that could not drill that zone. The result was a set of
IADC/SPE 87100 3

soft shale, and risk tooth breakage in the harder sandstones? Figure 2 shows the distribution of UCS values in the
Conversely, a 5-4 cutting structure, recommended for the example bit run. Simply calculating the minimum, mean and
hardest sandstone, would give sub-optimal penetration rate peak values does not capture the full impact of the variation
and may risk balling in the soft shale. Experienced judgment in UCS.
is necessary to arrive at an appropriate selection, which largely We found that we needed mechanisms that allowed the
negates the value of a foot by foot rule-based system as a tool system to reason about the relative proportions of different
to improve bit selection. localised drilling environments that produced conflicting bit
The same problem manifested itself for the zone-based selection constraints, in order to balance these concerns.
approach that our system took. It was often hard to identify In addition, a number of the environmental problems we
the best bit to use for the interval as a whole. In particular, in wished the system to reason about were cumulative:
interbedded formations, we could experience a series of hard abrasive wear is cumulative; there is no abrasion issue if
stringers which together warranted much more serious the bit has to drill only one foot of sandstone (although a
consideration than their individual thicknesses indicated. surprising short interval of hard sandstone can cause
abrasion and erosion issues over the following interval).
First Steps balling is also cumulative to a point. To establish the risk
Our initial attempts to resolve this problem involved adding of balling, and thus to recommend appropriate bit designs,
recommendations based on attributes that summarise the the system needs to calculate the full extent of balling
significant characteristics of the drilling environment for the conditions within the bit run. [It also needs to know the
whole of the bit run. These attributes can be calculated from maximum contiguous interval of such conditions
the descriptions of the zones which make up the bit run, or hundreds of feet of sand interbedded with shale may be
directly from the foot based data, if available. These attributes less of a concern than a single interval of 20 feet of highly
fall into two main categories: balling shale which could stop dead an inappropriate
minimum / maximum (peak) values we calculate the bit selection].
minimum and maximum values for various attributes over Our initial thought was to calculate the percentage of the
the interval to be drilled. If defined properly, these give bit run in which each drilling problem occurred. However, we
the same effect as the constraints from the individual concluded that a footage figure is easier to handle than a
zones. For example, if the peak UCS is 20ksi, we know percentage when debating the rules with human experts, and
that a bit with a 4-1 cutting structure code is likely to fail. less prone than a percentage to produce strange results for very
However, knowing what the peak value is also allows us to short bit runs. So, rather than using percentages, we calculate
make an additional inference if the peak UCS is 20ksi, we the total footage of such environmental conditions (e.g. total
do not need to consider cutting structure codes over 5-2; footage of highly balling conditions, total footage of abrasive
heavier cutting structures than that are not necessary, and rock, total footage of rock with UCS >25ksi).
indeed are likely to give penetration rate performance that For each index i for the interval a to b in the foot based
is substantially sub-optimal. data, the footage accumulated is
mean values We also calculate the mean value for some
(i +1) depth(i 1)
b depth

significant rock properties over the interval to be drilled.
This adds power to the reasoning process by allowing us to * f (i )
2
consider what the 'average' recommendation for the bit run i=a
would be. This is particularly powerful when combined
where the function f(i) is defined according to the
with the maximum/peak value for the same attribute; the
environmental conditions to be accumulated. For the
recommendation can be varied depending upon whether accumulated footage of rock with UCS over 25ksi the function
the peak hardness is much higher than the mean, or close
f(i) may be defined as:
to it.
1 if UCS (i ) > 25000
Further Steps f (i ) =
The min, mean and peak value mechanisms allow us to reason 0 otherwise
in ways that were not achievable in the original zone based Obviously considerable computational efficiency gains can
approach. However, while useful, we found that these be made on implementation by accumulating values for all the
mechanisms were not by themselves sufficient to capture all bands for a variable in a single pass through the data.
the reasoning performed by experts in bit selection. If foot based data (lithology and rock property logs) are not
In particular, while we could reason about the effect of available, the same footages can be accumulated from a
individual stringers of hard rock (or of soft balling rock), we description of the zones in the bit run entered by the user. In
were unable to account properly for the effect of a series of this case, the user is responsible for estimating the expected
similar stringers. While one 2 foot stringer of hard limestone values for lithology type and hardness for each zone. These
in an otherwise soft uniform marl is unlikely to have much values can then be used to calculate the indices for abrasivity,
effect on the bit, 100 such stringers in a 1000ft interval balling tendency, whirl potential and expected impact damage
introduce significant concerns about impact damage, and may for each zone; the accumulated footage figures for attributes
well affect the bit selection. such as rock UCS for the entire bit run can then be calculated
4 IADC/SPE 87100

from this data. So too can mean and peak values of generates another constraint, limiting the preferred
all attributes. recommendation to the range 5-2 to 6-3. Other hardness
The footage figures add more reasoning power to the bands add other constraints. When these constraints are
system by helping it to decide how much weight to place on intersected a preferred recommendation of 5-2 is made. In this
the constraints from the specified environment conditions and, example, there are no rocks over 35ksi, and no balling
thus, how to combine the recommendations from different concerns (we are using oil based mud). But note that cutting
parts of the bit run. structures 4-0 to 4-2 would not be preferred because of the
Of particular value in our example is a set of attributes that abrasive nature of the sandstone.
give the accumulated total footages of rock whose UCS lies in Note that our cutting structure recommendation (5-2)
a series of 5ksi bands (0-5ksi, 5-10ksi etc). Similar footages would not be obvious from examination of the foot by foot
are calculated for various levels of abrasivity, and for levels of cutting structure recommendations shown in Figure 1, where
balling tendency. Figure 3 shows the footage values in each of the untutored inclination is to select 4-3 or 4-4. Our
these UCS bands for the example bit run. recommendation corresponded with that of the local bit expert.
However, in this instance the system's knowledge base also
Use in Bit Selection recognized PDC drillability and recommended a heavy-set
Rules are defined which use these derived attributes (that is PDC bit, which was the operator's final selection. So,
the footages of rock in each UCS band) to provide constraints unfortunately, in this case we do not have final field validation
on the recommendation of cutting structure for both roller- of an optimal bit selection. Nevertheless, this example was
cone and PDC drill bits. For example, if the footage of rock chosen because it illustrates well the difficulties of bit
with UCS 15-20ksi is greater than 20ft, then TCI bits with selection in an interval with wide variation in rock properties.
cutting structure codes of 4-1, 4-2 and 4-3 should not be In this case, our rules have yielded an unique cutting
preferred. The cut-off footage of each hardness band can be structure code. This is not always the case. In non-
dependent on other attributes that characterise the bit run in challenging runs, the footage of hardness approach often
question, such as the bit rotary speed. The potential for recommends multiple cutting structure values. This indicates
breakage of teeth on TCI bits is increased in any given that our current knowledge of bit selection, as encoded in the
application as the rotary speed is increased. So the cut-off rules, would suggest that all the recommended cutting
footages for each cutting structure code should be lower at structures would be capable of performing efficiently for this
higher rotary speeds. run. The final selection should then be based on the balance in
In the rules for cutting structure code, we use the derived this specific application between the desire to maximise ROP
footage figure for each UCS band to add a constraint on the potential and the aversion to the risk of tooth breakage. The
preferred cutting structure code range for the interval to be appropriate balance will be influenced by factors such as
drilled (note that we do not recommend a single 'best' cutting operational costs and the operators drilling philosophy, which
structure code at this point). This constraint admits only those are beyond the scope of a bit vendors expert system.
cutting structures that would drill this footage of this hardness Occasionally, the intersection of the constraints is empty,
successfully, and excludes those that would probably fail, or in which case there is no ideal bit for the interval. This can
would yield significantly sub-optimal performance. This happen if the interval consists of a mixture of similar
approach ensures that bits that are likely to fail are rejected. proportions of very hard and very soft rocks. In many such
We regard this as one of the strengths of our algorithm; it is cases, the user can consider a separate set of wider 'acceptable'
inherently risk averse. constraints generated by the system, which exclude only
The constraints on cutting structure code generated from cutting structures which are expected to fail. These can be
the footage of each hardness band are wide if the footage is used to provide a recommendation for an 'acceptable' bit.
low, and become tighter as the footage of the hardness band These constraints are shown for our example in Figure 4,
increases. Note that these constraints focus on the hardness of where there is an acceptable cutting structure recommendation
the rock and the potential for tooth damage or breakage; of 4-4 to 6-2. In cases where there is no preferred
abrasion issues are dealt with separately, primarily by recommendation, but only an acceptable one, this does not
recommending features which provide abrasion resistance. mean that the interval cannot be drilled. Rather it means that
The peak UCS for the interval allows us to put a ceiling the best bit selection for the interval as a whole is a
constraint on preferred cutting structure; cutting structures compromise and would not be regarded as a preferred
heavier set than this will not be necessary. Additional selection for one or more of the component rocks in the
constraints are added from inclusions, mean abrasivity, and interval. As a result sub-optimal performance in some respect
balling and highly balling footage. may occur at some point in the interval, but for many
Intersection of these constraints gives a recommended applications this is to be expected. Because the system allows
cutting structure code range for the interval to be drilled. the user to explore the different constraints that eliminate the
Figure 4 shows how the constraints in the example preferred recommendation, he can make an informed decision
intersect. The cutting structure codes recommended as about which compromise to make.
preferred, or acceptable, or rejected by each rule, and overall, Of course, in some extreme cases even these acceptable
are shown with different patterns, which are indicated in the constraints do not intersect. Here the knowledge captured in
key. The footage of Very Soft Rock (0-5ksi) adds a constraint the system is suggesting that it is impossible to drill the whole
of 4-0 to 5-2 on the preferred cutting structure interval as a single bit run. In cases like this, the user can
recommendation. The footage of Very Hard Rock (20-25ksi) review why there is no bit recommendation, and may be able
IADC/SPE 87100 5

to identify operating parameters such as rotary speed, or even mechanisms readily yield a figure for the wear accumulated
elements of the drilling system such as BHA type, which can prior to each formation, and this allows us to produce rules
be varied to obtain a bit recommendation. Where the problem that assess whether the cutting structure is likely to be in a fit
is caused by the lithology to be drilled, the most obvious state to tackle the following formation.
response is to program two or more runs. A decision to drill Further, it should be possible to relax cutting structure
the interval without modifying the drilling plan should only be constraints if the limiting interval of rock is located at the very
made with extreme care, and with a detailed justification. end of the planned bit run. In that case, there is no concern
The same approach is taken to constrain recommended about sacrificing ROP potential over a long interval to section
values for other features of the bit, using many different TD, and it may not matter that the bits cutting structure is
characteristics of the interval to be drilled, including damaged if it is capable of reaching the planned section TD
trajectory, planned BHA, and pressure profiles. In this way a before failing completely. In many cases, there is an explicit
profile of the preferred bit for the interval can be built up. plan to run a bit into a formation for which it is not designed,
Figure 5 shows the full TCI bit recommendation for our and to pull it when it fails.
example. Note that the cutting structure recommendations
shown in this figure relate to bits with conventional chisel Conclusions
shaped teeth; cutting structures 6-1 and 6-2 would also be We have identified and documented weaknesses with
acceptable if the bits tooth shape were 90-degree chisel. straightforward foot by foot approaches to automated bit
We also calculate cumulative factors dependent on the selection that recommend bit features for each depth
length of the interval to be drilled to produce other attributes, station individually.
such as accumulated wear or expected bit revolutions, which By calculating characteristics such as mean and peak
have a direct relationship to bit life, and use these to constrain hardness, mean and accumulated abrasivity, and the total
further our selection. footage of different hardness bands, we have designed an
algorithm which makes subtle and reasoned decisions about
Current Status the best bit design to use for drilling a given interval. In real-
The expert system tool described in this paper is complete and world situations, where rock formations vary in hardness and
is being used in drilling optimisation projects in many are frequently interbedded, this algorithm achieves a quality of
applications and drilling environments. It has been recommendation that is impossible to achieve using reasoning
implemented in Allegro Common Lisp on a Sun workstation about instantaneous zone or foot-by-foot indices alone. This
running X-windows and Motif, using CLM and GINA algorithms recommendations are typically risk averse,
(publicly available from GMD in Germany) to provide a reflecting the character of the designers.
CLOS-based interface to Motif. The other approaches discussed need interpretation by an
The rule base has been field tested by domain experts expert user before a final recommendation can be made. Our
delivering consultancy in drill bit selection. Results have been algorithm typically yields a narrow selection of cutting
encouraging and significant savings in drilling cost have structure codes all of which can be expected to drill the
been demonstrated. interval successfully; any expert review of these
Table 1 gives a summary of the results achieved by our recommendations is providing fine-tuning of the decision to
algorithm when its cutting structure recommendations are optimise performance.
compared with those of the local Company Expert. The This algorithm has been implemented in a software tool,
column titled "Runs" shows the number of bit runs on which which makes automated bit recommendations of a consistently
the system was run for which the company expert high quality.
recommended a Roller Cone bit. The column titled "Right" We have validated our rule base in a number of
gives the number of bit runs for which the system cutting geographically dispersed and physically diverse drilling
structure recommendation matched the recommendation of the applications. Recommendation quality in these data sets
expert. The columns titled "Within 1 CS Code" and "Within 2 matches closely the recommendations of experienced local
CS Codes" give the number of bit runs for which the system drill bit engineers, although the system does not use any
cutting structure recommendation was within one and two localized geographical expertise. In roughly 80% of bit runs,
cutting structure codes respectively. Note that in at least three the system recommends the same cutting structure as the local
of these cases we believe that the system made a better expert, and in the remaining cases the recommendation is
recommendation than the local expert, and this can in one case heavier set (more conservative), and within 2 cutting structure
be demonstrated by the damage sustained by the bit actually codes of that of the expert.
run. In all but one bit run, where the system cutting structure
recommendation did not match that of the local expert, the Acknowledgements
system recommendation was more conservative than that of The authors would like to acknowledge the assistance of Hugh
the expert. Evans and Harvey Brown of Baker Hughes OASIS in the
However, we are in the process of developing our compilation of the validation data. Thanks must go to Eric
algorithm further. The system at the moment does not fully Bjornsson (Baker Hughes OASIS) for provision of the
take into account the effect of the order in which different example data set and to Canadian Natural Resources Ltd. for
environmental conditions are encountered. For example, the their permission to publish this information. Thanks also to all
recommendation does not vary depending on whether an the experts from the various research and application groups
interval of shale is followed by sandstone, or vice versa. Our within Baker Hughes who have contributed their valuable time
6 IADC/SPE 87100

and energy to this undertaking. Hughes Christensen now 72374, SPE Eastern Regional Meeting, Canton, Ohio,
funds this project, but initial developments were funded jointly October 2001.
by Hughes Christensen and BP Exploration; the authors would 6. M.J. Fear, N.C. Meany and J. M. Evans. An Expert System for
like to thank both companies, and further thank the Drill Bit Selection SPE/IADC paper no 27470, SPE/IADC
Drilling Conf., Society of Petroleum Engineers, Dallas 1994.
management of Hughes Christensen for permission to publish 7. J. M. Evans, M. J. Fear, N. C. Meany. A New Graphical
the paper. Last but not least, the authors would like to thank Representation for Rule Definition and Explanation in an Expert
Nigel Meany of Baker Hughes INTEQ for his stalwart System. Proceedings of Expert Systems 95, 15th Annual Tech.
championing of this project throughout its life. Conf. of the British Computer Society Specialist Group on
Expert Systems, Cambridge 1995.
References 8. K.L. Mason, "Tricone Bit Selection Using Sonic Logs" SPE
1. V. Uboldi, L. Civolani, F. Zauza. "Rock Strength Measurements paper no. 13256, 59th SPE Tech. Conf. , Houston 1984.
on Cuttings as Input Data for Optimizing Drill Bit Selection" 9. J. Sparr, L. Ledgerwood, H. Goodman, R. L. Graff and T. J.
SPE paper no. 56441, SPE ATCE, Houston, Texas, 1999. Moo. "Formation compressive strength estimates for predicting
2. B. Ivie, "Formation Evaluation for Drill Bit Analysis, Selection" drillability and PDC bit selection" SPE/IADC paper no 29397,
Drilling Contractor March-April 2002. SPE/IADC Drilling Conf. Amsterdam 1995.
3. T. Sorenson, O. Nordhus, "LOKE a Drill Bit Selection System", 10. C. H. Cooley, D. A. Curry and L. Ledgerwood, US patent no
in D.S.Moralee (ed). Research and Development in Expert 6,386,297, Method and apparatus for determining potential
Systems IV., Proc. Expert Systems 87, pp. 32-41, Cambridge abrasivity in a wellbore.
University Press, 1987. 11. N.C Meany, L. Ledgerwood, D.A. Curry and C. H. Cooley, US
4. R. Irrgang, C. Damski, S. Kravis, E. Maidla, K. Millheim, "A patent no 6,353,799, Method and apparatus for determining
Case-based System to Cut Drilling Costs" SPE paper no 56504, potential interfacial severity for a formation.
SPE ATCE, Houston, Texas, 1999.
5. H. Xiong, B.M. Robinson, S. Foh, "Using an Expert System to
Diagnose Formation Damage Mechanisms and Design
Stimulation Treatments for Gas Storage Wells" SPE paper no

Well Hole Size Application Runs Right Within 1 Within 2

CS Code CS Codes
A 8.75 Highly interbedded 5-10ksi shales with 15-20ksi sands 5 4 1
B 12.25 Very soft (0-5ksi) sand-shale mix. 1 1
8.5 Highly interbedded, including tuffs 2 1 1
C 17.5 0-5ksi high balling shale 1 1
D 17.5 5-10ksi siltstone with 15-20ksi sandstone stringers 2 2
14.75 5-15ksi carbonates, 20-25ksi limestones, 20-25ksi 3 3
sandstones, siltstone
E 12.25 Chalk with Chert bands 1 1
F 24 0-5ksi balling claystone with chert 1 1
17.5 0-10ksi chalk with chert 3 3
G 26 5-10ksi sandstones 1 1
16 5-10ksi Siltstones and 20-25ksi sandstones - tectonic 11 9 2
H 12.25 5-15ksi sandstones with some 10-15ksi limestone 2 2
I 16 5-15ksi sandstones with some 10-15ksi limestone 2 2
12.25 5-15ksi sandstones with some 10-15ksi limestone 2 2
J 17.5 0-5ksi balling claystone 1 1
K 8.5 Igneous rocks (20-25 ksi) 1 1
L 12.25 Highly interbedded 25ksi sands, 10-15ksi limestones, 3 3
claystones, metamorphosed
Totals 42 34 6 2

Table 1. Recommendations of the Drill Bit Advisor for Cutting Structure Compared with those of the local Company Expert
IADC/SPE 87100 7

Figure 1: Example Foot Based Data showing Cutting Structure Code index
8 IADC/SPE 87100

Metres of Rock in 1ksi UCS Bands

70

60

50
Metres of Rock

Coal
40
Conglomerates
Sand

30 Sandstone
Shale

20

10

0
10-11

12-13

14-15

16-17

18-19

20-21

22-23

24-25

26-27

28-29
0-1

2-3

4-5

6-7

8-9

Figure 2: Distribution of UCS values in the Example Bit Run colour coded by Rock Type.

Metres of Rock in each UCS Band

250

200

150

100

50

0
00-05ksi 05-10ksi 10-15ksi 15-20ksi 20-25ksi 25-30ksi

Figure 3: Hardness band figures obtained for the Example Bit Run
IADC/SPE 87100 9

Figure 4: Constraints on the Recommendation for Cutting Structure Code in the Example

Figure 5: Full TCI Recommendations for the Example