You are on page 1of 8

How to Improve Rate of Penetration

in Field Operations
M.J. Fear, SPE, BP Exploration Co. Ltd.

Summary nozzle arrangement.8,10 In summary, rock properties that influence


A method has been developed to identify which factors are con- ROP include at least mineralogy, strength, density, porosity, and
trolling rate of penetration ~ROP! in a particular group of bit runs. permeability.2,3,5,8,9,11,14 Interdependence between mechanical and
The method uses foot-based mud logging data, geological infor- hydraulic drilling parameter effects may also be significant, mean-
mation, and drill bit characteristics, to produce numerical correla- ing for example that the response of ROP to weight on bit ~WOB!,
tions between ROP and applied drilling parameters or other at- rotary speed and flow rate can depend on absolute values of these
tributes of drilling conditions. These correlations are then used to parameters.7 Bit design effects are also not simple; differences in
generate recommendations for maximizing ROP in drilling bit design effects on ROP with polycrystalline diamond compact
operations. ~PDC! bits appear only to become significant when bit cleaning
problems occur, or when cutters become worn.9 With roller cone
bits, varying jet nozzle arrangement may or may not affect ROP,
depending on which of bit or bottomhole cleaning is deficient.8,12
Introduction
Finally, complexity is increased by errors and inconsistencies in
Time spent drilling ahead is usually a significant portion of total
drilling data, meaning that correlations with ROP may be masked
well cost. In typical wells drilled by BP Exploration ~BPX!, this
without extensive data treatment.13 This latter point may explain
rotating time usually accounts for 10% to 30% of well cost.
why, despite publication of a number of analytically derived ROP
This means that the penetration rate achieved by the drill bit has
models,1922 none has yet become established as an operational
considerable significance for drilling cost reduction.
tool.
Despite this, the bit/rock interaction, and the ROP that results,
The complexity of the bit/rock interaction, and the difficulties
is not well understood in detail. Other drilling phenomena, such as
with implementation of analytical models, have encouraged BPX
torque/drag modeling or directional behavior, are typically man-
to adopt an empirical approach to optimization of ROP in drilling
aged with the aid of validated, quantitative models. Such tools are
operations. This method follows that approach.
conspicuously absent from planning or analysis of bit perfor-
Emphasis has also been placed in this work on understanding
mance.
the effects of controllable variables, i.e., those that can be readily
Within drilling operations, uncertainty over expected ROP
changed to optimize ROP. Other environmental effects are how-
clouds decisions on which bit types and downhole tools to select
ever incorporated into another ROP modeling technique devel-
to maximize ROP. That intuition is prominent in these decisions
oped by BPX and described elsewhere.13
demonstrates the inability of drilling data or predictive models to
provide anything more substantial.
Against this background of commercial significance and tech-
Description of the Method
nical difficulty, BPX is developing engineering methods to raise
ROP. This paper describes one such method, and its application The objective of this method is to quantify the effects of opera-
and benefits to one drilling operation. tionally controllable variables on ROP. To reveal the effects of
these variables, data sets must be constructed so as to minimize
variation in environmental conditions. The first step is therefore to
Background to the Method
select a group of bit runs made with the same bit size, through
Table 1 lists factors which are proposed to affect ROP.118 The
similar formations. Next, intervals of consistent lithology are
number of factors hints at the complexity of the bit/rock interac-
identified, with a preference for formations exhibiting lateral ho-
tion, something which is compounded by interdependence and
mogeneity. Formations such as shale and limestone are, in gen-
nonlinearity in some of these effects.6,7,11,15,18
eral, more suitable than variable lithologies such as sandstone.
Laboratory studies and modeling are however unraveling this
Rock property logs can of course be used to verify comparability.
complexity. For example, how ROP responds to changes in drill-
Depending on the objectives of each specific analysis, further sort-
ing parameters has been shown to depend strongly on rock prop-
ing can be made to separate bit runs made in different mud types,
erties. In permeable rocks for example, overbalance pressure in-
with different classes of bit, or to separate intervals drilled with
fluences ROP,1 giving way to a dependence on bottomhole
sharp bits versus those in worn condition. Each step helps to fur-
pressure as permeability decreases.24 Overbalance pressure ef-
ther expose the effects on ROP of bit design and mechanical or
fects however are subject to dynamic influences, either via filtra-
hydraulic drilling parameters.
tion effects on pore pressure in the bit/rock interaction zone,5 or
Once intervals have been selected and sorted, numerical aver-
via stress effects on pore pressure around the wellbore.6 Bit clean-
ages of the variables of interest are obtained. This is critical be-
ing effects while drilling hydratable formations in water base
cause many sources of error exist in drilling parameter measure-
drilling fluids ~muds! may also override the effects of me-
ments and, pending improvement in data quality, averaging to
chanical drilling parameters, so that rock mineralogy and mud
raise sample size is the most obvious method to minimize error
chemistry are obviously significant factors.7,8 These cleaning ef-
effects.13
fects are however themselves influenced by bit design,9 and jet
Fig. 1 shows a log where data have been extracted and aver-
aged from an interval of shale early in the bit run, prior to a drop
Copyright 1999 Society of Petroleum Engineers in ROP related to bit wear in a sandstone. This process would then
This paper (SPE 55050) was revised for publication from paper SPE 35107, first presented be repeated for other bit runs made through the same stratigraphic
at the 1996 IADC/SPE Drilling Conference held in New Orleans, 1215 March. Original interval, yielding a data set suitable for analysis.
manuscript received for review 16 May 1996. Revised manuscript received 30 November
1998. Paper peer approved 9 December 1998. Within BPX, customized petrophysical software is used to au-

42 1064-6671/99/14~1!/42/8/$3.5010.15 SPE Drill. & Completion 14 ~1!, March 1999


TABLE 1 FACTORS PROPOSED TO AFFECT ROP

Environmental Factors Controllable Factors

Formation type Bit wear state


Formation properties Bit design
Mud type Weight on bit
Mud density Bit rotary speed
Other mud properties Flow rate
Overbalance and/or Bit hydraulic horsepower
bottomhole mud pressure
Bit size Motor/turbine geometry
Bit nozzle arrangement

tomate the extraction and averaging of drilling data, though


manual processing from paper logs is equally successful.
Once the data are prepared, correlation analysis is performed in
conventional spreadsheets. Crossplots are used to seek visible cor-
relations between ROP and the independent variables, and sta- Fig. 1Example of bit-run data showing interval to be extracted
tistics functions are used to establish the degree of correlation and and averaged.
to build models for prediction of ROP.
The next section shows examples of the method as applied to
real drilling situations. processing. Note that mud density effects are quantified by use of
mud bottomhole pressure ~Mud BHP in the table!, since this
Application of the Method incorporates depth effects, and is probably the most influential
Table 1 is used as a checklist in this method; data are assembled pressure parameter in compacted shale drilling.4 Pore pressure in
so as to minimize the effects of the variables in the left column of the shales is, in any case, unknown to any degree of accuracy.
the table, and then further sorted to emphasize the effects of the The three columns labeled Shale Disc, Buffer, and
variables in the right column. Data Points are critical elements of the data sorting process.
Example 1. Table 2 shows a data set for a group of 17.1/2-in. The first shows the type and value of log data used to identify
insert bit runs made through compacted shale, in oil and water which data points lie in the required lithology for each bit run, so
base mud. The bits all had similar cutting structures, since that that only those are extracted and averaged. This minimizes the
type of bit was found to be most cost effective in terms of bearing effect of lithological variation on ROP, while allowing data from
life and gauge durability. In this case, the objective of the ROP various parts of a bit run to be grouped into one average response
analysis was to determine which mechanical and hydraulic param- for that lithology. If no such log data are available, top and bottom
eters were most influential to ROP with this favored bit type. depths for a formation can be used, providing the formation is
The columns in Table 2 show the bit type, the number and sizes relatively homogenous both vertically and laterally. This careful
of nozzles used, and average values of drilling parameters. These minimization of geological effects avoids the main disadvantage
are the controllable variables expected to be most influential to of ROP comparisons made from bit record data, that of an un-
ROP, given that the effects of bit diameter and type, lithology, and known amount of ROP variation due to geology.
mud type have all been accounted for during data selection and The Buffer value is the depth increment either side of a lithol-

TABLE 2 DATA SET FOR EXAMPLE 1

Bit Avg Avg Avg Mud Avg Avg Shale Data


Type Nozzles WOB Flow ROT BHP HSI ROP Disc. Buffer Points

Inch/32 klbs gpm revs/min psi hhp/sq. in. ft/hr Log value Feet Feet
WBM RUNS
X11 20,20,24 38.55 788 110 5425 1.24 3.48 ? ? ?
MAX-T11HD 18,18,16 40.11 1093 132 5806 6.91 7.39 ILD 5-30 8 131
MAX-11HD 16,16,20 36.14 946 156 4716 4.05 6.18 ILD 0-9 8 81
MAX-11HD 16,16,20 41.27 928 161 4975 3.99 6.83 ILD 0-9 8 37
MAX-F11HD 14,16,16 28.94 926 97 2886 5.6 6.73 GR 45-200 8 35
MAX-11HD 14,16,16 30.53 970 107 3048 6.43 6.43 GR 45-200 8 202
MAX-11HD 14,16,16 29.45 1004 114 3192 7.15 7.94 GR 45-200 8 98
MAX-T11HD 20,18,11 59.60 972 147 4123 5.59 6.20 Cased GR NO ?
MAX-F11HD 20,20,16 60.00 986 127 5128 3.85 5.80 Cased GR NO ?
MAX-11HD 3 * 20/14 46.53 1082 160 4861 2.84 4.40 CDR GR NO 16
MAX-11HD 2 * 20/16/14 40.21 1091 180 5145 3.59 5.27 CDR GR NO 35
OBM RUNS
ATM-T11 2 * 28 29.50 939 124 3651 1.41 8.40 GR 65-200 8 11
MAX-T11HD 24,24,16 37.50 894 182 3380 1.51 6.56 GR 65-200 8 81
MAX-T11HD 28,28,16 34.15 956 136 3918 1.13 5.76 GR 65-200 8 36
MAX-T11HD 28,28,16 34.45 943 143 4060 1.09 6.19 GR 65-200 8 34

M.J. Fear: Improve Rate of Penetration SPE Drill. & Completion, Vol. 14, No. 1, March 1999 43
Fig. 4Relationship of ROP to WOB from a drilloff test for an
insert bit in WBM/shale.

Fig. 2Relationship of ROP to HSI for insert bits in shale with ~WBM! data. This perhaps implies that bit and/or bottomhole
WBM and OBM. cleaning improves with higher HSI, raising ROP. This is not un-
expected in a WBM/shale application, though it is useful to deter-
mine that increasing HSI is more productive for ROP than flow
ogy change, from which data are not to be extracted. In other rate, given that a finite amount of hydraulic horsepower ~HHP! is
words, if, at one particular depth, the wireline log curve used for available at the rig. This interpretation is strengthened by the ob-
identifying the required lithology has a value outside that in the servation that the oil base mud ~OBM! data points show higher
Disc column ~implying another lithology!, data from both above ROP at low HSI values, reflecting the contribution that OBM
and below that depth will be excluded. This allows for any depth makes to bit or bottomhole cleaning and hence ROP.
offset between wireline and drilling data. One more surprising observation from Fig. 2 is that the nozzle
The last column in the table shows the number of data points configuration used appears to exert little influence on ROP. The
which met the criteria for lithology discrimination and buffer ef- uniformity of the HSI/ROP trend in WBM, despite very different
fects, and hence indicates the number of points used to derive the nozzle arrangements, implies that in this case, ROP is more de-
averaged drilling parameter values. Attempts are usually made to pendent on the amount of HHP delivered, rather than how it is
compile averages with not less than 20 to 25 data points, since this distributed around the bit. This is a good example of where pre-
appears to be a near optimal value for minimizing errors in ROP sumed behavior ~the benefit of extended nozzles! is different to
data.13 that observed, highlighting the value of empirical observations in
The next step is to interpret the observed ROP variation. This operational decisions.
starts by systematic crossplotting of each of the independent vari- Fig. 3 shows a crossplot of average WOB versus average ROP,
ables against ROP, seeking visible correlations. Apparent effects for these same bit runs. No clear relationship is apparent. While
on ROP can then be investigated further, for example by subdi- this is initially surprising, it highlights another key feature of this
viding the data to see if hydraulic horsepower delivered through method. Within a bit run, variables such as HSI typically vary
one nozzle configuration provides more ROP than through alter- little, varying much more when values from different bit runs are
native configurations. compared. If the effects of such variables on ROP are strong, as
Two crossplots from Table 2 are shown as Figs. 2 and 3. Fig. here, they may mask ROP variation due to other variables such as
2 shows the relationship between hydraulic horsepower per square WOB or rotary speed, when different bit runs are being compared.
inch of hole area ~HSI! and average ROP; a clear trend of That this masking effect is present in this group of bit runs is
rising ROP with HSI is apparent for the water base mud demonstrated by Fig. 4, where the true WOB/ROP relationship is
confirmed from the results of a drilloff test performed during one
of these bit runs. Through the drilloff test, flow rate and hence
HSI are constant, allowing the WOB effect on ROP to appear.
This drilloff test used a method previously found effective in de-
termining WOB effects on ROP.23
No other variables were found to influence ROP in this appli-
cation. Drilling guidelines would therefore recommend use of
maximum WOB and HSI, for ROP.
This example of combined bit run analysis and drilloff testing
highlights that, to obtain a complete description of factors affect-
ing ROP, various methods should be used. The method described
here can be used to quantify the effects of variables that change
significantly between bit runs ~such as HSI, flow rate, bit charac-
teristics, mud type, and density, etc.!, while another method exists
to quantify the effects of variables that change significantly within
one bit run ~such as WOB, rotary speed. etc.!.13 Finally, drilloff
tests can be used to confirm drilling parameter effects, and to
Fig. 3Relationship of ROB to WOB for insert bits in WBM/ identify phenomena such as excessive drag or bit wear.
shale. Example 2. Table 3 shows another data set, this time for

44 M.J. Fear: Improve Rate of Penetration SPE Drill. & Completion, Vol. 14, No. 1, March 1999
TABLE 3 DATA SET FOR EXAMPLE 2

Bit Avg Avg Avg Mud Mud Avg Avg Data


Type WOB Flow ROT Density BHP HSI ROP Points

klbs gpm revs/min ppg psi hhp/sq. in. ft/hr Feet


DS34HFU 16 564 158 10.70 8020 3.52 31.30 536
DS34HFU 18.40 645 143 14.20 8144 1.8 9.70 77
DS34HFU 19.40 644 157 14.50 8736 1.83 11.40 300
DS34HFU 28.20 645 76 14.40 8486 3.22 20.60 260
DS34HFU 25.90 642 147 14.40 8885 3.11 11.71 135
PD12 27.70 660 147 14.50 9649 3.86 28.50 399
PD12 20.70 650 119 14.50 1001 3.69 20.70 554
DS34HFU 6.50 702 117 9.25 3179 6.8 50.30 450
AR526 10.50 579 153 9.40 5219 6.4 53.60 521
AR545U 18.30 903 140 10.40 6174 6.72 47.40 90
DS47BHU 30.30 609 145 14.20 7968 1.81 8.30 107
TD19HBR 30.10 588 133 14.40 9594 3.58 8.83 303
TD19HBR 37.70 566 99 14.50 10374 3.24 6.67 178
DS77HU 31.21 720 171 15.10 9364 6.75 13.40 400
DS77HU 35.45 720 185 15.60 10761 6.28 4.40 289
AR545U 14.90 720 83 11.98 6641 7.884 30.90 100
AR526 19.16 804 148 15.00 10594 8.983 38.44 111

12.1/4-in. PDC bit runs made through compacted shale in WBM. point for users of this method; not every correlation implies a
As shown in Fig. 1, data were sorted to isolate shale sections from direct cause and effect relationship. In this case, because bit clean-
early in each bit run, where the bits could be assumed to be still in ing dominates ROP, runs made with low HSI and/or high BHP
sharp condition. This is important because, in this application, give low ROP, despite use of high WOB. Fig. 7 therefore reflects
cutter wear exerts a strong influence on ROP, yet the degree of the unsuccessful attempt to counter low ROP with high WOB,
wear is unknown at any particular depth once some abrasive for- when bit cleaning is poor. As in example 1, when bit cleaning
mations have been drilled. The effects of wear cannot therefore be effects are near constant ~within one bit run!, drilloff tests do
handled quantitatively and must effectively be edited out. reveal a positive WOB effect on ROP, though the slope of that
For these runs, sections of clean shale were identified from relationship varies from one run to another.
gamma ray logs where available, and then from formation top and Because of this apparent WOB effect, a statistical regression of
bottom depths in later wells when wireline logs were not run. This ROP against drilling parameters would not make sense; higher
is acceptable in homogenous formations. WOB would emerge as detrimental to ROP. Statistical analysis
Figs. 5 and 6 plot the only variables against which ROP could was however desirable to test the validity of the HSI and BHP
be correlated. These show that both HSI and the bottomhole pres- effects. For this reason, and because drilloff test results showed a
sure of the mud column ~BHP! are influencing ROP. The scatter linear WOB/ROP relationship with intercept close to zero, the
in the two crossplots exists because ROP is varying due to more ratio ROP:WOB was taken as the independent variable. A linear
than just the effect of the variable against which it is plotted. regression of this ratio against other drilling parameters confirmed
The effect of HSI and BHP is expected for a shale/WBM ap- the dominance of the HSI and BHP effects, these two effects
plication, since both can affect the efficiency of bit cleaning. As in accounting for more than 80% of the observed variation in ROP.
example 1 however it is again important to determine which of There are a number of other informative observations from this
flow rate and HSI to maximize when rig HHP is limited. data set. In Fig. 6, the data points at highest BHP are interesting.
There is one other correlation in this data set which is, initially, The points marked X were made with bits of low hydraulic
surprising. This is an apparent negative correlation between ROP
and WOB, shown in Fig. 7. This however highlights another key

Fig. 5Relationship of ROP to HSI for PDC bits in WBM/shale. Fig. 6Relationship of ROP to BHP for PDC bits in WBM/shale.

M.J. Fear: Improve Rate of Penetration SPE Drill. & Completion, Vol. 14, No. 1, March 1999 45
In this data set, the same gamma ray value was used as the
sandstone discriminant in all cases. In addition, the close proxim-
ity of a casing shoe as a marker for both mud logging and wireline
data allowed the latter to be confidently depth shifted into align-
ment with the former. Hence, the depth buffer facility was unnec-
essary. Last, given the potential effect of overbalance pressure
~OBP! in this permeable formation, static overbalance pressure
was calculated for all bit runs.
The drilling of hard sandstone poses particular problems for bit
optimization. Conventional methods for raising ROP, such as use
of PDC bits, are limited by rapid cutter wear in hard, quartz-rich
formations. In these nonhydratable formations, bit cleaning may
also exert a negligible effect on ROP, closing off another potential
source of improvement. In practice this means that options to raise
Fig. 7Apparent negative effect of WOB on ROP for PDC bits in
ROP in hard sandstone may be restricted to exploiting the effects
WBM/shale. of drilling parameters, with bit types capable of resisting rapid
abrasive wear.
Fig. 8 shows a crossplot of the clearest correlation between any
of the independent variables and ROP; that with OBP. For the
efficiency, and high HSI. Their ROP is low. Points marked Y
PDC bit runs in particular, the classic pattern of exponentially
are from runs made with lower HSI, but a very open-faced bit
design. ROP is higher. This implies that bit design can influence decreasing ROP with higher OBP is seen. However, as discussed
ROP under high BHP. The run marked Z was made with a previously, PDC bits are not cost effective in this application be-
hydraulically efficient bit and higher HSI. ROP is higher than cause of rapid abrasive wear. The remaining data in the graph do
points X and Y. This shows that, even when BHP is very high, however provide some pointers as to how bit performance can be
HSI can still influence ROP. These findings are significant in that, improved.
in some areas, high BHP while drilling shale in WBM is rendering First, it is apparent that natural diamond bits run on rotary
PDC bits uneconomic due to low ROP. bottomhole assemblies ~those with a rotary speed of 200 revolu-
Example 3. This last example addresses the drilling of hard tions per minute or less! fail to respond to lower OBP with higher
sandstones, and worked from a data set gathered for bit sizes from ROP. This class of bit therefore offers little potential for raising
6 to 8.1/2 in. One particular sandstone formation was chosen, and ROP on rotary bottomhole assemblies. The contrast between the
intervals of clean sandstone identified from gamma ray logs. The response of these bits and the PDC bits to low OBP is interesting,
averaged data are shown in Table 4. implying that larger cutting elements ~PDC cutters! are required to

TABLE 4 DATA SET FOR EXAMPLE 3

Bit Avg Avg Avg Mud Mud Mud Avg Avg Data
Size Type WOB Flow ROT Density BHP OBP HSI ROP Points

In. klbs gpm revs/min ppg psi psi hhp/sq. in. ft/hr Feet
8.5 B36 12.9 399 95 11.000 8312 2499 4.21 3.55 30
8.5 DS47H 15.00 388 73 10.460 7310 1934 6.08 2.94 45
6 M19B 13.20 315 85 9.020 7077 1042 6.96 5.19 19
6 M19B 18.80 314 88 9.020 7135 1050 6.88 3.82 50
6.5 TD13H 8.60 429 132 8.300 6343 464 2.45 13.00 18
6.5 R437GN 13.40 430 127 8.300 6371 466 4.31 18.00 20
6.5 EHP51H 24.50 438 177 8.400 6316 532 7.32 9.00 30
8.5 DE331G 17.30 315 129 11.000 8277 2489 2.05 2.80 40
8.5 D41G 24.90 393 150 10.600 8088 2219 1.46 3.50 35
6 TB16 16.40 288 133 9.350 7294 1293 4.72 5.42 22
6.5 SD248 18.90 400 134 8.300 6249 458 4.65 6.00 25
6.5 TB603J 15.40 435 147 8.300 6303 461 4.79 3.50 20
8.5 SD248 18.80 419 128 8.217 6249 399 2.128 4.19 35
6 SD248 20.70 336 160 8.660 6642 742 2.057 3.07 15
6 263ND 20.10 312 160 8.786 6779 844 1.672 4.50 36
6 263ND 19.00 288 162 8.443 6594 586 1.267 5.71 50
8.5 SD248 13.40 507 354 9.000 6286 913 1.97 7.10 31
6 462TS 16.1 260 720 8.688 6630 760 3.219 7.59 25
8.5 462XG 12.4 517 853 8.550 6381 640 2.04 9.8 50
8.5 462TS 8.73 503 830 8.350 6222 490 1.79 11.7 13
8.5 462TS 12.6 506 835 8.850 6111 799 1.93 12.4 50
8.5 462TS 10.7 538 888 8.400 649 540 2.202 12.3 39
8.5 462TS 10.5 538 998 9.022 6478 955 2.365 12.5 50
8.5 462TS 16.5 519 950 8.508 6107 585 2.007 7.8 25
8.5 462TS 11.2 533 983 8.697 6534 755 2.323 6.96 50
8.5 462TS 12.7 523 955 9.034 6850 1017 2.915 13.6 30

46 M.J. Fear: Improve Rate of Penetration SPE Drill. & Completion, Vol. 14, No. 1, March 1999
Fig. 8Relationship of ROP to overbalance pressure in sand- Fig. 10Jump in ROP after application of the method in well 4,
stone for various bit types. group A. Wells in group B did not use the method.

exploit the OBP effect. This suggests significant differences in Performance Benefits
cutting mechanics between the two classes of bit at low OBP.
Fig. 10 shows a comparison of ROP trends from the first eight
Second, the one run made with a roller cone bit at low OBP
wells drilled on two different development projects. The wells in
gave ROP that is almost competitive with PDC bits, given the
group A optimized drilling practices in accordance with results
lower bit price. This offers one route for reducing cost per foot if
from this method. This method was not applied on the other
the usual concerns with roller cone bits in hard formations ~prin-
project ~group B!. The data show changes in average penetra-
cipally cutting structure breakage and gauge wear!, can be over-
tion ROP through selected intervals of consistent geology, at three
come.
different hole sizes, for the first eight wells on the two develop-
The most promising indications from the data are however the
ments. Since geology and drilling conditions are different for the
penetration rates achieved with the two groups of bits run on
two projects, penetration rates are normalized to 100% at the first
downhole motor or turbine. These drilled considerably faster than
well. Improvement in ROP versus time is therefore being com-
the natural diamond bits run on rotary bottomhole assemblies,
pared for the two projects.
even though OBP was similar or higher. The most obvious expla-
In the group A wells, results of this method were first applied
nation is the higher bit rotary speed provided by the motor or
in well four, and thereafter on all wells. ROP shows a two- to
turbine. This is confirmed by the correlation shown in Fig. 9,
sixfold improvement in wells four to eight compared to wells one
where a group of runs with diamond and impregnated bits, all
to three. The value of this ROP improvement was approximately
made within a narrow range of OBP, show a clear trend of rising
U.S. $650,000 per well.
ROP with increasing rotary speed. Such a clear correlation assists
In the group B wells, some ROP improvement is evident, but is
greatly with justification of the additional cost incurred by running
more gradual and of lesser magnitude.
downhole motors or turbines, since overall cost per interval is
These data do not, by themselves, prove the impact of this
reduced.
method on drilling performance. However, drilling practices on
Thus, in this application, optimization recommendations now
the group A wells were modified in accordance with findings from
stipulate the use of turbines or motors for ROP, in tandem with
the method, primarily in terms of hydraulic horsepower, mechani-
diamond or impregnated bits to provide the required durability.
cal drilling parameters, motor and turbine usage, and preferred bit
Low OBP is encouraged, though of course mud density is often
characteristics. The expected benefit to ROP clearly occurred. In
dictated by other criteria.
contrast, the group B development lacked any systematic studies
of ROP, and therefore failed to identify either which factors were
controlling ROP, or how drilling practices could be optimized for
ROP improvement.

Discussion
The ROP analysis tool described here brings an engineering
method to a part of the drilling process that is technically com-
plex, commercially significant, and lacking in analysis techniques
and models. It is cheap to implement, requiring only standard
wellsite data and a spreadsheet package. As the above examples
show, results can be highly informative as to the underlying
mechanisms that control ROP.
For a drilling operations community, the method identifies op-
portunities by which ROP can be raised. The correlations are rel-
evant since they are drawn directly from local data. Their effect is
Fig. 9Relationship of ROP to rotary speed at the bit for rotary, to reduce reliance on intuition in favor of practices which are
motor, and turbine runs in sandstone. demonstrably most cost effective. Thus, disagreements over

M.J. Fear: Improve Rate of Penetration SPE Drill. & Completion, Vol. 14, No. 1, March 1999 47
which flow rate or HSI are most helpful to bit cleaning, or 11. Onyia, E.C.: Relationships Between Formation Strength, Drilling
whether downhole motors do or do not raise ROP, can be quickly Strength, and Electric Log Properties, paper SPE 18166, Presented
resolved and best practices standardized. at the 1988 Annual Technical Conference and Exhibition, Houston,
For drill bit suppliers, the method provides a means to make TX, 25 October.
12. White, D.B., Curry, D.A., and Gavignet, A.G.: Effects of Nozzle
ROP comparisons without results being distorted by geological
Configuration on Roller Cone Bit Performance, paper IADC/SPE
variations, how long a dull bit was persisted with, or any of the 17188, Presented at the 1988 IADC/SPE Drilling Conference, Dallas,
other influences that reduce the credibility and accuracy of a TX, 28 February3 March.
claimed bit effect. When assessment of bit designs is made from 13. Xu, H.: A Method for Modelling Rate of Penetration of Drill Bits
field data, it seems appropriate to at least ensure that performance Using Electric and Mud Logging Data, PhD dissertation, Heriott
comparisons are properly constructed and understood. Watt U., Edinburgh, Scotland ~1995!.
For analytic modeling of ROP, results from the method shed 14. Walker, B.H. et al.: Roller-Bit Penetration Rate Response as a
some light on why a particular model may be valid for the condi- Function of Rock Properties and Well Depth, paper SPE 15620,
tions under which it was developed, but impossible to extrapolate Presented at the 1986 Annual Technical Conference and Exhibition,
to other applications where factors such as bit design, mud type, New Orleans, LA, 58 October.
15. Tibbitts, G.A. et al.: The Effects of Bit Hydraulics on Full-Scale
or rock type may have changed. For example, a model for PDC bit
Laboratory Drilled Shale, paper SPE 8439, Presented at the 1979
response in OBM and shale could apparently exclude bit features Annual Fall Technical Conference and Exhibition of the Society of
and mud density, but would need to incorporate those variables in Petroleum Engineers of AIME, Las Vegas, NV, 2326 September.
WBM. This type of change is very difficult to predict without 16. Cheatham, C.A. and Nahm, J.J.: Effects of Selected Mud Properties
guidance from findings such as those reported here. Perhaps the on Rate of Penetration in Full-Scale Drilling Simulations, paper
greatest contribution this method could make to analytic ROP SPE/IADC 13465, Presented at the 1985 SPE/IADC Drilling Confer-
modeling is to indicate with which variables the model should be ence, New Orleans, LA, 68 March.
constructed for a particular application. 17. Black, A.D. et al.: Effects of Size on Three-Cone Bit Performance
Ultimately, the objective of these ROP investigations is to iden- in Laboratory Drilled Shale, paper SPE 11231, Presented at the
tify, for common combinations of bit size, mud type, bit class and 1982 Annual Fall Technical Conference and Exhibition of the Society
of Petroleum Engineers of AIME, New Orleans, LA, 2629 Septem-
rock type, which variables are most influential to ROP. Such gen-
ber.
eralized findings can then provide drilling operations communities 18. Holster, J.L. and Kipp, R.J.: Effect of Bit Hydraulic Horsepower on
with guidelines on how to maximize ROP, spreading the benefit the Drilling Rate of a Polycrystalline Diamond Compact Bit, paper
of the method to all applications under which its findings are SPE 11949, Presented at the 1983 Annual Technical Conference and
physically valid. Exhibition, San Francisco, CA, 58 October.
19. Burgess, T.M. and Lesso, W.G.: Measuring the Wear of Milled
Tooth Bits Using MWD Torque and Weight-on-Bit, paper SPE/
Acknowledgments IADC 13475, Presented at the 1985 SPE/IADC Drilling Conference,
The author wishes to thank BP Exploration, Total C.F.P., Triton New Orleans, LA, 68 March.
Energy Corporation, and Ecopetrol for permission to publish this 20. Winters, W.J., Warren, T.M., and Onyia, E.C.: Roller Bit Model
paper. With Rock Ductility and Cone Offset, paper SPE 16696, Presented
at the 1987 Annual Technical Conference and Exhibition, Dallas, TX,
2730 September.
References 21. Bourgoyne, A.T. Jr. and Young, F.S. Jr.: A Multiple Regression
1. Cunningham, R.A. and Eenink, J.G.: Laboratory Study of Effect of Approach to Optimal Drilling and Abnormal Pressure Detection,
Overburden, Formation and Mud Column Pressure on Drilling Rate SPE J 371 ~August 1974!; Trans., AIME 257.
of Permeable Formations, J. Pet. Technol. 9 ~January 1959!. 22. Detournay, E. and Defourny, P.: A Phenomenological Model for the
2. Detournay, E. and Atkinson, C.: Influence of Pore Pressure on the Drilling Action of Drag Bits, Int. J. Rock Mech. Min. Sci. Geomech.
Drilling Response of PDC bits, Rock Mechanics as a Multidisci- Abstr. 29~1!, 13 ~1992!.
plinary Science, Roegiers ~ed.!, Balkema, Rotterdam ~1991! 539. 23. Bourdon, J.-C. et al.: Comparison of Field and Laboratory-
3. Zijsling, D.H.: Single Cutter TestingA Key for PDC Bt Develop- Simulated Drill-Off Tests, SPE Drill. Complet. 329 ~December
ment, paper SPE 16529/1, Presented at the 1987 Offshore Europe 1989!.
Conference, Aberdeen, 811 September.
4. Gray-Stephens, D., Cook, J.M., and Sheppard, M.C.: Influence of AppendixHow to Use the Method
Pore Pressure on Drilling Response in Hard Shales, SPE Drill. This method analyzes ROP variation between bit runs, in such a
Complet. 263 ~December 1994!. way as to minimize the effects of geological variation on ROP.
5. Peltier, B. and Atkinson, C.: Dynamic Pore Pressure Ahead of the
This allows the effects of controllable drilling and hydraulic pa-
Bit, paper IADC/SPE 14787, Presented at the 1986 IADC/SPE
rameters to be clearly seen, together with the effects of any influ-
Drilling Conference, Dallas, TX, 1012 February.
6. Warren, T. and Smith, M.B.: Bottomhole Stress Factors Affecting
ential bit features. Specific subsections of bit runs can be further
Drilling Rate at Depth, JPT 1523 ~August 1985!. extracted to minimize ROP variation due to bit wear. The method
7. Black, A.D. et al.: PDC Bit Performance for Rotary, Mud Motor, uses the following steps:
and Turbine Drilling Applications, SPE Drill. Complet. 409 ~De- 1. Select a group of wells with comparable geology and casing
cember 1986!. programs.
8. Pessier, R.C., Fear, M.J., and Wells, M.R.: Different Shales Dictate 2. From the wells, select a group of hole sections of equal or
Fundamentally Different Strategies in Hydraulics, Bit Selection, and similar bit size, where similar formations were drilled.
Operating Practices, paper SPE 28322, Presented at the 1994 An- 3. Select an interval of homogenous lithology within the hole
nual Technical Conference and Exhibition, New Orleans, LA, 2528 sections.
September.
4. Identify the portions of bit runs that fall within the interval
9. Warren, T.M. and Armagost, W.K.: Laboratory Drilling Perfor-
mance of PDC Bits, SPE Drill. Complet. 125 ~June 1988!.
of homogenous lithology, and obtain averages of drilling/
10. Moffitt, S.R. and McGehee, D.Y.: Performance Comparison of hydraulic parameters and formation properties for those portions.
Rolling Cutter Bits With Alternate Nozzle Configurations, paper 5. Tabulate the averaged data in a spreadsheet, together with
SPE/IADC 18630, Presented at the 1989 SPE/IADC Drilling Confer- bit size, mud type, bottomhole assembly ~BHA! type, formation
ence; New Orleans, LA, 28 February3 March. name, interval top/bottom depths, and bit nozzle arrangement.

48 M.J. Fear: Improve Rate of Penetration SPE Drill. & Completion, Vol. 14, No. 1, March 1999
6. Separate the tabulated data by mud type, type of BHA, and 12. If clear, unambiguous correlations are found, attempt to fit
bit class ~e.g., PDC or roller cone!. a ROP model to the data. Use a step-wise regression in a package
7. Choose the dependent variable ~usually ROP!. such as SAS.
8. From a list of factors that could potentially influence ROP 13. If a model can be fitted, assess carefully the degree of fit
~the independent variables, see Table 1!, crossplot each of the ~from r squared, error in model parameter values, number of pa-
independent variables against ROP for each bit class in each mud rameters versus sample size, etc.!.
type. 14. If a reliable model was fitted, assess the accuracy of the
9. To avoid misinterpretation, check for any interdependence model in a predictive mode against future bit runs.
between supposedly independent variables ~e.g., HSI and flow 15. From comparisons of model predictions versus actual
rate, bit rotary speed and flow rate with downhole motors!. ROP, or from crossplotted relationships, examine the data for any
10. If bit wear state can be assumed known over the averaged bit design effects on ROP.
intervals ~e.g., in a nonabrasive formation, at the start or end of a 16. Combine the findings with those from other methods.
bit run!, crossplot ROP against wear state, for each bit class in
17. Generate recommendations for maximizing ROP.
each mud type.
11. If a good correlation exists between ROP and one other Martyn Fear is a bit-optimization specialist with BP Exploration
variable, attempt to normalize ROP for that variable before cross- in Colombia. He provides bit-optimization support to BP drilling
plotting the normalized values against all other variables ~this may operations worldwide. Fear holds a BS degree in geological
reveal correlations that are otherwise masked!. sciences from the U. of Birmingham.

M.J. Fear: Improve Rate of Penetration SPE Drill. & Completion, Vol. 14, No. 1, March 1999 49

You might also like