NOTICE CONCERNING COPYRIGHT RESTRICTIONS

This document may contain copyrighted materials. These materials have been made available for use in research, teaching, and private study, but may not be used for any commercial purpose. Users may not otherwise copy, reproduce, retransmit, distribute, publish, commercially exploit or otherwise transfer any material.

The copyright law of the United States (Title 17, United States Code) governs the making of photocopies or other reproductions of copyrighted material.

Under certain conditions specified in the law, libraries and archives are authorized to furnish a photocopy or other reproduction. One of these specific conditions is that the photocopy or reproduction is not to be "used for any purpose other than private study, scholarship, or research." If a user makes a request for, or later uses, a photocopy or reproduction for purposes in excess of "fair use," that user may be liable for copyright infringement.

This institution reserves the right to refuse to accept a copying order if, in its judgment, fulfillment of the order would involve violation of copyright law.

Developments at Schlumberger Cambridge Research

G. Cooper, Sedco Forex Schlumberger, Montrouge, France

SUMMARY

Research in Drilling Mechanics at Schlumberger

Cambridge Research is aimed at improving the

efficiency and safety of drilling in two ways. First, by

understanding the steady-state response of the drill bit

to weight on bit and rotary speed for any particular

combination of bit, rock and hydraulics, we hope to be

able to optimise the normal operating parameters and

also to obtain information about the rocks being

penetrated. Second, and, we believe, of equal value, we

have been studying the means of identifying a variety

of "Drilling Events". These are non-steady and

generally unwelcome occurrences such as bit bearing

failure, bit balling, influxes of various types etc. If

detected at an early stage, such events are relatively

straightforward to deal with, but if left undetected, they

can become costly to correct and very dangerous.

The main method of attack has been to study

torque and rate of penetration as a function of weight

on bit (WOB) and rotary speed, not only as steady-state

values, averaged over a period ranging from seconds to

hours, but in terms of their characteristic vibrations.

The latter have been shown to contain quantitative

information about the state of wear of the bit teeth and

bearings, and have been related to changes in rock

indentation resistance.

STEADY STATE MEASUREMENTS

Optimisation of operating parameters

The measurement and optimisation of the

steady-state drilling parameters for any combination of

bit, rock and operating conditions is of course not new.

It is, however, important for the driller in the field to

know what bit to select in any given conditions, and

although much information is available from the bit

manufacturers, it is not always complete. It is, for

example, difficult to obtain comparisons between bits

coming from different manufacturers, or to obtain

information about the performance of a bit in any wear

state other than new. In the latter context, for example,

it is known that the rate of penetration of a new

milled-tooth bit is superior to that of the nominally

equivalent tungsten carbide insert bit in spite of the fact

that the latter may cost perhaps double. It is also

known, however, that in most rocks, the wear of the

teeth of the insert bit will be negligible, and its

performance will be maintained, whereas the milled

tooth bit will wear progressively, and its rate of

penetration will fall.

gil.3

The question is thus posed'"After how long will

the overall performance of the insert bit overtake that

of the milled-tooth bit, and when will it become

economically advantageous ?" Clearly, the answer

depends upon many factors, including those related to

the rig type and operating environment. At SCR, we

have been using our full-scale drilling teit machine

(Cooper and Peltier, 1986) to measure the performance

of bits in different wear states in order to provide the

basic drilling information from which, when the

external factors are included, a rational choice can be

made.

Detection of Bit tooth wear.

Experiments on bits in different wear states also

serve another function, to support work on the

development of diagnostic methods for determining

the state of wear of the drill bit from the drilling

response. Analysis of the ratio of torque divided by

weight-on-bit to rate of penetration per bit rotation is

the basis of the Mechanical Efficiency Log (Burgess and

Lesso, 1985). This is a diagnostic technique which is

capable of determining tooth wear in milled tooth bits.

The method works well for shales, but is less accurate

when the lithology is changing rapidly, essentially

because there is difficulty in distinguishing increases in

rock hardness from additional wear of the teeth. Work

is under way at SCR to separate the two effects.

An alternative method for determining bit wear is

by the analysis of drilling vibrations. If a spectrum is

obtained of the vibrations occurring above the bit,

many peaks are seen, whether in torsion or vertical

movement force or acceleration. Many of these peaks

can be associated with the impact of successive teeth on

particular rows and on specific cones of the bit. Since

the number of teeth on any row is known a priori, the

measurement of the frequency of the relevant peak(s)

gives a measurement of the cone rotation speed. By

comparing this value with the bit rotary speed, one can

then infer the effective rolling radius of the cone. This,

of course, decreases progressively as the cone teeth

wear, and so a diagnostic for thi wear of the teeth may

be obtained. This method at present appears to give a

value which is independent of lithology, although in

some cases considerable signal processing is required to

obtain a reliable result. Figure 1 shows the result of

applying the technique in a favourable case to a milled

tooth bit for soft rock ( IADC code 1:3:6 ) (Cooper et

al., 1987). Here, the upward,shift in frequency of the

main peak can be clearly seen as the wear state changes

from T : l to T : 5.

1

Another method which we have attempted to use

to derive a diagnostic for tooth wear is the Drill-off Test.

(Lubinski, 1958). In this test, the • driller applies

weight-on-bit to some predetermined value, then sets

the brake. As the bit drills ahead, WOB decays at a rate

determined by the rate of penetration and the drill

string compliance. Meanwhile, the surface WOB is

recorded, and if ROP is proportional to WOB, it is

expected that WOB will decay to zero exponentially

with time (Bourdon et al.,1987). This is indeed the case

in many circumstances, but sometimes the weight does

not decay to zero (see Fig. 2 ). Under such conditions, it

was believed that the offset might be an indicator of

wear of the bit, the argument being that a certain

minimum weight would be necessary to cause the bit to

drill, and that this threshold would increase with the

size of the wear flat. To test this idea, careful

comparisons have been made with the same new and

worn bits tested in· the field and in the laboratory in the

drilling test machine. The latter tests have shown that

a threshold is never recorded in the laboratory, and so

we have been led to the conclusion that the observation

of the offset in field conditions is not an indicator of bit

wear, but is more likely to be associated with poor

weight transfer (eg stabiliser hanging or similar

problerns).

Detection of changes in lithology

While we are actively engaged in this area, there

is little which can be reported at present. Considerable

effort is being deployed to develop drilling models

which contain an explicit link between the observed

drilling parameters and the rock properties as they

would be measured in a conventional rock-mechanics

test. This is relatively straightforward for drag bits, but

roller-cone bits have a very complicated cutting action

which has to be accurately modelled.

At SCR, a combined theoretical and experimental

approach has been adopted. For the experiments, we

have built an "instrumented bit", which comprises the

three cones of a commercial roller cone bit which have

been removed from the bit body and reassembled on an

instrumented carrier. Two of the cones are attached to

shaft encoders, so that their angular positions may be

determined. The third cone has been cut so as to

separate the tooth rows, and each tooth row is carried

separately on strain-gauged supports. With this cone

also linked to a shaft encoder, we can obtain all the

information necessary to reconstitute the force-

indentation behaviour of each of the bit teeth as they

drill. Fig. 3 shows a record of the vertical forces on the

inner and outer tooth rows of one of the cones of an

IADC code 1:3:6 bit during one complete rotation of

the cone. The force peaks from each of the nineteen

teeth on the outer row and the nine teeth on the inner

row can be seen clearly. Knowing the angular relations

between the teeth, such data can then be used to

construct the force - indentation curves for the

individual teeth, and examples are shown in Fig. 4 for

four different rock types (Cooper et al. 1987). These

curves may now be compared with the equivalent data

obtained in single-point indentation experimer•ts, and

thence with the suite of more conventional rock

mechanics tests. The ultimate objective is, of course, to

infer the mechanical properties of the rock from its

drilling response.

EVENT DETECTION

As mentioned above, we believe that the timely

detection of abnormal behaviour during drilling is at

least as important as optimising the steady-state

operating parameters, and is probably more so

regarding questions of safety. Within the Drilling

Mechanics Group, we have been investigating various

ways in which some of these events may be detected.

One of the most annoying, and possibly the most

preventable events is the loss of one or more of the

cones of a roller cone bit following failure of the

bearings. If the cones are lost, then a fishing trip must

be undertaken, because drilling cannot, recommence

until the lost cones are recovered.

A technique which we have used to detect the

early stages of bit bearing failure (before the cones are

lost) is the analysis of torque and weight-on-bit in real

time using a moving window technique (Peltier et al.,

1987). The object is to detect the intermittent snatching

of the cone bearing which occurs during the early stages

of failure, before the cone locks or falls off. The method

consists of monitoring the ratio of torque to weight on

bit and looking for anomalies. in this ratio in a time

scale appropriate to the scale of the event which it is

hoped to detect (This is a few seconds in the case of

bearing failures). It should be realised that the ratio

may be affected by changes other than changes in

bearing condition - lithology, for example. In this case,

however, it is not expected that the rock quality will

change dramatically in a timescale of a few seconds, and

so the time constant of the diagnosis may be changed to

"tune out" such unwanted influences.

Fig. 5 shows an example of the application of the

method. It records a test in which an 8 1/2 " bit was run

at constant rotary speed, but with weight on bit varying

between approximately six and twenty tons. The bit had

been used in the field, but was believed to be still in

good condition. Before the test, holes were drilled in

each bit lug, and thermocouples were placed as close as

possible to the bearings to record bearing temperatures

during the test.

In the figure, the first two tracks show the changes

of WOB and torque during the test. Track three shows

one of the three thermocouple records, which was the

only one to show a large temperature rise during the

test (indicating failure in that one particular bearing

only). It will be seen that there were no problems in the

early part of the test, at low weight on bit, but that there

was a sharp rise in temperature after about 130 seconds,

2

:

and another at about 170 seconds. Clearly, these twOincreases in temperature indicate incipient bearingfailure at the higher loads. Track four shows the torquediagnosis signal derived from tracks one and two. Iteasily detects the two points of bearing failure, and withas much precision as the thermocouple. The advantageof the torque analysis, however, over the thermocouplemethod, is that with the former, it is not necessary tomodify the bit to obtain the measurement, as the inputvalues could be obtained by, for example, a MWD unitmounted above the bit.

This technique appears to be applicable to thedetection of a variety of events; so far we have been ableto use a very similar method in the detection of thebeginning of bit-balling, and other applications areunder investigation.

If the early stages of bearing failure have passedundetected, there is another diagnostic method whichwe believe may be of value, through the analysis of thetorque signal at a higher frequency (Cooper et al., 1987).Fig. 6 shows traces of torque and torsional accelerationtaken over about three seconds for a bit in goodcondition (right) and one with one of the cones locked(left). As might be expected, the bit with the locked coneshows a higher average level of torque (lower traces),but the signals are noisy, and not easy to distinguish.Compare, however, the records of torsional acceleration(above). Here it is immediately obvious that whereasthe bit which is in good condition gives a relativelyquiet signal, the bit with the locked cone shows a seriesof sharp one-sided spikes. These almost certainly comefrom the impact of the bit teeth of the locked coneagainst the roughnesses of the hole bottom. They are avery characteristic feature of the vibration signature of abit in this condition, and as such, form a very goodindicator of a locked cone.

Fig. 1. The upward shift in frequency of thecharacteristic tooth-generated vibration peak as anIADC code 1:3:6 bit wore from T: 1 (dotted) to T: 5(solid line).

Cooper, G. A., Lesage, M., Sheppard, M., and Wand, P.,1987. The interpretation of roller-cone drill bitvibrations for bit wear and rock type. Proceedings of1987 Rapid Excavation and Tunnelling Conference,New Orleans, LA., 14-18 June, 1987.

Lubinski, A., 1958. Proposal for future tests. ThePetroleum Engineer, B 50 - B 52, Jan 1858.

Peltier, B. P., Cooper, G. A., and Curry, D.A., 1987. Use oftorque analysis to determine roller-cone bit bearingfailure. SPE Paper 16698, presented at 62nd. SPEMeeting, Dallas, TX., 27-30 Sept., 1987.

0.5000E-01M/S-2

ASPEC

\6 :fl."-«. - -. -'0.0000 -» '0 0000 Norm Hz 40.00EFFECT OF WEAR ON VERTICAL ACCN

REFERENCES

Bourdon, J-C., Cooper, G. A., Curry, D. A., and McCann,D. P., 1987, Comparison of field and laboratory 5simulated Drill-off tests. SPE Paper 16162, presented 5 150at the 1987 SPE/IADC Drilling Conference, New •Orleans, LA., 15-18 March, 1987. &!

m 100Burgess, T. M. and Lesso, W. G., 1985. Measuring the 5wear of milled tooth bits using MWD Torque and w

Weight-on-bit. SPE Paper 13475, presented at the1985 SPE/IADC Drilling Conference, New Orleans,LA., 6-8 March, 1985.

0 15 30 45 60 75 90Cooper, G.A., and Peltier, B., 1986. Advanced techniques 105 120 135 150 1for laboratory full-scale drilling tests. SPE Paper TIMEs14783, presented at the 1986 SPE/IADC Drilling Fig. 2. Record of a field drill-off test, in whichConference, Dallas TX., 10-12 Feb., 1986. weight-on-bit did not decay to zero.

3

/'1/ 1/1111 111 1

250 -DRILL-OFF TESTDEPTH 1885MRPM 100200 - BIT. STRATABIT PD2

50-

0

4

20.00 300 1Weight on bitKN kN 200

100 -••-1 - ic07.5 2

0.0000 0.0000 Degrees •359.0 Torque 5.0 -kN.m5.000 2.5 -KN 0200 3

Temperature $*./tr'-1.C 100 -0.0000 Degrees 359.0 10AVERAGED VERTICAL FORCES Instrumented Bit Drilling Portland LS 500 4

Diagnosis -• 1Fig. 3. Vertical force on the outer (above) and inner N.mi(below) tooth rows as a function of rotation angle forone of the cones on an IADC code 1:3:6 bit. 1 1

0 25 50 75 100 125 150 175 200 225Time s

10.00 Fig. 5. Detection of the beginning of bit bearing failureas weight-onfbit was increased in five stages at constant

KN rotary speed. Tracks 1 and 2, weight-on-bit and torque.Track 3, record from a thermocouple in one of the bit

/\ lugs. Track 4, a torque diagnosis derived from tracks 11 14 and 2.

/ 1/n-/Al 'r\----

1 3 --i USED BIT NEW BIT1/- '' / 3/-/r y 50.00

- l'f »«* RAD/52-'lt>/

0.0000 1. 1o. oood MMAVERAGED FORCE PENETRATION CURVES -50.00Inner Row OF Instrumented Bit Drilling 1500.Different Lithologies NM

Fig. 4. Different force-penetration curves derived froman analysis of torque and weight-on-bit.

0.0000 0 0008 ' 2.000COMPARISON BETWEEN NEW ANO USED J22 BITDRILLING MARBLE 2T W. 0. B. 31 RPMUSED BIT HAS ONE CONE STUCK

Fig. 6. Detection of locked cone from an analysis oftorsional vibrations. Note the sharp spikes in torsionalacceleration from the failed bit (top left). This ti,aceshows a much more striking difference from that •of thenew bit ( top right) than do the corresponding torquesignals (bottom left and right)

OUTER ROW

.-IIINNER ROW

0 0000

L

2500 A

250

CARRERA MARBLEPORTLAND LIMESTONE

--- HOLLI NGTON SANDSTONE- KIMMERIOGE BAY SHALE

237\ 1\\ -+•114·*44414*

1«'11•1 d»,Adv.0 0000 2 000

4 000