Summary nents allows more accurate definition of friction for torque projec-
Excessive torque and drag can be critical limitations in extended- tions and allows proper prioritization for torque-reduction mea-
reach drilling (ERD). This paper details issues related to torque-and- sures. Frictional torque is generated by contact loads between the
drag prediction, monitoring, and management in ERD wells. Results drillstring and casing or open hole. The magnitude of contact loads
are presented from sensitivity analyses of extreme ERD trajectories is determined by drillstring tension/compression, dogleg severities,
such as 7- to 8-km departures at 1600 m true vertical depth (TVD). drillpipe (DP) and hole size, drillstring weight, and inclination. Pro-
Several such wells have now been successfully drilled at the Wytch file optimization and tortuosity control are therefore important mea-
Farm oilfield using results from these studies. In such high-angle sures to minimize contact loads. Lubricity is a major factor control-
ERD wells, compression generated in the drillpipe during tripping ling friction, and is itself largely controlled by mud and formation
and sliding operations can exceed the critical buckling load and cause types. With means of predicting bit torque, the implications of using
the drillpipe to buckle. As a result, buckling initiation and post-buck- different bit types can be assessed. Dynamic torques can also signif-
ling analyses are used to quantify the extent and severity of buckling icantly impact operations and should be minimized.4 Mechanical
and the associated increases in drag forces and pipe stresses. The pa- torque sources, such as cutting beds, borehole ledges, and stabilizer
per addresses the importance of drilling data in calibrating torque/ effects can be very significant and must also be minimized.
drag models to capture the continual changes in drilling parameters
and operating conditions. The paper presents a number of field case Drag and Buckling. Prediction of upward drag poses issues similar
studies where analyses have been conducted to directly assist drilling to torque prediction in terms of trajectory design, mud lubricity,
operations. This paper should be of high interest to engineers execut- wellbore condition, tortuosity, and mechanical influences. Down-
ing, planning, or evaluating ERD operations. ward drag prediction in ERD wells introduces a further level of com-
plexity as a result of potential buckling of the string under excessive
Introduction axial compression. Buckling is an important consideration in ERD
wells because, while tripping in or sliding, the drillstring and other
Torque and upward drag must be projected for ERD operations to
tubulars (e.g. liners, workstrings, tubing, perforating guns, coiled
ensure the rig’s rotary and hoisting equipment are adequately sized
tubing) are subjected to large compressive forces. Consequently,
and the drillstring is properly designed. Downward drag must be
projected to evaluate the limits for sliding-oriented drilling motors many operations in ERD wells may result in buckling. Buckling
and running tubulars. A key aspect of these projections is to ensure must be properly addressed to account for its impact on drag and its
good accuracy with some level of conservatism without incurring elevation of drillstring stresses.
excessive overdesign. Various components can contribute to the Fig. 1 illustrates string behavior under increasing compressive
buildup of both torque and drag in ERD operations. Identifying and loads. When compression is below the critical buckling load,5,6 the
quantifying these distinct components is an important part of prop- string will sustain this compression without buckling. Above the
erly projecting torque and drag. The acquisition and analysis of field critical buckling load, the string buckles into sinusoidal or “snaky”
data is critical in this process. When careful analysis and forecasting buckling. This buckling condition results in the string deforming
has identified torque or drag as a limiting factor, effective measures into a snaky configuration along the low side of the well. For com-
must be available to alleviate the specific operational constraint. pression loads above the critical helical buckling load,7 the string
These measures include both torque/drag reduction and alternative can no longer maintain its snaky configuration and it coils up against
means to achieve the desired operation. the wellbore and helically buckles. String lock-up immediately fol-
Torque/drag has been addressed in various industry publications lows the onset of helical buckling caused by a dramatic increase in
over the years. This paper provides additional understanding and ob- wall forces. Helical buckling should therefore be avoided. In high
servations on torque/drag issues as a result of theoretical and empiri- inclination wells, the magnitude of the helical buckling load (for a
cal analyses of extreme ERD operations. These analyses were per- conventional drillstring) is very high, hence its occurrence is not
formed in support of the Wytch Farm ERD development,1,2 which common. It is thus the growth of the snaky buckling and the
has set a number of world-record achievements. This paper presents associated increase in wall forces which predominantly need to be
both field data and modeling predictions to convey key torque/drag quantified.
issues. Topics covered include torque/drag projection, analysis, vari- The severity of snaky buckling is quantified by how far from low-
ability, control, and management. In addition to drilling torque/drag side (in degrees) the string is displaced. If the snaky buckling ampli-
issues, the paper covers other important operations such as liner rota- tude remains below about 40, the buckling is generally tolerable
tion while cementing and completion operations. The purpose of the and does not cause significant increases in drag. However, more se-
paper is to highlight key lessons and observations on torque and drag vere sinusoidal buckling should be avoided as it can cause large in-
as they impact world-class ERD operations. creases in wall forces, which can also lead to string lock-up or loss
of surface string weight. Both sinusoidal and helical buckling im-
General Torque-and-Drag Considerations for ERD pose additional stresses in the string. However, because of wellbore
Frictional and Mechanical Torques. The theory behind the “soft- confinement, the string remains elastic and is not damaged (i.e. it
string” model for basic torque/drag prediction is well-known in the does not yield) unless the hole section is heavily washed out. When
industry.3 Proper application of the model requires a full under- rotating, string-wellbore axial friction is considerably reduced and
standing of the factors influencing torque and drag in the field. To- string buckling becomes less likely, although it is still possible.
tal-surface torque is comprised of frictional string torque, bit torque, Fig. 2 illustrates typical buckling behavior in an ERD-type well
mechanical torques, and dynamic torques. Separating these compo- profile. Only the upper portion of the 80 tangent section is shown
where compression exceeds the critical buckling load. Buckling ex-
Copyright 1997 Society of Petroleum Engineers tends from the 80 tangent, where compression is a maximum up
into the near-vertical section. In the 80 section, stabilization forces
Original SPE manuscript received for review 10 April 1996. Revised manuscript received 6
January 1997. Paper peer approved 6 January 1997. Paper (SPE 35102) first presented at due to high inclination provide adequate support to restrain the
the 1996 IADC/SPE Drilling Conference held in New Orleans, 12–15 March. buckling to the snaky mode. In the build section, wellbore curvature
but they first must be recognized. For this reason, approaches in-
volving the “lumping” of mechanical and frictional effects into
Fig. 4—Overprediction of drillstring torque due to improved single factors is discouraged. Higher flow rates, careful rheology
drilling practices in 12¼-in. section.
control, and drillstring rotation can improve hole cleaning to mini-
mize cutting beds. Mechanical and chemical wellbore stability anal-
Actual bit torques vary dynamically and substantially during drill- ysis can result in mud weight/chemistry recommendations to mini-
ing and are influenced by many factors. These include weight-on-bit mize instabilities. Identification of excessive stabilizer torque can
(WOB), rev/min (RPM), formation characteristics (shear and com- lead to better equipment selection such as undergauge stabilizers,
pressive strengths), polycrystalline-diamond-compact (PDC) bit reamers, etc.
design variations, bit wear, and hydraulics. Bit torque models can As an example of mechanically-induced torque variations, Fig.
show substantial inaccuracies in nonductile formations such as 6 shows maximum 8½-in. surface torques from several recent
sandstones and carbonates since the ratio of shear/penetration Wytch Farm wells. A fairly well-defined 8½-in. torque trend is es-
strength varies dramatically from those in ductile shales which are tablished by wells F19, F20, and F21. That torque trend, however,
used for most laboratory tests. PDC bits, which come in a wide vari- indicated a maximum drillable depth of about 6750 m because of the
ety of designs (i.e. bit profile, cutting structure, and hydraulic de- top-drive limit of 45,000 lbf-ft. On well M2, lost-circulation prob-
sign) generally have higher torques and higher torque/weight ratios lems in the 8½-in. reservoir section led to substantial additions of
and have not been studied as thoroughly as tricone bits. Finally, fibrous lost-circulation material (LCM). The LCM formed a low-
most bit torque models do not currently account for the effects of bit side bed that, in combination with the OBM, produced a much lower
wear or bit cleaning hydraulics. frictional interface with the drillstring/BHA than the previous steel-
The implication of these issues for torque forecasting is that bit sandstone contact. The M2 torque “trend” is thus very complex. The
torque models should be used carefully. A more empirical approach “trend” involves a cyclic process of drilling openhole reservoir
is to use a conservative upper bound for bit torque, preferably from (with the associated steep torque trend) followed by the condition-
field data. Bit torque should be monitored in ERD wells using a drill- ing of that openhole section by first cleaning the fine sand grain cut-
ing-mechanics sub in the MWD to provide DTOB. Alternately, tings from the well while simultaneously adding LCM (with an
crude torque-on-bit (TOB) measurements can be taken by monitor- associated torque reduction). These behaviors underscore the varia-
ing off-bottom and on-bottom surface torques, although this is only tions in torque that occur as a result of wellbore condition and the
approximate since WOB causes variations in the drillstring tension/ dominant role that mechanical factors can play in torque projection
compression profile affecting wall forces and hence string torque over frictional considerations.
magnitude. Use of an upper bound for TOB is conservative and ap-
propriate for torque forecasting. However, since the majority of sur- Liner Torques During Cementing. A special torque-projection
face torques from the field will be taken when DTOB is below this problem is posed for the rotation of production liners during cement-
upper bound, torque projections can appear conservative unless this ing. In many ERD wells, the trajectory includes a high-angle or hori-
“safety margin” is recognized. zontal reservoir section. Rotation of the liner during cementing is an
As an example of the impact of bit type on torque projection, Fig.
5 shows 12¼-in. surface torques from a recent North Slope well. All
bit runs in this section were tricone with the exception of the bit run
over the interval from about 13,100 ft to 14,050 ft. As clearly shown,
the PDC bit run exhibits an offset of about 5,000 lbf-ft relative to the
tricone torque trends. This torque increase is substantial, particular-
ly for an ERD operation approaching rotary or drillstring torsional
limits. Also of interest in this figure are the erratic torques seen with
the tricone bit over the interval from about 11,150 ft to 12,100 ft.
This interval was accompanied by significant torsional dynamics
that, as mentioned above, elevate the normal torque trend.
effective mechanical measure to promote good fluid displacements initiated with regard to the liner connections, liner hanger, and liner
and enhance cement bonding. Rotation of the liner, however, requires running tools such that the entire system was rated to 20,000 to
an accurate projection of torques that will be experienced in the liner, 24,000 lbf-ft. In addition, work was undertaken to modify the ce-
as well as the hanger, the running tool, and the drillstring. Fig. 7 shows ment to a low-rheology blend and calibration analyses were con-
surface-torque measurements during the liner cement job on well F21. ducted for proper friction coefficients to account for the cement
The projected torque schedule starts at about 13,000 lbf-ft and increases torques. Fig. 8 shows the result of these efforts on the well M3 liner
to a maximum of 20,000 lbf-ft before ultimately falling to a surface cement job. As shown, the torque projection closely matches actual
torque of 10,000 lbf-ft. This projection is based on a constant set of fric- torques due to the updating of the annulus friction coefficient from
tion factors during the cement job with torque variations caused by 0.13 (with OBM in the annulus) to 0.40 as the cement is displaced.
buoyancy changes from the various fluids being pumped. The pre- The benefit of improved hole cleaning and the modified cement
dicted torque increase from 13,000 to 20,000 lbf-ft is caused by the lin- rheology is also seen as only a small (1,800 lbf-ft), transient (2-min-
er being filled with cement heavier than the OBM, while the subse- ute) torque spike was seen during this cement job.
quent reduction is caused by the additional buoyancy as that cement is
displaced into the liner annulus. After some calibration, agreement was Drag Pro¡ection
achieved between actual and projected torques until the cement enters Some conventional torque/drag simulators assume the drillstring re-
the annulus. A discrepancy is then observed which is attributed to ce- mains unbuckled and are therefore limited to modeling situations
ment effects in the annulus. This includes a general increase in friction where there is no buckling. When critical buckling loads are exceed-
between the liner and nonrotating centralizers as those oil-wet inter- ed, the drag model must have more sophisticated capabilities. These
faces are infiltrated by cement slurry. As shown in the figure, the “ce- include buckling and post-buckling analysis capability taking ac-
ment torque” generates up to 8,000 lbf-ft of additional torque which count of the 3-D wellbore geometry, curvature and wellbore friction
was previously unrecognized. In addition, a significant transient torque effects. A drillstring buckling simulator has been developed by BP
event is seen late in the cement-displacement process. This torque spike Exploration specifically for ERD application which is capable of pre-
is associated with the leading edge of the cement slurry passing by the dicting several effects: the onset of sinusoidal/snaky buckling; the
liner/hanger area. The leading edge of the slurry has undoubtedly transition from sinusoidal to helical buckling; the extent of buck-
picked up various solids as it passed along the annulus. In addition, that ling, i.e. specific intervals where the drillstring has buckled; the se-
interface could have been rheologically affected by mixing with the verity of buckling, i.e. sinusoidal or helical, and degrees of magni-
spacer and residual OBM. This torque spike was evident for about 6 tude; the associated wall forces and drag; the onset of string lock-up
minutes during the displacement and it created a torque increase of or loss of surface-string weight; and the resultant forces and stresses
6,000 to 7,000 lbf-ft. On well F21, these two effects combined to pro- in the string.
duce torques 15,000 lbf-ft above those projected. Fig. 9 shows results from a drag analysis of oriented drilling in the
As a result of these observations and a longer 2000- to 2400-m 8½-in. reservoir section of one of the record-breaking Wytch Farm
reservoir section planned for later wells, an upgrade program was ERD wells. Because of the extent of frictional drag and the very chal-
Fig. 9—Drag analysis showing buckling severity for oriented Fig. 10—Drag analysis showing buckling severity for running of
drilling in 8½-in. reservoir section. liner clean-out string.
lenging nature of providing WOB for sliding, the specific operation Fig. 12—Predicted vs. actual 95/8-in. casing running weights
analyzed includes several stands of drill collars (DC) which were showing mechanical weight losses much larger than frictional
picked up and run at surface. In addition, 30,000 lbs of traveling weight losses.
equipment weight, specifically from the top-drive and swivel, is be-
ing applied to the drillstring to assist sliding. This weight application picked up and thereby buckling was avoided. In case of buckling,
was conducted only after careful stress analysis of the involved load- rotation should not be initiated until the buckling is relieved by pick-
ing scenarios and various components’ capacities. As shown, the ap- ing up and then reaming back to bottom. Rotating a buckled string
plied surface compressive loads resulted in substantial buckling in the is not recommended as it can lead to high bending stresses, severe
drillstring (the shaded region). The DC at the surface do not buckle dynamics, and twistoffs.
because of their substantial thickness and bending stiffness. Immedi- As a final drag example, Fig. 12 shows running weight for the
ately below the DC section, the 5½-in. DP partially buckles. The 95/8-in. production casing. Running of the 95/8-in. casing does not
buckling mode in the 5½-in. section is sinusoidal and varies in ampli- introduce any buckling because of the substantial bending stiffness
tude from 43 to 57. A short section of the 5½-in. DP does not buckle of the casing. However, the 95/8-in. casing running is a critical opera-
because of the additional wellbore support provided by the curvature tion in these ERD wells and warrants discussion. As shown in the
in the build interval. The 5-in. DP section buckles completely inside figure, the casing down-weights were significantly below and up-
the 95/8-in. casing with a peak buckling amplitude of 74 in the sinusoi- weights significantly above the projected weights. As with drilling
dal mode. The buckling of the 5-in. DP ceases only within the 8½–in. torques, mechanical effects during casing running can dominate
open hole because of the higher inclination and hence wellbore sup- frictional effects. In this case, mechanically-induced weight losses
port in that section. Overall, about 75% of the drillstring is buckled caused by cuttings and wellbore sloughing reduce available running
according to this analysis. Despite this significant buckling and weight by about 70,000 lbs near 2000 m measured depth (MD). At
associated increase in drag, analysis indicated that sufficient com- that point, a casing-circulating tool was used to circulate the casing
pressive weight would be available to the bit, and oriented drilling while working the string. As shown, running weight was partially
was successfully achieved in this extreme ERD profile. regained, although it remained about 40,000 lbs below levels pro-
Fig. 10 shows a composite buckling plot for the clean-out string jected from frictional considerations. Casing circulation was used
following the running and cementing of the 5½-in. liner in the on several occasions during the casing running and eventually cas-
8½-in. reservoir section. The buckling amplitude is plotted for a se- ing-running weight recovered to projected levels below about 2700
ries of running depths from the initiation of buckling in the drill- m MD. Because of the potential for serious mechanically induced
string until reaching total/target depth (TD). Buckling is limited to weight losses, ERD casing-running procedures should include con-
the 27/8-in. DP inside the 95/8-in. production casing. The amplitude tingency measures. This may include circulation, utilization of top-
of the 27/8-in. buckling, which is sinusoidal, grows up to a maximum drive weight, full or partial flotation, and casing rotation. With prop-
of 70. The maximum buckling occurs when nearly all of the 27/8-in. er engineering, casing running systems can be designed to improve
DP is inside the horizontal 5½-in. liner, generating drag that must running capabilities and achieve successful casing runs into ex-
be sustained by the top of the 27/8-in. DP in the 95/8-in. casing above treme ERD well profiles. At Wytch Farm, 95/8-in. casing has been
the 5½-in. liner. Inside the 5½-in. liner, buckling is suppressed by successfully run to 6006 m MD, which involved 5164 m of 12¼-in.
smaller clearances and higher inclination.
openhole section at 80 to 82 of inclination.
Fig. 11 shows a buckling analysis for the perforating string. Buck-
ling is predicted to occur in both the 27/8-in. DP above the guns and
in the upper portion of the gun string. The buckling is sinusoidal and Torque Control and Management Measures
has a maximum amplitude of about 70. For added assurance in this Torque reduction should not necessarily be pursued if the operation
operation, the 27/8-in. DP section was not run and 3½-in. DC’s were has adequate drillstring and rotary capacity to handle high torques.
used instead to stiffen the section above the guns and eliminate buck- Where limits are being approached, various measures may be pur-
ling. In addition, the perforating-gun connections were modified and sued to reduce torque or to improve the capacity of the limiting
torqued up to increase their bending rigidity and also enable the con- equipment. Cased-hole torque reduction can be achieved with non-
tingency action of rotating the perforating string if required. The per- rotating DP protectors (DPP). These DPP are ideally run over long
forating string was successfully run to the target depth without rota- intervals where drillstring/casing contact loads are high. Likewise,
tion and no damage was incurred in any of the string sections. open-hole torque reduction can be achieved with subs that involve
The cases described here involve buckling while sliding strings. a nonrotating metal sleeve mounted on bearings. Optimization of
Buckling of rotated strings has also been reported in extreme ERD the well trajectory in terms of both general design and execution can
wells.9 Specifically, in some ERD wells with deep 17½-in. sections, be critical to torque reduction. This will be discussed in the last sec-
drillstring buckling has been detected while rotary drilling with a tri- tion of this paper. Increased mud lubricity and use of lubricants is
cone bit under high WOB. The buckling was detected because of another torque-reduction option. Effective lubricants are very lim-
higher-than-normal torques. To complete the section while avoiding ited when OBM is already being used; however, increasing the oil/
further buckling, a PDC bit (which can drill with a lower WOB) was water ratio (OWR) can improve OBM lubricity. Lubricating beads
Payne Abbassian