Session 21 - Project & Design Case Studies I

doi:10.3723/TXTC9076

A method for predicting the drilling rate of penetration for pile

foundations installed into rock
L. Jones
Independent Consultant, London, UK
R. McLean
Scottish Power Renewables, Glasgow, UK
L. Costa
Scottish Power Renewables, Paris, France

ABSTRACT: Installing pile foundations within rock strata at offshore sites poses a variety of challenges. For
sites where it is necessary to drill into rock to install piles, a key concern is the duration required to drill the
holes in which the piles will be placed, since this can have significant technical and commercial implications.
Utilising recent experience from an offshore wind development, this paper outlines a method for predicting
the drilling rate of penetration (ROP) for a range of rock types and rock strengths. The proposed method is
based on the well-established Mechanical Specific Energy concept and is shown to provide reasonable
agreement with site observations.

1 Introduction installation, equipment changes, ancillary opera-

Piled foundations remain the most commonly uti- tions, etc.) – is derived from ROP as follows:
lised foundation type for fixed-bottom offshore ∆𝑧𝑧𝑧𝑧𝑖𝑖𝑖𝑖
structures. These foundations are relatively well- 𝑡𝑡𝑡𝑡𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = ∑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑=1 (1)
𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖
understood by industry and can provide economic
advantages compared to other foundation options. where ∆𝑧𝑧𝑧𝑧𝑑𝑑𝑑𝑑 = thickness of stratum i, 𝑅𝑅𝑅𝑅𝑊𝑊𝑊𝑊𝑅𝑅𝑅𝑅𝑑𝑑𝑑𝑑 = ROP
Recently many offshore developments have sought within stratum i, i = stratum number, n = total num-
to install piled foundations into rock strata, either by ber of strata.
using driven or drilled and grouted (D&G) installa- The ROP models utilised within the oil and gas
tion methods. For driven piles, installation into rock industry were developed for well drilling. The drill
strata presents an increased risk of pile driving re- hole sizes associated with well drilling are much
fusal, therefore contingency drilling is typically smaller than for pile drilling. Furthermore, well drill-
planned for. Consequently, where piles are to be in- ing normally involves the use of drilling fluids to
stalled into rock, it is necessary to have reliable pre- maintain hole stability and assist with spoil recov-
dictions of drilling durations, since these are directly ery. Conversely, large diameter pile drilling offshore
related to development costs. is typically performed with just seawater present.
Predicting drilling durations has historically been These differences mean that the ROP models used
undertaken within the oil and gas industry by per- within the oil and gas industry are not directly appli-
forming drillability assessments which make use of cable for large diameter pile drilling.
rate of penetration (ROP) models. These ROP mod- The aim of this paper is to outline a method for
els relate pertinent factors – such as ground condi- predicting ROP for large diameter pile drilling in
tions, applied weight on bit (𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊), applied rota- rock. The focus of the paper is drilling performance
tional speed (𝑁𝑁𝑁𝑁), drill torque (𝑇𝑇𝑇𝑇), drill bit type, drill in rock only since this is usually the most critical as-
bit wear, drill bit hydraulics, etc. – to produce esti- pect of drilling, given that drilling in rock is usually
mates of drilling ROP in each stratum. Various ROP slower and generally would lead to greater drill bit
models have been developed (e.g. Maurer, 1962; wear than drilling in soil.
Bingham, 1965; Bourgoyne and Young, 1974; War-
ren, 1987; Pessier and Fear, 1992; Hareland and
Hoberock, 1993; Hareland et al., 2010), many of 2 Background
which attempt to account for these influences explic- 2.1 General
itly.
The drilling duration (𝑡𝑡𝑡𝑡𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 ) – which is the To develop a method for predicting ROP for pile
time taken just to drill (i.e. excluding time for casing drilling in rock, a dataset of 124 D&G pile installa-
tions across 41 locations from an offshore site was

2121
 Innovative Geotechnologies for Energy Transition | The Society for Underwater Technology

collated. This dataset comprised location-specific 3 Proposed method

geotechnical information along with drilling records
3.1 General
and data for each pile. The drill data included ap-
plied and measured drilling parameters, giving an To derive a method for predicting ROP in rock for
accurate record for each drilling operation. pile drilling, the general approach outlined by
Pessier and Fear (1992) was adopted. This relatively
2.2 Site conditions simple approach builds upon the long-established
The site is situated in water depths ranging from concept of mechanical specific energy (𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀), which
29 m to 42 m relative to lowest astronomical tide relates applied drilling parameters to ROP and in-
(LAT). The water depths at the D&G pile locations corporates empirical factors to account for the ef-
ranged from 32 m to 39 m LAT. fects of different drill tools and ground conditions.
The ground conditions at the site typically com- In addition, a hole cleaning limit to ROP is outlined
prise thin (<5 m thick) superficial deposits overlying based on API (2017) recommendations. The follow-
bedrock, with some locally thicker superficial layers ing subsections outline the key features of this ap-
where infilled paleochannels are encountered. Re- proach in greater detail.
sidual soil, derived from the weathering of the bed-
rock, is interpreted to be variably present across the
site, with a thickness ranging from 0 m to 17 m. The 3.2 Specific energy
bedrock geology predominantly comprises interbed- Teale (1965) defined 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 as the amount of energy
ded mudstones, sandstones and siltstones. Doleritic required to excavate a unit volume of rock. 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀
intrusions were also interpreted to be present. Local- (with units of kPa) is defined as follows:
ised occurrences of quartzite, breccia and conglom- 𝑊𝑊𝑊𝑊𝑅𝑅𝑅𝑅𝑊𝑊𝑊𝑊 120𝜋𝜋𝜋𝜋𝜋𝜋𝜋𝜋𝜋𝜋𝜋𝜋
erate rocks were also encountered. 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 = + (2)
𝐴𝐴𝐴𝐴𝑏𝑏𝑏𝑏 𝐴𝐴𝐴𝐴𝑏𝑏𝑏𝑏 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅
The rock is highly variable in strength and weath-
ering. The uniaxial compressive strength (𝜎𝜎𝜎𝜎𝑐𝑐𝑐𝑐𝑑𝑑𝑑𝑑 ) varies where 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 = weight on bit (kN), 𝐴𝐴𝐴𝐴𝑏𝑏𝑏𝑏 = area of drill
from 0.6 MPa (i.e. borderline soil) to >100 MPa. bit (m2), 𝑁𝑁𝑁𝑁 = rotary speed (revolutions per minute;
Considering the median values for each interpreted RPM), 𝑇𝑇𝑇𝑇 = drill torque (kNm) and 𝑅𝑅𝑅𝑅𝑊𝑊𝑊𝑊𝑅𝑅𝑅𝑅 = rate of
rock unit, rock quality designation (RQD) ranges penetration (m/hr).
from 14% to 85% and geological strength index Note that the value of 120 in Equation 2 includes
(GSI) ranges from 34 to 68. a factor of 60 to convert from minutes (𝑁𝑁𝑁𝑁 has units
of RPM) to hours (𝑅𝑅𝑅𝑅𝑊𝑊𝑊𝑊𝑅𝑅𝑅𝑅 has units of m/hr).
2.3 Equipment The concept of 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 was furthered developed by
authors such as Armenta (2008) and Mohan et al.
Due to the ground conditions at the site, it was nec- (2009) by incorporating a hydraulic term into Equa-
essary to have fully-cased drill holes, i.e. temporary tion 2 which accounted for the additional energy re-
casings would be installed along the full length of quired to remove a unit volume of rock. The ap-
the drill hole to maintain hole stability. The tempo- plicability of these concepts for large diameter
rary casings were oscillated into the ground using drilling is unknown, though the formulation pro-
oscillators mounted on a subsea drilling template. posed by Armenta (2008) suggests that the hydraulic
The subsea drilling template also served to ensure component becomes negligible for larger drill diam-
the D&G piles would be installed to required posi- eters. For these reasons, the conventional definition
tional tolerances prior to jacket installation. of 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 was adopted for the current work. This is
Drill holes were excavated using subsea drills. not to say that hydraulic effects are unimportant for
Different drill heads were available, though full-face large diameter drilling – evidently spoil recovery is a
cutter (FFC) drill heads were predominantly used critical concern – just that the contribution of this ef-
and are therefore the focus of this paper. The diame- fect to 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 is believed to be relatively small.
ter of the FFC drill heads was 2.95m, though this
could be increased to 3.15m by engaging under-
reamers. The FFC drill heads could be dressed with 3.3 Coefficient of sliding friction
different cutter types to suit the ground conditions. The main novelty of the Pessier and Fear (1992)
The predominantly used cutter types, which are model is the development of a relationship to calcu-
therefore the focus of this paper, were Mika late torque from applied WOB using a bit coefficient
milltooth rollers (denoted MTR), Mika button rollers of sliding friction (𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 ). This relationship is defined
(MBR), Palmieri bullet rollers (PBR1) and Palmieri as follows:
button rollers (PBR2). 3𝜋𝜋𝜋𝜋
Drill cuttings were removed using a suction 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 = (3)
𝐷𝐷𝐷𝐷𝑏𝑏𝑏𝑏 .𝑊𝑊𝑊𝑊𝑅𝑅𝑅𝑅𝑊𝑊𝑊𝑊
pump. Cuttings were pumped to the deck of the in-
stallation vessel prior to disposal, with sampling where 𝐷𝐷𝐷𝐷𝑏𝑏𝑏𝑏 = diameter of drill bit (m).
possible using a vibrating screen. Some publications state a factor of 36 rather than
3 in the numerator of Equation 3, though this is due

2122
 Session 21 - Project & Design Case Studies I

to converting units of inches (in the 𝐷𝐷𝐷𝐷𝑏𝑏𝑏𝑏 term) to feet where 𝜎𝜎𝜎𝜎𝜎1 = major effective principal stress at fail-
(in the 𝑇𝑇𝑇𝑇 term). The above formulation is stated in its ure, 𝜎𝜎𝜎𝜎𝜎3 = minor effective principal stress at failure,
original form to avoid confusion when metric units 𝑚𝑚𝑚𝑚𝑑𝑑𝑑𝑑 = Hoek-Brown slope constant for intact rock,
are used. 𝐺𝐺𝐺𝐺𝑀𝑀𝑀𝑀𝐺𝐺𝐺𝐺 = geological strength index, 𝐷𝐷𝐷𝐷 = disturbance
For the proposed approach, values of 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 were factor, and 𝑚𝑚𝑚𝑚𝑏𝑏𝑏𝑏 , 𝑠𝑠𝑠𝑠 and 𝑎𝑎𝑎𝑎 are rockmass material con-
empirically derived based on drilling measurements. stants.
The required geotechnical parameters for use in
the above equations were taken as the best estimate
3.4 ROP model design lines derived by the foundation designer us-
Rearranging Equation 3 and substituting into Equa- ing site-specific test data. The value of 𝐷𝐷𝐷𝐷 was equal
tion 2 gives the following expression: to zero, based on the recommendations of Hoek and
1 160𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 𝜋𝜋𝜋𝜋 Brown (2018).
𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 = 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 � + � (4) When determining the confining pressure, 𝜎𝜎𝜎𝜎𝜎3 ,
𝐴𝐴𝐴𝐴𝑏𝑏𝑏𝑏 𝐷𝐷𝐷𝐷𝑏𝑏𝑏𝑏 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅
the effects of overbalanced (i.e. differential pressure,
Furthermore, rearranging Equation 4 gives the ∆p, is positive) or underbalanced (i.e. ∆p is negative)
ROP model: drilling should be considered (e.g. Calhoun and
𝑅𝑅𝑅𝑅𝑊𝑊𝑊𝑊𝑅𝑅𝑅𝑅 =
160𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 𝜋𝜋𝜋𝜋
(5) Ewy, 2005; Shirkavand et al., 2009). The calculated
𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 1
𝐷𝐷𝐷𝐷𝑏𝑏𝑏𝑏 �𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊−𝐴𝐴𝐴𝐴 � ∆p should reflect the conditions at the bottom of the
𝑏𝑏𝑏𝑏
drill hole. Given the equipment and site conditions,
the effect of ∆p was estimated as being relatively
3.5 Drilling efficiency small, particularly given the natural scatter in the
Pessier and Fear (1992) derived 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 for input to rock geotechnical parameters. Therefore, the influ-
Equation 5 by determining empirical values of drill- ence of ∆p was discounted when determining 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 and
ing efficiency (𝑒𝑒𝑒𝑒𝑑𝑑𝑑𝑑 ). It is widely documented that the inclusion of a tension cut-off, as described by
when drilling efficiently, 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 reaches a minimum Hoek and Brown (2018), was not required.
value which corresponds to the rock strength. For In addition, to aid comparison at sites where
offshore drilling, the appropriate rock strength is the rockmass characteristics are unknown, relationships
confined compressive strength (𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 ; units of kPa) as between 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 and 𝜎𝜎𝜎𝜎𝑐𝑐𝑐𝑐𝑑𝑑𝑑𝑑 were also derived.
opposed to 𝜎𝜎𝜎𝜎𝑐𝑐𝑐𝑐𝑑𝑑𝑑𝑑 . Given the above, drilling efficiency
is defined as: 3.7 Hole cleaning limit
𝑆𝑆𝑆𝑆𝑐𝑐𝑐𝑐
𝑒𝑒𝑒𝑒𝑑𝑑𝑑𝑑 = ≤ 1.0 (6) During drilling it is necessary to remove cuttings
𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀
from the face of a drill hole so that the drill bit can
Chen et al. (2018) noted a wide range in assumed efficiently penetrate and cut fresh rock rather than
values of 𝑒𝑒𝑒𝑒𝑑𝑑𝑑𝑑 in published literature. They concluded repeatedly cutting rock which has already been cut.
that 𝑒𝑒𝑒𝑒𝑑𝑑𝑑𝑑 was both bit-specific and formation-specific, Failure to remove cuttings can also lead to a variety
and therefore should be inferred from real drilling of drilling problems and equipment breakdowns,
data. Furthermore, Pessier and Fear (1992) observed which would subsequently lead to extra time and
a factor of 2 difference in drilling efficiency between cost to drill a hole successfully. Ensuring that drill-
new bits and dull bits, with bit and bottom balling ing ROP remains below the rate at which the hole
more problematic for the dull bit. can be effectively cleaned is therefore an important
For the proposed approach, a relationship be- part of executing efficient drilling operations.
tween 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 and 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 was empirically derived. To calculate the hole cleaning limit, a cuttings
concentration approach is adopted based on the rec-
3.6 Rock strength ommendations of API (2017):
𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝐴𝐴𝐴𝐴𝑏𝑏𝑏𝑏
As large diameter drilling is a rockmass problem, 𝐶𝐶𝐶𝐶𝑐𝑐𝑐𝑐 = (11)
𝑄𝑄𝑄𝑄𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡
values of 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 were determined using the Hoek and
Brown (2018) approach, which accounts for both where 𝐶𝐶𝐶𝐶𝑐𝑐𝑐𝑐 = cuttings concentration, 𝑄𝑄𝑄𝑄𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 = pump
rockmass and confinement effects. This approach is flow rate (m3/hour) and 𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = cuttings transport ratio.
summarised as follows: Equation 11 can be rearranged to determine the
𝑎𝑎𝑎𝑎 hole cleaning ROP limit (𝑅𝑅𝑅𝑅𝑊𝑊𝑊𝑊𝑅𝑅𝑅𝑅ℎ𝑐𝑐𝑐𝑐 ) associated with a
𝜎𝜎𝜎𝜎𝜎3
𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 = 𝜎𝜎𝜎𝜎𝜎1 = 𝜎𝜎𝜎𝜎𝜎3 + 𝜎𝜎𝜎𝜎𝑐𝑐𝑐𝑐𝑑𝑑𝑑𝑑 �𝑚𝑚𝑚𝑚𝑏𝑏𝑏𝑏 + 𝑠𝑠𝑠𝑠𝑠 (7) target value of 𝐶𝐶𝐶𝐶𝑐𝑐𝑐𝑐 :
𝜎𝜎𝜎𝜎𝑐𝑐𝑐𝑐𝑖𝑖𝑖𝑖
𝐶𝐶𝐶𝐶𝑐𝑐𝑐𝑐 𝑄𝑄𝑄𝑄𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡
𝑚𝑚𝑚𝑚𝑏𝑏𝑏𝑏 = 𝑚𝑚𝑚𝑚𝑑𝑑𝑑𝑑 𝑒𝑒𝑒𝑒 (𝐺𝐺𝐺𝐺𝑆𝑆𝑆𝑆𝐺𝐺𝐺𝐺−100)⁄(28−14𝐷𝐷𝐷𝐷) (8) 𝑅𝑅𝑅𝑅𝑊𝑊𝑊𝑊𝑅𝑅𝑅𝑅ℎ𝑐𝑐𝑐𝑐 = (1−𝑑𝑑𝑑𝑑)𝐴𝐴𝐴𝐴𝑏𝑏𝑏𝑏
(12)

𝑠𝑠𝑠𝑠 = 𝑒𝑒𝑒𝑒 (𝐺𝐺𝐺𝐺𝑆𝑆𝑆𝑆𝐺𝐺𝐺𝐺−100)⁄(9−3𝐷𝐷𝐷𝐷) (9) API (2017) suggests 𝐶𝐶𝐶𝐶𝑐𝑐𝑐𝑐 should generally not ex-
1 ceed 0.05 (i.e. 5%) to avoid drilling problems,
𝑎𝑎𝑎𝑎 = 0.5 + �𝑒𝑒𝑒𝑒 −𝐺𝐺𝐺𝐺𝑆𝑆𝑆𝑆𝐺𝐺𝐺𝐺⁄15 − 𝑒𝑒𝑒𝑒 −20⁄3 � (10) though experience of large diameter drilling sug-
6

2123
 Innovative Geotechnologies for Energy Transition | The Society for Underwater Technology

gests values as high as 0.1 (i.e. 10%) may be permis- 4 Derivation of ROP model parameters
sible. Values of 𝐶𝐶𝐶𝐶𝑐𝑐𝑐𝑐 are likely to be sensitive to spe-
cific drill head and spoil recovery designs. 4.1 General
In addition, 𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 is defined as: The first step in deriving the ROP model parameters
𝑣𝑣𝑣𝑣𝑝𝑝𝑝𝑝 −𝑣𝑣𝑣𝑣𝑠𝑠𝑠𝑠
𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = (13) used in Equation 5 (i.e. 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 and 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀) was to identify
𝑣𝑣𝑣𝑣𝑝𝑝𝑝𝑝
the combinations of rock type, rock strength and cut-
where 𝑣𝑣𝑣𝑣𝑝𝑝𝑝𝑝 = velocity of drilling fluid (m/s) and 𝑣𝑣𝑣𝑣𝑠𝑠𝑠𝑠 = ter type at each location. After defining these unique
particle slip velocity (m/s). layers, average values of 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 and 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 were deter-
𝑣𝑣𝑣𝑣𝑝𝑝𝑝𝑝 is calculated as follows: mined for each layer and the data were collated into
𝑄𝑄𝑄𝑄𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 a database for analysis of trends. The following sub-
𝑣𝑣𝑣𝑣𝑝𝑝𝑝𝑝 = (14) sections detail the results of the trend analysis and
𝐴𝐴𝐴𝐴𝑟𝑟𝑟𝑟
subsequent derivation of the empirical parameters.
where 𝐴𝐴𝐴𝐴𝑑𝑑𝑑𝑑 = average cuttings return area (m2).
In accordance with API (2017) recommendations
for vertical drill holes, different formulations are 4.2 Coefficient of sliding friction
used to calculate 𝑣𝑣𝑣𝑣𝑠𝑠𝑠𝑠 depending on whether the flow Figure 1 presents derived values of 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 versus 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 for
is turbulent or not. The formulations, adjusted for each rock type and cutter type combination. The re-
the metric units presented hereafter, are given below. sults for all types of bullet and button rollers were
For particle Reynolds number (𝑁𝑁𝑁𝑁𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅,𝑝𝑝𝑝𝑝 ) >100 (i.e. grouped due to their similar design and performance.
turbulent flow): The data presented in Figure 1 may suggest a
𝑣𝑣𝑣𝑣𝑠𝑠𝑠𝑠 = 61.04
𝜏𝜏𝜏𝜏𝑠𝑠𝑠𝑠
(15) weak relationship between 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 and 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 , given higher
𝜌𝜌𝜌𝜌𝑓𝑓𝑓𝑓 0.5 values of 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 are observed at lower values of 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 . This
For 𝑁𝑁𝑁𝑁𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅,𝑝𝑝𝑝𝑝 <100 (i.e. laminar or transitional flow): could be related to bit wear given that lower values
of 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 are achieved in stronger rocks and quartzite,
0.5
𝐷𝐷𝐷𝐷𝑐𝑐𝑐𝑐 𝛾𝛾𝛾𝛾̇ which were more abrasive. However, a clear rela-
𝑣𝑣𝑣𝑣𝑠𝑠𝑠𝑠 = 1.511𝜏𝜏𝜏𝜏𝑠𝑠𝑠𝑠 � 0.5 � (16) tionship between observed cutter wear and 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 could
𝜌𝜌𝜌𝜌𝑓𝑓𝑓𝑓
not be discerned, therefore it was decided to adopt
where 𝜏𝜏𝜏𝜏𝑠𝑠𝑠𝑠 = shear stress due to particle slip (kPa), 𝜌𝜌𝜌𝜌𝑓𝑓𝑓𝑓 average values of 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 which were derived based sole-
= drilling fluid density (Mg/m3), 𝐷𝐷𝐷𝐷𝑐𝑐𝑐𝑐 = diameter of ly on rock type and cutter type combinations.
cuttings (mm), and 𝛾𝛾𝛾𝛾𝛾 = shear rate associated with 𝜏𝜏𝜏𝜏𝑠𝑠𝑠𝑠 Table 1 summarises the derived values of 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 ,
(s-1). which are also presented in Figure 1 as horizontal
For both of the above equations, 𝜏𝜏𝜏𝜏𝑠𝑠𝑠𝑠 is calculated dashed lines.
as follows:
0.5 Table 1. Derived values of coefficient of sliding resistance
𝜏𝜏𝜏𝜏𝑠𝑠𝑠𝑠 = 0.00217�ℎ𝑐𝑐𝑐𝑐 �𝜌𝜌𝜌𝜌𝑠𝑠𝑠𝑠 − 𝜌𝜌𝜌𝜌𝑓𝑓𝑓𝑓 �� (17) Rock type Coefficient of sliding resistance, 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏
where ℎ𝑐𝑐𝑐𝑐 = thickness of cuttings (mm) and 𝜌𝜌𝜌𝜌𝑠𝑠𝑠𝑠 = sol- Milltooth rollers Bullet or button rollers
Sandstone 0.15 0.12
id particle density (Mg/m3). Mudstone 0.17 0.13
Observed values of 𝐷𝐷𝐷𝐷𝑐𝑐𝑐𝑐 typically ranged from Breccia – 0.14
10 mm to 30 mm. Cuttings generally had high sphe- Dolerite 0.23 0.22
ricity, meaning values of ℎ𝑐𝑐𝑐𝑐 were similar to 𝐷𝐷𝐷𝐷𝑐𝑐𝑐𝑐 . Quartzite – 0.08
In addition, 𝛾𝛾𝛾𝛾𝛾 was calculated as follows: Notes:
– = No data
1000𝜏𝜏𝜏𝜏𝑠𝑠𝑠𝑠
𝛾𝛾𝛾𝛾𝛾 = (18)
𝜇𝜇𝜇𝜇𝑎𝑎𝑎𝑎

where 𝜇𝜇𝜇𝜇𝑎𝑎𝑎𝑎 = apparent viscosity (kg/ms), which was 4.3 Mechanical specific energy
assessed to be 0.00185 kg/ms based on the expected Figure 2 presents derived values of 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 versus 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐
salinity, seabed temperature and cuttings concentra- for the same combinations of rock type and cutter
tion. The value of 1000 in the above equation is re- type as presented in Figure 1. The results indicate a
quired to ensure consistency of units (i.e. to convert nonlinear relationship between 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 and 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 , which
𝜏𝜏𝜏𝜏𝑠𝑠𝑠𝑠 from kPa to Pa or kg/ms2). can be represented using a simple power law func-
Finally, the following equation is used to calcu- tion. Given that 𝑒𝑒𝑒𝑒𝑑𝑑𝑑𝑑 ≤ 1.0, 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 must also remain
late 𝑁𝑁𝑁𝑁𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅,𝑝𝑝𝑝𝑝 and determine which formulation of 𝑣𝑣𝑣𝑣𝑠𝑠𝑠𝑠 to greater than 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 , therefore this limit is also required.
utilise to calculate 𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 : Due to small datasets for some rock type and cut-
𝜌𝜌𝜌𝜌𝑓𝑓𝑓𝑓 𝑣𝑣𝑣𝑣𝑠𝑠𝑠𝑠 𝐷𝐷𝐷𝐷𝑐𝑐𝑐𝑐 ter type combinations, it was decided to derive a
𝑁𝑁𝑁𝑁𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅,𝑝𝑝𝑝𝑝 = (19) single correlation between 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 and 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 for all rock
𝜇𝜇𝜇𝜇𝑎𝑎𝑎𝑎
types and cutter types. It is possible that, with further
data, more refined correlations could be determined.
Equation 20 shows the form of the derived relation-
ship.

2124
 Session 21 - Project & Design Case Studies I

(a) (b)
0.4 0.4

0.3 0.3

µb [-]
µb[-]
0.2 0.2

0.1 0.1

0 0 Design
0 10 20 30 40 50 0 10 20 30 40 50
Cutter type:
Sc [MPa] Sc [MPa]
MTR
(c) (d)
0.4 0.4 PBR1
0.3 0.3 PBR2
µb [-]

µb [-]
0.2 0.2 MBR

0.1 0.1 Rock type:

0 0 Sandstone
0 10 20 30 40 50 0 10 20 30 40 50 Mudstone
Sc [MPa] Sc [MPa]
Breccia
(e) (f)
0.4 0.4 Dolerite
0.3 0.3 Quartzite
µb [-]

µb [-]

0.2 0.2

0.1 0.1

0 0
0 10 20 30 40 50 0 10 20 30 40 50
Sc [MPa] Sc [MPa]

Figure 1. Relationship between coefficient of sliding friction and rock strength for different rock type and cutter type combinations:
(a) milltooth rollers in sandstone; (b) bullet and button rollers in sandstone; (c) milltooth rollers in mudstone; (d) button rollers in
mudstone; (e) milltooth rollers in dolerite; (f) bullet and button rollers in breccia, dolerite and quartzite

(a) (b)
400 400

300 300
MSE [MPa]

MSE [MPa]

200 200

100 100 Design correlations:

Low estimate
0 0
0 10 20 30 40 50 0 10 20 30 40 50 Best estimate
Sc [MPa] Sc [MPa] High estimate
(c) (d)
400 400 Cutter type:
300 300 MTR
MSE [MPa]

MSE [MPa]

200 200 PBR1

PBR2
100 100
MBR
0 0
0 10 20 30 40 50 0 10 20 30 40 50 Rock type:
Sc [MPa] Sc [MPa] Sandstone
(e) (f)
400 400 Mudstone
300 300 Breccia
MSE [MPa]

MSE [MPa]

200 200 Dolerite

Quartzite
100 100

0 0
0 10 20 30 40 50 0 10 20 30 40 50
Sc [MPa] Sc [MPa]

Figure 2. Relationship between mechanical specific energy and rock strength for different rock type and cutter type combinations:
(a) milltooth rollers in sandstone; (b) bullet and button rollers in sandstone; (c) milltooth rollers in mudstone; (d) button rollers in
mudstone; (e) milltooth rollers in dolerite; (f) bullet and button rollers in breccia, dolerite and quartzite

2125
 Innovative Geotechnologies for Energy Transition | The Society for Underwater Technology

400
𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 = 𝑗𝑗𝑗𝑗𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 𝑘𝑘𝑘𝑘 ≥ 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 (20)
350
where 𝑗𝑗𝑗𝑗 and k are empirical coefficients. Note that
this relationship and the resultant values of 𝑗𝑗𝑗𝑗 and k 300
were derived with 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 and 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 in units of MPa ra- 250
ther than the units of kPa used in earlier equations.

MSE [MPa]
An initial regression analysis of the data resulted 200

in predictions of 𝑡𝑡𝑡𝑡𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 being generally lower than 150

actual measurements. This was believed to be due to
100
the possible relationship between 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 and 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 . There-
fore, an alternative curve-fitting approach was im- 50
plemented whereby the data were initially bounded 0
with low estimate and high estimate curves, with an 0 20 40 60 80 100
average of these curves taken as the best estimate. σci [MPa]
This has the effect of slightly biasing towards the Design correlations: Cutter type: Rock type:
lower values of 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 , thereby implicitly correcting for Low estimate MTR Sandstone
mispredictions of 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 at lower values of 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 . Figure 2 Best estimate PBR1 Mudstone
presents the derived curves, which fit the data well. High estimate PBR2 Breccia
MBR
Following the above approach, a correlation be- Dolerite
Quartzite
tween 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 and 𝜎𝜎𝜎𝜎𝑐𝑐𝑐𝑐𝑑𝑑𝑑𝑑 was derived for all rock types
and cutter types (substituting 𝜎𝜎𝜎𝜎𝑐𝑐𝑐𝑐𝑑𝑑𝑑𝑑 for 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 in Equa- Figure 3. Relationship between mechanical specific energy
and uniaxial compressive strength
tion 20). Figure 3 presents the derived correlations.
Table 2 summarises the derived values of 𝑗𝑗𝑗𝑗 and 𝑘𝑘𝑘𝑘.

Table 2. Derived values of empirical coefficients j and k 5 Results and discussion

Case Used with 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 Used with 𝜎𝜎𝜎𝜎𝑐𝑐𝑐𝑐𝑑𝑑𝑑𝑑
j k j k
Low estimate 30 0.25 25 0.25
5.1 Overall method performance
Best estimate 75 0.25 60 0.25 Figure 4 presents a comparison of measured and
High estimate 120 0.25 95 0.25 predicted 𝑡𝑡𝑡𝑡𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 in rock for each pile.

(a) (b) (c)

80 80 80
Predicted Drilling Duration [hours]

Predicted Drilling Duration [hours]

Predicted Drilling Duration [hours]

1:1 1:1 1:1

+30%
60 60 60

-30%
40 40 40

20 20 20

0 0 0
0 20 40 60 80 0 20 40 60 80 0 20 40 60 80
Actual Drilling Duration [hours] Actual Drilling Duration [hours] Actual Drilling Duration [hours]

(d) (e) (f)

80 80 80
Predicted Drilling Duration [hours]

Predicted Drilling Duration [hours]

Predicted Drilling Duration [hours]

1:1 1:1 1:1

+30%
60 60 60

-30%
40 40 40

20 20 20

0 0 0
0 20 40 60 80 0 20 40 60 80 0 20 40 60 80
Actual Drilling Duration [hours] Actual Drilling Duration [hours] Actual Drilling Duration [hours]

Figure 4. Comparison of measured and predicted drilling durations in rock: (a) using best estimate MSE-Sc correlation; (b) using
low estimate MSE-Sc correlation; (c) using high estimate MSE-Sc correlation; (d) using best estimate MSE-σci correlation; (e) using
low estimate MSE-σci correlation; (f) using high estimate MSE-σci correlation

2126
 Session 21 - Project & Design Case Studies I

As shown in Figure 4, overall performance of the until the end of drilling at 26.2 m BSF.
proposed method when using the best estimate The predicted 𝑅𝑅𝑅𝑅𝑊𝑊𝑊𝑊𝑅𝑅𝑅𝑅 was calculated using Equa-
𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀–𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 correlation is reasonable, with 𝑡𝑡𝑡𝑡𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 for tion 5 with the measured drilling parameters and the
more than 95% of piles predicted to within 30% of proposed empirical estimates of 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 and 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀. It can
the actual value. This is a comparable level of accu- be seen that the predicted 𝑅𝑅𝑅𝑅𝑊𝑊𝑊𝑊𝑅𝑅𝑅𝑅 follows the trends in
racy to other geotechnical installation assessments. the measured data, though tends to overpredict be-
In addition, the low estimate and high estimate neath 12.0 m BSF. It should be reiterated that the
𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀–𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 correlations achieve similar levels of relia- site is subject to significant variability in rock
bility when utilised for underestimating and overes- strength, so the misprediction may be the result of an
timating 𝑡𝑡𝑡𝑡𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 , respectively. An alternative ap- inaccurate estimate of 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 at this pile location.
proach to utilising the low estimate and high Also of note in Figure 5 is that the measured 𝑅𝑅𝑅𝑅𝑊𝑊𝑊𝑊𝑅𝑅𝑅𝑅
estimate correlations may be to simply calculate was not able to be sustained above the calculated
±30% around the best estimate prediction, since this 𝑅𝑅𝑅𝑅𝑊𝑊𝑊𝑊𝑅𝑅𝑅𝑅ℎ𝑐𝑐𝑐𝑐 (calculated using measured values) for signif-
would achieve a similar level of reliability. icant depth intervals. Consistent with this, there were
Furthermore, overall performance when utilising no observed spoil recovery difficulties at this loca-
the 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀–𝜎𝜎𝜎𝜎𝑐𝑐𝑐𝑐𝑑𝑑𝑑𝑑 correlations is also reasonable. How- tion. Examples such as this provide some reassur-
ever, because these correlations do not explicitly ance that the selected cuttings concentration ap-
take account of rockmass effects, predictions made proach provides reasonable estimates of 𝑅𝑅𝑅𝑅𝑊𝑊𝑊𝑊𝑅𝑅𝑅𝑅ℎ𝑐𝑐𝑐𝑐 for
on this basis may be less reliable for other sites with large diameter drilling.
different rockmass characteristics. It is anticipated
that the predictions based on the presented 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀–𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 5.3 Example from another site
correlations will be more generally reliable, though Figure 6 presents a comparison of measured and
further data from other sites would be required to predicted ROP for a location at another site where
verify this. monopile relief drilling was conducted using a pile
top drill rig. Predictions were again made using
5.2 Example from study site measured drilling parameters, to ensure compatibil-
Figure 5 presents a comparison of measured and ity. The drill head had a diameter of 5.2 m and was
predicted 𝑅𝑅𝑅𝑅𝑊𝑊𝑊𝑊𝑅𝑅𝑅𝑅 for the pile which had the greatest dressed with milltooth rollers. The ground condi-
misprediction within the database. Predictions are tions were mudstone with an average 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 of 0.8 MPa.
made using measured drilling parameters, to ensure Insufficient information was available for the spoil
compatibility. recovery system, so 𝑅𝑅𝑅𝑅𝑊𝑊𝑊𝑊𝑅𝑅𝑅𝑅ℎ𝑐𝑐𝑐𝑐 could not be estimated.
This location encounters sandstone from It can be seen in Figure 6 that the proposed meth-
1.6 m below seafloor (BSF), with 𝑀𝑀𝑀𝑀𝑐𝑐𝑐𝑐 averaging od performs well and broadly matches the measured
1.7 MPa from 1.6 m to 12.0 m BSF and 9.0 MPa data. Whilst this is encouraging, further case studies
from 12.0 m to 26.2 m BSF. Cutter types were MTR would be required to verify the extent of applicabil-
from the start of drilling to 14.4 m BSF, then PBR2 ity of the proposed method.

ROP [m/hour] ROP [m/hour]

0 1 2 3 4 0 0.2 0.4 0.6 0.8 1
0 0

5 5

10 10
Depth [m BSF]
Depth [m BSF]

15 15

20 20
Measured
Measured
Predicted
25 25
Hole Cleaning Limit
Predicted
Layer boundary
30 30
Figure 5. Comparison of measured and predicted ROP for Figure 6. Comparison of measured and predicted drilling
the pile with the greatest misprediction within the database ROP for monopile relief drilling at alternative site

2127
 Innovative Geotechnologies for Energy Transition | The Society for Underwater Technology

5.4 Limitations 7 References

Whilst the performance of the proposed method ap- API, 2017. API RP 13D Rheology and hydraulics of oil-well
pears reasonable, it is recognised that there are vari- drilling fluids. Washington, DC: American Petroleum Insti-
ous limitations which could be improved upon. A tute (API).
key improvement would be to explicitly incorporate Armenta, M., 2008. Identifying inefficient drilling conditions
a bit wear model into the method since this would using drilling-specific energy. Society of Petroleum Engi-
facilitate more robust planning of drilling operations neers. Proceedings of the Society of Petroleum Engineers
(SPE) Annual Technical Conference and Exhibition, Den-
and more reliable estimates of drillability. This ver, USA, 21–24 September 2008. SPE 116667.
would also ensure a more realistic relationship be- Bingham, M.G., 1965. A new approach to interpreting rock
tween 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 and rock strength, since the proposed drillability. Oil and Gas Journal, Technical Manual reprint,
power law function implies drilling efficiency will p. 93.
always increase as rock strength increases, whereas Bourgoyne, A.T. and Young, F.S., 1974. A multiple regression
approach to optimal drilling and abnormal pressure detec-
all cutter types will have an optimal range of rock tion. Society of Petroleum Engineers Journal, 14(4), pp.
strengths which they are suited for. 371–384.
In addition, the proposed approach does not ex- Calhoun, W. and Ewy, R., 2005. New confined compressive
plicitly consider the effects of drilling hydraulics on strength calculation improves bit selection and bit perfor-
large diameter drilling, either in terms of confining mance. Proceedings of AADE 2005 National Technical
pressure or 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀, since these effects were expected Conference and Exhibition, Wyndam Greenspoint, Hou-
ston, Texas, 5–7 April.
to be relatively small for the cases considered herein. Chen, X., Yang, J. and Gao, D., 2018. Drilling performance
There may be circumstances where such factors are optimisation based on mechanical specific energy technol-
of greater importance, therefore explicit inclusion of ogies. IntechOpen.
these effects may be warranted. Hareland, G. and Hoberock, L.L., 1993. Use of drilling param-
Furthermore, the correlations for 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 are gener- eters to predict in-situ stress bounds. Proceedings of
SPE/IADC Drilling Conference, Amsterdam, Netherlands,
alised for all rock and cutter types, and these correla- 22–25 February, SPE/IADC 25727, pp. 457–471.
tions may also correct for mispredictions of 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 at Hareland, G., Wu, A., James, J.A. and Rashidi, B., 2010. A
low values of rock strength. Further refinement of New Drilling Rate Model for Tricone Bits and Its Applica-
these correlations would seem appropriate, though it tion to Predict Rock Compressive Strength. Proceedings of
should be recognised that the proposed method and the 44th US Rock Mechanics Symposium and 5th US-
presented correlations are empirical, therefore site- Canada Rock Mechanics Symposium, 27–30 June, pp. 167–
171.
specific correlations would be preferable. Hoek, E. and Brown, E.T., 2018. The Hoek-Brown failure cri-
terion and GSI - 2018 edition. Journal of Rock Mechanics
and Geotechnical Engineering, 11(3), pp. 445–463.
6 Conclusions Maurer, W., 1962. The ‘perfect-cleaning’ theory of rotary drill-
ing, Journal of Petroleum Technology, 14(11), pp. 1270–
1274.
The following are concluded: Mohan, K., Adil, F. and Samuel, R., 2009. Tracking Drilling
• A method based on the concept of 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 was Efficiency Using Hydro-Mechanical Specific Energy. Pro-
developed to predict drilling durations when ceedings of SPE/IADC Drilling Conference and Exhibition,
undertaking large diameter drilling in rock; Amsterdam, The Netherlands, 17–19 March, SPE/IADC
• Empirical relationships were established to 119421, pp. 493–504.
estimate 𝜇𝜇𝜇𝜇𝑏𝑏𝑏𝑏 and 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀; Pessier, R.C. and Fear, M.J., 1992. Quantifying common drill-
ing problems with mechanical specific energy and a bit-
• Predicted drilling durations for over 95% of specific coefficient of sliding friction. Proceedings of 67th
pile locations were within 30% of the meas- Annual Technical Conference and Exhibition of the Society
ured values when applying the proposed of Petroleum Engineers, Washington, DC, 4–7 October,
method and derived correlations, which is SPE 24584, pp. 373–388.
comparable accuracy to other geotechnical Shirkavand, F., Hareland, G. and Aadnoy, B.S., 2009. Rock
mechanical modelling for an underbalanced drilling rate of
installation assessments; penetration prediction. Proceedings of the 43rd US Rock
• A modified version of the cuttings concentra- Mechanics Symposium and 4th U.S.-Canada Rock Me-
tion approach recommended by API (2017) chanics Symposium, Asheville, NC, 28 June–1 July.
provided reasonable estimates of the hole Teale, R., 1965. The concept of specific energy in rock drilling.
cleaning limit for large diameter drilling; International Journal of Rock Mechanics and Mining Sci-
• Application of the proposed method and de- ence & Geomechanics Abstracts, 2(1), pp. 57–73.
Warren, T.M., 1987. Penetration Rate Performance of Roller
rived correlations to a different site with dif- Cone Bits, Proceedings of the 59th Annual Technical Con-
ferent ground conditions and equipment indi- ference and Exhibition Houston, Texas. 16–19 September,
cated reasonable predictive performance; pp. 9–18.
• Further data are required to verify the extent
of applicability of the proposed method,
though the general approach appears suitable
for adaptation with site-specific correlations.

2128