Automated Monitoring of Torque and Drag in Real Time

Automated Monitoring of Torque
and Drag in Real-time
Master Thesis
Philipp Zöllner
Mining University Leoben

Department Mineral Resources and Petroleum Engineering
Drilling Engineering
Advisor:
Univ.-Prof. Dipl.-Ing. Dr.mont. Gerhard Thonhauser

(,'(667$77/,&+((5./b581*

,FKHUNOlUHDQ(LGHVVWDWWGDVVLFKGLHYRUOLHJHQGH$UEHLWVHOEVWlQGLJXQGRKQH
IUHPGH+LOIHYHUIDVVWDQGHUHDOVGLHDQJHJHEHQHQ4XHOOHQXQG+LOIVPLWWHOQLFKW
EHQXW]WXQGPLFKDXFKVRQVWNHLQHUXQHUODXEWHQ+LOIVPLWWHOEHGLHQWKDEH

$)),'$9,7

,GHFODUHLQOLHXRIRDWKWKDW,ZURWHWKLVWKHVLVDQGSHUIRUPHGWKHDVVRFLDWHG
UHVHDUFKP\VHOIXVLQJRQO\OLWHUDWXUHFLWHGLQWKLVYROXPH

BBBBBBBBBBBBBBBBBBBBBBBB BBBBBBBBBBBBBBBBBBBBBBBB
'DWH 3KLOLSS=|OOQHU

$&.12/('*(0(17

7KHDXWKRUZRXOGOLNHWRWKDQNWKHIROORZLQJ

8QLY3URI'LSO,QJ'UPRQW*HUKDUG7KRQKDXVHUIRUDGYLVLQJ
DQGVXSSRUWLQJWKLVWKHVLV

209IRUWKHVXSSRUWWKHSURYLGHGVHQVRUGDWDDVZHOODVWKHSRVVLELOLW\
WRWHVWWKHFRQFHSWLQUHDOWLPH

7KHVWDIIRI7'(7KRQKDXVHU'DWD(QJLQHHULQJ*PE+IRUWKHLUKHOSLGHDVDQG
WHFKQLFDODVVLVWDQFHGXULQJWKHZKROHSURMHFW

9HUHQDIRUKHUVXSSRUWDQGSDWLHQFH

Automated Monitoring of Torque and Drag in Real-time
7DEOHRI&RQWHQWV
Abstract .............................................................................................................................................. 4
Kurzfassung ...................................................................................................................................... 5
Real-time Monitoring of Torque and Drag.................................................................................... 7
Introduction...................................................................................................................................................... 7
Current Real-time Monitoring Approaches.................................................................................................... 9
‘On-Line T&D’ – Real-time Friction Factor Monitoring .................................................................................................. 9
Other Company Solutions ........................................................................................................................................... 18
‘PRS Test’ .................................................................................................................................................................... 19
Parameters influencing T&D Design ........................................................................................................... 22
Soft-String Model ......................................................................................................................................................... 22
Stiff-String Model.......................................................................................................................................................... 24
Buckling........................................................................................................................................................................ 26
Fluid Flow Effects......................................................................................................................................................... 28
Tortuosity...................................................................................................................................................................... 28
Conclusion on T&D Model Limitations ........................................................................................................................ 30
T&D Monitoring – A New Approach ............................................................................................ 31

Wellbore Health Monitoring – Concept and Purpose ................................................................................. 32
Phase I – Definition Phase........................................................................................................................... 34
Automated Operations Recognition ............................................................................................................................ 42
Phase II - Proof of Concept.......................................................................................................................... 47
Simulations, Improved ‘TaD_Rule’.............................................................................................................................. 48
Output Options – TaD Tool.......................................................................................................................................... 52
Phase III - Testing and Evaluation ............................................................................................................... 56
Phase IV – Implementation and Feedback................................................................................................. 63
Monitoring while Drilling ............................................................................................................................................... 64
Monitoring while Tripping ............................................................................................................................................. 65
Monitoring of Casing & Completion Runs ................................................................................................................... 67
Additional Feedback .................................................................................................................................................... 68
Conclusion....................................................................................................................................... 72
Appendix.......................................................................................................................................... 74
Friction........................................................................................................................................................... 74
Types of Friction........................................................................................................................................................... 75
Normal Force & Drag................................................................................................................................................... 77
Torque .......................................................................................................................................................................... 80
References....................................................................................................................................... 82
Nomenclature.................................................................................................................................. 84
Philipp Zöllner Page:

7DEOHRI)LJXUHV
Figure 1: Tension comparison ........................................................................................................... 8
Figure 2: Surface hook load curves calculated in the planning phase (for POOH)......................... 9
1
Figure 3: Global friction factor ........................................................................................................ 12
1
Figure 4: Incremental friction factor ............................................................................................... 13
1
Figure 5: Averaged incremental friction factor over depth ............................................................ 14
2
Figure 6: Buckling displayed over the well path (‘Buckling Mode’) .............................................. 15
2
Figure 7: ‘WOB Reserve Until Buckling’ ........................................................................................ 16
4
Figure 8: ‘PRS-Test’ developed for extended reach drilling ......................................................... 19
4
Figure 9: Comparison of ‘PRS Test’ and predicted hook load ..................................................... 20
5
Figure 10: Comparison of stiff sting and soft-string model ........................................................... 25
7
Figure 11: Hole spiraling or micro-tortuosity .................................................................................. 29
Figure 12: Simple plot of measured hook load versus time with one simulated curve ................. 35
Figure 13: Tension comparison – RIH for 5 different WOB curves and buckling limits................ 37
Figure 14: Tension comparison: POOH for a family of FF curves................................................. 37
Figure 15: Torque profile for a family of FF curves ......................................................................... 38
Figure 16: Hook load over depth...................................................................................................... 39
Figure 17: Simulated and actual hook load over depth .................................................................. 39
Figure 18: Hysteresis of hook load over depth................................................................................ 40
Figure 19: ‘Wellbore Aging’ plots ..................................................................................................... 41
Figure 20: First decision tree for plot prototype ............................................................................... 45
Figure 21: Standard simulation model structure ............................................................................. 48
Figure 22: Graphical output of simulated curves based on Excel export ...................................... 49
Figure 23: Final decision tree for the ‘TaD_Rule’............................................................................ 51
Figure 24: Real-time plot of several data channels showing OOS state ....................................... 52
Figure 25: Runs Management Tool................................................................................................. 53
Figure 26: Available hook load operations in TaD .......................................................................... 54

Figure 27: Trend analysis for surface hook load over depth – 12” phase ..................................... 55
Figure 28: Available torque operations in TaD ................................................................................ 55
Figure 29: Trend analysis for surface torque over depth – 17 ½” phase, 4 drilling runs............... 56
Figure 30: POOH and ROB while drilling ........................................................................................ 57
Figure 31: POOH - hook load verification (1) .................................................................................. 59
Figure 34: Hook load intersection of POOH and RIH ..................................................................... 62
Figure 35: Higher hook load while RIH compared to POOH.......................................................... 63
Figure 36: T&D monitoring of a 9 5/8” casing run ........................................................................... 68
Figure 37: Hook load analysis for drilling runs prior to hold up 9 5/8” casing run .......................... 70
Figure 38: Solid body on a horizontal surface with acting forces ................................................... 75
Figure 39: Transition from static to kinetic friction........................................................................... 76
Figure 40: Object on an inclined surface......................................................................................... 77
Figure 41: Lowering the drill pipe (RIH) ........................................................................................... 78
Figure 42: Raising the drill pipe (POOH) ......................................................................................... 79
Figure 43: Torque acting on a segment of pipe during ROB (approximation)............................... 81
Figure 44: Torque acting on a segment of pipe during ROB (accurate evaluation)...................... 81

$EVWUDFW
Torque and drag are two parameters in the well construction process that deserve
special concern, as they are ever-present factors during drilling and tripping
operations. Especially today’s increase in drilling and completing highly inclined and
extended reach wells, often results in situations where these drilling parameters are
pushed to their limits.
Everyone involved in the well construction process needs to be aware of the

challenges resulting from excessive torque and drag. Due to the difficult well paths to
be drilled, stuck and lost pipe situations may be encountered more easily, but need to
be avoided at all costs.
As a consequence, detailed monitoring of torque and drag is a key element in the

successful construction of a planned well path. Although this is already done in an
excessive manner, the important parameters to enable a reduction of lost and hidden
lost time are still taken manually and inconsistently. These parameters that need to be
tracked are hook load, while drilling and running in respectively out of the hole, as well
as torque, during pipe rotation. Trend analyses, based on this manually process, for
wellbore health status evaluation, are still uncommon.
In order to improve the monitoring process, it is essential to make use of the

mudlogging sensor data combined with an automated algorithm, recognizing the
ongoing rig operations. An improved approach of tracking torque and drag in real-
time, as well as the newly developed software application for this purpose, are
described throughout this thesis. In addition, already available monitoring approaches
have been evaluated and discussed, based on their advantages and limitations.
The main principle behind the used technique is a hook load and torque comparison
of actual versus planned (simulated) values. These actual values are calculated for
different operations (drilling, tripping, running casing, etc.) on a stand per stand basis
and plotted over the measured depth of the bit. The resulting trend analysis that can
be performed, allows identification of upcoming critical situations at an early stage.
Based on this information, the drilling crew is able to react immediately by executing
the appropriate counteractions, and expensive lost time situations can be prevented
thereof. In advance, wellbore conditioning operations can be optimized based on the
quality of the wellbore, which is evaluated without interfering ongoing rig operations.
The main goals were to keep this new automated real-time approach as simple as
possible and to focus on visualization methods the office as well as the field personnel
are used to. Wrong assumptions and misinterpretations due to new visualization
techniques had to be avoided any time.

.XU]IDVVXQJ
Zwei Parameter, die während des Bohrprozesses spezielle Aufmerksamkeit
benötigen, sind Torque und Drag. Der Grund hierfür liegt darin, dass diese
Bohrparameter ständig präsent sind, sei es ob gebohrt, verrohrt, ein- oder ausgebaut
wird. Da immer kompliziertere Bohrpfade benötigt werden um eine Lagerstätte zu
erschließen, werden speziell diese beiden Bohrparameter nahe an die Grenzen des
technisch möglichen gebracht.
Es ist daher essentiell, dass das gesamte Team, welches für die erfolgreiche
Abteufung von stark abgelenkten oder sogenannten ’Extended Reach’ Bohrungen
zuständig ist, sich über die Folgen von überschrittenen Werten von Torque und Drag
im Klaren ist. Auf Grund dieser schwierigen Bohrpfade sind Situationen, in welchen
Rohre durch Steckenbleiben verloren gehen, schneller erreicht als bei vertikalen
Bohrungen. Da mit den eben beschriebenen Situationen hohe Kosten verbunden
sind, liegt das Hauptaugenmerk auf ihrer vehementen Vermeidung.
Daraus ergibt sich, dass die Parameter Torque und Drag auf genaue Weise verfolgt
werden müssen. Obwohl dies bereits in umfangreichem Maße durchgeführt wird,
werden die benötigten Werte oft noch händisch ermittelt und manuell in Graphen
eingetragen. Der dadurch entstehende Mehraufwand sowie das notwendige Training
resultieren in unregelmäßigen Arbeitsschritten. Parameter, welche für Torque und
Drag verfolgt werden müssen, sind die Hakenlast während des Bohrens und des Ein-
und Ausbauens sowie das Drehmoment während der Rotation des Meißels. Eine auf
diesen Werten basierende Trendanalyse um die Qualität und den Zustand des
Bohrloches zu ermitteln ist im Kommen aber noch nicht alltäglich.
Um das Verfolgen der eben erwähnten Parameter in Echtzeit zu ermöglichen und so

die Überwachung auf eine höhere Ebene zu bringen, wurde im Rahmen dieser Arbeit
ein Projekt durchgeführt, bei welchem eine Software Applikation zu diesem Zwecke
entwickelt werden sollte. Basierend auf den Echtzeitdaten, welche vom Mudlogger an
der Bohrstelle gemessen werden, sowie der Einbindung einer automatischen
’Operations Recognition’ sollte dies ermöglicht werden. Die genauen
Entwicklungsschritte sowie die Vorteile dieser verbesserten Methode werden in der
Arbeit beschrieben. Des Weiteren wurden bereits vorhandene Methoden evaluiert und
kritisch auf Vor- und Nachteile geprüft.
Das Kernstück der entwickelten Applikation ist die graphische Darstellung von
Hakenlast bzw. Drehmoment aufgetragen über die Meißeltiefe. Bei den
durchgeführten Trendanalysen werden die mittels Simulator errechneten und
geplanten Kurven mit den aktuellen Echtzeitwerten verglichen. Speziell daran ist die
verwendete ’Operations Recognition’, welche es ermöglicht zwischen den einzelnen
unterschiedlichen Operationen (Bohren, Ein- und Ausbau von Gestänge, Verrohen,
etc.) automatisch zu differenzieren.

Die Überlagerung von aktuellen und geplanten Punkten bzw. Kurven ermöglicht es
mögliche kritische Situationen früher zu erkennen. Daraus ergibt sich der Vorteil, dass
die Bohrmannschaft früher reagieren und eventuelle Gegenmaßnahmen setzen kann
um so kostenintensive Folgewirkungen zu vermeiden. Weiters können die
Trendanalysen zur Optimierung sowie Reduzierung der Bohrlochbehandlungszeiten
herangezogen werden. Der dabei andauernde Bohrprozess wird währenddessen
weder beeinflusst noch gestört.
Die wichtigsten Vorgaben lagen darin, die Anwendung so einfach wie möglich zu
halten sowie den Fokus auf eine Visualisierungsmethode zu richten, die für Ingenieure
sowie Bohrpersonal vertrautes Terrain darstellen. Missinterpretationen und falsche
Annahmen auf Grund von mehrdeutigen Darstellungsmöglichkeiten mussten dabei
vermieden werden.

5HDOWLPH0RQLWRULQJRI7RUTXHDQG'UDJ
This chapter discusses current real-time monitoring approaches of torque and drag
(T&D) as well as friction factor (FF). It further gives an overview on the different
possibilities on how to measure and track these parameters and addresses the weak
points of such systems. In addition, a separate chapter deals with the necessary
changes needed in order to modernize T&D software packages which are based on
different friction factor types and drill string models.
,QWURGXFWLRQ
Because today more difficult wells are drilled as compared to the past, most of the
drilling parameters are pushed to their limits during drilling operations. This is
especially the case for highly deviated or extended reach wells where for example
torque, drag, wellbore hydraulics and cuttings transport are of major concern. This
thesis primarily deals with the parameters torque and drag as well as the resulting
friction factor.
These two parameters have a major impact on the planning of a well because with
improper T&D management, the desired well may not be able to be drilled to total
depth. As a consequence, proper T&D planning is compulsory when constructing an
inclined well and current methods of drilling performance optimization need to be
improved. Several possibilities for friction reduction have been developed up to now,
such as non-rotating drill pipe protectors, specialized drill pipes and new mud systems
utilizing lubricants.
In addition to these mechanical methods, a pre-calculation of the expected hook load

and surface torque is performed. The calculations are based on offset well data from
which different expected friction factors and mud weights are used. The results are
then shown on a load versus depth plot which is used as a comparison to the current
hook load and torque that is observed at the rig.
The figures below show two different possibilities of such pre-calculated hook load
curves, so-called ‘tension comparisons’. Both methods are based on an assumed
friction factor evaluated from previous wells in the same geological area. The first
method is an incremental approach where the hook load (respectively torque) is
calculated from bottom to top in incremental steps of 30 or 50 meters (representing a
section or a stand of the drill string).

Figure 1: Tension comparison
Such an analysis is done for the operations ‘Running In Hole’ (RIH) and ‘Pulling Out
Of Hole’ (POOH) as well as for ‘Rotating Off Bottom’ (ROB). In addition, the sinusoidal
and helical buckling curves are calculated to ensure that the proposed well path can
be drilled and completed with the selected drill string equipment. During the planning
of a well, it is necessary to model several drilling parameters such as the trajectory
(inclination, azimuth) as well as drill pipe and BHA components (heavy-weight drill
pipes, drill collars, etc.) in order to size the rig correctly and to optimize the drill string
accordingly. The outcome should assure that the final well design is feasible.
Another possibility is to calculate the surface hook load (respectively torque) for each
depth with the necessary equations stated in the Appendix. The results are again
plotted over depth for different expected friction factors (Figure 2).

Figure 2: Surface hook load curves calculated in the planning phase (for POOH)
Throughout recent years it has become increasingly obvious that T&D are the two
parameters where efforts need to be focused in order to improve drilling performance
and to extend the reach of highly deviated wellbores. Although many sensors are
used at the rig site to measure the process of constructing a well, it is common to
leave these data unused instead of trying to improve operations with it.
The following chapters describe how to enhance T&D measurements with the use of
real-time data. Furthermore, the state-of-the-art methods of real-time T&D monitoring
are presented. Finally a separate chapter deals with the problems and limitations of
such automated drilling systems and T&D modeling in general.
&XUUHQW5HDOWLPH0RQLWRULQJ$SSURDFKHV
¶2Q/LQH7 '·²5HDOWLPH)ULFWLRQ)DFWRU0RQLWRULQJ
Several years ago, a new method of monitoring T&D was introduced to the oil
industry1, 2. In contrast to the ‘old’ method of using only pre-calculated curves, this
system determines two different types of real-time friction factors. The exact method
of how this is accomplished is described throughout this chapter. In general, the
system works by the use of the downhole-measured weight on bit (DWOB) and
torque on bit (DTOB) coupled with surface load measurements.
The results, together with the real-time measured data, are then presented graphically
and numerically on a computer screen. This allows the driller and drilling supervisors

to get direct information on the current wellbore condition as well as enabling them to
predict upcoming critical situations at an earlier stage. The main purpose of being able
to take counteractions as soon as possible is therefore fulfilled, which should allow rig
operations to be performed to the limits of the drill string load capacities.
The method described up to now has been developed as part of a series of

performance enhancement tools and is included in the ‘On-Line T&D’ software
package provided by an oilfield service company. This technique is based on the
calculation of a force and torque equilibrium along the drill string that is placed in a
certain borehole. In addition, all the forces and torques acting on respectively within
the drill string are determined. These calculations are based on the use of measured
surface data and downhole values as input parameters.
For the equilibrium calculation procedure, the drill string needs to be made up of short
elements. These segments, with a length of 10 meters, have calculation points on
each end, so-called ‘knots’. The procedure starts at the bottom of the string with the
lower boundary conditions. As already mentioned, downhole WOB and TOB
information needs to be available. This information is required in order to start
calculations at the bit. With these two parameters, all the required properties can be
determined at each knot. The calculation algorithm is then repeated with a changed
friction factor until the calculated loads at the final (upper-most) knot equal the values
measured at the surface.
In general, this technique works for four operation modes which are Drilling, ROB,
‘Picking Up’ and ‘Slacking Off’.
‘Picking Up’ (PU) and ‘Slacking Off’ (SO) are simply other expressions for the
operations RIH and POOH which are explained in the Appendix. It is important to
note, that during such operations the drill string is just moved in axial direction without
any rotation. The terms ‘Picking Up’ and ‘Slacking Off’ weights are used very
commonly in the oilfield business, as these weights indicate if there is a lot of drag
existing due to bad wellbore condition or improper design of specific parameters. The
greatest importance of these two operations occurs at the beginning of the string
movement. As described in the Appendix, the static friction factor is in most cases
higher than the kinetic one. This means that whenever the string is intended to be
moved, the weight will be at a minimum or at a maximum, depending on the direction.
When the pipe is run into the hole, the higher friction factor will result in less weight for
the hook to carry. In the beginning of a drill string raising process, the static FF will
cause a higher over pull compared to the kinetic one.
During ‘Rotating Off Bottom’, the drill string is just rotated without any movement in
axial direction. As there is no absolute velocity in the vertical direction existing during

such operations, but instead in a rotational direction, the drag due to frictional forces
can be neglected. The consequence of this phenomenon is that in the ideal case, just
pure tension is acting without any over pull or slack off. The ROB hook load curve can
therefore be plotted as a centre line between different FF curves for RIH and POOH.
Especially for drilling operations, DWOB and DTOB information needs to be available.
For RIH, POOH and ROB, the lower boundary conditions can be set to zero as the bit
is off bottom and no forces and moments will be acting on it. Nevertheless, the upper
boundary conditions, the real-time measured sensor data, need to be evaluated for all
operation modes. This information needs to consist of surface torque, hook load, the
rotations per minute of the string and the calculated rate of penetration.
The output of the iteration algorithm described above is the calculation of two different
friction factors:
Global Friction Factor
The global or wellbore friction factor is, as its name indicates, valid for the complete
well. This is essentially the same as described in the Introduction. As there is no need
of real-time data to get a friction factor for a complete section of a well, it is quite easy
to calculate. The simple torque and drag equations can be used in order to back-
calculate the friction factor with the use of historic surface sensor data measured
during already drilled wells.
The real-time method shown in the next figure calculates a new FF valid for the total
hole whenever a new step in the analyzing process starts. This means that with each
new step, the old friction factors are not valid anymore and therefore not required for
further analysis. Any changes in the frictional environment are not over-expressed
because of the fact that the calculated FF is averaged over the total wellbore. As a
consequence the true interaction between the drill string and the borehole wall is
determined.

Figure 3: Global friction factor 1
When using an expected friction factor from offset well data for torque and drag
calculations, the responsible drilling engineer needs to be very careful. Assuming
friction factors can always lead to inaccurate results because the basic equations do
only take the pure mechanical friction into account. This means that the friction factor
is not only dependent on the formations drilled, but also on the rheology of the mud as
well as the condition of the wellbore. Wellbore condition has a major impact on the
actual friction factor, which can be completely different compared to the one assumed
by experience. Factors influencing the ‘health’ of a borehole are the occurrence of key
seats, ledges, wash-outs and cavings. All these parameters together need to be
considered when talking about the actual friction factor, the so-called ‘pseudo friction
factor’, in a well.
Other contributors to the true mechanical friction between borehole wall and drill string
are listed below:
• Stiffness of the tubular components
• Viscous drag of the drilling fluids
• Presence of stabilizers and/or centralizers
• Formation types
• Differential pipe sticking due to pore pressure
• Loss of circulation
• Micro-tortuosity (wellbore spiraling)

Although the global friction factor is easy to back-calculate out of surface measured
data, the determination of the real friction factor between the drill string and borehole
wall is not that simple. As there are no FF-logging methods available, one common
approach is the already mentioned friction factor estimation for the total well. For more
advanced torque and drag calculations, for example when a mass-spring model is
used for drill string modeling, a friction factor for each segment of the wellbore needs
to be available. This type is explained in the next chapter.
Incremental Friction Factor
In contrast to the global friction factor, the incremental one is much more difficult to
calculate. The first step in the algorithm is the determination of a friction factor for the
top hole section (indicated by FF1 in Figure 4) which goes down to the starting point of
the FF-log. When the analysis starts, a friction factor is calculated in incremental
steps. This means that for a certain interval drilled, a distinct friction factor is evaluated
with the algorithm described before. Whenever a new interval is drilled, the previously
determined friction factors are applied for the rest of the hole and a new incremental
one is calculated. In this case the friction factor in the upper part of the borehole
remains constant.
Figure 4: Incremental friction factor 1

Figure 4 indicates that the friction factors calculated with the incremental method are
only valid for one certain small interval of the well. Due to the special calculation
procedure, any change in the frictional environment of upper sections, results in an
over expression of the currently calculated friction factor. As a consequence, the
higher actual friction factor value makes it more obvious for the user that there is a
problem arising. Unfortunately, the incremental method does not show where such a
change in frictional environment originates along the wellbore, which is a big
disadvantage.
Case studies conducted by the international service company developing this software
have shown that it is necessary to average the incremental friction factor over a
certain depth interval. The reason for this is to compensate the problem of data
fluctuations due to drill string vibrations and sensor inaccuracies. The normal interval
for the incremental calculation algorithm is about one meter long as it depends on the
frequency of the downhole data set that is transported to the surface. The problem
with the results is that a compromise between high density FF-log and clear visible
friction factor trends needs to be found.
The way to present the results to the driller and the responsible rig personnel is shown
in Figure 5.
Figure 5: Averaged incremental friction factor over depth 1
For the global friction factor visualization, there is no need of averaging as the sensor
fluctuations do only have a significant influence on the incremental method.

Algorithm for Sliding Drilling
The algorithm for sliding drilling differs from the one in rotating mode in one specific
way. A special method is applied with an own sliding friction factor calculation. During
such operations the iteration process is stopped when the measured and calculated
hook load are equal. This differs from the algorithm used while in rotating mode as the
axial velocity is negligible compared to the velocity in rotational direction. As a
consequence, the iterations are stopped when the measured torque matches the
calculated one.
Additional Applications
There are several other applications included in this real-time monitoring approach
which result from the algorithm described. The two important ones are the so-called
‘Buckling Mode’ and the ‘WOB Reserve Until Buckling’. These are two further
visualization possibilities which should help the driller, together with the friction factor
plot, to go closer to the operating limits.
The first method, the ‘Buckling Mode’, visualizes the current buckling situation of the
drill string. Due to all the calculations mentioned with the friction factor determination,
the complete load and torque distribution on and within the string is known. As Figure
6 indicates, the responsible drilling staff gets a graphical output for every calculation
interval where the occurrence and type of buckling is displayed over the well path.
Figure 6: Buckling displayed over the well path (‘Buckling Mode’) 2

Green areas indicate that there is no tendency of the drill string to buckle, whereas
yellow sections should warn the user of the occurrence of sinusoidal buckling. Helical
buckling is represented with a red color. This graphical illustration should allow the
driller to react when any part of the drill string is likely to be buckled.
The second application, the ‘WOB Reserve Until Buckling’, should allow optimization
of the weight transfer to the bit. This is done by a graphical illustration of the hook load
over the weight on bit (WOB). This plot (Figure 7) is the result of a ‘WOB Reserve
Until Buckling’ calculation and shows the additional WOB that is created when the
hook load is lowered for the well path shown in the previous figure. The curve in this
plot shows that any additional lowering of the hook load below a value of about 125 kN
would create only minimal increase in WOB. Furthermore, the string would start to
buckle or even to lock up.
If the drill string would be rotated, the curve in Figure 7 would be a straight line as the
drag would be missing. In the example used, the rotary speed is zero which means
that the resulting non-rotating graph runs against an asymptote with increasing weight
on bit. The asymptote indicates the lock-up effect if the hook load is decreased too
much.
Figure 7: ‘WOB Reserve Until Buckling’ 2

Conclusions on ‘On-Line T&D’
Generally it needs to be stated, that this approach is one of few real-time software
packages found during the first phase of this thesis, where it was the aim to evaluate
the state of the art in T&D monitoring. Several other real-time tools used to monitor
these two critical parameters simply plot the actual hook load values over the
measured depth of the hole without any additional algorithm or data processing acting
behind the visualization. The ‘On-Line T&D’ tool seems to be a smart tool as it
includes several helpful features but there are also limitations that have to be
addressed.
One example of these limitations is certainly the need for the two downhole channels
(WOB and TOB) for the special calculation procedure. Because these two channels
are a prerequisite for getting calculation points, especially during drilling operations, it
is necessary to measure these parameters to have the total functionality included. In
contrast to surface-measured channels, these two channels will not always be
available as there is not always a tool included in the BHA to measure downhole
values. The absence of these parameters would result in a huge usability limitation of
the tool.
Another point that needs to be discussed is the visualization method of the results.
The ‘WOB Reserve Until Buckling’ and ‘Buckling Mode’ features seem to be very
useful ways to present the calculation results to the drillers and responsible field
personnel. Having upcoming problems while drilling and tripping presented in such a
graphical way can help them to react as early as possible and to conduct correct
counteractions.
Nevertheless, the friction factor visualization method seems to be quite complicated

for personnel that are not used to getting torque and drag calculations presented in
such a graphical way. As a consequence, it could easily lead to misinterpretations.
Especially the need of both friction factor types and the necessary averaging
procedure seems to be a more complicated way compared to hook load versus depth
curves. Generally the user has to decide on his/her own if the FF versus depth plot is
a proper way to get the results visualized. Nevertheless, the user has to know what
trends in friction factor curves mean, especially as the two types differ from each other
and the incremental one has the averaging functionality included.
Another point to be addressed concerns the operations differentiation between RIH

and POOH. The correct definition of these two parameters includes only movement of
the string without any rotation and pumping. Unfortunately, it is not always the case
that a string is simply run in or out of the hole. Concerning ‘On-Line T&D’, it is not clear
how the algorithm works while reaming up and down before making a drilling
connection, while tripping with pumps on, or while wellbore conditioning. There was no
clear statement of how the results are visualized for other operations than drilling
(rotary and sliding) and simple running in and out of the hole. It is possible that during

such operations no points are calculated or that simply no differentiation between the
completely different types is made. As described earlier, friction factors for a rotated
drill string are generally lower compared to ones without rotation. As a consequence, a
clear differentiation has to be made for these two types and visualized in different
ways.
Finally it needs to be mentioned that there was no historical mode for this tool
described in the literature. For evaluation and analysis of previously drilled wells it
would be great to have such a feature included. Planning engineers could use this
functionality to process several wells in the same field as the one to be drilled and use
the results for their planning purpose. It is possible that this feature is included in the
software package but it was not described in the accompanying literature.
2WKHU&RPSDQ\6ROXWLRQV
3
A second real-time T&D monitoring approach described is called ‘Drilltronics’ . The
key element of this model is the same as the one with ‘On-Line T&D’ which is based
on the continuous updating of real-time measured data. This methodology uses
‘Kalman’ filtering techniques which provide a calibrated model that can be compared
to the real-time data and allows quick detection of unwanted occurrences. The same
principle as with the method described before is used to calculate the actual friction
factor in the wellbore. During several operations, downhole weight and torque on bit
are used to start calculations at the bottom. When the actual values for hook load or
torque at the surface match the calculated ones, with an assumed friction factor, the
algorithm is stopped.
T&D monitoring can also be used to optimize completion string installations. Although
it is of special importance during a well that is being drilled to total depth, the possibility
of being not able to run such special tools does exist. Especially in open hole
horizontal wells where additional challenges may come up (e.g.: gravel pack
assemblies) proper T&D management is essential. Furthermore, downhole
completion hardware is not as durable as is drill pipe or drilling BHA hardware. Newly
developed drilling technologies (Rotary Steerable System, Measurement While
Drilling, real-time T&D modeling software, etc.) allow the construction of more
complicated wells where additional loads and stresses are acting on the tubular
assemblies. These forces may not be a problem for the drill string, but need to be
evaluated and understood prior to designing a completion string. Another fact
concerning completion strings is that it is not possible to rotate or float the equipment
to final target. If the string cannot be delivered down to the bottom of the well, the
drilling investment is partially or even entirely compromised.

¶3567HVW·
The ‘PRS Test’ was a test newly developed for the drilling of several extended reach
4
wells in the North Sea . As especially for extended reach drilling projects the
monitoring of friction and its influence on the success of the project are essential,
following test was devised. During the following operations, the crew at the rig had to
manually record the actual surface hook load and torque:
• During tripping in and out at the casing shoe
• After a stand was drilled down
• During picking up the string
• During slacking off the string
• While rotating off bottom
Every operation of the ‘P (Picking Up), R (Rotate), S (Slacking Off) Test’ had to be
performed with and without the pumps running.
Figure 8 is a plot of time versus depth and shows a time log section of 24 minutes
where a ‘PRS Test’ was performed while drilling a well in ‘Captain Field’. The interval
of the test is marked with green lines and is divided into its three parts.
Figure 8: ‘PRS-Test’ developed for extended reach drilling 4

In this special example, the test was performed for about ten minutes after a stand
was drilled down to a depth of 9643 ft. In the beginning of the record, the bit was
moved off bottom, which is indicated by a decrease in bit depth and an increase in the
height of the block. At the end of this ‘Pick Up’ interval, the hook load slightly
increased and decreased again as the block movement was slowed down. The
standpipe pressure stayed constant for the whole interval, whereas the torque channel
showed fluctuations, especially in the end. The next stage of the test was to hold the
string stationary, which resulted in lower hook load and continued fluctuating torque.
During slacking off with rotation, the hook load and torque dropped further as friction
works against the lowering movement. Finally, the rotation of the string was stopped
when moving the pipe. This operation resulted in much higher drag, whereas torque
dropped to zero.
As already mentioned, the test had to be performed during several operations in a

very consistent manner to assure proper T&D monitoring. The next figure shows a
comparison of a T&D prediction for an extended reach well and the actual values
gathered during the ‘PRS Test’. A slight discrepancy can be clearly seen between
these two types of analysis.
Figure 9: Comparison of ‘PRS Test’ and predicted hook load 4

Conclusions on ‘PRS Test’
Several problems arising with such a test are obvious and need to be discussed.
Besides the most obvious factor, the time it takes to conduct the test in a consistent
manner, several other factors can influence the effectiveness of the project:
• Insufficient training of the rig personnel
• No continuity in measuring data at the rig
• Failure of not using measured data
• No clear baseline established
It is clear that making measurements of sufficient quantity as is required for ‘PRS

Tests’ takes a lot of rig time. The time needed to make such a test is about the same
as for a directional survey, but tends to take longer. According to experience, it is
necessary to take PRS readings after every drilled stand, to assure proper outcomes
for T&D monitoring. Unfortunately, the temptation to skip tests or to take them only
after every third stand is very high. There is always a tendency to eliminate time-
consuming measurements to reduce rig time. This may result in data curves where
trends cannot be identified anymore. Incorrect conclusions could be made by the
responsible drilling engineer, based on improper measurement procedures. The
curves and trend comparisons in Figure 9 for example, could lead an inexperienced
engineer to wrong assumptions. The previously-mentioned discrepancy shows lower
hook load values for the PRS readings than for the normal (expected) curves. As
both, the RIH and POOH curves, are lower the discrepancy must have its origin in
improperly calculated simulated curves or a lack of continuity in the conduction of the
‘PRS Tests’.
It is therefore necessary to improve any lack of continuity and make sure that in
practice every rig operation is monitored by trained rig personnel. Communication
failures between different shifts may also lead to difficulties in diagnosing key
problems. The problem of letting different people perform the tests in different
intervals also needs to be discussed. Some crews might even skip tests to have time
for other operations. As a consequence, a clear baseline needs to be established to
assure a perfect result that can be worked with.
The last point that needs to be addressed is the fact that signs of problems are often
just read without any reactions taken. It is a common practice in the petroleum
industry to use sensors for taking many measurements at the rig. Although nearly
every operation is recorded and stored in a database, the available data is not further
used for optimization.
Summarizing all the facts and problems described in the paragraphs above, it seems
that there are too many issues coming up when using the ‘PRS Test’ method to track

T&D. Not only the time it takes to make the tests is a major issue, but there is also the
problem that the field personnel have to take the readings. The accuracy and
continuity of the results depend on the attitude and training of the people working at
the rig site. As it is a time-consuming issue to perform the tests properly, the results
will always suffer from a lack of continuity.
3DUDPHWHUVLQIOXHQFLQJ7 ''HVLJQ
This chapter deals with the limitations of current T&D models that are commonly used
in the planning phase and to predict and prevent critical situations during drilling and
completion. It should further give an overview on the different models and parameters
influencing T&D calculations. According to the literature, T&D software has already
existed for more than 20 years but some confusion still exists over the validity of the
5
models used to characterize drilling operations . The mathematical models behind all
the T&D software packages available, is still the same since its original inception.
Although these models did not change at all, the software user interfaces have
improved with the significant improvements in computer hardware and processor
power during the last few years.
Another section of this chapter will deal with the practical limitations of the existing
T&D models and focuses on the mistakes made by the user due to misinterpretations.
These core drill string models already mentioned are the soft-string and the stiff-string
models and will be discussed in detail.
6RIW6WULQJ0RGHO
The soft-string model is a very basic model and underlies the easy equations stated in
the Appendix (chapter Friction Eq.2 – 13). It is called ‘soft’ because it ignores any
effects caused by tubular stiffness. This means that the model assumes the pipe to be
a heavy cable that is lying along the wellbore which results in the further assumption
that any tension and torque in axial direction is supported by the drill string and all the
contact forces are supported by the wellbore.
As already mentioned, the model underlies the very basic equations which take only
gravity and frictional drag into account. Equation 5 for example states the formula to
calculate the normal force N in a very basic manner. This equation is modified for the
soft-string model to account not only for an inclined surface but also for any change in
azimuth.

As a consequence, N ( FN ) can be calculated the following:
FN = ( FG ⋅ cos θ ⋅ Δφ ⋅ sin θ ) 2 + ( FG ⋅ cos θ ⋅ Δφ + FG ⋅ sin θ ) 2 (Eq.1)
Where Δφ = Change in azimuth [°]
Where N or FN = Normal force [N]
Where FG = Gravitational force [N]
Where θ = Inclination [°]
The basic idea and procedure of the common T&D calculations during planning
phases have already been described in the chapter Introduction. The next paragraph
describes the procedure and assumptions in more detail.
In order to be able to calculate the torques and forces acting on the selected drill
string, the following assumptions have to be made. For calculations the drill string is
assumed to be made up of short elements of 30 or 50 m in length. For each of these
segments, the basic equations of friction are applied where it is possible to calculate
tension, compression and torsion. The only required input data for this calculation
procedure is the detailed BHA information as well as the survey and the mud weight.
This principle starts at the bottom with the necessary boundary conditions WOB and
bit torque. The parameters mentioned are then calculated for each segment from
bottom to top, where finally cumulative values can be calculated for surface hook load
and surface torque.
One limitation with the soft-string model is the missing bending moment in the
calculation process. As it is assumed that the drill string is in continuous contact with
the wellbore over the whole wellbore path, any bending moments as well as hole
clearance effects are neglected.

6WLII6WULQJ0RGHO
In contrast to the soft-string models, the stiff-string models account for additional
factors acting on a drill string situated in the wellbore. Besides the already-mentioned
bending forces, they also consider radial clearance issues. This is done by not
assuming that the string is in contact with the wall over the entire length of the
wellbore. Several further factors are implemented in stiff-string models:
• Stiff tubular forced around curves results in higher side wall forces
• Variation in contact area between string components and the wellbore
• Presence of stabilizers, tool joints and casing centralizers
Including these parameters in T&D analysis should allow the drilling engineer to
produce more realistic and valuable results. The mathematical model behind the stiff-
string approach is much more complicated and complex to solve due to the greater
amount of equations needed to include the additional parameters. Finite difference,
finite element and semi-analytical techniques are commonly used to deal with the
greater variety of numerical methods.
Generally, there are rules of thumb about whether to use a stiff- or a soft-string model
for T&D calculations. The stiff-string approaches are designed to be used when
dealing with following situations:
• Tortuous trajectories
• High dogleg severity
• Stiff tubular components in drill string
• Narrow radial clearances
Although one would expect more precise and better matching results with the stiff-
string models, there is still a high amount of discrepancy between theory and practice.
According to the literature and field tests, the stiff-string models fail under certain
circumstances, especially when talking about hole size and radial clearance effects.
The failure in properly accounting for these effects can lead the responsible engineers
to wrong assumptions leading to problems in the planning phase.
The next figure shows the problem which occurs with the models that do not properly
account for hole clearance effects. Generally, higher friction factors are examined
when comparing casing running with drill pipe tripping in the same hole. This can be

explained by the fact that due to the larger casing diameters the hole clearance is
decreased accordingly.
The figure below shows hook load versus depth for several models conducted for a 9
5/8” casing that was run in a 12 ¼” hole. Each line represents a different radial
clearance for the stiff-string model. In addition, the line for the soft-string model can be
seen in light green color.
Figure 10: Comparison of stiff sting and soft-string model 5

The reason for this comparison was to examine the effects on the stiff-string model
when decreasing the hole size, and therefore the clearance. As one can clearly see,
the stiff-string model with a radial clearance of only 0.0375” still shows a feasible trend
in hook load, meaning that the casing can be run to total depth, although this could
never be accomplished in reality.
Further investigations made in the field showed that severe doglegs in the top hole
region, with a well path of already more than 30° inclination, lead to unrealistic results
in the models.
When summarizing the limitations of the T&D modeling following list of factors must
be conducted:
• Hole size, casing size, clearance
• Inclination
• Casing stiffness
• Tortuosity effects
• Resolution of survey data
All these facts have to be accounted for in reality. Therefore, it is necessary to

understand that friction factors can not be easily re-used for different run types. For
example a friction factor of a drilling run cannot be used for the subsequent casing
that is run into the hole. In addition, the friction factors can not be interchanged
between the described soft- and stiff-string models.
%XFNOLQJ
The term buckling was already mentioned several times throughout this thesis. This
chapter will explain in more detail what the buckling phenomenon is and addresses
the importance of considering buckling already in the T&D modeling phase.
Buckling is the term used whenever a failure mode needs to be characterized that
occurs due to high comprehensive strength. In drilling engineering, buckling occurs
when the drill string is subjected to high loads of compression that cannot be
supported by the material anymore. This results in a structural failure of the tubular
assembly. Following different buckling modes exist:
• Sinusoidal buckling
• Transition buckling
• Helical buckling

The most general mode, the sinusoidal buckling, occurs already when the string
snakes along the wellbore. This causes an increase in contact force between the
tubes and the wellbore. Any further compression can lead the buckled portion to
transit into a helix. This mode is known as helical buckling.
Once a drill string is facing buckling of any type mentioned above, any further weight
released from the hook is not supported by the bit anymore. This happens due to the
increased contact area between the buckled part of the string and the formation.
Instead of transferring the weight down to the bit to optimize ROP, the string is
progressively supported by wellbore friction. Ultimately, this can cause the string to
lock-up in the well, where no weight can be transferred down to the bit anymore as it is
totally consumed by the buckled portion of the string. This may not only result in
stopped drilling operations but also in a stuck pipe situation. If wellbore friction
becomes too high, it is possible that the string cannot be run out of the hole anymore
within the hook load limits of the rig.
Several models and calculations for buckling have been designed during the last
centuries. The equations developed are well known to every drilling engineer who has
to design a drill string for a new project. Generally, every engineer tries to design a
scenario (survey, BHA, drill pipe) where buckling is completely avoided. Unfortunately,
this cannot be achieved all the time as more and more difficult well paths have to be
drilled. The increase in horizontal wells drilled in the last two centuries shows that due
to bringing the material’s properties to its limits, it is not always possible to stay
completely on the safe side. There is a general acceptance in drilling to tolerate
sinusoidal buckling if there is no other possibility. In contrast to this, helical buckling
should be avoided all the time. If even helical buckling is unavoidable, it is necessary
to calculate the additional drag that will act on the post-buckled portion of the string.
Generally, the behavior of a drill string, acting beyond the limits of buckling, needs to
be better understood. Of course failures will occur due to excessive local stresses in
the post-buckled region. Loss of WOB and the high potential of cracks in the steel are
technical issues that come up especially when going beyond the helical buckling limits.
Due to the cyclic stress reversals, resulting from pipe rotation, the fatigue life is
decreased dramatically.
As already stated, the awareness of buckling has to be included in T&D programs in a

more graphical way. This can allow the driller and the engineers to react earlier,
especially when using a program at the rig and in real-time. Drilling and completion
engineers especially would get a good possibility to see under which circumstances
and for how long a string can work in sinusoidal or helical buckling mode.
The buckling visualization method of ‘On-Line T&D’, described on page 15, seems to
be a good possibility to assist the responsible field personnel and planning engineers.
Nevertheless, there is a common problem with the models when trying to simulate
casing runs in highly deviated wells. Due to the high loads, the casing is often floated

to total depth. As a consequence the casing string is nearly weightless. The

advantage of this situation is that frictional drag can be minimized. Unfortunately, most
of the T&D models available indicate that the casing will helically buckle in such
situations. As casing has a high stiffness, buckling resistance would be expected
when floating the string into the hole.
)OXLG)ORZ(IIHFWV
Another area of interest in T&D modeling deals with the effects resulting from fluid
flow. There are some models existing which correctly account for these effects but
they are ignored by the majority.
Of course, fluid flow effects can only play a role during drilling and wellbore
conditioning operations. While circulating mud through the string and the annulus, a
loss of the normal component of the fluid pressure occurs due to the frictional contact
between the fluid and the string. Furthermore, the shear stresses exerted by the fluid
flow on the string have to be taken into account.
These two effects resulting from fluid flow only have to be considered for hook load
calculations as there is nearly no effect on torque. Due to the circulation, the effective
weight of the string can be reduced significantly. Depending on the wellbore
construction elements, the BHA and the flow rate, the size of the so-called uplift is
defined. Especially for coiled tubing and casing drilling operations this effect has to be
considered.
As already stated, only a minor part of available T&D software takes fluid flow effects
into account for model calculation. The absence of fluid flow effects while simulating a
scenario with running pumps can lead to meaningless results. This is especially the
case for operations in an 8 ½” hole.
The effects of fluid flow on torque and drag design clearly show that proper design
and management goes hand in hand with hydraulics monitoring to assist the
engineers and drillers during difficult drilling operations. As there is a great amount of
rheological and fluid flow models available, they should be included in T&D programs.
7RUWXRVLW\
The term tortuosity is used to describe the ‘crookedness’ of an as-drilled well profile. 6
It generally occurs when a wellbore deviates from a straight hole. There are two types
of tortuosity known in the oilfield industry, among which the most common one is the
local dogleg created when using a steerable motor. Due to the regular changes in
drilling mode between rotary and sliding drilling, the occurrence of doglegs is very

likely. The second type is often referred to as ‘micro-tortuosity’ which indicates a

spiraled hole axis instead of a straight one. This is often associated with the use of
short-gauge bits. Unfortunately, this type of tortuosity often results in poor hole quality
and is a major contributor to today’s friction factor problems. Field studies have shown
that hole spiraling can be easily reduced by the use of extended-gauge bits and rotary
steerable systems which decrease the ‘Tortuosity Index’ (T.I.) significantly.
The conventional dogleg is generally quite easy to detect by the use of survey data. In
contrast to that, the recognition of hole spiraling is more difficult. The problem coming
up with micro-tortuosity is that it cannot be detected by conventional ‘Measurement
While Drilling’ (MWD) tools. It can only be seen from image and oriented caliper logs
because with MWD tools, only the inclination and direction of the drift of the tool in the
wellbore is measured and not that of the wellbore itself. An image of micro-tortuosity
can be seen below.
Figure 11: Hole spiraling or micro-tortuosity 7
When using T&D software in the planning phase of a project, there is generally a
tortuosity model applied to the smooth planned survey in order to make the planned
well path to an as-drilled profile. In order to be as realistic as possible with the model,
the Tortuosity Index (a kind of rippling effect) is added to a portion of the borehole.
There is a common approach to ignore the tortuosity effects but to account for it
indirectly by inflating the friction factors.

&RQFOXVLRQRQ7 '0RGHO/LPLWDWLRQV
As stated above, there are several parameters in T&D model design which need
further improvement. Currently there are three main issues leading engineers to
wrong results:
• Hole size and the resulting clearance does not affect results
• Friction factors for drilling operations are used for running casing design
• Differentiation and consideration of static friction effects
Generally it needs to be stated, that many engineers only use inflated friction factors in
order to take all other ‘minor’ effects into account. According to the literature, there are
many parameters available which affect T&D design but it has become a common
practice to adjust for them by using correction factors. Most of these additional
parameters are only a minor contributor in the overall results, so that they need not to
be accounted for.
Nevertheless, there are some effects, as listed above, which may lead users to wrong
assumptions and should therefore be included in more improved T&D software
packages. An improved stiff-string model is one of the issues that would be required
most.

7 '0RQLWRULQJ²$1HZ$SSURDFK
Up to now, only already available T&D monitoring approaches have been described
and compared with each other. Besides the evaluation of several existing techniques,
the main purpose of this thesis was to develop a completely new real-time tool that is
capable of tracking the two critical drilling parameters, namely torque and drag, in a
highly improved way. Therefore, the study of existing systems was necessary in order
to evaluate the state of the art in T&D monitoring prior to development.
The general objective of the system to be developed was to provide a real-time drilling
monitoring solution that allows continuous evaluation of wellbore condition with critical
parameters especially for the drilling of highly deviated and extended reach wells. This
was the main goal to be achieved with the new tool. The following scope of work
outlining the functionality and monitoring of critical parameters was defined in the very
beginning:
• Wellbore Health Monitoring
o Hook load and Torque Monitoring
o FF Monitoring
o Hole Cleaning Monitoring
There was no clear specification in the start-up phase about how the new application
should visualize the results to the drilling personnel. The only stated objective was to
develop the tool in a way that even inexperienced computer and software users can
benefit from the provided results. Therefore, several iteration steps were necessary to
actually find out which method will satisfy the costumers’ wishes most, so that
everybody can benefit from the use of the tool. The first workshops together with
drilling engineers as well as the evaluation phase have shown that the T&D results
should be provided in a way the engineers and field personnel are used to. The
general feedback indicated a need to remain using the value over depth plots instead
of developing completely new visualization methods. To present the results to the user
in a familiar format turned out to be most appropriate.
A clear goal to achieve was to totally automate most of the tool’s functions in a way
that only a minimum of engineering intervention and assistance is required during the
well construction process.

The implementation of the outlined system was split into several phases:
• Phase I, Definition Phase including the development start
• Phase II, Proof of Concept
• Phase III, Testing and Evaluation of the monitoring tool
• Phase IV, Implementation and Feedback
All phases, the development progress as well as the tool itself will be described
throughout the following chapters. Prior to this description, the main purpose and
concept of the new approach will be described in more detail.
:HOOERUH+HDOWK0RQLWRULQJ²&RQFHSWDQG3XUSRVH
The basic idea behind monitoring wellbore health is to reduce lost and hidden lost time
during drilling operations. In the context of this project, these two flat time sources
often resulting from bad hole conditions and/or improper operational procedure, are to
be reduced. Especially today’s increase in the drilling of highly inclined and extended
reach wells often results in increased reaming and washing time. The amount of and
correct way to perform wellbore conditioning, including reaming up/down, washing
up/down and circulation operations, is often totally left to the driller at the break. It is
not yet a common and standardized approach to evaluate how much wellbore
conditioning is actually necessary to achieve a proper hole condition. Tools assisting in
analysis of the condition of a borehole are approaching, but they are infrequently used
as most of them are simply too difficult to use and very sophisticated. Of course,
highly advanced features and special applications need to be available for
performance and planning engineers, but there are often too many add-ons included
that would complicate things for the field personnel if they would have to use these
tools too. As a consequence, there should also be tools available that effectively help
the personnel at the rig site actually drilling the hole. For example, when they can be
made aware that there is a potential to save time by reaming only half as much as
they had been doing in order to avoid getting stuck or encountering other troubles, this
can prove to be a useful way of reducing unnecessary wellbore conditioning time
which positively affects the operator’s goal of operational efficiency. Unfortunately, this
is not yet the case and therefore the driller will always rely on his old proven methods
without reducing the amount of wellbore conditioning just to be faster.
As there is commonly a ‘stick to the old reliable methods’ way of thinking at the rig site,
important parameters are preferably taken manually instead of trusting automated
systems to define best practices for performance improvement. Therefore, a change
in the philosophy needs to be made on how to present results from automated
systems to the crew at the rig. It is very important, once real-time systems are used, to

adjust the workflow and the system in a way that both drilling teams, the one at the rig
site and the other one in the remote operating center, collaborate to manage the
challenges coming up with difficult wells. It is very important to establish a ‘no blame’
environment with continuous training and optimization.
One problem is that important drilling parameters are still recorded manually most of
the time. 8 This is especially the case for hook load while moving and torque while
rotating a tubular assembly. By analysis of these two parameters in real-time, which
are of most concern when drilling inclined and horizontal wells, a huge potential in
reducing lost as well as hidden lost time can be achieved. Nevertheless, this can only
be accomplished with the rigorous use of real-time sensor data measured at the rig
site. The common method of measuring dozens of channels but not using them for
further analysis needs to be changed. One advancing method is the totally automatic
operations recognition which can be run on the real-time sensor data. This automated
process allows continuous identification of the current rig operations (drilling mode,
type of wellbore conditioning, etc.) without any human intervention.
By using the automated system based on real-time drilling data, it is possible to

reduce lost and hidden lost time. How this is achieved can be easily explained by the
following example. As the operations recognition allows continuous identification of the
rigs’ status and ongoing operations, it is possible to get pick up, slack off and rotating
weights as well as torque values during tripping, reaming and drilling. The detection
and automatic identification of these relevant parameters allows real-time analysis
without interfering with the drilling operations or requiring extra time or work force.
Furthermore, when comparing actual torque and hook load values with simulated
(planned) ones, one can easily identify if further wellbore conditioning is necessary or
if it can be avoided. This means that excessive reaming and washing can be
eliminated if the visualized results show that there is already low torque and drag
which may be an indication for a good hole quality. As a consequence, the
unnecessary practice of reaming more to be on the safe side can be corrected and
hidden lost time can be saved. In addition, lost time can be avoided as instant
measures against increasing torque and drag can be taken before critical ranges are
succeeded. The necessary counteractions can be conducted immediately, as there is
no need to manually process the hand taken values to get a graphical illustration of
T&D results.
A more detailed description of the automated operations detection based on real-time

sensor data can be found in the chapter ‘Automated Operations Recognition’.
Some of the main expected benefits from monitoring wellbore health are listed below:
• Possibility to optimize flat times
o Utilize wellbore conditioning efforts in the best possible way

o Define best practice
o Excessive reaming can be avoided
o Immediate counteractions can be taken once a trend towards

critical ranges is observed
• Quick method to evaluate hole condition and degradation thereof
• Possibility to determine reasons for increased friction
• Root cause identification and evaluation of responsive actions
As already stated, the functionality of the tool to be developed had to be based on

real-time operations detection with a highly automated handling of the data, but also to
be very user-friendly with a minimum in human intervention necessary. A very detailed
description of the development steps and the functions can be found throughout the
next chapters.
3KDVH,²'HILQLWLRQ3KDVH
This chapter describes all the definitions and specifications of the tool to be
developed, as well as the start of development. Furthermore, the feasibility of the
approach to monitor wellbore condition in real-time had to be proved. It was necessary
to work through several ideas of how the new real-time drilling monitoring approach
should look like and in which way the results should be visualized to the field
personnel at the rig and to the drilling engineers in the office. Several different
possibilities have been worked out and proved for their applicability in real-time mode.
Generally it needs to be stated, that due to the evaluation study done in advance,
several possibilities and ideas of how to graphically present T&D results were
available. Nevertheless, it was necessary to start at the very beginning, as many
problems arose with the special T&D operations detection which was an absolute
requirement for real-time visualization.
The possibility of having a plot showing continuous values of hook load and/or torque
over depth/time as shown below was not an option. A clear differentiation algorithm for
RIH, POOH and ROB while drilling, tripping and wellbore conditioning for individual
BHA runs, had to be developed. As Figure 12 indicates, the continuous plotting of
values over time can get quite confusing due to a missing operations algorithm. In this
example only the real-time measured values are shown, with one simulated (planned)
curve.

Figure 12: Simple plot of measured hook load versus time with one simulated curve
During the first workshops with drilling engineers as well as drilling supervisors,
several different plots of interest have been defined:
• Tension Comparison / Tension Profile including following curves:
o ‘Running In Hole’ (RIH) with a family of WOB / FF curves (Figure

13)
o ‘Rotating Off Bottom’ (ROB)

o ‘Pulling Out Of Hole’ (POOH) for a family of FF curves (Figure 14)
o Critical buckling curves (Figure 13)
• Torque Profile:
o For a family of FF curves (Figure 15)
o Maximum allowable torque line (make-up torque of weakest

connection)
• Hook load (hkld) and torque (tq) over depth (Figures 16 and 17):
o For a family of FF curves
o Hysteresis (Figure 18)
• Delta hook load (Δhkld) plot (‘Wellbore Aging’) (Figure 19)
• FF over depth
For all of the possible plots listed above an example is shown throughout the next
several pages. Figures 13 to 16 show planned curves of the relevant parameters,
namely hook load and torque, during the operations RIH, POOH and ROB.
It was not in the scope of this project to program a new simulator to develop torque
and drag models that are most realistic and exact. The focus was on giving the user
clear information about the trend of the actual hook load and torque compared to
planned values. This can simply be done by overlaying the expected curves with the
real-time values based on sensor data and an operations recognition algorithm. A
proper illustration method can then allow the user to compare actual versus planned
hook load, torque respectively and to focus on the trend development. Based on a
trend analysis, conclusions on the wellbore health status can be made.
An example for such an actual versus planned trend comparison can be found in
Figure 17.
In order to avoid a continuous plot of values versus depth or time as shown before, it
was necessary to use an operations detection algorithm to be able to focus only on
the three T&D relevant operations (RIH, POOH, ROB). How this was achieved and
implemented will be described in more detail during the next two chapters.

Figure 13: Tension comparison – RIH for 5 different WOB curves and buckling limits
Figure 14: Tension comparison: POOH for a family of FF curves

Figure 15: Torque profile for a family of FF curves
The figure below differs from the three previous ones as it shows the surface hook
load over depth. The method used for the previous plots is an incremental approach
where the hook load (respectively torque) is calculated from bottom to top in
incremental steps of 30 meters.

Figure 16: Hook load over depth
Figure 17: Simulated and actual hook load over depth

For the next two examples, the ‘Hysteresis’ and the ‘Wellbore Aging’ plots, only hand-
drawings of possible real-time curves are shown.
The next plot again shows curves for the actual hook load over a certain depth
interval. In this case a hysteresis without any simulated friction factor curves is used to
present the results. Again, only the operations RIH, ROB and POOH for one certain
BHA run are used. In contrast to the hook load over depth plots shown before, the
sequence of the operations in time is considered. As it can be seen on the figure
below, the plot starts with the RIH line, continues with the ROB and finishes with the
operation POOH.
Figure 18: Hysteresis of hook load over depth
A delta hook load over depth plot (or ǻhkld plot), as shown below, was discussed as a
possibility to get a clear indication about the wellbore health status over time. It allows
focusing on the condition of particular open hole sections. Formations causing
troubles can then be tracked in more detail and open hole times can be optimized.
As the next figures indicate, the delta of the actual hook load values is plotted over
depth. In order to be able to track the condition of the open hole, the time dimension
was added to the X-axis. Discrepancies from the zero base line, highlighted with
different colors, will allow the user to get an indication about hole sections that need
more concern.

Figure 19: ‘Wellbore Aging’ plots
All of the plots discussed up to now have a good potential in assisting the engineers
and field personnel in monitoring T&D in real-time. Nevertheless, the system to be
developed was intended to give clear and easy information about the condition of the
wellbore during the drilling operation, expressed in hook load and torque. In particular,
drillers and drilling supervisors could then directly relate the results to the sensor
information they visually get from their data provider’s screens. The whole package

together, monitoring of T&D and the real-time sensor data over time, could then be
used to react with proper counteractions as early as possible.
As a consequence, the approach of monitoring trends in hook load and torque

seemed to be the most appropriate one for both operating teams. In contrast to all the
other methods, the surface hook load and torque over depth plots, combining actual
and planned values, met the highest approval among the engineers and field
personnel. In comparison to the other approaches, this one has two major
advantages:
• Less real-time simulation efforts required
• Approach does not lead to signatures that are much more complex to
translate into actions especially for the driller at the break (e.g. FF over
depth)
$XWRPDWHG2SHUDWLRQV5HFRJQLWLRQ
As already addressed, the method of choice was to visualize the data in value over
depth plots for trend monitoring. By getting all the real-time data from the rig in a 1 Hz
frequency, it would be very easy to plot the entire values over depth in the raw one
second data. As shown in Figure 12 this can result in a very confusing plot.
As a consequence, it was necessary to find a solution to this problem as trend

analysis would not really be possible with such a plot. In order to get a clearer picture
to see emerging trends, the amount of data points in the plot had to be reduced. The
decision was made that each stand of pipe had to be represented by one data point
instead of thousands of values. The detection of such stands out of the real-time data
was one of the greatest challenges in the whole project as it had to be completely
automatic and very smart. In addition, the particular T&D related operations RIH,
ROB, and POOH had to be detected automatically without any human intervention.
The next paragraphs describe the principle of the automated operations recognition
(OpRec) used for the T&D monitoring approach.
The OpRec is a rule-based system that has been developed for autonomous analysis
of real-time surface sensor data measured by the various data providers at the rig-
site. The system allows evaluation of the sensor data and acquires crucial process
9, 10
information as a basis for further analysis . The common procedure to analyze
drilling operations is still based on conventional activity breakdowns that are reported
in daily morning reports. As these reports are reflecting human observations and
judgment, the accuracy of the results is quite limited. In addition, the coarse level of
detail (only operations with 15 to 30 minutes are reported) together with subjective
coding systems limit any analysis on drilling performance.

Due to all these limitations, the need of a system capable of recognizing drilling
operations such as tripping, making connections, reaming, washing, drilling mode etc.
became more and more obvious. Therefore, the operations recognition was
developed as a system to extend and enhance standard reporting for further analysis.
As already stated, the automated OpRec is based on logical rules applied to several
real-time data channels. For each operation to recognize (e.g. drilling) a set of rules
has to be defined. Each rule depends of a condition part and a conclusion part and
can consist of several other rules in addition to the sensor data channels. This means
that based on several data channels, their values and their interaction, the type of
operation is definite. As an example the WITS channels ‘Bit Depth (meas)’, ‘Hole
Depth (meas)’, ‘Surface RPM‘ and ‘Flow In Pump‘ are necessary for detecting rotary
and sliding drilling. All of these channels need to be available to get a maximum of
information about the drilling mode. Of course the ‘Hole Depth’ channel alone would
be sufficient enough to detect if there is hole made or not, but only the combination of
all of the listed channels allows differentiation between drilling modes and wellbore
conditioning while drilling.
Beside the availability of certain necessary channels, data quality as well as data
frequency play a very big role in the automated operations detection. For the analysis
based on real-time data a frequency of 1 Hz is recommended with an upper limit of
0.2 Hz. With 10 seconds data the recognition accuracy for certain operations starts to
decrease significantly. With data updated in intervals of 30 seconds, which is still
standard for several mud logging companies, several operations can not be detected
at all (e.g. fast slip to slip connections, pipe spinning times etc.). Data quality is crucial
to be on a very high level for any type of analysis based on sensor data. Misaligned
hook load sensors, bad depth tracking and wrong thresholds for calculated channels
are only three issues that have to be addressed very frequently. In order to avoid any
of these issues to influence analysis results, an automatic and manual quality control
check is performed including following steps: 11
• Data Range Check
• Gap Filling
• Outlier Removal
• Noise Reduction
• Logical Checks
Especially the use of different filters (mean, median, conservative smoothing and
wavelet decomposition) to remove outliers from the dataset is of high importance
when working with a large amount of sensor data.

T&D OpRec Rules
As already stated, it was necessary to develop several new rules for the automatic
detection of T&D related operations. The following basic states had to be detected for
the plot prototype:
• RIH, ROB and POOH (T&D operations) on a stand by stand basis

(considering difference between drilling, tripping and running
casing/liner/tubing) Æ ‘TaD_Rule’
• Stands of tubular assembly moved up or down in the hole
• Hook load and torque during T&D operations
• One filtered value of torque and hook load representing a stand of pipe
(including the corresponding stand depth) Æ ‘HxD’ and ‘TaD_Rule’
In order to be able to test the newly developed rules, a dataset of an 18° inclined
3000m well was used. All screenshots below showing real-time data are referring to
this well, named ‘Testwell’ throughout the thesis.
First step in designing rules, beside the selection of an appropriate dataset for testing,
is the design of a decision tree. As such rules are based on logical statements and
interactions between data channels; a decision tree as shown below is a very good
approach in starting the iteration process.
The intention of the ‘TaD_Rule’ was to detect following states: RIH, POOH, and ROB.
During these operations the bit is off always off bottom. As no points while actually
making hole had to be plotted, the first decision in the rule was set to be if the bit is on
bottom or not. From there on, several other channels and logical decisions were
implemented to find out if one of the T&D operations is active or not. A list of these
decisions as well as the necessary sensor channels can be found below the decision
tree.

Figure 20: First decision tree for plot prototype
Based on the decision that the bit is off bottom, the following paths in the tree with
additional parameters (thresholds) are possible:
• Path with string movement (RIH or POOH) – left path
o Thresholds for:
Velocity of string movement
Minimum traveling distance of the string
Minimum time of string movement
o ‘Bit Depth’ (mdbit) slope for differentiation between RIH (negative)

and POOH (positive).
• Path without string movement (ROB) – right path
o Thresholds for:
Velocity of string movement
Minimum time for a ROB operation
Minimum flow rate through the pipe
Minimum surface revolutions of the pipe

Following real-time data channels are necessary for the ‘TaD_Rule’:
• Hole Depth (meas) (mdhole)
• Bit Depth (meas) (mdbit)
• Surface RPM (rpmav)
• Flow In Pump (flowinav)
• Block Position (posblock)
• Surface Hook Load (hkldav)
• Optional backup channels:
o Bit Rotation (bitrpm)
o Total Pump Pressure (prespumpav)
Once all the logical decisions are finished, one of the states RIH, POOH, ROB, NO or
UNKN must exist for the ‘TaD_Rule’. In the case of missing channels or bad data
(flagged with the WITS NULL value ‘-999.25’), the rule will automatically jump into the
unknown (UNKN) state.
The next step was to test this rule in very detail with the use of the ‘Testwell’ dataset.
The operations recognition algorithm was run over the data in historic mode. This was
followed by analysis of the exactness of the results as the thresholds and parameters
had to be fine-tuned. In addition, appropriate buffer sizes and delays for channels and
other rules had to be found. As OpRec rules are based on the interactions between
rules and channels, exact availability times are crucial (e.g. ‘TaD_Rule’ can start once
the rule ‘Rotate’, which decides if the drill string is rotated or not, is finished; due to this
delay the ‘Rotate’ rule needs to be buffered in the system so that it is available when
the ‘TaD_Rule’ needs it for deciding if ROB is active or not).
During this testing phase, appropriate defaults for the variable parameters had to be
found. Generally, all of them can be adjusted, but the standards listed below seem to
be the most appropriate ones based on the dataset used.
• Slope in bit depth to separate ROB from RIH and POOH: 0.0001 m/s
• Min ǻmdbit (A m): 3 meters (minimum delta in ‘Bit Depth’ to activate rule)
• Min time (X sec): 10 seconds (minimum time POOH or RIH has to last)
• Min time (Y sec): 20 seconds (minimum time ROB has to last)
• Min flowinav (B l/min): 30 l/min
• Min rpmav (C rpm): 10 rpm

For the stand detection a user-definable trigger was included which is the ’min
ǻmdbit’. This parameter defines the depth difference in bit depth the string has to
travel in order to trigger the calculation of one median value. If it seems useful to
trigger the calculation not after the traveling of the entire stand, but also if the string
has only traveled 5 meters (about 15 ft), this parameter can easily be adjusted.
As one can see from the values above, a data frequency of 1 Hz is highly
recommended for the T&D operations detection. As this rule depends on actions very
short in time, data update intervals of more than 3 to 4 seconds already highly
decrease the accuracy of the results.
The next step was to develop the two rules calculating the hook load and torque
values for each detected stand with their corresponding bit depth (‘HxD’ rule). Based
on the successful detection of a stand (RIH, POOH or ROB), hook load and torque
respectively are calculated by using a median filter as default. It is optional to use the
mean filter instead of the median. For the corresponding depth, always the middle of a
stand is used as reference value.
In addition to the use of a median filter for outlier elimination, a trimming value had to
be included. This value was implemented to assure that acceleration forces in the
beginning and in the end of a moved stand do not affect the calculated values.
Basically it represents the time in seconds (7 seconds as default) for which all hook
load and torque values are skipped at the beginning and in the end of a stand for the
filter calculation.
Once all these parameters and thresholds had been optimized, the next step was to
specify and develop the interface of the torque and drag visualization software (TaD
tool). This will be described throughout the next chapter.
3KDVH,,3URRIRI&RQFHSW
The second phase of the project was split into the following two parts:
• Specifications of the required features for prototype development
• Field test of T&D rules and prototype in real-time mode
Both phases were run in parallel as the real-time field test allowed continuous testing
of every new feature added to the prototype.
As stated above, the decision was made to focus on the surface hook load and torque
over depth analyses. The specified plot type described in the previous chapter is

based on the concept of plotting actual values for hook load and torque over different
friction factor curves.
6LPXODWLRQV,PSURYHG¶7D'B5XOH·
It is necessary to model the simulated curves for the operations RIH, POOH and ROB
with a torque and drag simulation software. The output of this software needs to be
three separate reports in Excel format containing tables of measured depth, hook load
and torque for five different friction factors. First step was to define a standard
structure for these reports to allow development of a ‘Simulations Loader’. As there
were several different exports delivered in the beginning, a standard had to be set for
each model to assure that the data can be imported into the tool’s database.
This standard export with the defined data structure for the three Excel files is shown
below:
Figure 21: Standard simulation model structure

It is important to mention, that one model consists of three Excel files and is only valid
for one certain BHA, mud weight and wellbore geometry. There is no limit in the
amount of models that can be loaded into the system.
The graphical output of the data exported was set to be the following:
• RIH: blue (lighter with increasing FF)
• POOH: red (lighter with increasing FF)
• ROB: green (bold line to indicate centre)
The next figure shows such simulated curves which have been plotted based on an
export for the ‘Testwell’.
Figure 22: Graphical output of simulated curves based on Excel export
Once the simulations loader was implemented, the first RIH, POOH and ROB were
plotted over the planned friction factor curves.
During the first tests of the rule and the visualization method, it was immediately
obvious that the detection of real ‘Slack Off’ and ‘Pick Up’ weights during drilling
operations is not that easy. As a stand of pipe is drilled down it is an uncommon

operation to move this stand up and down afterwards without any rotation. Therefore,
it was not always possible to detect the specified operations as they were simply not
conducted. Generally the string is moved with rotation after a stand was drilled down.
In order to avoid a time-consuming test (e.g.: ‘PRS Test’) to get T&D values, the
decision was made to include these operations with rotation as a separate state in the
‘TaD_Rule’ for subsequent visualization. There was no intention to introduce a new
standard rig operation after each drilled stand just to get real values for RIH and
POOH. Due to the higher drag during such operations, stuck pipe situations could be
the consequence.
For tripping operations this problem did not occur in the very beginning as the stands
were typically moved without rotation. As it is also possible to ream in or out of the
hole, this type of operation had to be implemented in the ‘TaD_Rule’ as well. The list
below gives an overview on all the operations that can be visualized in the final version
of the TaD prototype. Furthermore, the adopted decision tree for the ‘TaD_Rule’ can
be seen in the next figure.
• POOH - ‘Pulling Out Of Hole’ while drilling a
• POOH Rotate - ‘Pulling Out Of Hole’ with rotation while drilling
• RIH - ‘Running In Hole’ while drilling
• RIH Rotate - ‘Running In Hole’ with rotation while drilling
• POOH Trip - ‘Pulling Out Of Hole’ while tripping b
• POOH Trip Rotate - ‘Pulling Out Of Hole’ with rotation while tripping
• RIH Trip - ‘Running In Hole’ while tripping
• RIH Trip Rotate - ‘Running In Hole’ with rotation while tripping
• ROB - ‘Rotating Off Bottom’ while drilling
• ROB Trip - ‘Rotating Off Bottom’ while tripping
a
Distance of mdhole and mdbit is closer than a certain threshold (default 35 meters)
b
Distance of mdhole and mdbit is larger than a certain threshold (default 35 meters)

'ULOOLQJ
7D'B5XOH <(6 12
0'+ROH±0'%LWP
12
VORSHRIǻPGELW
!PV
<(6 12
USPDY!USP PLQWLPH!VHF
12
<(6
<(6
12 12
PLQǻPGELW
IORZLQDY!OPLQ
!P
12 12
<(6 <(6
12 12
PLQWLPH!VHF USPDY!USP
12 12
<(6 <(6
12 12
PGKROH±PGELW!IW
VORSHRIPGELW!
12 <(6 12 <(6
52% 52%7ULS
PGKROH±PGELW!IW PGKROH±PGELW!IW
12 <(6 12 <(6
5,+ 5,+7ULS 322+ 322+7ULS
7ULPPLQJYDOXH VHFRQGV
USPDY!USP IURPHDFKVLGH
5,+5RWDWH
<(6 5,+7ULS5RWDWH
322+5RWDWH
322+7ULS5RWDWH
Figure 23: Final decision tree for the ‘TaD_Rule’
In addition to the standard median (or mean) calculation for the operations listed
above, the minimum value for RIH and the maximum value for POOH are calculated
by default and can be displayed in the TaD tool.
Finally, also the out of slip (OOS) hook loads were of great interest. For this operation
the hook load after each slip to slip connection is measured for an interval of 20
seconds. From these values, the maximum is used for visualization over depth in the

TaD tool. The figure below shows real-time data for a slip-to-connection (dark grey)
and the subsequent OOS state (light red).
Figure 24: Real-time plot of several data channels showing OOS state
A list with all possible operations for visualization in the TaD tool can be found in the
chapter below. Furthermore, the output options of the newly developed tool are
described.
2XWSXW2SWLRQV²7D'7RRO
For proper field testing, the graphical visualization interface had to be built according
to the ideas and specifications summarized during the first phases. This chapter
describes the output options of the prototype according to the already described and
additional new specifications.
The interface developed can be seen below. It is important to mention that the system
was designed to act as a focal tool for all T&D related monitoring and optimization.
The only requirement to run the tool is a computer with an internet connection. This
makes the tool completely independent of where the user is located, as the data is
access able all time. To do so, the user only has to select the well (scenario) of
interest from the ‘File’ menu. Once a well has been selected, there are three plots
available: the two ‘Analyses Windows’ (for hook load and torque) and the ‘Runs
Management Tool’ (RMT).

Runs Management Tool (RMT)
In order to avoid confusing unclear plots due to too many data points (especially for
long wells), a grouping of the different operations was necessary. The concept of
having T&D values grouped for individual BHA runs seemed to be the most
appropriate solution for monitoring trends. In order to avoid manual interventions by
the user, an additional OpRec rule was developed. This rule was designed in a way to
detect each BHA run (casing, drilling, conditioning, etc.) of a well automatically. For
visualization of the runs, the RMT was implemented in the TaD tool. The tool allows
manual adjustments of runs and its parameters (e.g. start time, duration, type, etc.) as
well as defining new runs manually. Furthermore, it shows the actual hole and bit
depth, which can be tracked in real-time as every 60 seconds a new value for each
channel is added. The RMT for the ‘Testwell’ is shown below:
Figure 25: Runs Management Tool
Hook Load Analysis
In the ‘Hook Load Analysis’ the specified load over depth plots can be generated. In
this view the simulated curves and the actual values can be plotted above each other
for the available BHA runs.
This is done by picking the desired runs and operations in the properties window on
the right side of the ‘Hook Load Analysis’ screen.

The figure below shows the window with the operations that can be visualized for
analysis:
Figure 26: Available hook load operations in TaD
Figure 27 is a plot created with the TaD tool. It shows simulated curves for a 12” hole
phase with a friction factor range of 0.0 to 0.5 for both, RIH and POOH. In addition,
the actual operations RIH and POOH while tripping with a drilling BHA are plotted in
blue and red circles. The plot area is automatically refreshed every several seconds.
Every operation conducted at the rig site is then automatically added to the plot
without any need of human intervention.

Figure 27: Trend analysis for surface hook load over depth – 12” phase
Torque Analysis
The ‘Torque Analysis’ works very similar to the ‘Hook Load Analysis’. One difference
is that instead of hook load, the actual measured values for torque are plotted over the
imported simulation models. As torque cannot be present without drill string rotation,
the differentiation between the operations RIH, POOH and ROB with and without
rotations was not necessary. The next figure shows the available operations for trend
analysis:
Figure 28: Available torque operations in TaD

Figure 29: Trend analysis for surface torque over depth – 17 ½” phase, 4 drilling runs
As one can see in the figure above, different intensities in colors are used for the
values plotted. Values for more recent BHA runs are darker in color than ones further
in the past. This applies for both types of analyses.
In addition to these curves, annotations and formations can be entered into the
system for visualization. This was a highly required feature as T&D mostly depends on
the formation drilled and tripped through. Furthermore, annotations for the highlighting
of critical points or trends can allow different users at different locations to exchange
their interpretations and subsequent decisions. A plot including formations and
annotations can be seen on page 62.
3KDVH,,,7HVWLQJDQG(YDOXDWLRQ
The final part of the tool development consisted of an extensive testing and evaluation
phase. This field test was conducted by two drilling engineers, one software tester and
the responsible software engineer. The well selected for the test was a three months

land drilling operation with suitable inclination (> 20°) and depth (~ 5500 m). Special
focus was placed on the real-time applicability and the T&D results.
In order to assure that the results generated with the new rule and plotted in the TaD
tool are correct, a real-time data plotting software was used.
During the first testing of the prototype by the responsible drilling engineers, the
following screenshot came up:
Figure 30: POOH and ROB while drilling
The actual points for the operations POOH and ROB were plotted over a set of friction
factor curves for the 12” phase. In the figure above a clear trend can be seen for these
two operations towards higher friction factor curves. From 4000 to 4400 meters the
actual points were consistently around a friction factor of about 0.23. As depth
increased, also the friction factor increased to more than 0.4. This trend in getting
more and more overpull after a stand was drilled down, indicated with the TaD tool,
needed to be verified.

Due to badly mismatching values for actual versus simulated POOH and ROB, the
engineers believed that that the rule did not fit the specifications. An extensive test
with the use of the real-time sensor data (sequence of screenshots below) was used
for the purpose of showing that the values are correct.
The same increase as in the TaD tool can be seen in the hook load channel
measured by the mud logger. Starting with values around 225 to 230 tons at a depth
of 4500 m, the hook load increased to more than 260 tons. The red lines in the
screenshots show the value of the median calculation of the OpRec rule developed for
the TaD tool.

Figure 31: POOH - hook load verification (1)



Furthermore, the following behavior of the POOH and RIH curves could be observed
while drilling and tripping in depths below 1500 m:
Figure 34: Hook load intersection of POOH and RIH
As the figure above indicates, the curves for POOH and RIH intersect at a certain
depth below which the hookload while running out is lower compared to the one while
running in hole. As this is in contrast to the physical laws stated in the Appendix of this
thesis and also because this phenomenon did not occur on every rig tracked, further
investigation was necessary. The real-time plot of sensor data shown below, verified
that the calculation is correct and that there must be some problem with the hook load
sensor or measurement process itself.

Figure 35: Higher hook load while RIH compared to POOH
Figure 35 above verifies that the hook load was constantly higher for the two times
where the string was pulled out of hole (red section) compared to where it was run in
hole (blue section).
A possible reason for this phenomenon could be that the hook load is measured at the
dead line. This can result in lower values for POOH due to the lesser amount of drilling
line that is released off the draw works when the block is on top. In contrast to that,
when the block is at the bottom, a maximum of drill line is released off the draw works.
If the sensor, which is attached to the dead line, does not take the weight if this
additional drill line into account, such an incorrect pattern for T&D can result. The
magnitude of this shift in hook load is therefore based on the size and weight of the
cable used. One issue which could not be explained is that the hook load remains
nearly constant while the string is moved. In no case was an expected steady increase
or decrease in hook load due to the weight of the cable observed.
As this problem does not seem to be considered in the measurement process of

some data providers, an additional parameter was included in the T&D rules. This
allows compensating such a phenomenon by simply adding or subtracting a pre-set
value from the calculated one.
3KDVH,9²,PSOHPHQWDWLRQDQG)HHGEDFN
As the intensive field testing showed that the generated results were correct as well as
that the real-time applicability is given, several analyses were made and feedback was
gathered. Furthermore, the tool was finalized and released to a company where it was
used for real-time monitoring of T&D for another two 5000 m wells.
In particular the workflow between the drilling engineers and the field personnel had to
be optimized. During the evaluation and implementation phase, the operations centre

with the drilling (planning and optimization) engineers developed to the main work
station accessing the real-time system. It furthermore turned out to be the focal point
for all relevant information about the status, the history and the planning of the current
operation related to T&D. Basically the engineers in the office have to gather all the
necessary information for the simulation models as well as to identify situations that
could result in upcoming problems during drilling, tripping and casing operations.
Finally, the flow of information had to be defined. Following a particular line of
communications was specified once the user has identified a critical situation:
The reported information would have to contain information about the type of the
emerging situation, including the fact why the operation differs from the normal
process. The situation is then discussed within the drilling team and a reaction to this
critical development needs to be worked out. This reaction is communicated to the
company man by the drilling engineer. Then the user of the TaD tool has the
responsibility to monitor the effects of the countermeasures in real-time. The user
would also be in the ideal position to be responsible for the production of lessons
learned after the effects of the decisions made have been tracked. These lessons
could then be used by the drilling engineer for planning purposes of upcoming wells.
0RQLWRULQJZKLOH'ULOOLQJ
During drilling operations it is important to track the RIH and POOH weight without
rotation. Unfortunately, these operations are rarely done after a stand has been drilled
down as it can lead to critical situations. Generally the best procedure is to move the
string while rotating and pumping (reaming). Nevertheless, to monitor and judge the
downhole situation it would be much better to have the loads from string movement
without rotation available. Only then full friction is acting on the string and the hook
load is on its maximum. Generally it needs to be stated that proper optimization while
drilling can only be done if values for real RIH and POOH are available.
The series of chapters below gives an overview on the problems that can be detected
in the wellbore due to increased friction. Based on the feedback gathered from drilling
engineers, possible reactions and counteractions are mentioned.
Cuttings Bed Build Up
A build up of a cuttings bed in the annulus can lead to severe problems, especially in
highly inclined and horizontal wells. Such an accumulation of cuttings can act on the
drill string in a way that the friction permanently increases. In the worst case this can
result in a stuck pipe situation where the string gets totally locked up in the hole. In

order to avoid such costly trouble time, constant real-time monitoring of trends in
torque and drag is a key element. Proper counter measurements and decisions can
then be made at an early stage.
Differential Sticking
Differential pipe sticking is another phenomenon that increases the friction (pseudo
friction) between drill string and wellbore wall. Due to the pressure difference between
wellbore pressure and formation pressure during overbalanced drilling operations, it
can happen that the string is forced to the wellbore wall. Such a situation would
increase the normal force and therefore the friction.
As stuck pipe situations can develop over hours, it is important to follow the trend of
the out of slip hook loads in real-time. As already mentioned, the OOS is the
maximum hook load value recorded over a time frame of 20 seconds after a slip to
slip connection was made.
Monitoring of these values over measured depth can allow early detection of an
increasing trend and early counter actions can avoid expensive sidetracking.
Wellbore Conditioning
A common, but necessary operation is to ream a section that has been drilled in order
to improve the downhole condition. In order to avoid such time consuming operations
to be conducted in an excessive way, the actual improvement needs to be monitored.
The TaD tool allows real-time tracking of these operations as they are conducted.
At the beginning of an extended wellbore conditioning operation, the hook load is

expected to be quite significantly away from the center model line representing the
hook load during ROB. After every reamed stand (up- or downwards), the measured
hook load should move more and more to the center model line. Once the measured
hook load values are at a level that is already indicating a very smooth downhole
situation, time may be saved by stopping the reaming operation earlier.
0RQLWRULQJZKLOH7ULSSLQJ
Real-time monitoring of T&D is not only important while actually making hole but also
during tripping operations. The reason for this is that vital information about the
wellbore aging and the behavior of trouble zones over time can be gathered. Following
problems related to wellbore aging can be tracked:

Forming Ledges
Ledges and their trend over time can easily be detected while running in the hole with
a BHA. As a stand of pipe is run into the hole, a ledge would suddenly decrease the
hook load for this stand due to the additional resistance acting. Also the minimum
hook load measured during the running of this particular stand (‘RIH Trip min’) would
be much lower compared the average and the normal.
Especially the trend over several trips is of importance when monitoring the forming of
ledges. If the median and the minimum hook load for ‘RIH Trip’ further decrease in
that certain depth over later trips, the problem is getting worse. If the hook load gets
back to normal over later trips, the situation is getting better.
Emerging Key Seats
In contrast to ledges, key seats can be detected while tripping out of hole. This can be
done by monitoring any increase in hook load while pulling the string. The reason for
such an increase in drag when running through a section with a key seat is that the
string can hang up in this critical area.
Especially when depths with a high risk of key seats are known, these sections shall
be closely monitored in real-time. Again, it is important to monitor the development
over time to get information if the problem gets better or worse.
Tight Hole Problems
Problems due to a tight wellbore are quite similar to the ones of key seats. A tight hole
that closes the annular space around a tubular assembly that is located in the wellbore
can also be detected by monitoring trends in hook load. In contrast to the key seats
where the detection is only possible while pulling out of the hole, wellbore instabilities
can also be observed while tripping in. During both operations, the hook load is
expected to diverge from the normal trend. When pulling the string, the hook load will
increase, whereas while tripping in, the hook load will be lower than normal. If such a
behavior can be observed, the particular formation at this depth shall be watched with
more attention. As with the other problems described before, the situation downhole
gets worse if the deviations from normal become larger.
A possible solution for such a tight hole problem is to ream through the particular
section to improve the hole’s quality until the hook load is back to normal (at ROB
level). The improvement can be tracked with the real-time TaD tool and the time frame
for this operation can be optimized.

0RQLWRULQJRI&DVLQJ &RPSOHWLRQ5XQV
Before a casing or completion string is run into the open hole, it is recommended to
analyze the T&D tripping data of previous drilling and conditioning runs. This is
especially the case for highly inclined and horizontal wells where the casing string can
easily be locked up in the hole if the wellbore is in a bad condition. In addition, the
analysis of all available well data can give crucial information about the necessity of
spending additional time on wellbore conditioning before actually running the casing.
With the use of the TaD tool it is very easy to analyze the tripping of BHA’s that were
run through the open hole section to be cased. Importantly, these runs can give
information on zones where the casing run can reach its limits. As a consequence, it is
very important to analyze the trends in measured hook load compared to the
simulated models.
If any inconsistencies are detected, further information from other drilling data can be
gathered (e.g. caliper log) to analyze the reasons. This additional information can
assist in more accurate identification of where the casing run may encounter
problems. With this information at hand, actions can be discussed to improve the
situation in an identified critical zone.
Once the decision is made to run the casing or completion into the hole, TaD allows
real-time monitoring of the run in the same manner as already described in the
chapter Monitoring while Tripping.
The drilling engineers who tested the tool during the evaluation phase gave the
following feedback on the casing monitoring procedure. They recommended running
every casing job optimization directly at the rig site. The reason for this is that while
running casing the interaction between drilling engineers and supervisors needs to be
at a frequent and high level. Furthermore, during running casing, additional factors
need to be accounted for. Such parameters (e.g. partially filled casing string, changing
fluid level in casing, etc.) have a high impact on the simulation models that are loaded
into the TaD tool. In order to be able to adjust the models whenever necessary, close
cooperation with the supervisors is of high importance. Otherwise it may become quite
hard to distinguish changes in friction factor from changes induced by operational
measures. Once real changes in hook load are observed, the casing run can already
be stopped at an early stage. A stuck casing is something that really needs to be
avoided because of the high tubular costs and the lost time. Furthermore, a casing
cannot be rotated which would result in lower drag while moving the joints. As a
consequence, analysis always needs to be made on the RIH (no rotation) curves.
Once sharp decreases off the normal hook load trend are detected, the casing is
under threat to be held up or to get stuck in the hole. Real-time monitoring and early
counteractions are crucial tasks to successfully run a casing to total depth.

$GGLWLRQDO)HHGEDFN
During the implementation phase further feedback was gathered from the responsible
drilling engineers. As the tool was used to monitor T&D for two deep wells (> 5000 m),
the expectations on the feedback were very high. For one well, the tool was
unfortunately implemented five days too late as following example shows:
The whole monitoring system was installed and provided during the logging phase
prior to the casing of the 12 ¼” section. The purpose was to monitor T&D for the
imminent 9 5/8” casing run. As the 12 ¼” open hole section was drilled without any
problems from 3003 to 4430 m MD, no additional time was spent on wellbore
conditioning. Once the casing float shoe passed the 3003 m, circulation was broken
and the string was washed down further. At a depth of 3300 m the string was held up
as the figure below indicates. It took 17 hours to get the casing string beyond this
depth.
Figure 36: T&D monitoring of a 9 5/8” casing run

As one can see in Figure 36, the trend (gradient) in hook load decreased immediately
once the bottom of the casing entered the open hole section. The casing was run
further into the hole but got totally held up in a depth of 3300 m (indicated by the
horizontal series of blue dots at the same depth and the blue circle).
Finally the caliper log was interpreted in order to find the reasons for this casing string
lock up. As there was no obvious concern mentioned by the drillers who drilled and
tripped through this section, the reason for this situation was quite unclear in the
beginning.
The caliper log conducted before the casing was run in the hole, showed a clear
increase of hole size with interbedding ledges at a depth of about 3290 to 3300 m.
The maximum caliper measurement increased from 12 ¼” up to 16” whereas the
minimum caliper showed much smaller and fewer deviations. According to the
responsible drilling engineer, this indicated washouts in an elliptical shape. The rest of
the wellbore seemed to be in a good condition. Furthermore, this critical section had a
5.5° inclination due to the close kick off point at 3200 m. As a consequence the dogleg
severity increased up to 1.7 °/30m. The combination of these conditions (washouts
together with a high building angle) seemed to be the reason for the casing to hold up.
Due to the fact that this critical area with ledges was not detected with ‘normal’
methods, the decision was made to load the data for the whole well into the T&D
database. As the system is also capable of loading the necessary sensor channels in
history mode, even previously drilled wells can be re-processed and analyzed for T&D
related problems. This was done for the well with the trouble zone in the 12 ¼” section
and took only a few hours. If the results shown below would have been available for
the responsible engineers and supervisors, the decision to run the casing might have
been reviewed.

Figure 37: Hook load analysis for drilling runs prior to hold up 9 5/8” casing run
The figure above shows a TaD ‘Hook Load Analysis’ of the three drilling runs prior to
the casing run. For all of them the RIH while tripping curves are displayed (‘RIH Trip’ –
different intensities of blue). In addition, the POOH while drilling (POOH) curve for the
run where this critical interval was drilled through, is shown in orange. Both types of
curves clearly show the presence of some irregularities between 3290 and 3300 m
MD.
As stated before, forming ledges can be detected during running in hole. Once the bit
reaches the critical zone, the hook load for the current moving stand will suddenly
decrease and deviate from the normal trend.
This behavior is indicated with a blue circle in the figure above. Although the situation
seemed to become better, due to the less sudden decrease in hook load over time,
the casing string was held up in the wellbore. Furthermore, the hook load values while
POOH during the drilling of this section showed increased friction (orange circle).
According to the drilling engineers, operations would have been re-designed if this
information would have been available already prior to the casing run.

This is only one of several cases where the TaD tool proved its correctness and
operational applicability. For all of the wells monitored, the time spent on wellbore
conditioning was optimized and critical trends were detected at an early stage.

&RQFOXVLRQ
There are several approaches available in the oil-field industry allowing successful
real-time monitoring of torque and drag. Nevertheless, there are weak points and
limitations which have been addressed throughout the thesis. The three most
important ones are: The visualization method used to present the results to the user,
the absence of an automated algorithm differentiating between the various rig
operations based on real-time sensor data, and the amount of rig time it takes to
actually get the raw values for further calculations.
All three of them have been taken into account during the design of the new software
application. Because the tool had to be developed not only for drilling engineers, but
for all personnel involved in the well construction process, the importance of
developing a straightforward visualization method was very high. In order to prevent
misinterpretations and wrong assumptions about the wellbore health’s status, it was
preferred to stick with proven T&D visualization techniques instead of showing the
same information simply in other, more complex ways.
Besides the surface value over measured depth plot, the automated operations
differentiation is the key element of the newly developed application. The possibility to
get torque and drag values while drilling, tripping (drill pipe, casing, tubing, liner) and
wellbore conditioning differentiated, confirmed the real-time applicability. All this was
achieved by the implementation of an automated routine using real-time sensor data
as input. The main advantage is that there is no human intervention or additional rig
time needed to run the system. As a consequence, the main requirement of avoiding
a time-consuming on-site test was fulfilled. It is important to mention that although the
used concept of tracking torque and drag is straight forward, the developed system is
very complex. The combination of real-time sensor data, torque and drag simulation
models and the automated operations recognition algorithm based on logical rules,
resulted in an extensive software application.
It was very important, that the values visualized with the newly developed tool have
been proved for their correctness. Because of the ongoing tendency in using hand
taken values, manually drawn into a graph, instead of switching over to new
approaches, the tool is examined with care. Drilling engineers currently using the tool
in their daily monitoring routine for drilling and completing a campaign of several
extended reach wells, confirm this tendency. Although they have proven several times
that the results presented with the tool show real values from the driller’s console, the
challenge is not yet over. Even the saved rig time as well as the automatic
visualization is not yet reason enough for drillers and tool pushers to trust an
automated software routine.
Nevertheless, the monitoring of trends in actual versus planned met with optimization
engineer’s approval in order to improve wellbore conditioning operations. Excessive

reaming and washing can be avoided as low deviations from planned torque and drag
may be an indication for a good hole quality. As a consequence, hidden lost time can
be saved.
An additional advantage is that the application is not limited to be run from the rig-site.
The possibility to broadcast the data to the office as well as to real-time operating
centers all over the world, allows drilling engineers to actively monitor trends in torque
and drag. As a consequence, upcoming critical situations can be detected earlier,
allowing faster feedback on necessary counteractions.
The extensive field test has shown, that it is recommended to run the torque and drag
tool in combination with a real-time sensor data plotting software in order to improve
cause studies on shifting trends.

$SSHQGL[
)ULFWLRQ
Friction is a complicated phenomenon that is not fully understood up to now 12. The
origin of this force between two surfaces in contact is based on the molecular
attraction of the atoms at the contact area. As friction is derived from the
electromagnetic forces between those atoms, it is not a fundamental force.
At every surface area where two solid objects are in contact with each other, frictional
resistance is present. This resistance depends on two factors, namely the force that
presses the surfaces together as well as the roughness at the contact area. Because
the pressing force is always perpendicular to the moving direction, respectively friction
resistance, it is called the normal force FN. The roughness is also referred to as the
coefficient of friction or friction factor ( μ ). This parameter is an empirical property of
the contact materials. Therefore, the friction factor depends on the condition and
composition of the contact areas.
Based on this information, the equation for the friction force between two solid
surfaces, also known as Coulomb friction, can be written as following:
FF = μ ⋅ N (Standard Friction Model) (Eq.2)
Where FF or F friction = Force exerted by friction [N]
Where μ = Friction factor [ ]
Where N or FN = Normal force [N]
As one can see, the friction force is independent of the area between the two objects
in contact. Figure 38 shows a solid body on a horizontal surface with all the forces
acting on it. Besides the previously mentioned friction and normal forces, there are
also the applied force F and the gravitational force FG acting.

Figure 38: Solid body on a horizontal surface with acting forces
The following additional equations apply when talking about friction:
FG = m ⋅ g (Eq.3)
FG = N (Eq.4)
Where FG = Gravitational force [N]
Where m = Mass/buoyed weight of solid body [kg]
Where g = Gravitational acceleration [ 9.81 m s 2 ]
As already stated, the frictional force is presumed to be proportional to the coefficient

of friction. However, the amount of force required to move an object that is in rest is
usually greater than the force required to keep it moving at a constant velocity. This
fact requires the definition of three different types of friction with their own different
friction factor.
7\SHVRI)ULFWLRQ
Static Friction
13, 14
Static friction applies if two bodies in contact are at rest relative to each other .
This type of friction force increases with the applied force in order to prevent any
motion up to some certain limit where motion will occur. This threshold of motion due
to the interlocking of the surface irregularities is characterized by the coefficient of

static friction. As it can be seen on Figure 39, the static friction factor is higher than the
kinetic one where the surfaces are in relative motion to each other. Below the
transition point, there is a linear relationship between the applied force F and the
friction resistance, which means that every increase in F results in a counteraction by
Fs with the same magnitude. When the critical point, the threshold of motion, is
reached, the applied force is equal to the maximum static friction.
Figure 39: Transition from static to kinetic friction
Kinetic Friction
Once the applied force is larger than the maximum static force the friction type
changes into kinetic mode. Due to a force F acting on one side of the solid body,
which is now large enough to overcome the resistance, the body starts to move. As
already stated, the kinetic friction factor μk is smaller that the static one. This margin is
indicated with grey color in Figure 39. Due to the smaller kinetic friction factor, it is
easier to keep a body in motion than to bring it out of rest.
The resulting friction resistance is almost constant over a wide range of low speeds
and therefore independent of the velocity. As a consequence, the standard friction
model (Eq.2) applies also for the kinetic mode.
Rolling Friction
The third mode of friction is the rolling type. This mode is associated with the rotational
movement of a circular object along a surface. Generally, the frictional force in rolling
mode is less than that associated with kinetic friction. A typical value for the rolling
friction factor is 0.001.

1RUPDO)RUFH 'UDJ
Up to now, only the case of objects on horizontal surfaces was discussed. In such
situations, the normal force is equal to the weight of the body and perpendicular to the
frictional resistance.
If the object is on an incline, the normal force is not equal to the weight anymore. The
same applies if the applied force acting on the object has components that are
perpendicular to the surface.
The next figure shows a solid body on an inclined surface with all the forces acting on
it in such a situation.
Figure 40: Object on an inclined surface
It is obvious that because of the inclination with a certain angle ș, the normal force
does not equalize with the force that is induced by the mass of the body. As this
weight needs to be separated into its two resulting forces, the normal force is lower
compared to the gravitational force.
Following equations apply if an object slides on an inclined and straight surface:
FN = FG ⋅ sin θ (Eq.5)
FT = FG ⋅ cos θ (Eq.6)
FF = μ ⋅ FG ⋅ sin θ (Eq.7)

Where θ = Inclination [°]
If the component of the gravity force down the incline is equal to the frictional
resistance, then:
m ⋅ g ⋅ cos θ = μ ⋅ m ⋅ g ⋅ sin θ (Eq.8)
cos θ
μ= = cot θ (Eq.9)
sin θ
All these equations are of importance in drilling engineering when talking about torque
and drag, as these parameters can become very critical. Whenever the drill string is
lowered or raised in the wellbore, friction resistance is induced based on the theory
described up to now. The next two figures are similar to Figure 40, but include the
conditions for lowering and raising a section of pipe in an inclined wellbore.
Figure 41: Lowering the drill pipe (RIH)
The picture shows the forces acting on the segment of pipe during the operation
‘Running In Hole’, which means that the drill pipe is lowered into the well. During this
operation the frictional resistance works against the lowering process and reduces
therefore the load carried by the hook. The following equation applies for this
operation with respect to having a straight surface where only mechanical friction is
acting:

FRIH = FG ⋅ cos θ − μ ⋅ FG ⋅ sin θ (Eq.10)
Where FRIH = Force acting on the hook during lowering the drill string [N]
The opposite of lowering the string is the operation ‘Pulling Out Of Hole’, where the
pipe is raised, as shown in the next figure, in an inclined well. Again the friction works
against the desired process as the friction resistance works downwards and applies
additional load that needs to be carried. The equation for this operation only differs
from the previously formula above by the plus sign, as the friction resistance force is
now added to the resulting weight.
FPOOH = FG ⋅ cos θ + μ ⋅ FG ⋅ sin θ (Eq.11)
Where FPOOH = Force acting on the hook during raising the drill string [N]
Figure 42: Raising the drill pipe (POOH)
For all the equations above, the angle of inclination ș has a large impact on the force
F that is exerted on the hook.

Practical examples:
• If a complete vertical well is drilled, which means that there is absolutely no

inclination, then the mechanical friction resistance can be neglected, as
the friction term is zero ( sin 0° = 0 ).
• For wells with inclined sections ( > 0° ), the weight of the pipe in the hole to
carry by the hook will decrease and the frictional resistance force will
increase with higher angles.
• Finally, the case of having horizontal sections (e.g.: extended reach wells),
results in frictional resistance forces only, as the weight of the pipe will be
supported by the formation ( cos 90° = 0 ).
7RUTXH
The concept of torque is based on a force that is applied to a lever, as well as on the
distance from the lever’s base to where this force is acting.
In order to calculate the torque acting on a drill string when it is ‘Rotated Off Bottom’
(ROB), following equation can be used which is based on this torque = ( moment
arm) ⋅ force concept:
d d
Tq = FF ⋅ = μ ⋅ FG ⋅ sin θ ⋅ (Eq.12)
2 2
Where Tq = Torque [ N ⋅ m ], [ ft ⋅ lbf ]
Where d = Pipe diameter [m], [ft]
The force acting in such a situation is the frictional resistance force induced by the
contact of the pipe with the borehole wall. This means that in a theoretically straight
and vertical wellbore, no torque will appear as the string would hang on the hook
without touching the formation at the side.
Figure 43 shows the simplified case of where pipe and wellbore wall are in contact
with each other. Furthermore, the forces and moments acting during such a situation
are displayed.

Figure 43: Torque acting on a segment of pipe during ROB (approximation)
This simplified case assumes that the pipe lies at the lowest point of the inclined
borehole wall while it is rotated off bottom. In contrast to this approximation, the real
case will look like as shown in Figure 44. Due to the rotational movement, the pipe will
always tend to move up the borehole wall in order to balance out the forces. As one
can see in the figure above, there is no force that balances the friction force FF. The
second case below, accounts for this and assumes equilibrium at an offset angle ĳ.
Figure 44: Torque acting on a segment of pipe during ROB (accurate evaluation)
The formula to calculate the torque for this more accurate evaluation is the following:
d
Tq = FG ⋅ sin θ ⋅ sin ϕ ⋅ ( μ = tan ϕ ) (Eq.13)
2
Where ϕ = Offset angle [°]
Based on the formulas stated, the solutions of the two cases differ by a factor of
cos ϕ .

5HIHUHQFHV
1. Reiber, F. et al.: “On-Line Torque & Drag: A Real-Time Drilling Performance
Optimization Tool,” paper SPE 52836 presented at the 1999 SPE/IADC Drilling
Conference, Amsterdam, The Netherlands, 9-11 March.
2. Vos, B.E. and Reiber, F.: “The Benefits of Monitoring Torque & Drag in Real
Time,” paper SPE 62784 presented at the 2000 IADC/SPE Asia Pacific Drilling
Technology, Kuala Lumpur, Malaysia, 11-13 September.
3. Iversen, F.P. et al.: “Monitoring and Control of Drilling Utilizing Continuously

Updated Process Models,“ paper SPE 99207 presented at the 2006 IADC/SPE
Drilling Conference, Miami, Florida, USA, 21-23 February.
4. Rae, G. et al.: “Understanding Torque and Drag: Best Practices and Lessons
Learnt from the Captain Field’s Extended Reach Wells,” paper SPE 91854
presented at the 2005 SPE/IADC Drilling Conference, Amsterdam, The
Netherlands, 23-25 February.
5. Mason, C.J. and Chen D.C.-K.: “Step Changes Needed to Modernize T&D
Software,” paper SPE 104609 presented at the 2007 SPE/IADC Drilling
Conference, Amsterdam, The Netherlands, 20-22 February.
6. Gaynor, T. et al.: “Quantifying Tortuosities by Friction Factor in Torque and Drag

Model,” paper SPE 77617 presented at the 2002 SPE annual Technical
Conference and Exhibition, San Antonio, Texas, USA, 29 September - 2 October.
7. http://www.halliburton.com/public/ss/contents/Brochures/images/
03075_811_b.gif, queried in Oct. 2008
8. Kucs, R. et al.: “Automated Real Time Hookload and Torque Monitoring,” paper
SPE 112565 presented at the 2008 IADC/SPE Drilling Conference, Orlando,
Florida, USA, 4-6 March.
9. Thonhauser, G.: “Using Real-Time Data for Automated Drilling Performance

Analysis,” European Oil and Gas Magazine, Edition 4, 2004, page 170ff.
10. Thonhauser, G. et al.: “Use of Real-Time Rig Sensor Data to Improve Daily
Drilling Reporting, Benchmarking and Planning – A Case Study,” paper SPE
99880 presented at the 2006 SPE Intelligent Energy Conference and Exhibition,
Amsterdam, The Netherlands, 11-13 April.
11. Mathis W. and Thonhauser G.: “Mastering Real-Time Data Quality Control – How
to Measure and Manage the Quality of (Rig) Sensor Data,” paper presented at
the 2007 11th International Conference on Petroleum Data Integration,
Information and Data Management, Amsterdam, The Netherlands, 19-20 April.

12. Physik Für Wissenschaftler und Ingenieure, P.A. Tipler, Elsevier Spektrum
Akademischer Verlag, New York and Basingstoke (2003), Chapters 4 and 5
13. http://hyperphysics.phy-astr.gsu.edu/hbase/frict.html#coe, queried in Dec. 2007
14. http://www.roymech.co.uk/Useful_Tables/Tribology/co_of_frict.htm, queried in

Dec. 2007

1RPHQFODWXUH
T&D Torque and Drag
FF Friction Factor
RIH Running In Hole
POOH Pulling Out Of Hole
ROB Rotating Off Bottom
DWOB Downhole Weight On Bit
DTOB Downhole Torque On Bit
WOB Weight On Bit
TOB Torque On Bit
PU Pick / Picking Up
SO Slack / Slacking Off
BHA Bottom Hole Assembly
MWD Measurement While Drilling
PRS P (Picking Up), R (Rotate), S (Slacking Off)
ROP Rate Of Penetration
OpRec Operations Recognition
WITS Wellsite Information Transfer Specification
TaD Torque and Drag (name of the software and OpRec rule developed)
OOS Out Of Slips
RMT Runs Management Tool
MD Measured Depth

Automated Monitoring of Torque and Drag in Real Time

Uploaded by

Copyright:

Available Formats

Automated Monitoring of Torque and Drag in Real Time

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Automated Monitoring of Torque and Drag in Real Time

Uploaded by

Copyright:

Available Formats

Automated Monitoring of Torque

and Drag in Real-time

Mining University Leoben

T&D Monitoring – A New Approach ............................................................................................ 31

Philipp Zöllner Page: 

Figure 14: Tension comparison: POOH for a family of FF curves................................................. 37

Figure 15: Torque profile for a family of FF curves ......................................................................... 38

Figure 16: Hook load over depth...................................................................................................... 39

Figure 18: Hysteresis of hook load over depth................................................................................ 40

Figure 19: ‘Wellbore Aging’ plots ..................................................................................................... 41

Figure 20: First decision tree for plot prototype ............................................................................... 45

Figure 21: Standard simulation model structure ............................................................................. 48

Figure 23: Final decision tree for the ‘TaD_Rule’............................................................................ 51

Figure 25: Runs Management Tool................................................................................................. 53

Figure 26: Available hook load operations in TaD .......................................................................... 54

Philipp Zöllner Page: 

Figure 28: Available torque operations in TaD ................................................................................ 55

Figure 30: POOH and ROB while drilling ........................................................................................ 57

Figure 31: POOH - hook load verification (1) .................................................................................. 59

Figure 32: POOH - hook load verification (2) .................................................................................. 60

Figure 33: POOH - hook load verification (3) .................................................................................. 61

Figure 34: Hook load intersection of POOH and RIH ..................................................................... 62

Figure 35: Higher hook load while RIH compared to POOH.......................................................... 63

Figure 36: T&D monitoring of a 9 5/8” casing run ........................................................................... 68

Figure 39: Transition from static to kinetic friction........................................................................... 76

Figure 40: Object on an inclined surface......................................................................................... 77

Figure 41: Lowering the drill pipe (RIH) ........................................................................................... 78

Figure 42: Raising the drill pipe (POOH) ......................................................................................... 79

Figure 43: Torque acting on a segment of pipe during ROB (approximation)............................... 81

Philipp Zöllner Page: 

Everyone involved in the well construction process needs to be aware of the

As a consequence, detailed monitoring of torque and drag is a key element in the

In order to improve the monitoring process, it is essential to make use of the

Philipp Zöllner Page: 

Um das Verfolgen der eben erwähnten Parameter in Echtzeit zu ermöglichen und so

Philipp Zöllner Page: 

Philipp Zöllner Page: 

In addition to these mechanical methods, a pre-calculation of the expected hook load

Philipp Zöllner Page: 

Figure 1: Tension comparison

Philipp Zöllner Page: 

Philipp Zöllner Page: 

The method described up to now has been developed as part of a series of

Philipp Zöllner Page: 

Global Friction Factor

Philipp Zöllner Page: 

Figure 3: Global friction factor 1

• Stiffness of the tubular components

• Viscous drag of the drilling fluids

• Presence of stabilizers and/or centralizers

• Differential pipe sticking due to pore pressure

• Micro-tortuosity (wellbore spiraling)

Philipp Zöllner Page: 

Incremental Friction Factor

Figure 4: Incremental friction factor 1

Philipp Zöllner Page: 

Figure 5: Averaged incremental friction factor over depth 1

Philipp Zöllner Page: 

Algorithm for Sliding Drilling

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page:

Philipp Zöllner Page: