Sie sind auf Seite 1von 24

SPE/IADC 166801

Precise Gain and Loss Detection Using a Transient Hydraulic Model of the
Return Flow to the Pit
Eric Cayeux, SPE, Benot Daireaux, IRIS
Copyright 2013, SPE/IADC Middle East Drilling Technology Conference and Exhibition
This paper was prepared for presentation at the SPE/IADC Middle East Drilling Technology Conference and Exhibition held in Dubai, UAE, 79 October 2013.
This paper was selected for presentation by an SPE/IADC program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not
been reviewed by the Society of Petroleum Engineers or the International Association of Drilling Contractors and are subject to correction by the author(s). The material does not necessarily
reflect any position of the Society of Petroleum Engineers or the International Association of Drilling Contractors, its officers, or members. Electronic reproduction, distribution, or storage of any
part of this paper without the written consent of the Society of Petroleum Engineers or the International Association of Drilling Contractors is prohibited. Permission to reproduce in print is
restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE/IADC copyright.

Abstract
Abnormal variations of pit volume are a prime indicator of mud losses and formation fluid influxes. However the
detection of gain and loss at the pit level can be difficult because of the transient behavior of several components in the
hydraulic loop. For instance, during circulation the compressibility of the drilling fluid changes the total volume of mud
needed to fill the well compared to static conditions. Similarly, changes of the drilling fluid temperature modify the volume
of mud due to its thermal expansion. Furthermore, the transport and separation of cuttings also influences volume variations
in the pit. Finally, the retention capacity of some of the mud transport and treatment equipment (e.g., return flow-line,
shakers, sand-trap, degasser, transfer pit) has a direct impact on the active pit level. All of these effects are transient and can
cause substantial variations of the active pit volume that may be interpreted as gain or loss while they actually have natural
origins.
Most of these effects are well known and are dealt with in a pragmatic way by using finger printing between the current
pit volume variation and a reference pattern obtained in similar drilling conditions. Nevertheless, the finger printing method
has its limitations when the current drilling condition variation does not have an obvious reference pattern. In addition, it
necessitates a selection of the reference pattern by an operator and therefore is rather difficult to automate.
This paper describes in detail a transient hydraulic model of the downhole and surface equipment effects on the pit
volume variations. Several use cases are also described including early kick detection with small margins during pump
shutdown and detection of tiny gains during draw-down tests used in an MPD (Managed Pressure Drilling) operation.
Introduction
During a drilling operation, a certain number of unexpected events, related to the flow of drilling fluid in the well, may
happen rather quickly. Examples of such events are formation fluid influx, mud loss to the formation, pipe washout, plugging
of the drill-string or bridging of the annulus (Aldred et al. 2008). To minimize the impact of such incidents on the drilling
operation, it is important to detect and classify them as soon as possible, in order to initiate counter actions that can reduce
the risk of escalation of an abnormal situation. The common factor for the above mentioned complications is a perturbation of
the flow of drilling fluid between the inlet and the outlet of the system composed by the drill-string and the wellbore. This
flow disturbance can be due to a mass imbalance (influx or efflux), or a change of the pressure loss through the system (leak
from a pipe to the annulus, partial or complete obstruction in the drill-string or the annulus). As a consequence, the amount of
fluid in the pit is affected for either a short period of time (pressure loss related event) or in a more permanent way (mass
imbalance) characterized by an abnormal gain or loss of fluid.
There are mainly four methods for detecting fluid gain and loss in a drilling operation:
1. Abnormal variations of active volume: monitoring abnormal pit volume variation is the most common way of
detecting gain and loss problems (Anfinsen and Rommetveit 1992). Especially during run in hole and pull out of
hole, the balance between theoretical and measured trip tank volumes provides an accurate way to observe whether
the well takes the correct amount of fluid. However during drilling, the interpretation of active pit volume variations
may be difficult when the flow-rate in is changed or when the mud pumps are stopped or started because a large
amount of mud is buffered in the return flow lines, shakers, sand traps and other transfer pits. A standard solution for
pit volume variations is to use fingerprinting from a previously observed mud pump stop and start. But each time a
different flow-rate is used or if the duration for starting and stopping the mud pumps is not identical to the

SPE/IADC 166801

2.

3.

4.

comparison sample, it may be difficult to spot a gain or loss problem (Tarab et al. 2013). Additionally, change in the
pit configuration modifies abruptly the total active volume rendering it difficult to read the pit volume variations.
The direct addition of barite, base oil or brine into the active pit may look like a gain. Transfer of mud from the
active pit to another tank can be misinterpreted as a loss. Proper communication with the mud logger can avoid
unnecessary false alarms but these direct additions or transfers of fluids and materials to the active pit make it
difficult to completely automate gain and loss detection based on that sole parameter.
Difference between in and out flow, so called delta-flow (Maus et al. 1979): the comparison of the flow-rate in and
out of the well gives a direct and quick indication of whether there are gains or losses of fluid inside the well.
However, the method necessitates relatively precise measurements of both flow-rates, in and out, in order to obtain
reliable results (Orban et al. 1987, Orban and Zanker 1988, Shafer et al. 1992). In addition, natural discrepancies
between the flow-rate in and out may generate false alarms such as may occur when starting and stopping the mud
pumps or when axially moving the drill-string. The variations between flow-rate in and out may also be induced by
heave movement on a floater as the result of the natural variations of annular volume due to the expansion and
retraction of the slip-joint (Speers and Gehrig 1987). The flow-rate measurements are often quite noisy and subject
to calibration problems. Several filtering methods have been used to extract more reliable parameters including
simple low-pass signal filter (Speers and Gehrig 1987), cumulative sum (CUSUM) method using the Hinkley page
algorithm (McCann et al. 1991) or a method based on Bayesian probability calculations (Hargreaves et al. 2001).
Since the delta-flow method does not consider the dynamic hydraulic processes occurring inside the well, it can only
be used in steady state hydraulic conditions and a rig state algorithm is often used to turn on and off the alarm
system.
Variations of the standpipe and annular discharge pressures (Reitsma 2010 and 2011, Mills et al. 2012): the
variations of the pressure at the inlet (Stand Pipe Pressure, SPP) and at the outlet (Annular Discharge Pressure,
ADP) of the system are used to identify when the flow is being abnormally disturbed. For instance, a simultaneous
increase of both the SPP and ADP is characteristic of a kick situation while an increase of SPP accompanied by a
decrease of ADP is typical of a plugged drill-string or annulus bridging. This method can only be applied in steady
state flow conditions without any substantial pipe movement, otherwise many false alarms are generated. It also
requires that the ADP can be measured in a proper way. This is relatively simple in a back-pressure MPD context
because the annulus is sealed by a rotating control device (RCD) and therefore any annular flow-rate or annular
pressure loss variations are reflected on the upstream pressure of the MPD choke. The comparison of the MPD
choke opening and the pressure before the choke can easily be used to spot abnormal flow conditions. However in
conventional drilling, it is necessary to put the ADP sensor upstream of an element causing enough pressure loss
such that it is possible to observe the pressure variations due to the changes of the flow-rate out. It should be noted
that the delta-flow method and the analysis of the variations of standpipe and annular discharge pressures are in fact
equivalent. The second one consists in observing a side effect of flow (the pressure drop) instead of comparing the
flows directly. We will from now on only refer to both of them as a delta-flow method.
Processed differential flow out (Le Blay et al. 2012): the measured flow-rate out is compared with the expected one
so as to detect any deviations that cannot be explained by a normal response of the well to the drilling parameters
such as the mud pump rate or the axial drill-string movement. This method enables monitoring under all flow
conditions including transient periods but depends on accurate hydraulic models to avoid too many false alarms.

It should be noted that other techniques than the volumetric ones, have been investigated to detect influxes in particular. For
instance, a method based on measuring the travel time of pressure waves through the drill-string and annulus have been
documented by several authors (Bryant and Wallace 1991, Codazzi et al. 1992, Stokka et al. 1993). The principle of that
method relies on the fact that acoustic waves propagate much slower in liquid that contains some amount of free gas. The
demonstration and validation of this technique has been made in full scale experiments but its usage has been limited due to
the fact it does not work in oil or synthetic based mud since gas enters in solution when the in situ pressure is larger than the
bubble point pressure which thereby masks the effect of the gas influx on the acoustic wave velocity.
Each of the methods listed above have advantages and drawbacks. The first three methods have the advantage of relative
simplicity because they only rely on sensor measurements and do not require any advanced description of the drill-string,
fluid or wellbore configuration in order to operate. However, they are not adapted to deal with transient phenomenon such as
those that occur during connections or when substantial axial pipe movement takes place. During such events, the gain and
loss detection is left to a large extent to human interpretation. The fourth method has the advantage of coping with dynamic
conditions but at the cost of requiring a detailed description of the drilling setup and running advanced drilling hydraulic
models.
The scope of the presented work is to expand all of the described methods to cope with not only steady conditions but also
any transient periods. However, the extensions that are provided in this paper have greater benefits for the active pit volume
monitoring and the processed differential flow out. To describe how the developed model works, we will first discuss
different drilling methods and how they can affect the return flow in various ways. We will then discuss the possible issues

SPE/IADC 166801

associated to the different measurements available on a drilling rig. The third part of the paper consists of a description of the
transient hydraulic models that govern the flow of drilling fluid inside the well and in the return flow line. We will thereafter
present two examples that illustrate how the use of these models can significantly improve the existing technique to detect
gain and loss situations, by accounting for the expected transient behavior of the hydraulic system associated with both the
wellbore and the rig.
Drilling Methods and Flow-setup
For the purpose of predicting the expected flow-rate out of a well in both transient and steady state conditions, it is necessary
to model the flow of the drilling fluid from the inlet of the hydraulic system to its outlet. In its simplest from, the pumping
system can be considered as a U-tube, with one side being the drill-string and the other side being the annulus. But in many
instances, the hydraulic circuit is in fact a network of different conduits with several junctions, additional pumps and valves.
In this section, we will review how the different drilling methods and downhole drill-string components are put together to
create the effective hydraulic network. We will start with the downhole hydraulic network considerations since they are
common to all drilling methods and continue with a quick description of the impact of three drilling methods (conventional
drilling, back-pressure MPD and dual-gradient drilling) on the hydraulic network.
Hole Geometry:
The most common way to drill uses a simple bit configuration; the open hole dimension is therefore easily determined by the
bit size. However, it is possible to use hole openers or under-reamers to create a larger hole size. There are several
alternatives for enlarging the borehole including:
A pilot hole is first drilled, then the BHA is replaced with a hole opener and the hole is enlarged,
The hole is drilled and enlarged at the same time using one or several hole openers,
A pilot hole is drilled until some depth is reached, then the drill-string is pulled back and the under-reamer is
activated, this cycle may be repeated several times
Furthermore, there is a diversion of the flow of drilling fluid at the level of the hole opener or under-reamer cutter arms in
order to clean cuttings from the tool. This flow diversion together with changes of pressure loss in the annulus due to the
placement of the drill-string within the different diameters of the borehole affects the total pressure loss through the system.
This must be accounted for in gain/loss detection methods that are based on the monitoring of the pressure difference
between the inlet and the outlet of the system.
Downhole Motors:
When using a PDM (Positive Displacement Motor) some of the flow is diverted directly to the annulus to lubricate the
bearings of the motor. This flow diversion has an impact on how much flow is actually flowing through the bit and therefore
on the pressure drop through the system. In addition the pressure drop inside the PDM depends not only on the flow-rate but
also on the bit torque. As a consequence, the pump pressure can vary substantially when using a downhole motor and this
needs to be accounted for when monitoring pressure variations between the inlet and outlet of the drilling system to detect
potential gains and losses.
Float-valve:
In order to avoid flow through the drill-pipe when making a connection during a kick situation, a float-valve is usually placed
inside the BHA (it has also the benefit of preventing cuttings from entering the BHA with the potential consequence of
blocking the bit nozzles or the MWD). The float-valve closes when the pressure inside the drill-string is lower than below the
tool and it opens in the opposite case. As a result of the float-valve action, no fluid enters the drill-string when running in
hole. Therefore the liquid level in the drill-string falls when running in hole. It is necessary to fill the drill-string at regular
intervals in order to limit the effect of buoyancy and reduce the pressure difference between the interior and exterior of the
pipe. As a consequence, drilling fluid volume calculations shall account for the side effects of the float-valve both when
running in hole (the displacement volume is based on the external dimension of the drill-string) and when filling the pipe (the
pit volume decreases while compressing the air column and return flow does not occur until the pipe is filled).
Circulation Sub:
When a circulation sub is activated, the flow is totally diverted to the annulus at the level of the sub, therefore shortcutting the
rest of the drill-string below the tool. This causes an abrupt change in pressure loss both inside the drill-string and in the
annulus. This drastic change in pressure shall be accounted for in the case of a gain/loss detection method based on
differential pressure between the inlet and the outlet.

SPE/IADC 166801

Conventional Drilling:
In conventional drilling, the mud pumps are used to circulate the drilling fluid into the drill-string which thereafter returns to
the surface via the annulus. However this simple configuration can be altered when drilling from a floater with a marine riser
which can be long and have a large diameter. When drilling a small section (e.g., 8 in. or 6 in.) the flow-rate is usually
low and consequently the fluid velocity in the riser can be insufficient to transport the cuttings to surface. In order to clean
the riser, it is usual to circulate at a higher flow-rate through the kill-line of the sub-sea BOP (or the booster line of some
marine risers) so as to increase the fluid velocity inside the riser (see Fig. 1). The gain/loss detection methods need to account
for the additional flow from the outlet. The delta flow method needs to sum up the flow from the booster pump and from the
mud pumps in order to work properly. The method based on comparing the SPP with the ADP needs to incorporate the
additional pressure losses due to the extra flow from the booster pump on the ADP measurement. This also impacts the pit
volume variation method because there is a larger amount of mud buffered in the return flow-lines due to the additional
booster pumping (Ali et al. 2013). Obviously the method based on comparing the expected flow-rate out with the measured
one shall account for the additional flow from the booster pump in the calculation (Le Blay et al. 2012).

Fig. 1: Equivalent flow-diagram for conventional drilling

Pressurized MPD:
When using managed pressure drilling based on back-pressure, the pressure created by the weight of the mud column is
combined with an additional pressure maintained at surface. To be able to apply the pressure at the surface, the annulus is
sealed with a RCD. This allows active control of the downhole pressure by changing the surface back-pressure. The backpressure is generated by a choke which creates a pressure drop when drilling fluids are circulated through it. If there is no
circulation or if the flowrate through the well is too low to generate the desired back-pressure then an additional pump is used
to circulate the drilling fluid directly through the choke in order to create the required back-pressure (see Fig. 2). This method
provides a quick way to adjust the downhole pressure either at bottom hole or any other specific depth. A float-valve is
needed in the BHA to prevent the additional pressure applied to the annulus causing the drilling fluid to flow back through
the drill-string whenever the top-drive is disconnected.

SPE/IADC 166801

Fig. 2: Equivalent flow-diagram for back-pressure MPD

Dual-Gradient and Low-Level Annulus Return Methods:


When drilling in deep water, the hydrostatic pressure caused by the drilling mud in the riser may have a strong influence on
the relatively weak formations near the seabed and require additional strings of casing to be run. An alternative approach that
provides a better match between the mud weight and the formation pressure gradient is a drilling technique called a dualgradient method. The annulus is divided into two intervals, with the top section filled with a fluid with a different density
compared to the section below. The top fluid is often called the blanket fluid. An extraction pump is used to lift the normal
drilling fluid up to surface. This method can be extended to an MPD method (Zeigler et al. 2013) by actively varying the
level of the interface between the blanket fluid and the mud in order to change the hydrostatic component of the pressure at
the point to be controlled. A pump is also required to fill the marine riser with blanket fluid if the interface level sinks. The
blanket fluid will flow out of the riser if the interface level rises. In order to raise the interface level, especially when the mud
pumps stop, a booster pump is connected to the riser at the level of the lift pump suction point (see Fig. 3).

Fig. 3: Equivalent flow-diagram for dual-gradient drilling

A similar technique to the dual-gradient MPD solution consists of using air as a blanket fluid and is called low-level annulus
return. In this case, it is possible to let the interface level go back to the bell nipple level when circulation stops, thus when
there is no flow this is not a dual gradient method anymore. The liquid level in the annulus is lowered only to compensate for
the effect of the pressure losses during circulation (Falk et al. 2011).

SPE/IADC 166801

For steady state conditions, it is straightforward to detect gains or losses as the controller of the lift pump is designed to react
automatically to variations of the reference pressure which is typically measured at the level of the lift pump inlet. In the case
of an increase or a decrease of the return flow-rate, there will be a variation of the interface liquid level of the blanket fluid
and the lift pump will immediately change its flow-rate to compensate. Thus any variation of the lift pump flow-rate, when
everything else is in a steady condition, is a sign of influx or loss to the formation.
However, one difficulty associated with dual gradient solutions is related to the U-tube effect which occurs each time the
mud pumps stop. Since the drill-string is filled with drilling fluid that is usually heavier than the blanket fluid, the weight of
the mud column in the drill-string is larger than the dual-gradient fluid column in the annulus. This results in a fluid flow, socalled free fall flow, which continues after stopping the mud pumps until a weight balance between the drill-string and the
annulus is attained (see Fig. 4).

Fig. 4: Example of a fee-fall flow after a mud pump stop in a dual-gradient drilling operation

Depending on the water depth, mud weight, viscosity, pipe diameter, drill-string length, pressure losses through the BHA and
bit, the U-tube flow can last from a few minutes to, in the worst case, perhaps half an hour (Schubert et al. 2006). This is
unfortunate as it may be much more complicated to detect a gain or loss situation during this very transient period.
Furthermore, if a kick is detected, it is necessary to wait until the U-tube effect is finished before shutting in the well
otherwise, the hydrostatic head of the mud column contained in the drill-string would reduce the margin to the fracturing
pressure gradient. To limit the impact of the U-tube effect, an anti U-tube effect valve has been developed (Goldsmith 1998).
Schubert et al. (2006) give a detailed description on how to perform a well control in a dual gradient drilling operation with
and without the anti U-tube valve. Monitoring gain or loss during transient periods is however both important and difficult
due to the complex setting of this drilling technique.
Measurements
We provide in this section a review of the different measurements that can be used for gain and loss detection. For each type
of measurement, we discuss their reliability for this task.
Pit Volume:
To obtain the volume of drilling fluid contained in the pit, the level of mud in the tank is measured. Traditional float level
instruments use a magnetostrictive linear position sensor but are sensitive to the buildup of solids on the surface of the
device. Ultrasonic level sensors do not have this problem but need to be compensated for temperature variations (Orban et al.
1987). In addition on a floater, it is necessary to compensate the level readings for the effects of the rig not staying perfectly
horizontal (Schuh 1979).
In normal operation, the mud flow is either directed to the active pit system or to the trip tank. In most cases, there are no
signals telling which of the two is used. Furthermore, the hydraulic circuit to the trip tank is usually much shorter than the
one connected to the active pit and therefore the delays and quantities of mud buffered in the return line are quite different
depending on which one is lined up.
During mud displacement, another pit system is used in addition in order to receive the mud displaced from the well. Again
in most cases there are no signals telling which pit is lined up for the inflow to or outflow from the well. Similarly, when a
pill is pumped into the well, it is contained in an alternate pit system like the slug pit. All these alignments of pits can be the
source of misinterpretations when attempting to automate gain or loss detection. The lack of computer readable information
related to the pit alignment has been reported by Parigot (1992), but since this publication, little has changed and this remains
a challenge when implementing automatic gain or loss detection and other drilling automation solutions (Cayeux et al. 2013).

SPE/IADC 166801

Flow-rate In:
The flow-rate in is usually derived from the mud pump rate by multiplying the pump stroke frequency by the stroke volume
and the mud pump efficiency. In most cases the mud pump rate is calculated by measuring the time between two consecutive
strokes, however this method suffers from lag at low flow-rates because it can take a substantial amount of time between two
strokes being counted. This is typically illustrated by the pump pressure starting to increase even before any pump-rate is
actually reported (see Fig. 5).

Fig. 5: The SPP increases before any flow-rate is reported because of the delay necessary to record the flow-rate based on stroke
count at low pump rate.

A better flow-rate estimate can be derived from the actual mud pump motor speed (Shafer et al. 1992) which is usually
higher than the piston movement simply due to the gear box reduction factor. When an AC motor is used to drive the mud
pump, it is sufficient to acquire the frequency of the VFD (Variable Frequency Drive) drive controlling the motor as this is
tightly connected to the actual rotational speed of the motor. For a DC motor drive, it is necessary to install a tachometer or
an encoder on the shaft of the motor to measure the actual angular velocity of the motor.
The next important parameter used to convert the mud pump rate into volumetric flow-rate is the mud pump efficiency.
Traditionally the pump efficiency is considered to be constant and is calibrated once in a while, typically during a cement job.
However, Shafer et al. (1992) reported that in their test setup, where an accurate flow-meter (electro-magnetic) was
connected to the inflow of the mud pumps, the pump efficiency had varied from 86% to 96% during the experiment with
daily variations reaching 5%. The pump efficiency reflects the leakage between the pistons and the liners as well as the fluid
flow back of the inlet and outlet valves, dependent on the valve design, closing speed and condition of the valve springs.
Therefore the pump efficiency is in fact not constant and depends upon the flow-rate, pressure and properties of the drilling
fluid.
Alternatively, the pumped flow-rate can be directly measured. To really gain advantage of that measurement, the flow-meter
must be accurate. Ideally it should be placed on the outlet of the pump in order to measure the real flow of fluid into the well.
Simple to install solutions based on clamp-on ultra-sonic flow-meters have demonstrated poor accuracy (Shafer et al. 1992).
On the other hand, high accuracy flow-meters are seldom certified to work at high pressure and therefore, when they are
available, are positioned upstream of the pumps.
The Coriolis flow-meter is an example of an accurate instrument that keeps its precision even for fluid containing a high
solids concentration. It actually measures the mass flow-rate and the density of the fluid, which can thereafter be combined to
obtain a volumetric flow-rate. However even a Coriolis flow-meter does not measure a perfect flow-rate which is illustrated
in the following example. In this particular case, a Coriolis flow-meter measures the flow-rate produced by a back-pressure
pump during a MPD operation while another Coriolis flow-meter measures the flow-rate after the MPD choke. Note that one
Coriolis flow-meter is placed upstream of the back-pressure pump while the second is placed downstream of the MPD choke
and they are both therefore operating at pressures close to atmospheric. When the mud pumps are stopped, the only
circulation through the MPD choke is due to the back-pressure pump. In such a context, the measurements shown on Fig. 6
indicate a clear systematic bias. Note that there is no pit volume increase or decrease, thus ruling out an eventual leakage.

SPE/IADC 166801

Fig. 6: Bias between two Coriolis flow-meters measuring a volumetric flow-rate in between a pump and a choke

Flow-rate Out:
The measurement of the flow-rate out is usually done using a flow paddle, i.e. a sensor that measures the height of drilling
fluid in the return channel close to the bell nipple or the diverter. This is not a direct measurement of flow-rate and is usually
only considered as indicative. Indeed, the level of fluid in the flow-line does depend on the fluid velocity and is in that way
related to the flow-rate, but it is also affected by the viscosity and the density of the fluid as well as the shape of the channel
cross-section and the roughness of the flow-line walls. Some attempts have been made to measure not only the height of fluid
in the return flow-line but also the fluid velocity inside the conduit (Orban et al. 1987, Shafer et al. 1992) in order to derive a
volumetric flow-rate by accounting for the cross-sectional shape of the return flow-line. However, in a vast majority of cases,
only a channel filling proportion is measured. This value can be biased by the accumulation of cuttings at the bottom of the
gutter changing its cross-sectional area or variations induced by the angular movement of the rig when drilling from a floater.

Fig. 7: Paddle proportion against flow-rate in (left) and estimated flow-rate out (right), in steady state conditions.

Fig. 7 to Fig. 10 illustrate some of the difficulties encountered when analyzing the return flow-rate when using a flow paddle.
All the measurements come from a conventional drilling operation in the 8 -in. section of a 4000m long well with an oilbased mud of density 1.33 sg, and correspond to steady state drilling conditions (constant flow-rate in, block velocity and
RPM). The paddle measurements as a function of the flow-rate in are displayed in the left chart of Fig. 7, while they are
displayed as a function of the estimated flow-rate out (using the transient hydraulic model describe below) on the right chart.
As can be expected, the paddle proportion is more strongly correlated to the estimated flow-out than the flow-rate in.

Fig. 8: Effect of the change of fluid rheology on the height of mud in the return channel.

SPE/IADC 166801

However, the correlation is far from perfect. The mud viscosity has a non-negligible effect on the flow inside the return
channel. During the considered drilling operation, the rheology of the mud was reduced (while keeping the density constant):
the plastic viscosity (PV) and the yield point (YP) values went from respectively 41/35 to 27/20. The effect of the change of
rheology is clear in Fig. 8.

Fig. 9: Paddle proportion against flow-rate in and estimated flow-rate out. Mud continues to flow from the well even after mud pump
stops.

One of the major inconvenience of using the flow-rate in as reference when analyzing the return flow is that transient
phenomena are disregarded. Those effects come mostly from the mud compressibility. Fig. 9 shows the paddle proportion as
a function of both the flow-rate in and the estimated flow-rate out for low flow-rates. All those measurements correspond to
pump stops: even if the flow-rate in is 0, there is still an observed return flow, because of the compressibility related delay.
One can see that this effect is not visible when considering only the flow-rate in, while it is accounted for (although not
perfectly) by numerical transient simulations.
Finally, only considering the flow-rate in as a control parameter hides the influence of other drilling related actions on the
return flow. For example, the drillstring movement affects the flow pattern in the annulus, and thus the return flow. One can
observe on the top graph of Fig. 10 that although the flow-rate in is constant the return flow varies. The bottom plot of Fig.
10 provides an explanation: these measurements correspond to a ream up and down sequence and the axial movement of the
drillstring disturbs the velocity of the fluid in the annulus. Again, those effects are accounted for when using numerical
estimation of the return flow.

Fig. 10: Effects of block movements on the return flow.

Instead of only using a paddle, a precise measurement of the flow-rate out of the well can be made using an electro-magnetic
flow-meter but only with water based mud since the measurement principle requires that the fluid is electrically conductive
(Orban et al. 1987, Orban and Zanker 1988, Shafer et al. 1992, Le Blay et al. 2012, Reitsma 2010). A Coriolis flow-meter
can precisely measure the flow-rate out with a precision which is just as good for water, oil or synthetic based mud. Early
models of Coriolis flow-meters had a relatively small internal diameter which caused some pressure loss that resulted in a
liquid level rise into the bell nipple or diverter while circulating and consequently caused an increase of the hydrostatic head
in the annulus which was unfavorable for the downhole pressure. However in a back-pressure MPD context this additional
pressure drop is not a problem as it can be accounted for in the calculation of the MPD choke opening which maintains

10

SPE/IADC 166801

constant pressure at the depth of interest. Newer Coriolis flow-meters can be produced with a larger internal size reducing
much of the pressure loss through the metering tool and rendering it as a more practical instrument for conventional drilling
(Le Blay et al. 2012).
Pump pressure:
Pressure sensors placed downstream to mud pumps need to have a wide measurement range. This has a consequence for the
achievable precision of the sensor. The best pressure sensors qualified for operating at up to 10000psi (680bars) have a
resolution of 5psi (0.34bar) and an accuracy of 25psi (1.72bar) (Reitsma 2010).
Discharge pressure:
Discharge pressure does not usually change as much as pump pressure, however it is fairly common to use sensors with a
span of 3000psi (200bar). A much greater resolution and accuracy can be obtained if the sensor is calibrated for an operating
span of 100psi (6.9bar and obtain a resolution of 0.05psi (0.0035bar) and an accuracy of 0.25psi (0.017bar) (Reitsma 2010).
Dynamic Flow in the Well
In order to extend the application of delta-flow, variations of the standpipe and annular discharge pressures or the processed
differential flow out under transient conditions, it is necessary to apply a detailed model of the dynamic flow occurring inside
the well.
Conservation equations:
The equations governing single phase homogenous three dimensional flow are well known, and are obtained by applying
successively conservation of mass, Newtons second law and the first law of thermodynamics to a fluid passing through an
infinitesimal, fixed control volume. One then obtains the continuity equation, the momentum equation (Navier-Stokes
equation) and the energy equations:

(1)
+ = 0,

+ = + !" ,

(2)

+ ! =
+ + (!" )

(3)

where is the fluid density, the fluid velocity vector, is the second order tensor obtained by the dyadic product of
the velocity vector, is the body force per unit mass vector (usually gravity forces) and !" denotes the stress tensor, and
! is the total energy per unit volume ! = ( +

!!
!

+ potential energy) being the internal energy per unit mass, is the

heat flux, usually assumed to follow Fouriers law for heat transfer = , being the thermal conductivity and the
temperature, and finally is the heat produced by external sources.
Boundary conditions:
Those equations can be used to model the flow through the entire wellbore, given appropriate boundary conditions. In this
three dimensional formulation, one can typically associate flow conditions at the inlet of the pipes (typically the pump rate), a
zero flow at the various walls (pipe wall or open hole/casing wall), and pressure conditions at the outlet of the annulus
(usually the atmospheric pressure in conventional drilling). Note that when cuttings are produced by the drilling process,
when those cuttings exit the hydraulic system by settling into a bed, when formation fluid enters the wellbore or when
drilling fluid is lost, one has to update accordingly the flow boundary conditions.
One dimensional formulation:
The three equations above, together with appropriate boundary conditions fully describe the three dimensional behavior of
homogeneous single phase flow. However, a three dimensional resolution of the problem is computationally very expensive,
and for the purpose of analyzing the return flow from a well, it is enough to restrict the analysis to a one dimensional
problem. This can be done by cross sectional averaging of the main quantities. When doing so, the various flows through the
pipe wall (cuttings, formation fluid, losses, etc.) are no longer treated as boundary conditions, but as source terms in the
conservation equations. We will focus from now on the continuity and momentum equations. The equations become in their
one dimensional form, for a pipe flow (tube or annulus):

+
= ,

(4)

SPE/IADC 166801

11

+ ! +
= + + ! ,

(5)

where is the pipe cross sectional area, is the averaged density, is the density weighted axial velocity (i.e. center of mass
velocity), is the pressure, is the axial component of the body forces per unit mass (equal to cos for an inclined well
with inclination ), and the term () represents the shear stress per unit length along the pipe walls, being a positive
function dependent on the fluid rheology, the geometry and movements of the pipe. The mass flux (per unit length) through
the wall now carries a momentum and ! denotes the associated velocity.
When considering the entire hydraulic network, one also has to consider the junction points between several branches. The
flow entering one branch from the extremity of another can be treated as a source term in the continuity and momentum
equation. In particular, by using the continuity equation at all the junction points, one can build a system of equations whose
resolution provides the repartition of the mass fluxes between the different branches of the system.
Mud properties:
The properties of the mud play a central role in the above equations, and provide constitutive laws that are used in the
resolution of the equation. In particular, the mud compressibility and the mud rheology are of major importance. The role of
!!
!"
the mud compressibility appears naturally in the continuity equation if one notices that = . Thus the variations of
!"
!"
pressure with time modify the mass transport via the mud compressibility, and may result in oscillations of the velocity field.
Oil based muds have a larger compressibility than water based mud, and a proper characterization of the compressibility is
therefore necessary to correctly estimate the fluid velocity at the outlet of the well when using such fluids. The rheology of
the mud is used to describe the shear stress along the pipe, via the function . Drilling fluids are non-Newtonian. Several
models exist to determine the shear stress response to the fluid velocity, and parameterization of these models is done using
the Fann35 measurements made at regular intervals during drilling operations.
Drill-string and BHA:
The design of the BHA affects the behavior of the flow. First, the variations of diameter (inside and outside the BHA, and
!
around the connections of the drill pipes) have a visible effect on the term
of the continuity equation (note that
!"
discontinuities in the cross-sectional area need careful treatment). The induced variations in fluid velocity entail variations in
momentum that finally result in pressure losses. Then, the presence of particular elements of the BHA such as a hole opener,
under-reamer, down-hole motor, float valve, circulation sub etc... modify the geometry of the hydraulic network, and thus the
entire flow distribution in the wellbore.
Drilling method:
The type of drilling method is of course reflected in the hydraulic network, since the addition of mud pumps, back-pressure
pumps, lift pumps, fill pump can be either seen as modifications of the network structure, as the introduction of source terms
in the equations, or as a dynamic management of the boundary conditions, for example, if one decides to use the backpressure pump pressure as a varying boundary condition for the entire system.
Axial Drill-string Movement:
!
The effects of the drillstring movement are visible for eq. (4) and (5): the time derivative of the cross sectional area

!"
(the first terms of the continuity and momentum equations) captures the change in area observed when the pipe is moving.
These changes are more significant at the bit level, but are to be accounted for at the level of other elements where large
changes of diameter occur such as around the BHA elements especially but also around the pipe connections. The movement
of the drillstring also carries its own momentum, which should be included in the shear stress function .
U-tube effect:
When using some special drilling techniques based on annulus fluid level manipulations (e.g., Dual-Gradient, Low-level
annulus return) one observes a U-tube effect when the mud pumps are stopped due to the difference of height between the
annulus and string mud columns. This phenomenon is created by the term of the momentum equation, where
= cos corresponds to the gravity force. But the U-tube effect can also occur in conventional drilling when pumping the
same fluid as the one in place but at a different temperature. This is simply due to the temperature dependence of fluid
density due to thermal expansion which therefore causes a weight imbalance until the temperatures of the fluids in the drillstring and annulus have equilibrated (see Cayeux 2012 for the adverse effects of such a situation).
Multiphase/multicomponent fluid:

12

SPE/IADC 166801

So far, for the sake of simplicity, only single phase fluids have been considered. However, the fluids in a wellbore may
contain gas, liquids and various solids of different sizes and natures: bentonite, barite, cutting, etc. One common formulation
of the equation in the multiphase case is the drift-flux one that considers that each phase is circulated in separate tubes
whose area is given by the volume fraction of each phase (see Trapp et al., 1986 or Liles et al., 1978 for example). By
assuming common pressure for each phase, one uses mass conservation equations for each component and a mixture
momentum equation:

(6)
+
=
!"
! !
! ! !
!

! ! ! ) +
!

! ! !! ) +
!

= (

! ! ) + ! , . . , ! + !

(7)

where the index denotes the phase, the variables ! , ! , ! are respectively the volume fraction, density and velocity of
phase , the term !" represents the interphase mass exchange, with the convention that !! = ! , the mass flow-rate of
phase through the pipe walls, and the term (! , . . , ! ) represents the mixture shear stress per unit length function,
accounting for slip velocities (i.e. the phase velocities differences).
Cuttings:
While drilling, cuttings are generated and provide mass and momentum source terms. However, while being transported,
cuttings can settle into a cuttings bed, and go back from the bed to being in suspension in the mud. In both cases, this
corresponds to additional mass and momentum source terms, but also has the effect of modifying the cross-sectional area in
the annulus. Then, dynamic effects such as those described in the paragraphs Drill-string and BHA and Axial Drill-string
Movement have to be accounted for.
Fluid exchange with the formation:
In case of an influx, formation fluid enters the wellbore. In case of losses, drilling mud exits the system. In both cases, this
corresponds to an additional source term (positive or negative) in the above equations. Depending on the pressure regime, the
gas in the annulus can either be dissolved in the mud (above the bubble point pressure), in which case it is part of the liquid
phase, or it can be separated from the mud, in which case it is free. When this phenomenon occurs, inter phase mass
exchange takes place and is included in the equations via the term !" .

Fig. 11: Comparison between calculated and measured values when reaming down one stand in a long horizontal well (bit depth
8080mMD).

Fig. 11 illustrates the importance of transient hydraulic calculations to accurately detect gains and losses and avoid false
alarms. This example is for a reaming operation performed in an 8100mMD deep well. Each time the mud pumps are started,

SPE/IADC 166801

13

it took more than five minutes before the flow-rate out reached the same level as the flow-rate in. This is due to the relative
compressibility of the mud combined with the large amount of fluid contained in the drill-string and annulus. Similarly, when
the mud pumps were stopped, it took about two minutes before the mud flow-rate out stopped. Any gain and loss detection
techniques not accounting for those naturally explainable transient effects would raise false loss or gain alarms while reaming
down this well. It should be noted that, at this rig site, the paddle did not indicate any values unless the flow-rate was above
500l/min. Despite this threshold, the graphs show a good agreement between the amplitude and timing of the calculated flowrate out and the paddle measurements, at least when the flow-rate out was greater than 500l/min. The additional flow-out
resulting from the downward movement of the drill-string is also clearly visible on the paddle readings and on the calculated
flow-rate out.
Modeling of the Return Flow to the Pit From the Well Outlet
As discussed above, the evolution of the pit volume is an important parameter to detect mud losses and formation fluid
influx. However the pit volume evolution depends on the pump rate, the rate of penetration, the drill-string movement,
cuttings transport and the lag time for the mud to return to the pit. The wellbore hydraulic calculations account for the time
necessary for the mud to return to the annulus outlet, but there is also a need to model the flow of the mud from the wellbore
annulus outlet to the pit.
The hydraulic circuit from the wellbore annulus outlet to the pit is composed of four main components:

The mud return line

The shale shakers

The sand trap, degasser and transfer pit with associated flow lines

The active pit

Some of the required information is not well known. It can be possible to estimate these parameters through calibration
during actual drilling operations. Most of these calibration parameters are characteristics of the drilling installation at a given
rig site and therefore they can be reused from one drilling operation to another one (at least as long as the installation has not
been changed).
Flow Modeling in a Channel or a Flow-line:
The drilling fluid can return to the pit through a pipe or an open channel. The typical channels used at the rig site are semicylindrical or rectangular (see Fig. 12). Assuming that the level of fluid in an open channel never exceeds the total height of
the conduit, a semi-cylindrical channel or a flow-line is described by its radius () and a rectangular channel by its width
(). The height of drilling fluid in a pipe, when not completely filled, or in an open channel depends on the flow-rate () at
its entrance, the geometrical dimensions of the channel or pipe ( or ), the surface roughness of the wall sides (), the slope
of the channel () and the effective viscosity of the fluid (!"" ).

Fig. 12: Typical cross section of mud return channels or flow-line

The flow in those conduits is subject to very little pressure and it is therefore reasonable to consider that the fluid is
incompressible. The generalized Bernoulli equation for an unsteady but incompressible flow in an open channel, i.e. with a
constant boundary pressure, in between two sections indexed and + 1 respectively (see Fig. 13), distant by a length , is:
+ ! + !

! !
+
2

!!

1
!!! !
! = !!! + !!!
+ ! +

2

1
!!!

!!!!

(8)

14

SPE/IADC 166801

where is the slope of the conduit, ! and !!! are the depths of liquid at sections and + 1, ! and !!! are the average
fluid velocities across the sections and + 1, ! and !!! are the kinetic energy correction factors at the respective sections,
! is the head loss, and ! and !!! are the cross sectional area of the fluid at sections and + 1 (they are therefore
functions of ! and !!! and the shape of the channel).

Fig. 13: Energy balance on a control volume of the conduit

The term

!! !
!!

in eq. (8) originates from the fact that the Bernoulli energy equation is derived using the assumption of a plug

flow across the conduit, i.e. the velocity is identical anywhere in a cross-section of the channel. That is true for non-viscous
fluid or in fully turbulent flow. But in case of a laminar or intermediate flow of a viscous fluid, there is a gradient of velocity
from the wall of the conduit to the central part of the flow. To compensate for that approximation, the kinetic energy
correction factor has been introduced by Coriolis in the Bernoulli equation. The definition of this coefficient is:
=

(9)


!
For a Newtonian fluid in laminar flow into a fully filled circular pipe, = 2. But for non-filled conduits, as soon as the fluid
velocity is not too small, the coefficient tends back to unity. Since the term

!! !
!!

tends to be very small when ! is small and

since the difference of bulk velocities ! and !!! on the two sides of considered cross sections is also not much different, it
does not influence much the accuracy of the calculations to consider 1 in all cases.
If we neglect the acceleration effects, eq. (8) simplifies into:
+ ! +

! !
!!! !
= !!! +
+ !
2
2

(10)

It can be used together with the continuity equation, which takes the following form in this context:
(11)

= ! ! ! !!! !!! !!!

where ! is the fluid volume contained in between sections and + 1, is the average density of the fluid contained in that
volume, ! and !!! are the densities of the fluid at the sections and + 1.
!

After discretization of the channel in sections of length , and knowing the density and volumetric flow-rate of the fluid at
the entrance of the channel, it is possible to integrate eq. (9) and (10) over the length of the channel at any time and therefore
derive the depths and velocities of the fluid along the channel as a function of time. This is of course conditioned by having
an estimation of the head loss in between two sections.
To determine the head loss, the Darcy-Weisbach equation can be used, considering that for an open flow in a channel or a
pipe, the equivalent pipe diameter is the hydraulic diameter:
! = !

!
.
! 2

(12)

where ! is the pressure head loss over the length , ! is the Darcy friction factor, ! is the hydraulic diameter, is the
average fluid velocity. The hydraulic diameter is defined by the ratio of the cross-sectional area () of the fluid in the channel

SPE/IADC 166801

15

or pipe to the wetted perimeter (! ):


! =

(13)

The derivation of the hydraulic diameter for the rectangular, semi-cylindrical channels and for a partially filled pipe is given
in appendices A, B and C.
The Darcy friction factor is found by solving the Colebrook-White equation for open surface flow:
1
!

= 2 log!"

2.51
+

3! ! !

(14)

where ! is the Reynolds number. When flowing into a fully filled pipe, the Colebrook-White equation is:
1
!

= 2 log!"

2.51
+

3.7! ! !

(15)

!

!""

(16)

The Reynolds number is defined by:


! =

Drilling fluids are shear thinning with yield stress fluids. Their rheological behavior can be well approximated by a
Robertson-Stiff model (Robertson and Stiff 1976):
= ( + )!

(17)

where is the shear stress, is the shear rate and , and are the coefficients of the model. The shear rate at the wall is
then:
=

3 + 1 8

+
4
! 3

(18)

Fig. 14 shows how drilling fluid flows into an inclined open channel when the flow-rate out of the well increases. Similarly
Fig. 15 shows the drainage process of the inclined channel when the flow-rate out of the well stops.

Fig. 14: Propagation of a fluid front into an inclined channel when circulation is established.

Fig. 15: Drainage of an inclined channel when the circulation into a well is stopped.

16

SPE/IADC 166801

Flow in the Solid Control Equipment:


After flowing through the flow-line, the mud goes into the shale shakers so that cuttings can be separated from the drilling
fluid. The drilling fluid has to pass through screens with a given mesh size and thickness (!"#$$% ). This is equivalent to a
flow through a porous medium (ASME Shale Shaker Committee 2005) and the Darcys law can be used:
!"#$$% =

!"#$$% !"#$$%

!"" !"#$$%

(19)

where !"#$$% is the velocity of the flow through the screen, !"#$$% is the permeability of the screen, !"#$$% is the pressure
!
loss through the screen and !"" is the effective viscosity of the fluid. The factor !"#$$% = !"#$$% , also called conductance, is
!!"#$$%

a characteristic of the screen.

By integrating the continuity eq. (1) over the volume of mud retained by the screen, we obtain, at any time, the volume
(!"#$$% ) of drilling fluid retained on a screen and the flow-rate out of the shaker:
!"#$$% !"#$$%
= ! !

1 !"## !"#$% + !"## !"##

!"#$$% !"#$$% !"#$$%



!""

(20)

where !"#$$% is the density of the mud retained on the screen, ! is the density of the mud arriving on the screen, ! is the
volumetric flow-rate of the drilling fluid arriving on the screen, !"#$% is the density of the cleaned mud, i.e. without the
cuttings, !"## is the density of the cuttings separated from the mud and !"## is the volume fraction of cuttings separated from
the mud.

Fig. 16: variations of active pit volume due to cuttings arriving at surface and being separated from the mud

It should be noted that the volume fraction of cuttings removed from the mud is not exactly the same as the volume fraction
of cuttings inside the mud, simply because a film of mud coats each cutting particles and therefore some mud is also removed
in the separation process. As a consequence, the density of the cuttings particles separated from the mud is not the density of
the clean cuttings particles, it is in fact:
(21)

1
!"#$% +
!
1+
1+
where is the volume ratio of mud to cuttings and ! is the density of a clean cuttings particle. The volume ratio of the mud
coating around the cuttings particles depend on the size of the particles and the properties of the mud. Typically it is around
1, i.e. around each cuttings particle removed from the mud there is about the same amount of mud coating the particle. Fig.
16 shows how pit volume variations calculated using this principle matches relatively well the calculated arrival of cuttings
on the shale shakers.
!"## =

Calculation of the Volume in the Pit:


Using the continuity eq. (1) over the volume of mud contained in the active tank system, we can derive the change of volume
in the active pit system. The cross-sectional area of the tank is the same at any depth inside the tank. If we assume that, at any
time, the fluid contained into the tank is homogenous and if we consider that the fluid contained in the tank is incompressible
in view of the small range of pressure encountered inside the pit, then we can consider that the density of the fluid is the same
at any depths of the tank (!"#$ ). By integrating over the height of liquid contained into the tank and by using the boundary

SPE/IADC 166801

17

conditions, we obtain:
(22)
!"#$ !"#$
= ! ! !"#$ !

where !"#$ is the volume inside the tank, ! and ! are respectively the density and volumetric flow-rate of fluid entering
the pit, and ! is the volumetric flow-rate of fluid being pumped out of the tank.
However, it should be noted that the density of the fluid contained in the pit is not constant as a function of time because the
temperature in the tank changes and the fluid contained in the tank is mixed with the returned fluid that does not have
necessarily the exact same formulation as the one already contained into the pit.
Fig. 17 illustrates the end result of those calculations. In this example, until 21.38 the mud pump rate was constant at
2200l/min and since there were no axial movements of the drill-string (the block position is constant throughout the whole
period), the active volume stayed unchanged. But then the mud pump rate was reduced to 1000l/min, an apparent gain of
mud could be observed in the active pit. This was the result of less mud being buffered in the return flow-line and in the shale
shakers. At 21:39:30, the mud pump flow-rate was increased to 1900l/min and as a consequence the active volume decreased,
simply because more mud was accumulated into the return flow equipment. However, the active volume did not return to the
initial value because the new flow-rate was slightly lower than at the start of the sequence. It should be noted that during the
whole sequence the calculated active volume matched relatively well the measured one, not only when steady state
conditions were reached but also during the transient periods.

Fig. 17: This time-based log shows how the calculated active volume matches relatively well the measured pit volume when the mud
pump rate is first decreased and thereafter increased.

Use Case 1: Kick Detection During Connection


This first example is a classical problem of detecting a kick at connection time. In this particular case, a 9 -in hole,
enlarged to 10 5/8-in using an under-reamer, had reached the depth 2230mMD (last casing shoe, 10 -in, at 1786mMD) with
an average ROP of 15m/h. For the last 3 stands, the maximum gas reading had been 2%. At 12:01, the mud pumps were
stopped to make a connection (see Fig. 18). At first, the active volume increased as expected due to the normal flow-back
from the well, the flow-line and the solid control equipment. But at 12:05, the active pit volume continued to increase while
transient calculations of the return flow showed that the volume should have stabilized. A few minutes later, it was clear that
there was an influx and the well has been shut-in.

18

SPE/IADC 166801

Fig. 18: Detection of an influx at connection time.

Use Case 2: Draw-Down Tests During an MPD Operation


For the second use case, we will present the real-time analysis of draw-down tests made during an MPD operation. A well
with a narrow geo-pressure window (see Fig. 19) was drilled (8 -in bit with a 9 -in under-reamer) using a back-pressure
MPD technique.

Fig. 19: Geo-pressure prognosis

In order to measure the real value of the pore pressure gradient of the open hole section, draw-down tests were made at
regular intervals. These tests consisted of reducing the back-pressure slightly for about 10 minutes and check whether a gain
was observable, this procedure was repeated with lower back pressure until formation fluid started to flow in the well
therefore giving a clear indication of the maximum pore pressure gradient of the open hole section.
The preparation of the tests consisted in cleaning the well carefully by maintaining the circulation while reaming up and
down the drill-string. The observable slight decrease of the pit volume was due to temperature variations in the pit and the
well, and the current drill-string axial movement. These naturally explained pit volume variations were well simulated by the
above described hydraulic calculations (compare the calculated curve, green, on the relative pit volume track of Fig. 20 with
the actual measurements, blue curve before the start of the test procedure, i.e. prior to 18:00).

SPE/IADC 166801

19

Fig. 20: Draw-down tests in an MPD operation

For the first draw-down test, the back pressure was reduced by 3 bars. The calculated flow-rate out and the measured one
deviated very slightly during this test, but the discrepancy between the calculated active pit volume and the measured one
was more visible because of the cumulative effect of the influx flow-rate. However, it should be noted that just looking at the
active volume itself, i.e. without comparing it with the calculated one, would not give any indication of an influx since the
volume remained almost constant during the draw-down test. This tiny influx was confirmed by an increase of the gas
proportion when the contaminated mud reached the mud treatment equipment (see Fig. 21).
A second draw-down was conducted, this time reducing the back-pressure by one additional bar. This time the deviation
of the flow-rate out measured by the Coriolis flow-meter was more visible than in the first test. The gain in the pit was also
readily visible. The proportion of gas measured in the drilling mud when it left the well was also substantially larger than the
first time.

20

SPE/IADC 166801

Fig. 21: Confirmation of small influxes as a consequence of two draw-down tests.

Conclusion
The detection of gain and loss during drilling operations can be difficult because of the lack of accurate measurements but
also because many naturally explained phenomenon may cause false alarms. Depending on the used drilling technique, those
phenomenon can be of different natures, and it usually requires human interpretation to detect possible influx or losses.
However, real-time accurate modeling of the downhole hydraulics provides a way to predict the expected flow of the well.
When coupled to transient simulations of the flow of drilling fluid in the flow-lines and solid control equipment, it becomes
possible to compare different measurements to modeled values, so that one can adapt some of the different existing
methodologies to obtain quite precise detections of gains and losses both for conventional drilling but also for more complex
drilling methods such as MDP.
Nomenclature
AC-Alternate Current
ADP- Annular Discharge Pressure
BHA-Bottom Hole Assembly
BOP-Blow Out Preventer
CUSUM-Cumulative Sum
DC-Direct Current
LL-Liquid Level
MPD-Managed Pressure Drilling
MWD-Measurement While Drilling
PDM-Positive Displacement Motor
PV- Plastic Viscosity
RCD-Rotating Control Device
ROP-Rate Of Penetration
RPM-Rotation Per Minute
SPP- Stand Pipe Pressure
VFD-Variable Frequency Drive
YP-Yield Point

SPE/IADC 166801

Symbols

cross sectional area [L2](m2)


!"#$$% surface area of the screen [L2](m2)
!"#$$% screen conductance [L](m)
!
hydraulic diameter [L](m)

fluid depth in the channel or flow-line [L](m)


!"#$$% screen thickness [L](m)
!
total energy per unit volume [ML-1T-2](J/m3)

internal energy per unit mass [L2T-2](J/kg)

body force [MLT-2](N)

axial component of the body forces [MLT-2](N)


!"##
volume fraction of cuttings removed from the mud
!
Darcy friction factor

heat flux [MT-3](W/m2)

shear stress per unit length along the wall [MT-2](pa/m)

shear stress per unit length along the wall for multiphase fluid [MT-2](pa/m)

thermal conductivity [MLT-3-1](W/(m.K))


!"#$$% screen permeability [L2](m2)

hydraulic mean depth [L](m)

mass flux per unit length through the wall [MT-1L-1](kg/(s.m))


!
mass flow rate per unit length of phase through the pipe walls [MT-1L-1](kg/(s.m))
!
wetted perimeter [L](m)

pressure [ML-1T-2](pa)

volumetric flow-rate [L3T-1](m3/s)


!
volumetric flow-rate entering a system [L3T-1](m3/s)

heat produced from external sources [ML2T-2](J)

radius [L](m)

slope of the channel [L3](m3)


!"#$$% volume of drilling fluid retained on a screen [L3](m3)
!"#$
volume of mud in the active pit [L3](m3)

fluid velocity vector [LT-1](m/s)

velocity tensor [LT-1](m/s)

density weighted axial velocity [LT-1](m/s)

average velocity [LT-1](m/s)


!
velocity of the mass flux per unit length through the wall [LT-1](m/s)
!"#$$% velocity of flow through a shale shaker screen [LT-1](m/s)

width of a rectangular channel [L](m)


Greek Letters
!
volume fraction
!"
interphase mass exchange [ML-1T-1](kg/(ms))

shear rate [T-1](s-1)

surface roughness [L](m)

inclination (rd)

fluid compressibility [M-1LT2](pa-1)

volume ratio of the film of mud to the volume of cuttings.


!""
effective viscosity [ML-1T-1](pa.s)
!"
stress tensor [ML-1T-2](pa-1)

density [ML-3](kg/m3)

average density [ML-3](kg/m3)


!
density of the mud entering a system [ML-3](kg/m3)
!
density of a clean cuttings particle [ML-3](kg/m3)
!"#$% density of the mud after removal of the cuttings [ML-3](kg/m3)
!"##
density of the separated cuttings [ML-3](kg/m3)
!"#$$% density of the fluid retained on the screen [ML-3](kg/m3)
!"#$ density of the fluid contained in the active pit [ML-3](kg/m3)

shear stress [ML-1T-2](pa)

21

22

SPE/IADC 166801

References
Aldred, W., Hutin, R., Luppens, J., Ritchie, G., 2008. Development and Testing of a Rig-Based Quick Event Detection System to Mitigate
Drilling Risks. Paper IADC/SPE 111757 presented at the IADC/SPE Drilling Conference held in Orlando, Florida, USA, 4-6 March
2008.
Ali, T.H., Haberer, S.M., Says, I.P., Ubaru, C.C., Laing, M.L., Helgesen, O., Liang, M., Bjelland, B., 2013. Automated Alarms for Smart
Flowback Fingerprinting and Early Kick Detection. Paper presented at the SPE/IADC Drilling Conference and Exhibition held in
Amsterdam, The Netherlands, 5-7 Mar 2013.
Anfinsen, B.T., Rommetveit, R., 1992. Sensitivity of Early Kick Detection Parameters in Full-Scale Gas Kick Experiments With Oil- and
Water-Based Drilling Muds. Paper IADC/SPE 23934 presented at the IADC/SPE Drilling Conference held in New Orleans,
Louisiana, USA, Feb 18-21, 1992.
ASME Shale Shaker Committee, 2005. Drilling Fluid Processing Handbook. Published by Elsevier, ISBN-13: 978-0-7506-7775-2.
Bryant, T.M., Wallace, S.N., 1991. Field Results Of An MWD Acoustic Gas Influx Detection Technique. Paper SPE 21963 presented at
the SPE/IADC Drilling Conference held in Amsterdam, The Netherlands, Mar 11-13, 1991.
Cayeux, E., 2012. Safe Mud Pump Management while Conditioning Mud. Paper presented at the 2012 IFAC Workshop on Automatic
Control in Offshore Oil and Gas Production, Trondheim, Norway, 31 May-1 June, 2012.
Cayeux, E., Daireaux, B., Dvergsnes, E. W., Florence, F., 2013. Toward Drilling Automation: On the Necessity of Using Sensors That
Relate to Physical Models. Paper SPE 163440 presented at the SPE/IADC Drilling Conference and Exhibition held in Amsterdam,
The Netherlands, Mar 5-7, 2013.
Codazzi, D., Till, P.K., Starkey, A.A., Lenamond, C.P., 1992. Rapid and Reliable Gas Influx Detection. Paper SPE 23936 presented at the
IADC/SPE Drilling Conference held in New Orleans, Louisiana, USA, Feb 18-21, 1992.
Falk, K., Fossli, B., Lagerberg, C., Handal, A., Sangesland, S., 2011. Well Control When Drilling With a Partly-Evacuated Marine Drilling
Riser. Paper SPE 143095 presented at the IADC/SPE Managed Pressure Drilling and Underbalanced Operations Conference and
Exhibition held in Denver, Colorado, USA, 5-6 April 2011.
Goldsmith, R., 1998. Mudlift Drilling System Operations. Paper OTC 8751 presented at the Offshore Conference Technology Conference
held in Houston, Texas, USA, 4-7 May, 1998.
Hargreaves, D., Jardine, S., Jeffryes, B., 2001. Early Kick Detection for Deepwater Drilling: New Probabilistic Methods Applied in the
Field. Paper SPE 71369 presented at the SPE Annual Technical Conference and Exhibition held in New Orleans, Louisiana, 30 Sep-3
Oct, 2001.
Le Blay, F., Villard, E., Hilliard, S., Grns, T., 2012. A New Generation of Well Surveillance for Early Detection of Gains and Losses
When Drilling Very High Profile Ultradeepwater Wells, Improving Safety, and Optimizing Operating Procedures. Paper SPE 158374
presented at the SPETT 2012 Energy Conference and Exhibition held in Port of Spain, Trinidad, 11-13 June 2012.
Liles, D.R., Reed, Wm.H., 1978. A Semi-Implicit Method for Two-Phase Dynamics J. Computational Physics 26 1978
Maus, L.D., Tannich, J.D., Ilfrey, W.T., 1979. Instrumentation Requirements for Kick Detection in Deep Water. Paper SPE 7338 published
in Journal of Petroleum Technology, vol. 31, No 8, pp. 1029-1034, Aug. 1979.
McCann, D.P., White, D.B., Marais, L., Rodt, G.M., 1991. Improved Rig Safety by Rapid and Automated Kick Detection. Paper
SPE/IADC 21995 presented at the SPE/IADC Drilling Conference held in Amsterdam, The Netherlands, 11-14 March 1991.
Mills, I., Reitsma, D., Hardt, J., Tarique, Z., 2012. Simulator and the First Field Test Results of an Automated Early Kick Detection System
That Uses Standpipe Pressure and Annular Discharge Pressure. Paper SPE/IADC 156902 presented at the SPE/IADC Managed
Pressure Drilling and Underbalanced Operations Conference and Exhibitions held in Milan, Italy, 20-21 March 2012.
Orban, J.J., Zanker, K.J., Orban, A.E., 1987. New Flowmeters for Kick and Loss Detection During Drilling. Paper SPE 16665 presented at
the SPE Annual Technical Conference and Exhibition held in Dallas, Texas, USA, Sept. 27-30, 1987.
Orban, J.J., Zanker, K.J., 1988. Accurate Flow-out Measurements for Kick Detection, Actual Response to Controlled Gas Influxes. Paper
SPE 17229 presented at the IADC/SPE Drilling Conference held in Dallas, Texas, USA, Feb 28-Mar 2, 1988.
Parigot, P., 1992. Quality Assurance/Quality Control for the Sake of Hole Safety. Paper SPE 25043 presented at the European Petroleum
Conference held in Cannes, France, 16-18 November 1992.
Reitsma, D., 2010. A Simplified and Highly Effective Method to Identify Influx and Losses during Managed Pressure Drilling Without the
use of a Coriolis Flow-meter. Paper SPE 130312 presented at the SPE/IADC Managed Pressure Drilling and Underbalanced
Operations Conference and Exhibition held in Kuala Lumpur, Malaysia, Feb. 24-25, 2010.
Reitsma, D. 2011. Development of an Automated System for the Rapid Detection of Drilling Anomalies using Standpipe and Discharge
Pressure. Paper SPE/IADC 140255 presented at the SPE/IADC Drilling Conference and Exhibition held in Amsterdam, The
Netherlands, 1-3 March 2011.
Robertson R.E., Stiff H.A., 1976. An Improved Mathematical Model for Relating Shear Stress to Shear Rate in Drilling Fluids. SPE
Journal, February 1976, vol. 16, No 1, pp. 31-36.
Schubert, J.J., Juvkam-Wold, H.C., Choe, J., 2006. Well-Control Procedures for Dual-Gradient Drilling as Compared to Conventional
Riser Drilling. Paper SPE 99029 published in the SPE Drilling & Completion Journal, vol. 21, No 4, pp. 287-295, Dec. 2006.
Schuh, F., 1979. Surface Mud Volume Measurements on Floating Rigs. Paper SPE 7405 published in the Journal of Petroleum
Technology, Vol. 31, No 12, pp. 1497-1501, December 1979.
Shafer, D.M., Loeppke, G.E., Glowka, D.A., Scott, D.D., Wright, E.K., 1992. An Evaluation of Flowmeters for the Detection of Kicks and
Lost Circulation During Drilling. Paper SPE 23935 presented at the IADC/SPE Drilling Conference held in New Orleans, Louisiana,
USA, Feb 18-21, 1992.
Speers, J.M., Gehrig, G.F., 1987. Delta Flow: An Accurate, Reliable System for Detecting Kicks and Loss of Circulation During Drilling.
Paper SPE 13496 published in SPE Drilling Engineering, vol. 2, No. 4, pp. 359-363, Dec. 1987
Stokka, S., Andersen, J.O., Freyer, J., Welde, J., 1993. Gas Kick Warner An Early Gas Influx Detection Method. Paper SPE/IADC 25713
presented at the SPE/IADC Drilling Conference held in Amsterdam, The Netherlands, Feb 23-25, 1993.
Trapp, J.A. Riemke, R.A., 1986. A Nearly-Implicit Hydrodynamic Numerical Scheme for Two-Phase Flows J. Computational Physics 66
1986

SPE/IADC 166801

23

Ziegler, R., Ashley, P., Malt, R., Stave, R., Toftevg, K., 2013. Successful Application of Deepwater Dual Gradient Drilling. Paper SPE
164561 presented at the IADC/SPE Managed Pressure Drilling and Underbalanced Operations Conference and Exhibition held in San
Antonio, Texas, USA, 17-18 April 2013.

Appendix A Hydraulic Diameter for a Rectangular Channel


In the case of a rectangular gutter the cross-sectional area is:
=

(A-1)

! = + 2

(A-2)


+ 2

(A-3)

and the wetted perimeter is:


therefore the hydraulic diameter is:
! =

Appendix B Hydraulic Diameter for a Partially Filled Pipe


Case where :
If the liquid height is within the bottom half of the pipe ( ) then we can calculate the area occupied by the fluid by
deriving the surface of a circular segment AB as defined on see Fig. 22.

Fig. 22: Partially filled pipe

The angle between the two rays CA and CB is:


(B-1)

The area of the circular segment is the difference of the area of the circular sector and the area of the adjoin triangles:
= 2 cos !!

!
sin
2

(B-2)

The wetted perimeter is then:


! = 2 cos !!

(B-3)

Thus the hydraulic diameter is:


! =

Case where > :

sin

4 cos !!

(B-4)

24

SPE/IADC 166801

When the liquid height is above the median section of the pipe, the derivation of the hydraulic diameter is simply done by
taking the complementary area and perimeter defined in eq. B-2 and B-3:
! =

2 sin

2 cos !!

, = 2 cos !!

(B-5)

Appendix C Hydraulic Diameter for a Semi-cylindrical Channel


If the liquid level is within the bottom half, then the hydraulic diameter is the same as the one defined in eq. B-6. Otherwise
( ), the cross-sectional area is:
=

1 !
+ 2
2

(C-1)

and the wetted perimeter is:


! = + 2

(C-2)

1 !
+ 2
! = 2

+ 2

(C-3)

Therefore, the hydraulic diameter is:

Das könnte Ihnen auch gefallen