Sie sind auf Seite 1von 11

SPEIIADC 25710

Advanced Influx Analysis Gives More Information Following a Kick


John Billingham and Martin Thompson, * Schlumberger Cambridge Research, and
D.B. White, * Sedco ForexlSchlumberger
lADe Members
'SPE Members

Copyright 1993, SPE/IADC Drilling Conference.


This paper W\lS prepared for presentation at the 1993 SPElIADC Drilling Conference held in Amsterdam 23-25 February 1993.
This paper was selected for presentation by an SPElIADC Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper,
as presented, have not been reviewed by the International Association of Drilling Contractors or the Society of Petroleum Engineers and are subject to correction by the author(s). The
material, as presented, does not necessarily reflect any position of the SPE or IADC, their officers, or members. Papers presented at SPElIADC meetings are subject to publication
review by Editorial Committees of the SPE and IADC. Permission to copy is restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should
contain conspicuous acknowledgment of where and by whom the paper is presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A. Telex, 183245 SPEUT.

ABSTRACT

Industry standard kick control and kill procedures are very


simple and rely on calculations which can be performed by
hand under conditions of extreme stress (see [2]). The den.sity of the influx, and. hence its type, can ~_ estimated.from
a simple hydrostatic calculation using the measured pit gain
at shut in and the difference between the stabilised surface
pressures in the drillstring and the annulus. In addition,
the formation pressure can be calculated from the stabilised
drillpipe pressure. Finally, the maximum casing shoe pressure during the control/kill procedure can be calculated by
considering the change in hydrostatic pressure as the influx
moves up the well. These analyses assume that the influx
fills the annulus and exists as a single bubble.

We present an efficient, new procedure for characterising


a kick by analysing changes in surface pressures, surface
mud flow rates and pit level. This procedure allows us to
determine, faster and more reliably than is possible with
conventional methods: the times when the influx started
and stopped, the formation pressure, the influx type (even
for a horizontal well), the formation permeability and the
maximum casing shoe pressure. The performance of this
procedure has been tested using data from both a gas kick
simulator and a series of controlled experiments in a 4500 ft
test well. We present examples of our results, which show
good agreement between predictions and observations. This
analysis has been implemented in a rig based monitoring
system, [1].

There are a number of disadvantages to these calculations:


1. Under severe stress, errors can be. made during the

hand calculation of the various important quantities.


2. The determination of the time when the influx ceases
is made by simply looking at a plot of the surface
pressures, or even just a mechanical gauge. This
method can be inaccurate, and the driller may even
miss the end of the influx altogether.

INTRODUCTION
It is well-known that an influx of gas into the wellbore
from a permeable formation (a kick) can be extremely dan-

gerous. An uncontrolled kick can develop into a blowout,


which threatens the life of rig personnel, destruction of the
rig and significant environmental damage. It is very important that a kick is detected and the blowout preventors
(BOPs) closed as rapidly as possible in order to stop the
formation from flowing and secure the well. The influx
must then be circulated out and the well killed. During the
kill, bottomhole pressure must be maintained high enough
so that no further influx occurs, whilst the pressure at the
casing shoe must remain below the fracture pressure. Estimates must be made of the formation pressure, the influx
volume and type (gas/liquid/mixture) and the greatest casing shoe pressure during the kill. The accuracy of these
estimates is vital to the success of the operation.

3. Inaccuracies in pressure measurements can give erroneous, and even negative, estimates of the influx
density.
4. The determination of the influx density, and hence
influx type, relies on a pressure difference between
drillpipe and annulus. This will not work for horizontal wells.
5. The assumption that the influx exists as a single bubble can lead to a significant overprediction of the maximum casing shoe pressure and an inaccurate estimate
of influx density.
6. There is much more information available in the surface measurements than is currently used.

333

ADVANCED INFLUX ANALYSIS GIVES MORE INFORMATION FOLLOWING A KICK

We present a new method of analysing surface pressures and


mud flowrates which addresses each of these points. By formulating simple models for the behaviour of the downhole
fluids during and after a kick we can determine the expected
behaviour of surface measurements. Simple curve-fitting
then reveals various important downhole quantities.

SPE/IADC 25710

appendix.

Pit gain/delta flow during the kick


By using a simple fonnation model, we can show that the
pit gain, G. is given by,

The analysis can be used for kicks taken whilst drilling with
water-base mud. The method works in verticaVdeviated/
horizontal and slimjnonnal holes. The analysis, which gives
an estimate of fonnation pressure and the influx type. is also
valid in oil-base mud. This procedure has been automated
and include4 in a rig-based monitoring system (see [1]).

Here, Go is the initial pit volume before the influx starts


at time t = t K By fitting this curve to the pit gain, or
equivalently the integrated delta flow, we can determine the
time when the kick started, t K, and the constant, CK, from
which we can calculate, kf).Pdl J1.in, the product of the formation penneability and the drilling underbalance divided
by the influx viscosity. A typical curve is shown in figure 1. This analysis is very similar to that applied to the
draw down and build up of a Drill Stem Test

SEQUENCE OF EVENTS DURING A


KICK
Let us now consider the sequence of events and the qualitative behaviour of surface pressures, mud flow rates and pit
gain during a kick in water-base mud taken while drilling.

Analysis of surface pressures during shut in

Kick: As the bit penetrates an overpressured fonnation,


fluid starts to enter the well. The delta flow and pit gain
increase as the influx fluid enters the annulus.

We use a model for Darcy flow,

Flowcheck: Once the kick has been detected, the pumps are
shut off and a visual check is made for flow out of the well.
During this period, mud may flow down the drillstring and
up the annulus, pulling a vacuum at the top of the drillstring.
This is known as V-tubing.

qJ
qJ

= F f).p
=O

for PJ > Pw
fior PJ :$ Pw

(2)

which simply states that the fonnation flowrate, qJ' is proportional to the underbalance, f).p, whenever the fonnation
pressure, PJ' exceeds the bottom hole pressure, Pw. We
refer to the constant F as the fonnation producibility.

Shutin: The BOP is closed and, if V-tubing has occurred,


the flow from the fonnation will drive mud up the drillstring
until it is full again, whilst the annulus pressure increases.
Both annulus and drillpipe pressures will then increase along
with bottomhole pressure until the fonnation stops flowing.
The surface pressures are then affected by any migration of
the influx. If the influx is gas, surface pressures will rise
as the gas migrates upwards. If the influx is liquid, surface
pressures will not rise significanLly, since the influx only
migrates slowly, if at all.

We analyse the build up of surface pressures after the well


is shut in to detennine the compressibility of the downhole
fluids. The drillpipe pressure is given by,

Pd
CS 1 (I - e--'cs,(t-t u )
for tu :$ t :$ t e

. .(3)
Pd =
CS 1 [1- e-Cs,(tc-tu) {I- C S2 (t - t e )}]
for t > t e

Kill: The well is killed using either the wait and weight
or driller's methods. In each case the influx is circulated
out through the choke, which is used to maintain a sufficient
back-pressure to keep bottomhole pressure above fonnation
pressure.

Here, CS 1 and CS2 are constants which are related to properties of the wellbore fluids and the fonnation (see the appendix). A typical curve given by equation 3 is shown in
figure 2. Once the fonnation flow has shut off, the pressure
rise becomes linear as any free gas migrates. By fitting this
curve to the drillpipe pressure, Pd, we can determine the
time when the fonnation stopped flowing, t e This is the
time at which to take shut in pressures and start to control the well. It is difficult to determine t e accurately using
the standard method, which is simply to wait for surface
pressures to stabilise.

ANALYSIS OF SURFACE MEASURE


MENTS
We now describe the advanced analysis of surface measurements during the kick, flowcheck and shut in phases, along
with the benefits it brings. Full details can be found in the

334

SPE/IADC 25710

J. BILLINGHAM, M. THOMPSON" AND D. WHITE"

. From the drillpipe pressure at t


t e we can calculate the
fonnation pressure, PI' from simple hydrostatics. It is this
calculation of t e and PI which remains possible in oil-base
mud. The other calculations are strongly affected by the
high solubility of gas in oil.

pressure occurs when the leading edge of this bubble of gas


reaches the casing shoe. The expansion of the gas can be allowed for by assuming ideal behaviour. These calculations
overestimate the true pressures. If the shoe is weak, pressure may be bled off to prevent breakdown when it would
not actually have occured.

Next, by estimating the proportion of the measured frictional


pressure loss during drilling which occurs in the drillstring,
we can calculate the bottomhole pressure, and hence the
drilling underbalance, Llpd.

In reality, the influx gas does not exist as a single bubble


but as a distributed, bubbly mixture of gas and mud which
occupies a greater length of the annulus. This means that
the leading edge of the gas reaches the casing shoe earlier
than is usually assumed. In addition, the gas will have
expanded less since the centre of mass of the gas cloud is
lower, so the maximum casing shoe pressure will be smaller
than that calculated with the standard method. Indeed, if the
gas has reached the casing shoe before the start of the kill,
the maximum casing shoe pressure will occur at the start of
the kill. This is often the case when the open hole section
is short.

The delta flow, Llq, just before the pumps are turned off is
a good estimate of the rate of influx from the fonnalion, ql'
From LlPd, Llq and the simple model given by equation 2,
we can deduce the fonnation producibility, F.
From our estimates of the constant, CS2, and the fonnation
producibility, F, we can calculate the influx volume compressibility, Cin Vin. Since we know the influx volume Vin
from the pit gain, we can detennine the influx compressibility, Cin'

An example of the effect of gas distribution on casing shoe


pressure, as calculated by a simple gas kick simulator is
shown in figure 4. In each case the void fraction has a
. different, unifonn value whilst the mass of gas remai,.ns the
same. The lower the void fraction, the greater the length of
the influx. The effect described above can clearly be seen.

The compressibility of a gas is at least an order of magnitude greater than that of a liquid. The calculated value
of Cin is therefore an excellent indicator of the type.orinflux. In particular, this method still works in horizontal
wells where the standard field method for calculating influx
density fails. This has been found to be much more reliable
than conventional indicators in the examples which we have
analysed.
We have calculated LlPd and can estimate /lin from our
knowledge of the influx type. Therefore, we can estimate
the fonnation permeability, k, by using the calculation of
kLlpd/ /lin from the pit gain curve-fit. Whilst this information is not needed for well control, the permeability, k, can
be considered a "risk factor". If the permeability is high, the
well could flow strongly if any mistakes are made during
well control.

Recent studies of the dynamics of a rising cloud of influx


gas in drilling mud enable us to estimate the downhole gas
distribution, [3]. This estimate can be used in conjunction
with the standard assumption of ideal gas behaviour to improve the estimate of the maximum casing shoe pressure.
This lower estimate may be crucial in deciding whether a
given kill procedure is likely to fracture the fonnation. The
details of the calculations are given in an appendix. It is
now possible to use the advanced analysis of the kick development and shut in phases to estimate the casing shoe
pressure at any time during the kill.

The above sequence of calculations is illustrated in figure 3.

The analysis for the determination of:

If a continuous drillpipe pressure measurement is not available (i.e. there is a non-return valve in the drillstring), this
curve fit must be made to the annulus pressure. A single
measurement of drillpipe pressure needs to be taken, as is
nonnal practice. This allows the drillpipe pressure at t t e ,
and hence the fonnation pressure, to be estimated.

shutin pressures,
influx type, independent of wellbore inclination,

earliest time to start controlling the well,

MAXIMUM CASING SHOE PRESSURE

is valid for all mud types. The other calculation~, for formation penneability and shoe pressure, are strongly affected
by the high solubility of gas in oil.

Before starting the kill circulation after a gas influx it would

ANALYSIS VALIDATION

be extremely useful to have an estimate of the maximum

casing shoe pressure during the kill in order to minimise


the risk of fracturing the fonnation. The standard method,
which is not nonnally used on the rig, assumes that the gas
exists as a single bubble which fills the annulus at bottomhole at the start of the kill circulation. The maximum shoe

Validation is a key aspect of any safety related innovation.


Initially, two distinct methods have been used, which are
described below. Further validation is continuing with field
data.

335

ADVANCED INFLUX ANALYSIS GIVES MORE INFORMATION FOLLOWING A KICK SPE/IADC 25710

Data from a commercial gas kick simulator


.Firstly, we consider data from a numerically simulated methane
kick. The wellbore geometry and physics are summarised
in table 1. The kick is take by drilling into an overpressured
fonnation with an underbalance of around 300 psi. the influx is detected at a pit gain of 10 bbl, the pumps shut off
and a hard shutin perfonned. This shut in is rapid enough
that no V-tubing occurs.
Figure 5 shows the pit gain whilst the kick is taken, along
with the curve fit of equation 1. This non-linear curve fit
was perfonned using a standard method (see [4]). This
calculates the start time of the kick, tk, to be 306 s, in
excellent agreement with the actual time, 305 s.
Figure 6 shows the drillpipe pressure after shut in, along
with the curve fit of equation 3. The curve fit detennines
the time when the influx ceases, t e , to be 917 s, within one
minute of the actual time, 865 s. The fonnation pressure is
then calculated to be 8520 psi, in good agreement with the
actual value of 8500 psi. From this we estimate the underbalance just before the pumps are turned off to be 380 psi.
At this time the delta flow was 3.9 bbl/min, so we calculate
the fonnation producibility, F, defined in equation 2, to be
1.0 X 10- 2 (bbl/min)/psi. This then allows us to calculate
the influx compressibility from the curve fit shown in figure 6. We find that CS 2
2.7 X 10- 2 8- 1 , and hence that
cin
1.2 X 10- 4 pSi-I; about 25 times that of oil. The actual compressibility of the influx at bottomhole temperature
and pressure is 8.2 x 10- 5 psi-I.

In order to test the validity of our analysis we perfonned a


series of experiments ina well at the International Drilling
and Downhole Technology Centre (lDDTC) in Aberdeen.
This test well has been completed with 14 inch easing, with
a second 9 5/8 inch easing hanging freely within it There
is extensive instrumentation, both on surface and downhole,
including all nonnal oilfield sensors.
During the experiments, nitrogen was injected into the well
through coiled tubing. This ran from the surface to bottomhole between the two concentric easings and into the
inner easing via a non-return valve. Great care was taken
to ensure that the nitrogen injection rate would represent a
real fonnation flow. We assumed Darcy flow, so the experiments did not test the validity of the fonnation modelling
in the analysis.

The difference between our estimate and the actual value


is due to the simplifications made in the analysis. In spite
of the difference between the estimated and actual influx
compressibilities, it is obvious that the influx is gas rather
than liquid. This illustrates the robust nature of the method.
It is interesting to compare our new technique with the standard method used to detennine the influx type. This relies
on an estimate of the influx density from,
Pg

= Pm -

Pa - Pd
I
0
9 cos

fI

By estimating the maximum easing shoe pressure with the


method described in the appendix, we find that the expansion of the gas as it moves from bouomhole to the easing
shoe is negligible. Both the standard method and our modified method give a maximum casing shoe pressure of around
6540 psi. The actual maximum easing shoe pressure after
the start of the kill is about 6450 psi. The 90 psi difference
is due to the frictional pressure loss across the gas cloud,
which we have neglected in our analysis. This means that
the estimate is conservative. It is worth noting that the easing shoe pressure, illustrated in figure 7, was at its highest
during the shut in period before the start of the kill.

Data from a full-scale experimental kick

fore estimate the fonnation permeability. Our estimate is


60 mdarcy, whilst the actual value is 79 mdarcy. Although
this will not replace well testing, we have. a good estimate
of the order of magnitude of the risk associated with any
mistakes made during the well control operation.

Many experiments were perfonned for different "fonnations" and kick volume. Let us now consider in detail a
typical experiment. The wellbore geometry and physics are
given in table 2. Note that there was no easing shoe the casing ran from surface to bottomhole. However, since
there was a pressure tapping at a vertical depth of 2304
f1, where a measurement was made, we have assumed a
nominal casing shoe depth of 2304 ft

(4)

The difference between the shulin drillpipe and casing pressures was 368 psi, and an influx of 36 bbl occupies a vertical height 529 fl in the annulus. With these values, the
conventional analysis gives a negative influx density. The
error arises from the assumption that the gas exists as a
single bubble at bottomhole around both the drillstring and
the collars. This lcads to an overestimate of the volume of
gas around the coIlars and hence of the hydrostalic pressure
change across the gas cloud.

Gas was injected at an appropriate rate whilst mud was


circulating at 6.9 bbl/min. Once the pit gain reached 5
bbl the pumps were stopped and a hard shut in performed.
V-tubing of mud down the driIlstring and up the annulus
occurred before the BOPs were fuIly closed. A total of 7.5
bbl of gas was injected at bottomhole.

We now know the drilIing underbalance and thal the influx


is gas. From the curve fil to the pit gain, we have an estimale of the product of fonnalion penneabilily and drilIing
underbalance divided by influx viscosity. We can lhere-

Figure 8 shows the pit gain whilst the kick was taken, along
with the curve fit of the equation I. The calculated start time
of the kick is 656 s, in good agreement with the actual time,
650 s, when the gas injection began.

336

SPE/IADC 25710

J. BILLINGHAM, M. THOMPSOW AND D. WHITE-

Figure 9 shows the annulus and drillpipe pressures after


shutin, along with the curve fit of the equation 3. The rise
in the annulus pressure for 1120s < t < 1160s whilst the
drillpipe pressure remains constant clearly indicates that Utubing occurred and that the drillstring is refilling during
this period. The time when the influx ceased, teo was calculated to be 1246 s (actual value, 1230 s). The apparent
formation pressure was then calculated to be 2250 psi, with
an apparent drilling underbalance of 220 psi.

- The ratio of gas and liquid compressibilities is


much greater than the ratio of their densities compressbility is a more robust measure of influx type.
- The standard calculation of influx density is inaccurate since the volume of the influx in the
annulus around the collars is overestimated. In
addition, this calculation is sensitive to small errors in the surface measurements. The estimate
of the influx compressibility is insensitive to the
detailed distribution of the influx, a small error
in the measured pit gain and to a constant error
in the drillpipe pressure measurement

The delta flow just before the pumps were turned off was
1.4 bbVmin, so the apparent formation producibility, F, was
6.4 x 10- 3 bbVmin/psi. The constant, CS 2 , was estimated
to be 4.3 X 10- 2 S-I, and hence we calculate the influx
compressibility, Cin' as 2.5 x 10- 4 psi-I; about 50 times
that of oil. Clearly the influx was gas.

- The estimate of the influx compressibility works


in horiwntal wells where the standard method
fails.

The difference between the shutin drillpipe and casing pressures was 63 psi, and an influx of 7.5 bbl occupies 259 ft
in the annulus. The field calculation gives an influx density of 3.9 ppg, compared to an actual influx density of
1.4 ppg. The standard method wrongly suggests a inixed
gas/liquid kick. This error can be attributed to small errors
in the measured pit gain and surface pressures, to which the
calculation given by equation 4 is very sensitive.

We are able to estimate the formation permeability


from our analysis of pit gain and drillpipe pressure
measurements.
Recent studies of gas dynamics during a kick enable
os to estimate the. distribution of the influx. This allows us to estimate the maximum casing shoe pressure
during the kill. This is more accurate than the estimate obtained using the single bubble approximation.

By calCUlating the position of the influx using the method


described in the appendix, we find that the gas reaches the
casing shoe at about the same time as the kill circulation
begins. We estimate a maximum casing shoe pressure of
1390 psi which decreases during the kill as the gas passes
the shoe. The measured casing shoe pressure is shown in
figure 10, where it is clear that the maximum casing shoe
pressure during the kill is about 1400 psi. The standard
method predicts a maximum casing shoe pressure about
50 psi higher.

These analysis techniques have been implemented in


a rig-based monitoring system which can advise the
driller during the course of a kick.

Nomenclature
A
Aa

Validation of the analysis methods presented here continues


with the collection of field data.

A coil

AN
C

CONCLUSIONS

CK
CS 1
CS 2

Standard kick calculations can be significantly improved and currently do not make efficient use of the
information available in surface measurements.
A reappraisal of possible analysis methods both during and after a kick leads us to new techniques.
The measured pit gain during a kick can be analysed
to deduce when the kick started.
The drillpipe or annulus pressure during shut in can
be analysed to deduce when the influx stopped, the
formation pressure and the influx compressibility.
This estimate of influx compressibility is a better indicator of influx type than the standard estimate of
influx density.

337

area of a section of drillstring or annulus


cross-sectional area of annulus around drillstring
cross-sectional area of annulus around collars
total area of bit nozzles
gas distribution factor
constant in pit gain curve fit
constant in drillpipe pressure curve fit
constant in drillpipe pressure curve fit
influx compressibility
mud compressibility
total vertical depth
bit nozzle discharge coefficient
formation producibility
pit gain
pit gain offset
gravitational acceleration
vertical extent of influx
vertical distance from bit to shoe
formation permeability
length of a section of drillstring or annulus
axial length of collars
axial length of influx

I..
n

Pd
Pd/r
p!

Plr
Ph
Phe

Pmax
Pnozz!r
Pw
pwU

qdr
V
r

Th
The

t
to
te
tv
tv
tK
to!!
tu
V;n

Vm
Vg
V,l
X

!::J.p
!::J.Pd
!::J.q

pin
pge

Pin

Pm
Ldr
Ldr+ann
w

ADVANCED INFLUX ANALYSIS GIVES MORE INFORMATION FOLLOWING A KICK SPE/IADC 25710

axial distance from bit to casing shoe


constant in pit gain curve fit
drillpipe pressure
frictional pressure loss in drillstring
fonnation pressure
total frictional pressure loss
pressure at midpoint of influx
pressure at midpoint of influx when t t e
maximum casing shoe pressure
frictional pressure loss across bit nozzles
bottomhole pressure
bottomhole pressure when t tu
volume flowrate
pump rate during drilling
volume flowrate from fonnation
bit radius
temperature at midpoint of influx
temperature at midpoint of influx when t t e
time
fonnation time scale
time when fonnation stops flowing
dimensionless fonnation time
dimensionless fonnation time since penetration
time when influx started
time when pumps are stopped
time when U-tubing has recovered
influx volume
volume of mud in well
gas velocity
gas slip velocity
distance from top of fonnation
fonnation underbalance
fonnation underbalance during drilling
delta flow
inclination of open-hole section
influx viscosity
gas density when fonnation stops flowing
influx density
mud density
sum over all sections of drillstring
sum over all sections of drillstring and annulus
fonnation porosity
rate of penetration

[4] Press, W.H., Flannery, B.P., Teukolsky, S.A. and Vetterling, W.T.: Numerical Recipes. Cambridge University Press. (1986) 523.
[5] Dake, L.P.: Fundamentals of reservoir engineering,
Elsevier (1978) 154.

[6] Govier, G.W. and Aziz, K.: The flow of complex mixtures in pipes. Krieger (1972) 322.
[7] Collins, R., De Moraes, F.F., Davidson, J.F. and Harrison, D.: "The motion of a large gas bubble rising through liquid flowing in a tube," J. Fluid Meek.
(1978) 89, 3,497.

APPENDIX
Derivation of pit gain curve fit
We assume that the bit enters a large. horiwntally-bedded,
over-pressured reservoir, where the flow from the fonnation
is radial and axisymmetric. The analysis given in [5] shows
that, to a good approximation, the flow rate per unit height
of producing fonnation is,
21rk

q -;::, -

pin

Ap
1 (
)
2' In tv + 0.81

(A-I)

Here, the dimensionless time is tv


(t - tK) Ito, the
JPinCinr2 I k and the fonnation
reservoir timescale is to
starts to flow at t
t K. This result can be used to show
that the rate at which the flow rate changes with penetration
into the fonnation is

dq
-d
X

where, tv

21rk

= -Pin

= {w (t -

Ap
1 (.

2' In tv

+ 0.81

) .......

(A-2)

tK) - x} Iwto.

There is no analytical expression for the integral of this


equation. However, we can show that,
k!::J.p
n-l
q ex: (t - tK)

References

...

(A-3)

Pin

By integrating this expression with respect to time we obtain


the pit gain as,

[1] Jardine, S.I., White, D. and Billingham, J.:


"Computer-aided real-time kick analysis and control,"
paper SPE/IADC 25711.

G -;::, CK (t - tK

[2] Moore, L.P.: Drilling Practices Manual, PennWell,


Tulsa (1986) 497.

(A-4)

This equation suggests that we can fit a curve of the fonn


given by equation I to the measured pit gain data. In equation 1 the constant Go allows for a possible constant offset
in the pit gain data.

[3] Johnson, A.B. and White, D.: "Gas rise velocities during kicks," paper SPE 20431.

338

J. BILLINGHAM, M. THOMPSON" AND D. WHITE-

SPE/IADC 25710

= Pw -

Derivation of drillpipe pressure curve fit after


shut in

From simple hydrostatics Pd


at equation 3.

After the well has been shut in and V-tubing has recovered,
the well is sealed and full of a mixture of drilling mud
and influx fluid. The fonnation continues to flow until the
bottomhole pressure is higher than the fonnation pressure.
The rise in wellborepressures is driven by two mechanisms:

Note that the constant C S2, from which we derive the compressibility of the influx after curve-fitting, is independent
of the rate of gas migration. The gas rise velocity only
appears in the constant C S1 . The constant CS 2 remains
independent of V,I when the effects of gas solubility and
wenbore temperature gradients are included in the analysis.
We can still estimate Cin Vin from the constant CS2

PmgD and we arrive

1. The continuing influx "pumps up" the closed system

of wellbore fluids.

Estimation of the frictional pressure loss in the


drillstring

2. The migration of the influx fluids carries the bottomhole pressure upwards.

The pump pressure during drilling gives an estimate of the


total frictional pressure loss in the system, PJr. We can
estimate the pressure loss across the bit nozzles using the
correlation,

This can be expressed mathematically as


dpw/ dt = dpd/dt =
(cmVm + Cin \l;n)-1 (qJ

+ Cin \l;nPmgv.d

. (A-5)

PJrnozz

Note that, when qJ = 0 (no flow from the fonnation), equation A-5 simply gives the rate at which the drillpipe pressure
rises because of the migration of the influx. If we then neglect the compressibility of the drilling mud (cm = 0) we
obtain the standard equation which is used to deduce the
influx migration rate, V.I, from the rate of rise of drillpipe
pressure, [2].

e-Cso(l-I"

(~)2
DNAN

(A-8)

This allows us to calculate the bottomhole pressure at any


time before the pumps are turned off.

Maximum casing shoe pressure

If we now use the simple formation model given by equation 2 to give the formation flowrate, qJ in equation A-5
and solve the resulting equation, we obtain

pwu + CS 1 (1 for tu ~ t ~ t c

The discharge coefficient, D N , is approximately 0.95. We


now assume that the rest of the frictional pressure loss is
distributed between each section of the drillstring and the
annulus in proportion to the length of the section and the
inverse of the area squared. The frictional pressure loss in
the drillstring is then given by

Equation A-5 shows that mud compressibility reduces the


estimate of the rate of rise of drillpipe pressure for a given
influx migration velocity. In the two examples described in
this paper the reduction is about 50% for the numerically
simulated kick and 75% for the experimental kick. In addition, gas solubility in drilling mud (even for water-base
mud), the dependence of gas density on wellbore temperature and other effects reduce the rate of rise of drillpipe
pressure for a given migration velocity.

Pw

= .!.Pm

The experimental work described in [3] has shown that slug


flow (see, for example, [6]) occurs during a gas kick. In
slug flow, gas rises as a series of large bubbles, each of
which spans the whole of the annulus. In such flows, gas
velocity can be detennined from the correlation,

J)
(A-6)

Pw =

Pwu
for t

+ CS 1 [1 -

e-Cso(lc-lu)

{l- C S2 (t -

Vg =

ten]

> Ie

(A-IO)

Here, q is the total fluid flowrate, A is the annulus crosssectional area, C is a distribution factor and V.I is a bubble
slip velocity. The correlation is discussed in [7]. This allows us to estimate the length of the cloud of influx gas
after the formation has stopped flowing as,

Here, Pwu is the bottomhole pressure at time tu when Vtubing recovers, and

CS1 = PJ - PwU + Ci" ~i"PmgV.I/I


CS 2 = F/ (c;n \.~" + Cm \'~n)

Cq/A + V.I

... (A-7)

339

ADVANCED INFLUX ANALYSIS GIVES MORE INFORMATION FOLWWING A KICK SPE/IADC 25710

GEOMElRY DATA

Ig

= V_I (tc -

for Ig

to!!)

+ (ACqdr + V_I) (toff


co"

OD
ID
Length
OD
ID
Length
ID
OD
Length
Diamelec
Nozzles
Nozzle size
ID
Length

Drillpipe

- tK)

< I coll

Hcaviweighl-<lrillpipe

(A-ll)
Collars

Bil

If Ig ~ 1_. the maximum casing shoe pressure occurs at the


start of the kill. Next. we:

Casing
Casing shoe vertical depth
TOlaI vertical depth
Openltole inclination

neglect friction in the annulus.

5"
4.28"
14708 ft
5"
3"
897 fl
2.62"
633"
203 fl
8.5"
5
15/32"
8.53"
lOooO ft
8351 ft
11996 ft
51

PHYSICS DATA

assume ideal gas behaviour,

Mud

take the gas pressure to be that at the midpoint of the


gas cloud.

Gas

When the formation stops flowing at time t i c the pressure, Phc. at the midpoint of the gas cloud is,

Densily
Compn:ssibilily
Volume
Type
Densily
Viscosily
Compn:ssibilily
Volume

Fomwion permeabilily
BOllOmhole Iempel3tUle
Tc:mpecature gradienl
ROP
Drilling pump I3le

123 ppg
2.1xl0- psi
lOoo bbl
Mcd1ane
1.7ppg
3.8x 10-' Pa s
8.2x 10 -. psi36 bbl
79 mdarcy
3261"
2.141"/100 ft
100 ft,lhr
480p1/min

EVENTS

(A-12)

Start of kkIc
Pumps off
I'onnaLion SlOps flowing
SIaI1 of Idll

306s
665s
865 s
1300s

Table 1: Physical properties and wellbore geometry for the


numerically simulated kick.
When the gas first reaches the casing shoe. the pressure, Ph,
at the midpoint of the gas cloud is,

GEOMElRY DATA
OD
10
Length
ID
OD
Length
DiameleC
Nozzles
Nozzle size

Drillpipe

Ph
Pw - Pm9 (h. - tlg cosO)
V;n9 (cmPhcTh _ P c) cosO
2A.
Ph T hc
9

+
.... (A-B)

Collars

Bil

This is a simple quadratic equation for Ph. The maximum


casing shoe pressure is,
Pmax

= Pw

- Pm!lh.+

Casing ID
Nominal casing shoe vertical depth
Tolal vertical depth
Inclination

PhThc

This is identical to the standard calculation. except that we


derive the length of the gas cloud from equation A-II, rather
than from a single bubble modr.I.

10132"
8.5"
2304 fl
4638fl

9"

PHYSICS DATA

~ (cmPt.rTt. _ p c) cosO' . (A-14)


Aa

5"
4.Z"
4428fl
2.5"
6.5"
uO t
8.25"
5

Mud

Gas

Densily
Comprcssibilily
Volume
Type
Densily
Viscosily
Compn:ssibilily
Volume

Bouomhole leCnpetature
Tc:mpetature gradiClll
Drilling pump rate

8.5 ppg
3.2x 10 -. psi
289 bbl
Nitrogen
1.4 ppg
1.9xl0- Pa s
psI
4.5xl0
7.5 bbl
70"1"
negligible
292 p1/min

EVENTS
SIaI1 of kick
Pumps off
Recovery from V-wbing
FormationslOpsfiowing
SIaI1 of kill

650s
1095 s
1160s
1230s
1850 s

Table 2: Physical properties and wellbore geometry for the


experimentally simulalC(; kick.
340

J. BILLINGHAM, M. THOMPSON AND D. WHITE

SPE/IADC 25710

V-Iubing recovers
I=I U

.....
+-'

0..

slart of kick
1= I

0=00

fonnalion SlOpS
producing

t
t::===================------------

time, t

= Ic

time, t

Figure 1: The curve used to fit the pit gain during the kick.

Figure 2: The curve used to fit the drillpipe pressure after


shut in.

PIT GAIN
CURVE FIT

,
CK

~,

1------------11

Figure 3: The sequence of calculations. Estimates of important quantities are shown in circles.

void fraction = 0.15


void fraction = 0.25
void fraction = 0.5
void fraction = 0.75
void fraction = 1
_.-._-_.-

Figure 4: The casing shoe pressure during a simulated kill


for various uniform void fractions with the same mass of
gas.

341

10

ADVANCED INFLUX ANALYSIS GIVES MORE INFORMATION FOLLOWING A KICK SPE/IADC 25710

25
data
r-..
.......

.0
.0

'-"

.S
C':j

bf)

....
.......

0..

20

curve fit

.......

t5
.

.......'

'

.'

10
start of kick

5
0

200

100

....

....

400

300

.'

500

600

700

time (s)
Figure 5: The curve fit to the pit gain in the numerically
simulated kick.

.-

r-.. 1000

estimate

en

0..

'-"

data
curve fit

750

,r

;:j

en
en

500

0..

.Q)

0..
250
0..
.......

formatIOn stops
producing

.......

'1::

'"0
800

850

900

950

1000

1050

1100

time (5)
Figure 6: The curve fit to the drillpipe pressure in the numerically simulated kick.

Q6600
en

......._ _.._ predicted maximum


...maximum shoe pressure
after start of kill

0..

'-"

6400

;:j

en 6200
en

start of

0.. 6000

kill

.8

5800

<Zl

bf)5600

.S
en
cd

5400

L-_~---'-

_ _~_..l-_~_---.JL.._~_-L_~~--.J
2000
4000
6000
8000
1oo

time (s)
Figure 7: The casing shoe pressure in the numerically simulated kick.
342

J. BILLINGHAM, M. THOMPSOW AND D. WHITE

SPE/IADC 25710

data
4

,.-...

.......

,.0
,.0
.........,

...

curve fit

.5ro

OJ)

......

.0..

stall of kick

................ -

400

700

600

500

1000

900

800

1100

time (s)
Figure 8: The curve fit to the pit gain in the experimentally
simulated kick.

300

.-

,.-...

V-tubing

250

recovers

en
0.. 200
.........,
(1)
l-<

en
en

BOPs shut

ISO

::l

---------------------------

//
,/

/-

V/' .

I
'V

drillpipe pressure

100

// /

0..

0
1100

curve fit

'0="'00 "0"

!:

SO

---------------

estimate

annulus pressure

producing

-----------------

1150

1200

1250

1300

1350

1400

time (s)
Figure 9: The curve fit to the drillpipe pressure in the experimentally simulated kick. The annulus pressure is also'
shown to demonstrate that V-tubing has occurred.

;,:;- 1500

en
0..
.........,
~

1400

predicted maximum
standard method

new method

::l

en
en
~

1300

0..
~ 1200

..c
en

.-

OJ) 1100

l::
en
~

1000 '-._ _~_---ll_ _~_ ___L!--~ _ _-I!L-_~_ ___.JI


o
1000
2000
3000
4000

time (s)
Figure 10: The casing shoe pressure in the experimentally
simulated kick.

343

11

Das könnte Ihnen auch gefallen