Вы находитесь на странице: 1из 14

SPWLA 45th Annual Logging Symposium, June 6-9, 2004

FORMATION TESTING
IN THE DYNAMIC DRILLING ENVIRONMENT
Mark Proett, Halliburton Energy Services, Douglas Seifert, Saudi Aramco,
Wilson Chin, Solveig Lysen and Paul Sands, Halliburton Energy Services

Copyright 2004, held jointly by the Society of Petrophysicists and pressure changes can also occur due to drill pipe
Well Log Analysts (SPWLA) and the submitting authors.
th
This paper was prepared for presentation at the SPWLA 45 Annual
movement causing a swabbing effect. These pressure
Logging Symposium held in Noordwijk, The Netherlands, June 6–9, transients can be transmitted through the mudcake
2004. and into the formation. The magnitude of these
pressure transients are simulated using the forward
ABSTRACT modeling methods developed in this paper.
Formation Testers While Drilling (FTWD) tools have Field data from the new GeoTAP* tool are
recently been introduced and offer an alternative to presented to demonstrate the practicality of FTWD.
wireline testing. Several questions remain regarding The new tool requires the drill pipe to stop
the accuracy and repeatability of these measurements. rotating/sliding for about 7 minutes per pressure test.
First, the drilling environment is dynamic with During this time the pumps can be turned on or off.
hydrostatic pressure constantly changing. Some The field example demonstrates the robustness of the
FTWD tools offer the capability to perform pressure measurement. The field example included three
tests with the mud pumps on or off. Invasion and the repeat pressure tests taken at three different depth
resultant supercharge effect can also be factors points where, in each case, the pressures recorded
affecting FTWD measurements. The emphasis of this were within a few psi of each other. Final
paper is to consider these factors through forward conclusions are drawn regarding FTWD technology
modeling methods as well as field examples. and its future direction.
N
Several forward models are developed in this INTRODUCTION
paper. The simplest assumes single phase invasion
where the mudcake buildup is loosely coupled to the When wireline formation testers (WFT) were
filtrate invasion. The second is a two-component introduced decades ago there was a debate in the
fluid model where the mud filtrate invades and industry about the significance of these pressure
displaces the formation fluid. The third is a measurements. How did they compare to well testing
multiphase finite difference model that considers results and could they be relied on for formation
miscible and immiscible invasion that also couples evaluation? After many years and a number of
mudcake growth to invasion. improvements in WFT technology, the tables have
turned; and WFT pressure measurements are now
Invasion can result in FTWD or Wireline considered the standard for formation evaluation.
Formation Testers’ (WFT) pressure measurements One of the contributing factors is the consistent
being elevated above formation pressure. This creates manner in which wireline formation tests are
a supercharging effect as a result of mud filtrate conducted as opposed to production or drill stem
invasion increasing the sandface pressure just behind testing where procedures and equipment can vary
the mudcake. A time lapse technique is used that widely. Also, wireline tests can be repeated many
enables the FTWD tool to estimate the mudcake times over a zone of interest and the pressures used to
properties and invasion rate. This information can be determine fluid gradients and flow barriers. More
used in conjunction with the simulators to determine recently, WFT tools are being viewed as replacement
the supercharge magnitude. for DST and production testing. WFT tools are
Pressures in the borehole are constantly changing inexpensive, safer, and have improved ability to
due to several factors. It is usually desirable to keep obtain clean samples and to estimate permeability
the mud pumps on while testing to avoid hole
deterioration and pipe sticking. Slower hydrostatic
* A mark of Halliburton Energy Services

1
SPWLA 45th Annual Logging Symposium, June 6-9, 2004

(Proett, et al., 2000). pressure stability is studied over several days of


FTWD testing.
With the introduction of FTWD tools a new
debate has begun about the significance of these
PRESSURE STABILITY IN THE DRILLING
tools’ pressure measurements. How do they compare
ENVIRONMENT
to wireline measurements and can they be relied on
for formation evaluation? This debate centers on the Pressure variations near the wellbore are primarily
fact that FTWD tools can take measurements much influenced by mud filtrate invasion and mudcake
sooner in the life of a well in a potentially more formation. Considerable progress has been made in
dynamic environment. Invasion has always been a understanding how mudcakes form and influence the
concern in all types of pressure testing, but FTWD near wellbore pressure stability (Wu, et al., 2001;
pressure measurements have brought these concerns Dewan, et al., 2000; Sharma and Jiao, 1994). Based
into focus. on this research, miscible and immiscible multiphase
Invasion simulations and their effects on simulators have been developed that predict the
formation pressures have been subject to increased filtrate invasion for OBM and WBM (Chin, 2002). In
interest in the last decade. The primary efforts have reviewing this work, it is apparent that, even though
been directed to wireline sampling and, to some the problem is complex, it is possible to make
degree, to pressure testing. Most papers dealing with simplifying assumptions and still obtain a reasonable
this problem use detailed numerical simulators to estimate of the near wellbore pressures.
study these effects (Wu, et al., 2001, Proett, et al., In order to control production of formation fluids
2002). While these analyses can produce useful into the wellbore, wellbore pressure is normally
results, they are usually cumbersome to use and, as a maintained at a pressure substantially greater than the
result, are not routinely used for day-to-day formation pore pressure. When a producing zone is
evaluation of formation tester data. penetrated, the wellbore sandface is exposed to
It is the objective of this paper to develop a range hydrostatic pressure and filtrate immediately invades
of methods to analyze pressure tests for the effects of the near wellbore. Mudcake is formed when drilling
invasion and supercharging. The first two are fluid flows into the formation and solids are
analytical and consider single and multi-component deposited on the surface of the wellbore. This process
fluid invasion. These methods are closed form is normally referred to as static filtration. As the
solutions and can be used for spreadsheet analysis. mudcake grows it eventually stabilizes to a maximum
The analytical solutions can yield a first order thickness. Stabilization is a result of the shearing
estimate that in many cases is adequate to evaluate action of the mud circulation in the annulus as well as
the conditions where supercharging can be mechanical action of the rotating drill pipe. This
significant. process is referred to as dynamic filtration. During
these processes a pressure gradient is established in
A detailed finite element simulator that is
the formation (see Fig. 1). The pressure in the
optimized for detailed invasion analysis is introduced
wellbore near the surface of the mudcake is at
with this paper. This simulator can be used for either
hydrostatic (Pmh) but drops rapidly across the
miscible or immiscible invasion. The mudcake is also
mudcake and then gradually reduces in the formation,
simulated as with the analytical models but in this
approaching formation pressure (Pf) some distance
case is tightly coupled to the invasion process.
away from the wellbore. The supercharge pressure
The emphasis of these simulators is to predict the (Psc) can be defined as the difference between
dynamic pressure changes that can occur during sandface pressure (Pss) and formation pressure (Pf).
openhole drilling operations. These dynamic pressure
changes affect not only FTWD tools but WFT
measurements. Single Phase Supercharge Model.
Assuming single phase Darcy flow, the supercharge
Log examples are used to illustrate how FTWD
pressure can be predicted using the familiar radial
tools perform and are compared to the predictions.
flow equations for an infinite homogeneous reservoir.
Repeat pressure measurements are presented where

2
SPWLA 45th Annual Logging Symposium, June 6-9, 2004

⎡ q µ ⎤ ⎡ 4k f t ⎤ Eq. 3 can be reduced to a simpler form so that the


∆Psc = Pss − Pf = ⎢ m ⎥ ln ⎢ 2⎥
(1) effect that these non-dimensional parameters have
⎢⎣ 4πhk f ⎥⎦ ⎣ γφµcrw ⎦ can be studied.

τ Dmc
p Dsc (t ) = ln(t D ) (7)
2
The dimensionless supercharge pressure pDsc is
the relative degree of supercharging normalized to
the apparent overbalance. The apparent overbalance
∆Pob is the difference between hydrostatic mud
pressure and the sandface pressure. The term pDsc is
the ratio of the actual supercharge ∆Psc to the
apparent overbalance ∆Pob which can be measured
using a formation tester.
Dimensionless time tD determines the transient
response of the supercharging. Its definition is the
Fig. 1—Supercharge effect where hydrostatic pressure same as that used for transient well testing.
and filtration loss cause the sandface pressure Pss to be
The mudcake transmissibility ratio τDmc is the
elevated above formation pressure Pf.
most important dimensionless constant because it
Assuming the mudcake is relatively thin determines the overall supercharge effect based on
compared to the wellbore diameter (i.e., lmc << rw), the mudcake and formation properties. It is a
the flow through the mudcake can be modeled as a measure of the relative resistance to filtrate invasion
linear Darcy flow with the pressure differential from the mudcake versus the formation resistance. If
between the borehole mud hydrostatic (Pmh) and the the transmissibility ratio is small, the mudcake
sandface supercharge pressure (Pss) creating the mud dominates the filtrate invasion and supercharging is
filtrate loss (qm). small. If the transmissibility ratio is large, invasion is N
primarily influenced by the formation; and
2πrw hk mc
qm = ( Pmh − Pss ) (2) supercharging is relatively high.
µ l mc
Now using Equations 1 and 2, an expression for Dynamic Mudcake Growth Model. Two primary
the supercharge pressure can be determined in terms models exist for predicting mudcake growth. The
of the hydrostatic pressure and sandface supercharge most general model was developed for radial flow by
pressure as well as the formation and mud properties. Chin (2002).
⎛ ⎛ lmc ⎞ 1 ⎞ 1
2
r k ⎡ 4k f t ⎤ 1 ⎛ lmc ⎞
∆Psc = ( Pmh − Pss ) w mc ln ⎢ ⎥ (3) ⎜1 − ⎟⎟ ⎜ ln⎜1 − ⎟ − ⎟ +
2l mc k f ⎣ γφµcrw2 ⎦ 2 ⎜⎝ rw ⎠ ⎜ ⎜
⎝ ⎝ rw ⎟⎠ 2 ⎟⎠ 4
(8)
Using the following non-dimensional parameters, k λ ∆P t
Pss − Pf
= mc mc 2 mc
∆Psc µ rw
p Dsc = = (4)
∆Pob Pmh − Pss The derivation of Eq. 8 assumed that mudcake
4k f t differential ∆Pm was constant, but it is not limited to
tD = (5)
γφµcrw2 this constraint. As a mudcake forms, the pressure
differential changes. In this case the integral
rw k mc ∫ ∆Pm(t)dt would simply appear in place of ∆Pmt. In
τ Dmc = (6)
lmc k f this general form, Eq. 8 can be used as a boundary
condition for a multiphase reservoir model where the

3
SPWLA 45th Annual Logging Symposium, June 6-9, 2004

mudcake growth is coupled to the filtrate invasion periods and then peaks as the mud cake grows and
(Proett, et al., 2000). chokes off the invasion. This early invasion spurt
occurs in less than a minute after the formation is
Now, by assuming the mudcake is small relative
exposed to the mud hydrostatic pressure. Then the
to the wellbore radius (i.e., lmc /rw→ 0), it can be supercharge pressure declines as the mudcake grows
shown that Eq. 15 can be reduced to the following to its maximum thickness of 0.5 cm (0.2-in.). At this
simpler expression. point the pressure starts to increase at a slow rate and
2k mc λmc ∆Pmct approaches the dashed line showing the results for
lmc (t ) = (9) the static mudcake model. The static mudcake model
µ assumes a mudcake of 0.5 cm was formed instantly
This equation is the well-known lineal filtration when the wellbore was exposed to hydrostatic
model where the filter cake grows with the square pressure. This curve was calculated using Eq. 3, and
root of time. It can be shown that the √t the mudcake thickness is constant. Otherwise, the
approximation is quite satisfactory for values of mudcake would continue to grow, and the
lmc/rw < 0.20; the error is less than 7% (Chin, 2002). supercharge pressure would decline at a reduced rate.
This conclusion applies to radial and linear mudcake This simulation demonstrates that supercharge
buildup but does not apply to cake buildup on pressures predominately decrease when the mudcake
formations where the mudcake and formation have is growing and increase at a reduced rate when the
comparable permeabilities. Fortunately, the later mudcake has stabilized. In this case, the mudcake
situation is rarely the case for most producing zones; stops growing after 30 minutes of invasion. The
and Eq. 17 is a reasonable short-hand method to transition from static to dynamic filtration is probably
estimate mudcake growth. smoother than the modeling results show because the
model assumes an abrupt stop in the mudcake
The linear mudcake model can be incorporated
growth. This change in the mudcake growth tends to
into the general supercharge equation, Eq. 7, by stabilize the supercharge effect. In fact, from 30
applying superposition to the incremental time minutes to 1000 minutes (17 hours) the pressure
periods used to predict the mudcake growth.
changes about 3 psi. Because of the position of the
p Dsc (t ) =
1 n
2 i =1
[
∑ ( Ai − Ai−1 ) ln(t D − t D (i−1) ) ] (10)
tool in the string 30 minutes is probably the minimum
time that it would be possible to perform a pressure
test.
Where:
Table 1—Supercharge Base Example
Ai = τ Dmc (t i ) (11) Sensitivity variable Units Base
Formation permeability kf (md) 1.0
This model assumes that the supercharged pressure is
Formation porosity φ 0.25
loosely coupled to the mudcake growth.
Formation & filtrate
µ (cp) 1.0
Assuming that we know the formation pressure, viscosity
the following relationship can be developed to predict Compressibility c (1/psi) 3x10-6
supercharging. Mudcake permeability kmc (md) 0.0001
Mudcake compaction mc 10.0
Pf + p Dsc Pmh Mudcake max thickness lmc (cm) 0.5
∆Psc = − Pf (12)
∆Pob (psi)
1 + p Dsc Overbalance pressure 1000
Wellbore radius rw (cm) 10.0
Single Phase Supercharge Base Example. A
base example was chosen to illustrate the supercharge Formation Permeability Sensitivity. By varying
effect with invasion time using the variables shown the permeability from 0.01 to 10 md from the base
in Table 1. The base example has an overbalance of example, the curves in Fig. 3 were determined. As in
1000 psi combined with a formation permeability of Fig. 2, the solid lines represent the dynamic mudcake
1 md; and, as Fig. 2 illustrates, there are considerable growth model and the dashed lines for mudcake with
dynamic supercharge effects. The supercharge constant thickness. These examples clearly show that
pressure increases rapidly at the very early time lower-permeability-formation pressure tests can be

4
SPWLA 45th Annual Logging Symposium, June 6-9, 2004

heavily influenced by invasion and supercharging. pressure is that a formation tester probe can introduce
While this conclusion is well known, in practice it an error. The probe introduces a disturbance to the
has been difficult to easily quantify. The examples invasion which distorts the sandface pressure slightly
show that supercharge would be less than 1 psi in (see Fig. 4). The pad element can block mud seepage
zones greater that 10 md. These examples have an around the probe and cause a disturbance. This
overbalance of 1,000 psi, which would be considered disturbance is related to the velocity of the filtrate in
high. Normally, overbalance is less than 500 psi. the near wellbore region:
qm kf
vm = = ( Pss − Pbu ) (13)
70 0.6
2πrw h λe re µ
Dynamic Mudcake Growth Model (psi)

The packer element shape factor λe is a local


60
0.5

geometric correction accounting for non-spherical


Supercharge Pressure (∆Pss)

50

Mudcake Thickness (cm)


Mudcake Thickness (cm)
0.4
effects and can be determined both analytically and
40

0.3
numerically (Proett, et al., 1996). The analytical
30 solution for potential flow around a circular flat disk
0.2
can be used, which suffices for simple estimates.
20
Alternatively, finite element simulations can be used
10
0.1 to determine this shape factor which can consider the
well bore curvature.
Static Mudcake Model (psi)
0 0
0.01 0.1 1 10 100 1000 10000
Time (minutes)

Fig. 2—Base example of supercharge effect using the


variables from Table 1. The dynamic mudcake growth Packer element directs mud
model shows pressure increasing rapidly after exposure filtrate flow around probe
to hydrostatic pressure and as the mudcake grows the
pressure decreases. The static model shown with the
dashed line illustrates how the supercharge pressure
Pbu - sandface build pressure
lower than actual Pss
N
would increase if the mudcake were formed instantly.

1000 Pss - actual sandface


k=0.01 supercharge pressure
Pmh - mud hydrostatic pressure

k=0.1
Supercharge Pressure (psi)

100

k=1.0
Fig. 4 — The sandface buildup pressure Pbu is actually
10
somewhat lower than the undisturbed sandface
pressure Pss because of the shielding action of the
k=10.0 packer element.

1
1 10 100 1000 10000
Introduction of this relationship, along with Eqs.
Invasion Time (min) 1 and 2, defines the measured dimensionless
supercharge pressure pDsm in terms of the measured
Fig. 3 — Supercharge sensitivity to formation
buildup pressure Pbu and the mud hydrostatic
permeability. The dynamic mudcake model results are
shown with solid lines and the static mudcake model in pressure Pmh
dashed lines. The dynamic results approach the static
∆Psc Pss − Pf
results when mudcake growth stops. The supercharge p Dsm = = (14)
effect is reduced to less than 1 psi when the formation ∆Pmob Pmh − Pbu
permeability is greater the 10 md.
1 ⎛ τ Dmc ⎞
p Dsm (t ) = ⎜⎜ ⎟ ln (t D )
2 ⎝ 1 + τ Dmc rDre ⎟⎠
Formation Tester Measurements. An additional (15)
complication to the measurement of sandface

5
SPWLA 45th Annual Logging Symposium, June 6-9, 2004

where ∆Pmob is the overbalance determined using the characterize static mudcake growth (Jones, 1937).
wireline buildup pressure and rDre is the Since its inception, similar devices have been
dimensionless constant for the packer element. developed to measure filtration properties at wellbore
temperatures and pressures. The mudcake
λe re permeability properties, kmc, and the compaction
rDre = (16)
rw factor, λmc, that appears in Eqs. 18 and 19 can be
measured from the static filtration press as follows
The packer element dimensionless radius rDre (Dewan, et al., 2000; Sharma and Jiao, 1994).
determines the relative degree that the measured
lmc (t ) fs
sandface buildup pressure Pbu deviates from the
λmc = = (18)
actual sandface or supercharged pressure Pss. It is h(t ) (1 − f s )(1 − φmc )
primarily dominated by geometric constraints of the
system and not influenced by mudcake or formation ⎛ kmc ⎞ lmc (t ) h(t )
properties. ⎜⎜ ⎟⎟ = (19)
⎝ µ ⎠ 2t ∆p
Now, using the ∆Pmob measured by the
The quantities lmc(t) and h(t) are the measured
formation tester, the supercharge can be determined
mudcake thickness and filtrate fluid height,
from Eqs. 14 and 15, and then formation pressure,
respectively, at time t while maintaining a constant
using Eq. 12.
differential pressure ∆p across a filter used to grow
⎛ 1 + p Dsc ⎞ the mudcake. The mudcake compaction factor
Pf = Pmh − ∆Psc ⎜⎜ ⎟⎟ (17)
(Eq. 18) is a dimensionless parameter that can be
⎝ p Dsc ⎠ related to the porosity φmc and the solid fraction fs of
180 the “mud.” This relationship was developed
160 Sandface Pressure k=0.1 md Example considering the filtration of a fluid suspension of
140 solid particles by a porous but rigid mudcake. While
mudcakes may not behave as ideal solutions with
Supercharge Pressure (psi)

120
solid particles, the compaction factor is a measured
100 Formation Tester
Measured Pressures property that characterizes the mudcake growth in
80
downhole conditions. Additionally, this test is run
60
routinely to test mudcakes in the drilling process and
40 data is normally available.
Static Mudcake Model (psi)
20
The compaction factor primarily relates to the rate
0 that mudcakes form. This relationship is illustrated in
0.1 1 10 100 1000 10000
Invasion Time (min) Fig. 6 where an increase in the compaction factor
increases the mudcake growth rate and reduces
Fig. 5—Formation tester measurement deviation from supercharging. For low values for the compaction
sandface pressure for a 0.1 md formation. The
factor, the mudcake can continue to grow for days
difference declines to less than 1 psi when the
maximum mudcake thickness is reached at the curve after invasion starts. More dynamic effects would be
inflection points. For permeabilities greater than 1 md, seen during pressure testing. The case shown is for a
the difference is less than 1 psi. 1 md zone. A higher permeability would show a
reduced dynamic effect (see Fig. 3).
In most cases, the difference between the
sandface pressure and the buildup pressure measured Mudcake permeability also plays a key role in
by the formation tester is small. Even at low invasion and supercharging, as shown in Fig. 7. Here
formation permeabilities (i.e., < 0.1 md) the the mudcake permeability is changed from the base
difference is usually less than 1 psi after about 10 case, and the results show supercharge changes
minutes of invasion (see Fig. 5). nearly proportional to the mudcake changes. This
figure clearly shows the importance that the mudcake
Mudcake Properties. Mudcake properties can be plays in obtaining accurate formation pressures either
determined from a static filtration press, originally for FTWD or WFT tools.
designed by P.H. Jones, and has long been used to

6
SPWLA 45th Annual Logging Symposium, June 6-9, 2004

Pumps On and Swabbing. Pumps on operation Then, to simulate swabbing, the overbalance is
are common in FTWD testing so that data can be increased linearly. A development similar to the
transmitted to the surface in real-time. Furthermore, supercharging model provides a modified form of Eq.
pumps-on helps prevent pipe sticking. Having the 10 that can appropriately simulate sandface pressures.
pumps on can produce pressure pulses in the However, for this modification, mudcake is constant;
wellbore as high as 100 psi at a 1Hz frequency. and A varies as follows.
⎛ τ Dmc ⎞ G ti
Ai = ⎜⎜ ⎟⎟ sin( 2π f ti ) + (20)
⎝ 1 + rDreτ Dmc ⎠ ∆Pob
1000
λmc=0.1

λmc=1.0 Using a mud pulse frequency of 1 Hz for f and a


linear gradient of 5 psi/min for G, Fig. 8 shows the
Supercharge Pressure (psi)

100
λmc=10
results for the probe pressure fluctuations over a 10
λmc=100 second time period. This simulation used the same
input variables as the supercharge base case
10
presented previously (Table 1). Pressures vary as
much as ± 0.3 psi due to pressure pulses, and there is
a gradual increase in pressure of less than 0.1 psi over
1 10 seconds.
0.1 1 10 100 1000 10000
Invasion Time (min)

0.5
Fig. 6—The mudcake compaction λmc influences the rate
0.4
the mudcake is formed. As the compaction factor
Probe Sandface Pressure Variations (psi)

increases the cake is formed more rapidly, thus 0.3

reducing the supercharge pressures. 0.2

0.1
100
0

kmc= 1.0 µ darcy -0.1


N
-0.2
Supercharge Pressure (psi)

-0.3
kmc= 0.1 µ darcy

-0.4

10
-0.5
0 1 2 3 4 5 6 7 8 9 10
Time (sec)

kmc= 0.01 µ darcy


Fig. 8 Probe pressure variations considering a 100 psi
mud pulse at 1 Hz in the presence of an increasing
hydrostatic pressure of 5 psi/min.
1
1 10 100 1000 10000
Invasion Time (min) Because quartz gauges average data between
updates, the mud pulses are sufficiently small such
Fig. 7—The mudcake permeability influences the that they may not be observed. Also, depending on
magnitude of the supercharge as well as the rate at
which the mudcake is formed.
the length of time that swabbing continues, these
pressure changes could be muted as well. The
Additional hydrostatic variations can be produced magnitude of these pressure changes depends on the
due to the swabbing action of the drill pipe. When mudcake transmissibility ratio, much like the
depth changes are made the friction of the pipe can supercharging. Therefore, if the permeability of the
create a pressure dynamic. The pipe movement can formation increases above 1 md, these borehole
create hydrostatic pressure changes of as much as 50 dynamic effects should be well isolated from the
psi over the duration of a pressure test. To single out FTWD testing.
and observe these pressure fluctuations, we can
assume the constant mudcake thickness model. We
also assume that the overbalance varies sinusoidally.

7
SPWLA 45th Annual Logging Symposium, June 6-9, 2004

MULITPHASE INVASION By incrementing the invasion depth rin and then


solving for time, these equations determine an
Here we investigate a new model for radial invasion
invasion history. The sandface supercharge pressure
where the growth of the mudcake and position of the
is determined using the general equation for
moving invasion front are strongly coupled. The
incompressible radial flow with the following
model considers an invasion consisting of three
constants.
layers, each having unique properties. The mudcake
is assumed to have zero thickness when invasion Pss = α 2 ln (rw ) + β 2 (26)
starts (i.e., rmc = rw at t=0). Assuming piston
displacement, the invasion front can be defined as the Where:
region between rw to rin. The front radius rin also
k mc
starts at the wellbore sandface rw and grows radially α2 = α1 (27)
(i.e. rmc = rw at t=0). This invaded region can have kin
different permeabilities and viscosities from the
⎛ rw ⎞
uninvaded region between rin and the reservoir radius
β 2 = phm + α1 ln⎜⎜ ⎟⎟ − α 2 ln (rw ) (28)
rr (see Fig. 9)
⎝ rmc ⎠

α1 =
(k mc / µ mc )( pr − phm )
Phm Pss Pin Pr
k mc (µ f + k mc )
(29)
⎛ kf

⎜ ⎛ rw ⎞ in ⎛ rin ⎞ µ in2
µ ⎟
1 2 3 ln⎜ ⎜⎜ ⎟⎟ ⎜⎜ ⎟⎟ ⎟
rmc ⎜ ⎝ rmc ⎠ ⎝ rw ⎠ ⎟
rw rin rr
⎝ ⎠
This moving boundary model may be compared
Fig. 9— Multi fluid moving boundary model to the previous single phase compressible model (see
Fig. 10) using the variables shown in Table 2.
As shown previously, the mudcake thickness is Additionally, a similar model can be developed
governed by the general form of Eq. 8. Assuming where the mudcake is assumed to be a constant
incompressible flow, a closed form solution for this thickness but the invasion boundary continues to
three-layered moving boundary problem has been grow (see Chin, 2002). These two models are
developed in Chin, 2002. The problem is solved in combined in Fig. 10 where the transition to constant
terms of the time it takes for the invasion front rin to mudcake thickness is reflected in the curve inflection
reach a given depth. points around 30 minutes.
φ eff R22
t= [A + B + C ] (21)
( pmh − pr ) 4 Table 2—Multi Fluid Examples
Where: Sensitivity variable Units Base

[( ) (2 ln( )− 1)− 1]
Formation permeability kf (md) 1.0
µ in rin 2 Formation porosity φeff 0.25
A=
rin
(22)
kf rw rw Mud & filtrate viscosity µmf (cp) 1.0
Formation viscosity µf (cp) 1, 3, 10

B=−
µf
kf
[( rw
) (2 ln( )− 1)− (2 ln( )− 1)]
rin 2 rin
rr
rw
rr
(23)
Mudcake permeability
Mudcake compaction
kmc (md)
mc
0.0001
10
Overbalance pressure ∆Pob (psi) 1000
µin ⎡
( ( ) )(1 − ln(D ))⎤⎥
Wellbore radius Rw (cm) 10
ln (D ) − 1 −
1 rin 2
C= ⎢ (24) Reservoir radius Rr (cm) 300
⎢⎣φeff λmc
rw
kmc ⎥⎦

( ( )) rin 2
In the first case, where the filtrate and formation
D = 1 + λmcφ eff 1 − rw
(25) viscosities are the same (i.e., µmf=µmf ), the
supercharge pressure converges to the single phase
model case after several minutes of invasion. When

8
SPWLA 45th Annual Logging Symposium, June 6-9, 2004

the formation viscosity is increased, the supercharge convects the fluid into the probe while mud invasion
increases. This change is similar to the single phase at the sandface is occurring into the rock. If the flow
case with increased viscosity. If the mud filtrate is immiscible, diffusion is based on relative
viscosity is reduced, the supercharge pressure is permeability and capillary pressure effects “ride”
increased in a similar manner because filtrate invades above this convection. If the flow is miscible,
faster. diffusion based on molecular processes is important,
Because the multi-fluid model does not include e.g., high diffusion smears the diffusion front so
compressibility affects, the dynamics of the early- much that the second model of the above section no
time spurt loss seen in Fig. 3 are not simulated. In longer applies.
most cases, this spurt occurs in the first few minutes An invasion pumpout simulator was therefore
of invasion and tends to dissipate quickly and may developed and is used here to illustrate examples of
not play an important role in estimating the pressures invasion and pumpout modeling (see Proett, et al.,
recorded by formation testers. Even FTWD tools 2002).
usually cannot record pressures until at least 20 Immiscible Flow Simulator. Real-world flows
minutes after a zone has been drilled. are neither purely cylindrical nor purely spherical
(elliptical if anisotropic). Formation tester pumping
1000
while invasion is in progress requires a nonlinear
µf = 10 description of superposed flows. An immiscible flow
model was designed which solves the coupled
µf = 3
pressure and saturation equations as functions of the
Supercharge Pressure (psi)

100
µf = 1
radial “r” and vertical “z” coordinates and time “t.”
The model allows fully anisotropic, heterogeneous,
multilayer media with special sandface boundary
Single Phase µf = 1
10 conditions.
The equations used in reservoir engineering for
Constant mudcake thickness µf =1
oil and water production, to include relative N
1
permeability and capillary pressure effects, were
0.1 1 10 100 1000 10000 solved using an “implicit pressure, explicit
Invasion Time (min)
saturation” finite difference, time-marching scheme,
Fig. 10—The solid blue line is comparable to the solid in which our special boundary conditions are used to
red line where both estimate a single phase invasion update pressure and saturation fields at the end of
using two different modeling approaches. As the each numerical sweep of the flow field. Example
formation viscosity increases, the supercharge pressure
calculations are described in this paper, however, for
increases with the results being similar to increasing
viscosity in the single phase model. brevity, qualitative results are presented without
detailed numerical input summaries to provide an
FINITE DIFFERENCE MULTI PHASE idea of general capabilities.
SIMULATOR Miscible Flow Model. The immiscible flow
model requires inputs on relative permeability and
The simulators described above for simplicity assume
capillary pressure. In contrast, the miscible model
single-phase flow. For rough invasion estimates and
assumes molecular diffusion, although the model
for permeability prediction from pressure transient
allows the diffusion coefficients to differ in the r and
analysis, these simple models often suffice. However,
z directions.
they do not offer detailed information about real-
world fluid displacement processes in the reservoir, An example of a numerical simulation of miscible
e.g., water vs. oil, fresh vs. salty water, or oil mud vs. invasion is shown in Fig. 11. This plot illustrates
reservoir oil. In other words, the simpler models are miscible cylindrical radial invasion that occurs during
sometimes inapplicable. drilling, and this invasion ultimately slows due to
mudcake growth. The pressures are shown on the left
For the above problems, two mechanisms are at hand contour plot of Fig. 11; and saturations, after 1
work. At the basic level, the Darcy flow created by minute of invasion, are shown with the right hand
the sucking action of the formation tester piston

9
SPWLA 45th Annual Logging Symposium, June 6-9, 2004

contour plots. The left side of each figure represents Miscible Invasion Simulations. Using the
the sandface while the right side represents the miscible flow simulator, examples of supercharging
reservoir far field. Fig. 11 indicates an invasion that are presented in Fig. 14. These simulations use the
is uniform with the height variable z. If gravity and same input variables shown in Table 2. The results
layered properties such as porosity and permeability are similar to those shown in Fig. 10 where a closed
were considered the invasion would not be as form analytic model was used to model supercharge.
uniform. In both cases the overall level of supercharge
increases with viscosity of the formation fluid. The
inflection points on these curves are obvious.
Interestingly, the times associated with these
inflection points are about the same as the times for
the single phase solution and the closed form multi
phase analytic models shown earlier (Figs. 3 and 10).
This inflection point is a result of limiting the
mudcake growth which means all three models
exhibit a similar mudcake dynamic growth. Early
Fig. 11 — Miscible cylindrical radial flow invasion. time supercharge pressures are higher although all
three models converge to similar results. This is
Considering this example at 1 minute, the probably due to the fact that the early time invasion
formation tester is allowed to withdraw fluid in a front is influenced by the mud viscosity in the
pumping mode. After pumping for 1 minute, Fig. 12. numerical simulatons.
shows pressure and saturation distributions. The
movement of the formation fluid into the probe is
1000
clearly seen. Still deeper invasion is shown in
Fig. 13. at 5 minutes despite the fact the tester µf = 10
µf = 3
µf = 1
continues to withdraw fluid through the probe. In
Supercharge Pressure (psi)

both figures, note the blue low pressure zone 100

associated with the probe and the high pressure red


zones in the pressure plots adjacent to the probe,
indicating continuous invasion through the sandface. Single Phase µf = 1
10

Constant mudcake thickness µf =1


1
0.1 1 10 100 1000 10000
Invasion Time (min)

Fig. 14 — Multi phase simulation examples are


developed using inputs from Table 2. The solid blue line
is comparable to the solid red line where both estimate a
single phase invasion using two different modeling
Fig. 12 — Formation tester pumping at 2 min.
approaches. As the formation viscosity increases, the
supercharge pressure increases with the results being
similar to increasing viscosity in the single phase model
and the examples shown in Fig 10.

Comparison of the single phase model to the


multi- phase models shows that they have similar
behaviors. By changing the formation permeability or
mobility in proportion to the change in the formation
viscosity, the single phase model produces results
similar to the two phase models. For example, if the
Fig. 13 — Formation tester pumping at 5 min.
formation permeability is 1 md with a viscosity of 10

10
SPWLA 45th Annual Logging Symposium, June 6-9, 2004

cp for the formation fluid, mobility is decreased The primary adjustable mudcake properties are
from 1 to 0.1 md/cp. Using 0.1 md in the single phase the mudcake permeability and the compaction factor.
model produces similar supercharge pressures to the Both affect the mudcake growth rate (see Figs. 6 and
multi-phase case. Results, however, depend basically 7). While it may not be possible to determine a
on how fast the invasion front develops and mixes unique solution for both properties, the shape of the
due to diffusion of the two phases. supercharge pressure curve can be matched and an
estimate of the supercharge magnitude can be made.
LOG EXAMPLES In the case shown in Fig. 16, there is less than 1 psi
of supercharging.
When recording pressures with FTWD tools as well
as wireline test tools, pressure changes can occur
over time. These changes are indications of the
presence of supercharging. In many cases, the
changes are small. The forgoing simulations indicate
that small changes in test pressures are a good sign
that supercharging is negligible. However, if the
changes are measurable, that is, they exceed 1 psi,
then it is possible to estimate the supercharge effect
using these time lapse measurements.
In the case of FTWD operations, pressure
measurements can be made early in the drilling
program and then while tripping out of the hole days Fig. 15— Pressure test 1 taken at X,957.8 TVD with a
later. Pressure tests can be repeated at the same final buildup pressure of 3,320 psi.
depths. An example of this is shown in
Table 3 where three repeat measurements were made
several days apart. In two of these repeat tests the
Pstop or final buildup pressure changed by 4 and 5 psi, N
respectively. In the third case the Pstop pressure
repeated.
Plots of the first repeat pressure tests are shown in
Figs 15 and 16. Based on these pressure transients,
the formation mobility was determined to be 15.0 and
13.4 md cp, respectively. This information and other
data such as the hydrostatic overbalance make it
possible to construct a time lapse plot of the
supercharge pressure using the single phase model. Fig. 16— Pressure test 1 taken at X,958.0 TVD with a
Then, by adjusting the mud properties, an estimate of final buildup pressure of 3,316 psi 3.5 days later.
supercharge pressure can be made as shown in
Fig. 17.

Table 3 GeoTap Repeat Test Examples


Test TVD Repeat Pstop Pstdev Phyds1 Phyds2 Mobility
(ft) (days) (psi) (psi) (psi) (psi) (md/cp)
1 X,957.8 3,320 2.9 4,014 4,005 15.0
2 X,958.0 3.50 3,316 <1 3,991 3,983 13.4
3 X,914.8 3,298 1.54 3,992 3,985 75.1
4 X,913.9 3.47 3,293 <1 3,971 3,964 26.0
5 X,803.5 3,508 5.88 3,934 3,925 18.1
6 X,802.1 3.45 3,508 0.63 3,916 3,909 17.3

11
SPWLA 45th Annual Logging Symposium, June 6-9, 2004

10 0.6 tested, most of the pressure tests are within this


9 Dynamic Mudcake Growth Model (psi) range. It is also noted that all the GeoTap test
8
0.5
pressures were offset in the same direction indicating
Mudcake Thickness (cm)
a systematic gauge offset. The mobilities in the two
Supercharge Pressure (∆Pss)

Mudcake Thickness (cm)


7
0.4
6
zones tested are above 400 md/cp. All of these
5
Test 1
0.3
factors provide evidence of little or no supercharging
in both sets of pressure tests.
4
0.2
3
CONCLUSIONS
2
Test 2 0.1
1
1. Supercharge estimates were made using three
0 0
different modeling techniques to demonstrate the
10 100
Time (minutes)
1000 10000 dynamic pressure environment present while
drilling.
Fig. 17— Supercharge analysis of tests 1 and 2.
2. The single phase supercharge model was
Greater pressure changes are evidence of higher compared with an analytic model and a finite
levels of supercharge. More accurate estimates of difference, multi phase model.
supercharge could be made if more time lapse
measurements are made. One such possibility is to 3. All three models exhibited similar behaviors and
make measurements with the FTWD tool and then converged with similar mudcake growth times
with a wireline tool many days after the well is and supercharge pressures after the initial spurt
drilled. loss.
4. In certain circumstances, the single phase model
2650
parameters can be adjusted to match multi phase
X000
GeoTap 10 psi behavior.
MDT
2670
X020
508 md/cp
5. A log example was used to demonstrate how the
± 317 md/cp models can be used to estimate supercharge
2690
X040
pressures when time-lapse measurements are
TVD (meters)

made.
2710
X060
6. The models show that supercharge pressures can
2730
X080 affect both wireline and FTWD tools.
461 md/cp
± 243 md/cp
Nevertheless, it is possible to obtain reliable
2750
X100 10 psi measurements in both conditions.
XX3450 XX3500 XX3550 XX3600 XX3650 XX3700 3750
XX
Pressure (psi)
REFERENCES

**
Chin, W. C., 2002, Quantitative methods in reservoir
Fig. 18— Comparison of GeoTap and MDT pressure
engineering, Gulf Professional Publishing, Elsevier
tests show that recorded pressures are within ± 5 psi.
This is considered within the accepted range of error Science USA
when comparing two different tools.
Dewan, J. T., Chenevert, M. E., and Yang, X., 2000,
Many field test examples exist where FTWD A model for filtration of water-based mud during
measurements are compared to wireline drilling I. Determination of mudcake parameters,
measurements. In most cases, the pressures compare paper presented at the 41st Annual SPWLA meeting,
within ± 5 psi. Fig. 18 is an example where GeoTap Dallas, Texas, June 4-7, 2000.
pressures are compared against an MDT tool. When Jones, P. H., 1937, Field control of drilling mud, API
comparing one tool’s pressure measurements with Drill. Prod. Prac., pp. 24-29.
another, a ±5 psi deviation is expected due to gauge
accuracy and depth control errors. For the two zones Proett, M. A., and Chin, W. C., 1996, Supercharge
pressure compensation using a new wireline testing
** A mark of Schlumberger Oil Field Services

12
SPWLA 45th Annual Logging Symposium, June 6-9, 2004

method and newly developed early time spherical ∆Pmc mudcake pressure differential (psi)
flow model, SPE Paper 36524, presented at the 71st Pmh mud hydrostatic pressure (psi)
SPE Annual Technical Conference and Exhibition Pss sandface supercharge pressure (psi)
held in Denver, Colorado, U.S.A., 6-7 Oct., 1996. ∆Psc supercharge pressure differential (psi)
Proett, M. A., Chin, W. C., and Mandal, B. (2000) qm mud filtrate flow rate (cc/sec)
Advanced dual probe formation tester with transient, rDe dimensionless radius
harmonic, and pulsed time delay testing methods re packer element radius (cm)
determines permeability, skin, and anisotropy, paper rin invaded zone radius (cm)
SPE 64650, presented at the SPE International Oil rmc mudcake radius (cm)
and Gas Conference and Exhibition in China held in rr reservoir radius (cm)
Beijing, China, 7–10 November, 2000. rw wellbore radius (cm)
lmc mudcake maximum thickness (cm)
Proett, M. A., Chin, W. C., Wu, J. and Manohar, M., t buildup time (sec)
2002, Sample quality prediction with integrated oil tD dimensionless time
and water-based mud invasion modeling, Paper SPE σ standard deviation (psi)
77964, proceedings of the SPE Asia Pacific Oil and α1 invasion constant
Gas Conference and Exhibition held in Melbourne, α2 invasion constant
Australia, 8–10 October 2002 β2 invasion constant
Sharma, M. M., and Jiao, D., 1994, Mechanism of λe packer element shape factor
cake buildup in cross flow filtration of colloidal λmc mudcake compaction factor
suspensions, Journal of Colloid and Interface Science µ viscosity (cp)
V. 162, pp. 454-462. µf formation fluid viscosity (cp)
Wu, J., Torres-Verdín, C., Sepehrnoori, K. and µmf mud filtrate fluid viscosity (cp)
Delshad, M., 2001, Numerical simulation of mud φ formation porosity
filtrate invasion in deviated wells, Paper SPE 71739, φeff effective porosity
proceedings of the SPE Annual Technical φmc mudcake porosity N
Conference and Exhibition held in New Orleans, LA, γ Euler’s constant (1.78)
Sept. 30– Oct. 3, 2001. µ viscosity (cp)
τDmc mudcake transmissibility ratio

Nomenclature
A variable, dimension varies ABOUT THE AUTHORS
B variable (cp/md)
Mark A. Proett is a Senior Scientific Advisor in
C variable (cp/md)
Services for Halliburton Energy Services in the
D variable, dimensionless
Strategic Research group. He received a BSME
G pressure time gradient (psi/sec)
degree from the University of Maryland and a MS
c compressibility (1/psi)
degree from Johns Hopkins. He has been involved
f frequency of mud pulse (Hz)
with the development of formation testing systems
fs mud solid fraction
since the early 1980’s, and has published extensively.
h reservoir unit length (ft)
Proett holds 18 patents, 16 of which deal with well
kf formation permeability(md)
testing and fluid flow analysis methods. He has
kin invaded zone permeability(md)
served on the SPWLA and SPE technical committees
kmc mudcake permeability (md)
and served as the Chairman for the SPE Pressure
Pbu buildup pressure (psi)
Transient Testing Committee.
pDsc dimensionless supercharge pressure
pDsm dimensionless measured supercharge Douglas J. Seifert is a Petrophysicist with Saudi
Pf formation pressure (psi) Aramco in Dhahran, Saudi Arabia specializing in
∆Pob overbalance pressure differential (psi) LWD log interpretation, geosteering and logging
∆Pmob measured overbalance differential (psi) operations. He received a BS in Statistics and a MS
in Geology from the University of Akron. Doug was

13
SPWLA 45th Annual Logging Symposium, June 6-9, 2004

previously Western Hemisphere Regional has worked extensively in formation tester fluid flow
Petrophysicist for Pathfinder Energy Services in simulation and pressure transient permeability
Houston and Eastern Hemisphere Regional prediction. He is co-inventor with Mark Proett on
Petrophysicist, in Stavanger, Norway. He has worked three Halliburton pressure transient analysis patents
for Mærsk Olie og Gas in Denmark as the Head of and also works actively in MWD hardware and
Petrophysics, for Halliburton Energy Services in software design.
Operations, Research and Technical Support
Solveig Lysen is a Country Technical Adviser for
functions and also for Texaco in Technical Services
Sperry-Sun, Halliburton Energy Services in
and Production Operations. He has served on the
Stavanger, Norway.
SPWLA Technology Committee (1985-1988), as an
officer in the Houston Chapter of the SPWLA (1986- Paul Sands received a B.Sc. in Mechanical
1988) and as President of the Houston Chapter of the Engineering in 1977 from Brighton Polytechnic in
SPWLA (1989-1990). the UK. After a number of years of oilfield wireline
logging assignments and operations management, he
Wilson C. Chin, who earned his Ph.D. at the
joined NUMAR in 1993. He is currently based in
Massachusetts Institute of Technology, is author of
Aberdeen as Sperry-Sun’s Europe and Africa
the book "Quantitative Methods in Reservoir
regional GeoTap & MRIL-WD product champion.
Engineering" (Elsevier Science, 2002). He is a
consultant to Halliburton Energy Services, where he

14

Вам также может понравиться