Sie sind auf Seite 1von 11

Modeling of Transient Cuttings Transport

in Underbalanced Drilling (UBD)


Q.T. Doan,* SPE, Vincano Inc.; M. Oguztoreli,** Mustafa Oguztoreli Inc.; Y. Masuda, SPE, U. of Tokyo;
T. Yonezawa, SPE, TRC/JNOC, and A. Kobayashi,† TRC/JNOC; S. Naganawa, SPE, U. of Tokyo;
and A. Kamp, SPE, PDVSA Intevep

Summary chanical energy equations were obtained. Allowance was also


Underbalanced drilling (UBD) holds several important advantages made for the changes in the foam, caused by the influxing of
over conventional drilling technology. These include minimization reservoir fluids. The average error between simulation results and
of formation damage, faster penetration rate, and ability for evalu- Chevron’s results and test data was 11.2%. Foam quality was
ation of reservoir productivity during the drilling process. As UBD restricted to 0.97 in this model.
technology matures, it has also been used more and more in dif- Wang et al.3 pointed out the importance of monitoring and
ferent applications. However, many aspects of UBD technology controlling downhole pressure, as this affected reservoir fluids
remain poorly understood. The model presented in this paper seeks influx and, therefore, foam rheological properties. Field data from
to understand the mechanisms involved in the transport of cuttings a Brazilian underbalanced well revealed that the equivalent circu-
in UBD. lating densities, which were needed in the numerical model, did
The model simulates the transport of drill cuttings in an annulus not reach steady state and fluctuated by up to 50% during the time
of arbitrary eccentricity and includes a wide range of transport required to drill one section of pipe. It was postulated that changes
phenomena, including cuttings deposition and resuspension, for- in gas and liquid density, void fraction, surface control of choke
mation, and movement of cuttings bed. The model consists of for two-phase flow, and disturbances attributed to drillstring con-
conservation equations for the fluid and cuttings components in the nections and tripping operations caused the pressure fluctuations.
suspension and the cuttings deposit bed. Interaction between the Wang et al.,4 in a follow-up study, reported the application of
suspension and the cuttings deposit bed, and between the fluid and their model to two field cases. Dynamic effects such as circulation
cuttings components in the suspension, are incorporated. Solution start-ups and shutdowns, tripping, and gas injection were included.
of the model determines the distribution of fluid and cuttings con- In the first case, nitrogen gas was injected into the annulus as a
centration, velocity, fluid pressure, and velocity profile of cuttings means of controlling underbalanced conditions. The simulator
deposit bed at different times. overpredicted the bottomhole pressure by 3.7%, compared to field
The model is used to determine the critical transport velocity measurement. In the second case, nitrogen foam was used as a
for different hydrodynamic conditions. Results from the model drilling fluid in a naturally fractured reservoir exhibiting serious
agree quite closely, qualitatively, with experimental data obtained loss of circulation. It was found that the simulator overpredicted
from a cuttings transport flow loop at the Technology Research Cen- bottomhole pressures, and the overprediction increased with higher
ter of the Japan Natl. Oil Corp. (TRC/JNOC)’s Kashiwazaki Test gas injection rates.
Field in Japan. These results show the importance of slippage in the Langlinais et al.5 presented experimental data of annular fric-
formation of the cuttings deposit bed. The model is useful in evalu- tional pressure drops for mud-gas mixtures flowing in vertical
ating the minimum flow rate for effective cuttings removal in UBD. wells. The mixtures were composed of water-base drilling mud
and nitrogen gas. Both the Bingham plastic and power-law mod-
Introduction els (for slot flow) showed deviation from measured data. Several
Rommetveit et al.1 developed a numerical model for UBD with different two-phase flow correlations were used to calculate the
coiled tubing. The main features of this transient, 1D model in- frictional pressure drop. The Hagedorn and Brown correlation,
cluded reservoir-wellbore interaction, alternative geometries for coupled with a power-law rheological model and equivalent diam-
gas injection, and rheology of different fluids, among others. The eter, was found to provide the best fit for the experimental data.
model consisted of seven mass conservation equations (for free Luo et al.6 combined dimensionless groups with experimental
produced gas, free injected gas, mud, dissolved gas, formation oil, data to develop a model for critical flow rates. The model did not
formation water, and drill cuttings) and one overall momentum con- consider the effects of penetration rate and drillpipe rotation. The
servation equation. Simulation results showed that a lower gas injec- difference between the predicted critical flow rate and laboratory
tion rate was required in the case of gasified drilling fluid, compared data (for concentric annulus) showed a difference of 15.9%, on the
with annular gas injection strategy. Gas-oil ratio of the reservoir pro- average. Comparison was also performed against field data ob-
duction into the wellbore was found to have important effects on the tained for three different hole sizes from wells in several North Sea
ability to maintain downhole conditions underbalanced. fields. The model was found to overpredict field data for the larger
Liu and Medley2 compared results generated from their com- holes, and underpredict in the case of smaller holes. Hole cleaning
puter model with results from Chevron’s Foamup program and test charts were then developed from numerical results of the model,
data. Two different equations of state for foam were derived: for different hole angles, maximum rates of penetration, and trans-
downward flow in the drillstring, and upward flow in the annulus. port indexes.
For the upward flow, the model allowed for three phases: gas and Sharma and Chowdhry7 presented an isothermal, steady-state
liquid in the foam, and solid cuttings. Correspondingly, two me- model for annular flow of gas-solids suspension. Some of the main
assumptions of the model included: no rotation of inner and outer
pipes, and negligible momentum transfer owing to particle-particle
collisions and particle-wall collision. A friction coefficient for the
* Now with APA Petroleum Engineering Inc. solids-gas mixture was defined, based on frictional pressure drop
** Now with Shell Trading Gas and Power.

Now with Telnite Co., Japan.
between pipe wall and air, and drag forces between the fluid and solid
particles. Results from this model were compared against indepen-
Copyright © 2003 Society of Petroleum Engineers
dently obtained data of simulated air drilling experiments. The dif-
This paper (SPE 85061) was revised for publication from paper SPE 62742, first presented ference between calculated friction coefficient and measured value
at the 2000 IADC/SPE Asia Pacific Drilling Technology Conference, Kuala Lumpur, 11–13
September. Original manuscript received for review 15 September 2000. Revised manu-
was within 6%; however, experimental data were obtained for a nar-
script received 14 March 2003. Manuscript peer approved 24 March 2003. row range of particle sizes, and there were no reported pressure data.

160 June 2003 SPE Journal


Hemphill and Larsen8 compared experimental data on hole interfacial shear stress exerted on it by the flowing suspension. The
cleaning performance of oil- and water-based drilling fluids. Ef- model predicted deposit bed height as a function of mud rate,
fects of fluid rheological properties on critical flow velocities for cuttings diameter, mud viscosity, pipe eccentricity, and so on. It
various hole angles were the main objective of this study. Bingham was determined that, compared to independently obtained experi-
plastic, power-law, and yield power-law models were all used to mental data, the model overpredicted cuttings transport for a given
describe the cuttings transport data. Two different flow rates were mud flow rate and larger particles. The authors suggested that
defined: critical flow rate (at which a cuttings bed begins to form) separate momentum equations be solved for the solids and mud in
and subcritical flow rate (at which cuttings accumulation takes the suspension layer in inclined wells.
place). Oil-based and water-based mud had similar hole cleaning
performance under equivalent velocity and rheological conditions; Development of the Model
additionally, fluid velocity had a more pronounced effect than mud The following assumptions are made in the formulation of the
weight on cuttings transport, especially at high angles of inclina- model. The transport of drill cuttings inside the annulus during the
tion and for equivalent fluid viscosities. UBD process is modeled as a two-phase flow process. The con-
Guo and Rajtar9 calculated the volume of air required to opti- tinuous fluid phase is uniform in properties; it could be 100%
mize hole cleaning performance for aerated drilling. The model liquid, 100% gas, or a mixture of liquid and gas. In the last case,
incorporated cuttings transport, pipe sticking, hole stability, and the mixture has uniform transport properties insofar as the settling
formation damage. The required air and liquid (water) volumes behavior of drill cuttings is concerned. What this means is that the
were dependent on the optimum flowing bottomhole pressure effects of gas bubble size and liquid droplet size on the settling
(which affected hole stability, pipe sticking) and cuttings transport tendency of drill cuttings are not considered in this model. The
capacity of the aerated mud (which affected formation damage). disperse solid phase is composed of uniform-size and spherical
Increasing liquid flow rate, while fixing air injection rate, led to drill cuttings. It is further assumed that the drill cuttings are ini-
lower cuttings transport performance for the aerated mud because tially uniformly dispersed in the fluid phase. Also, reservoir fluids
of higher mixture density and large reduction in its velocity (to (oil, gas, and/or water) produced into the annulus because of the
below the cuttings slip velocity). underbalanced condition are assumed to instantaneously combine
Gao et al.10 tried to establish minimum transport velocity for with the drilling mud to become a new continuous phase with
cuttings removal. Experimental data and analytical consideration uniform and homogeneous transport properties. The solid particles
were used to define the minimum transport velocity for cuttings (drill cuttings) are under the influence of several forces, including
rolling and cuttings suspension. The Hershel-Bulkley rheological gravity, drag force, and lift force. These forces and others affect
model was selected. The fluids were assumed incompressible and the distribution of the drill cuttings inside the annulus. The flow
inelastic; the flow was laminar, isothermal, and steady. The mass inside the annulus is isothermal, transient, and 1D (from the drillbit
and momentum conservation equations were solved to determine to the surface). Empirical correlations in the literature are assumed
the velocity profile, shear rate profile, and flow rate-pressure drop to be valid and applicable to describe the interphase and intraphase
relationship in the annulus. Different cases were investigated, in- momentum transfer processes, within limitations which arise out
cluding those with concentric or eccentric flow with and without of different experimental conditions.
pipe rotation.
Larsen et al.11 developed correlations from experimental data Features of the Model. The model has the following main features:
of cuttings transport in horizontal and high-angle wells. They de- • Arbitrary geometry (i.e., the well/borehole profile is not fixed
fined the critical transport fluid velocity to be minimum velocity to be either vertical or horizontal; rather, it could assume any
for cuttings transport, and subcritical fluid-flow velocity to be the inclination). This is accomplished through coordinate transforma-
annular fluid velocity below which cuttings begin to accumulate in tion technique, which relates segments of the borehole in the (x,z)
the wellbore. One of the derived correlations was found to over- Cartesian coordinate system with their transformed counterparts in
predict cuttings concentration in the annulus under subcritical the (s,n) coordinate system, as shown in Fig. 1.
transport fluid velocity, and the overprediction increased with • Arbitrary eccentricity (i.e., the drillstring and borehole are
higher-viscosity muds. It was believed that flow was diverted to not necessarily always concentric with one another). Within the
the underside of the annulus as the diversion of flow increased annulus between the drillstring and the borehole, there can be, at
with higher-viscosity muds. A correction factor was introduced to maximum, two regions. In Region 1, there is a cuttings deposit
account for the overprediction. bed. This deposit could either be stationary or moving, depending
Kamp and Rivero12 presented a 2-layer model for steady-state on hydrodynamics conditions. In Region 2, which is the portion of
cuttings transport in highly inclined wells. The two layers were a the annulus above the deposit bed, drilling fluids and drill cuttings
homogeneous suspension above and a deposit bed below. Cuttings are transported through, as seen in Fig. 2.
resuspension was based on friction velocity, and governed by a • The deposit bed is assumed to have packed concentration.
heavyside function. The deposit bed movement was influenced by Continuing deposition and/or re-entrainment of drill cuttings affect

Fig. 1—Quasi-stationary borehole-aligned axial coordinate system (s, n).

June 2003 SPE Journal 161


1
␶1 = ␭1u1 ␭1 = f1␳1|u1|
2
1
␶2 = ␭f 2uf 2 ␭f 2 = ff 2␳f |uf 2|, . . . . . . . . . . . . . . . . . . . . . . (4)
2
where hydrodynamic wall shear stress in the suspension Region 2
is assumed to be a function of the fluid hydrodynamics only. The
hydrodynamic wall friction coefficients, f1 and ff2, depend on the
bed and suspension Reynolds numbers, Re1 and Ref2, respectively.
They assume the form of the well-known Blasius formula for
turbulent tube flow, as shown in Eq. 5:
f1 = ␣1Re1−␤1 and ff 2 = ␣2Ref 2−␤2, . . . . . . . . . . . . . . . . . . . . . . . . (5)
Reynolds numbers are determined using power-law form for non-
Newtonian flow with fluid flow index, n,
␳1ⱍu1ⱍ2−n Dn1
Re1 =
8n−1␮1
Cf 2␳f |uf 2|2−n Dn2
Ref 2 = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (6)
8n−1␮f
Fig. 2—Two-layer annulus model geometry, with arbitrary an- and hydraulic diameter, D,
nulus eccentricity.
4A1
D1 =
共L12 + L1,outer + L1,inner兲
the height of the deposit bed, but not its concentration (i.e., solid
4A2
phase concentration is fixed and constant). As a result, the interface D2 = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (7)
area between Region 1 and Region 2 varies with the flow process. 共L12 + L2,outer + L2,inner兲
• In Region 2, the flowing fluid phase and solid phase expe- The bed mixture viscosity, ␮1, is correlated to the fluid viscosity,
rience slippage (i.e., they move at different velocities). Their con- ␮f, and an interaction function (Kirkwood et al.15), fi, at high bed
centrations vary with space (axial location, height) and time be- solids concentration, Cs1.
cause of deposition and/or re-entrainment.

Conservation Equations 冋
␮1共Cs1兲 = 4␮f Cs1 .8 + .767共 fi − 1兲 +
1
共 fi − 1兲
册 . . . . . . . . . . . . (8)
There are three mass conservation equations: one for Region 1 The coefficients ␣1, ␣2, ␤1, and ␤2 in Eq. 5 for turbulent and
(deposit bed), one for the fluid phase, and one for the solid phase laminar flow are to be calibrated using experimental data.
(cuttings) in Region 2. These equations, together with the three Hydrodynamic interfacial shear stress (existing at the interface
momentum conservation equations, are listed in Appendix A. In between the two layers) is assumed to be a function of the fluid
the momentum conservation equations, the following aspects of hydrodynamics only. It is expressed in terms of relative velocities
multiphase flow were considered. Multiparticle drag forces, Fsf, between fluid flowing in suspension and the moving bed regions.
between solid and fluid are considered in the suspension Region 2
(Shook and Roco13). ␶12 = ␭12 共uf 2 − u1兲
1
Fs,2␴ = Cs2F2␴ + Fsf ␭12 = f12␳f |uf 2 − u1|. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (9)
2
Ff,2␴ = Cf2F2␴ − Fsf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (1)
The interfacial friction coefficient, f12, in Eq. 9 is evaluated using a
For 1D flow, modified Colebrook formula for rough pipes (Televantos et al.16).

Fsf = A2Cs2␭sf 共uf2 − us2兲⌬s,


3CD␳f |uf 2 − us2|
1
公f12
= −.86ln 冉 dp
+
2.51
3.7 D2 Re12 公f12 冊
. . . . . . . . . . . . . . . . . . (10)

␭sf = , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2) The interface Reynolds number, Re12, is evaluated using power-
4dp 共1 − Cs2兲1.65 law form for the non-Newtonian flow associated with the fluid
flowing in suspension relative to the moving bed’s rough surface
where the drag coefficient CD is determined from the Reynolds
(of roughness dimension equal to the particle diameter, dp), as
number, Rep.
shown in Eq. 11.
Cf 2␳f |uf 2 − us2 |2−n dnp Cf 2␳f |uf 2 − u1|2−n Dn2
Rep = Re12 = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (11)
8n−1 ␮f 8n−1␮f
D = .44
C* Rep ⱖ 989 Deposition and entrainment rates, vDEP and vENT, determine the
volumetric material transfer rate across the interfacial boundary,
24共1 + 0.15Rep 0.687
兲 aligned in the direction normal to bulk flow. The deposition rate,
D=
C* Rep ⬍ 989
Rep vDEP, is determined as the hindered terminal settling velocity of a
single particle, vp. A force balance on a single particle equates the
CD = ␣DC*
D. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (3) drag force FD to the gravitational force Fg.
A drag factor, ␣D, is assumed as a corrective measure to be cali- ␲d3p
brated with experimental results. The particle Reynolds number is Fg = 共␳s − ␳f兲g
determined using power-law for non-Newtonian flow with fluid 6
flow index, n. Hydrodynamic shear stresses are modeled by the 1 ␲d2p
FD = CD␳f v2p , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (12)
following expressions (Doron et al.14): 2 4

162 June 2003 SPE Journal


where the drag coefficient, CD, is calculated using the power-law locity difference, the more re-entrainment of deposited cuttings
non-Newtonian particle Reynolds number, Rep, with respect to the into the flowing suspension layer. The slope, mENT, and critical
terminal settling velocity, as shown in Eq. 3. * , are tuning parameters to match simulation results
threshold, u12
The single-particle terminal velocity is iterated using the fol- with experimental data for cuttings concentration.
lowing expression: The sum of driving forces in Region 1 must overcome the dry
friction resistance force, F1, in the direction opposing flow for the

vp = 冑 4共␳s − ␳f兲gdp
3␳f CD
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (13)
bed to move (Doron et al.14). The definition for this dry friction is
given by Eq. A-7 in Appendix A.
The system of six nonlinear first-order partial differential equa-
tions for mass balances and momentum balances completely de-
Doron et al.14 suggest the following equations to consider the fine the behavior of six unknown variables, Xⴕ, including the area
concentration effects on the hindered terminal settling velocity: (height) of the deposit bed, deposit bed velocity, cuttings concen-
tration in the suspension layer, mud velocity and cuttings velocity
in the suspension layer, and pressure.
vDEP = vpCf2m
m = 4.45ReDEP−0.1 ReDEP ⬍ 596 X⬘ = 关A2,u1,Cs2,us2,uf 2,p兴T. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (17)
m = 2.39 ReDEP ⱖ 596, . . . . . . . . . . . . . . . . . . (14)
The equations were differenced and solved simultaneously using New-
where ReDEP⳱the Reynolds number iterated with respect to the ton’s method with relaxation to obtain stable numerical solutions.
hindered terminal velocity deposition rate, vDEP .
The entrainment rate, vENT, is assumed to be a function of the Discussion of Results
interfacial shear velocity u12, as suggested by Doron et al.14 The numerical model was run to generate results for a variety of
hydrodynamic conditions, including fluid rheology, mud rate and

冑 冑
viscosity, and cuttings size. The results were analyzed and com-
␶12 ␭12|uf2 − u1|
u12 = = . . . . . . . . . . . . . . . . . . . . . . . . . . (15) pared to experimental data obtained from a cuttings transport flow
␳f ␳f loop system (CTFLS) at TRC/JNOC’s Kashiwazaki Test Facility
in Japan. Masuda et al.17 provided a comprehensive description of
For this study, a simple linear dependence on the interfacial shear the experimental apparatus and procedure to determine the cuttings
velocity with slope, mENT, and a critical threshold, u12
* , is assumed. velocity and concentration values inside the test section. Fig. 3 is
a schematic drawing of the flow loop. The annular test section was
vENT = 0 u12 ⱕ u12
* 9 m long, and consisted of a 5-in. (127.0 mm) ID transparent
vENT = mENT共u12 − u12
*兲 u12 ⬎ u12
* . . . . . . . . . . . . . . . . . . . (16) acrylic outer pipe (borehole or casing). A 2.063-in. (52.4 mm) or
2.875-in. (73.0 mm) OD steel inner pipe (drillpipe) was used. The
test section was mounted on a movable frame, and the inclination
As such, it is seen that below the critical threshold, no deposited of test section could be set at an angle ranging from 0 (horizontal)
cuttings are re-entrained into the flowing suspension layer. Above to 90° (vertical) in 15° increments. Drillpipe eccentricity could be
the critical threshold, the re-entrainment rate is assumed to be changed from 0 to 100%. Inner drillpipe could rotate at speeds
directly proportional to the velocity difference: the larger the ve- ranging from 0 to 80 rpm.

Fig. 3—Schematic drawing of TRC/JNOC’s Kashiwazaki Cuttings Transport Flow Loop System (CTFLS) at Kashiwazaki Test
Facility.

June 2003 SPE Journal 163


Mud containing a determined concentration of drill cuttings Results of Typical Simulation Run
was circulated in the flow loop by a PC pump. The flow rate of Figs. 4 through 9 provide a history of a simulation run exhibiting
mud-cuttings mixture was measured by an electromagnetic flow- critical behavior. In this case, the borehole was horizontal; the
meter. The mud flowed through the annulus in the transparent test distribution of pressure, cuttings concentration and velocity, mud
section and returned to the mud tank. The mud circulation lines velocity, deposit bed velocity, and normalized height with time
were made of 3-in. (7.62 mm) fixed and flexible pipes. Four pres- were determined with the distance measured downstream from
sure transducers and one temperature sensor were mounted along the drill bit. A non-Newtonian drilling mud (Mud 2) was used
the test section. Other pressure and temperature sensors and flow (␮⳱216 mPa•s, n⳱0.61, ␳f⳱1.19 g/cm3). The simulation was run
meters were positioned at different points in the loop. Water and for a duration of 201 seconds of simulated time. The mud flow rate
polyacrylamide solutions were used to simulate drilling mud, and was 40 m3/hour, and drill cuttings were introduced into the test
ceramic beads with average diameter of either 0.732 cm (∼0.25 in.) loop at a rate of 2.148 l/min (corresponding to a prescribed pen-
or 0.366 cm (∼0.125 in.) were used to simulate drilled cuttings. etration rate). The cuttings concentration worked out to be approxi-
The moving behavior of cuttings at both steady-state and unsteady- mately 0.3% (volume). Initially, there was no deposit bed inside
state flow conditions was recorded by a CCD video camera sys- the annulus, and the cuttings were fully suspended at the inlet. Fig.
tem. The captured images were analyzed to obtain the velocity 4 shows the pressure profile in the 9-m test section throughout the
profile, cross-sectional distribution, and average velocity of cut- run. The pressure drop reached a maximum value of 8.2 Pa at the
tings in the annulus. The LabVIEW™ program was used to display end of the run. It can be seen that after about 90 seconds, the
pressure, temperature, flow rate data in real time, and to save these pressure drop started to increase, corresponding to an increase in
data in hard disks at intervals of less than 0.1 s. Table 1 provides the height of the accumulating and moving deposit bed (Figs. 8 and
a summary of the experimental conditions, as well as data used in 9). The cuttings, after being introduced, began to deposit down-
the simulation runs. stream, as shown in Fig. 5. There was negligible slippage between
the mud and the few still-suspended cuttings in Region 2 because
of the high mud viscosity. Therefore, the two velocity profiles
tracked one another closely, as shown in Figs. 6 and 7.

Fig. 4—Pressure distribution profile for a simulation run (Mud 2 Fig. 5—Solid-phase concentration distribution profile for a
drilling fluid, ␳f=1.19 g/cm3, µ=216.0 mPaⴢs, n=0.61, ␳s=2.40 simulation run (Mud 2 drilling fluid, ␳f=1.19 g/cm3, µ=216.0
g/cm3; cuttings size=0.366 cm), mud rate=40 m3/hour; cuttings mPaⴢs, n=0.61, ␳s=2.40 g/cm3; cuttings size=0.366 cm), mud
injection rate=2.148 L/min. rate=40 m3/hour; cuttings injection rate=2.148 L/min.

164 June 2003 SPE Journal


Fig. 6—Solid-phase velocity distribution profile for a simulation
run (Mud 2 drilling fluid, ␳f=1.19 g/cm3, µ=216.0 mPaⴢs, n=0.61,
Fig. 7—Fluid-phase velocity distribution profile for a simulation
␳s=2.40 g/cm3; cuttings size=0.366 cm), mud rate=40 m3/hour;
run (Mud 2 drilling fluid, ␳f=1.19 g/cm3, µ=216.0 mPaⴢs, n=0.61,
cuttings injection rate=2.148 L/min.
␳s=2.40 g/cm3; cuttings size=0.366 cm), mud rate=40 m3/hour;
cuttings injection rate=2.148 L/min.
In Fig. 8, there was no deposit bed for the first 90 seconds of
the run. The deposited cuttings were initially dragged slowly along velocity, and pressure distribution. As the deposit bed height in-
the bottom of the annulus by the dilute suspension flowing above creased, the flow area became constricted. This, in turn, led to an
it. Their velocity was about 40% of the mud flow rate (velocity). increase in the mud velocity, and thus an increase in the pressure
As the cuttings continued to deposit, the bed became less mobile drop across the constricted interval. Furthermore, the increase in
and stagnated at a point about 4.0 m downstream from the drill bit. mud velocity also led to an increased interfacial stress and mo-
There, the cuttings accumulated to form a stationary bed, which mentum transfer to the bed mass accumulating at the stagnation
grew and spread with time. After 90 seconds, the height of the point. Locally, the bed mass moved again to approach a local
stationary deposit bed increased to its maximum value of 30% of steady-state velocity which balanced the driving forces with fric-
the annulus, at 7.5 m downstream from the drillbit. Over the length tion. Consequently, the stagnation point and the leading edge of
of the deposit bed (between 4.0 and 8.0 m), its average height was the deposit bed shifted downstream.
about 25% of the annulus (Fig. 9).
The following is offered to explain the phenomena associated Matching Simulation Results With
with the behavior observed in the simulation results. A stationary Experimental Data
bed formed locally as friction forces (owing to increased wall Figs. 10 through 13 show the comparison between simulation
contact) countered the momentum of a growing deposit bed. The results generated by this model and experimental data. In Fig. 10,
onset of a stationary bed developed at a stagnation point. At this water was the drilling fluid, and its flow rate was 40 m3/hour. Two
point, the friction forces were greater than the local driving forces different experiments were carried out, each having a different
(from interfacial stress, bulk momentum transfer, and pressure cuttings rate (0.92 and 1.076 l/min, respectively). The measured
gradient). Bed particles accumulated locally as a result of particles cuttings velocity for these two cases was 48.2 and 42.2 cm/s,
deceleration. The formation of a stationary deposit bed had an correspondingly. The simulator calculated average cuttings veloc-
impact on the profiles of the suspended cuttings velocity, mud ity of 48.5 and 44.7 cm/s for the same two runs (giving a maximum

Fig. 8—Deposit bed velocity distribution profile for a simulation Fig. 9—Deposit bed height distribution profile for a simulation
run (Mud 2 drilling fluid, ␳f=1.19 g/cm3, µ=216.0 mPaⴢs, n=0.61, run (Mud 2 drilling fluid, ␳f=1.19 g/cm3, µ=216.0 mPaⴢs, n=0.61,
␳s=2.40 g/cm3; cuttings size=0.366 cm), mud rate=40 m3/hour; ␳s=2.40 g/cm3; cuttings size=0.366 cm), mud rate=40 m3/hour;
cuttings injection rate=2.148 L/min. cuttings injection rate=2.148 L/min.

June 2003 SPE Journal 165


Fig. 10—Comparison of experimental data and simulation results, horizontal borehole, water drilling mud (␳f=1.00 g/cm3, µ=0.65
mPaⴢs, ␳s=2.40 g/cm3); mud rate=40 m3/hour; cuttings injection rate=0.920 and 1.076 L/min.

deviation of 6% of simulated velocity from measured data). It is to extend the experimental range and establish critical conditions
noted that in the experiments, cuttings velocity and concentration for each of the cases. The results were plotted to produce an
were (essentially) measured at a single point along the 9-m test envelope of minimum and maximum velocity observed over the
section (where the CCD video system was stationed). The numeri- entire length of the test section at the terminal simulation time.
cal model, on the other hand, enabled the cuttings velocity and Within the envelope, velocities averaged over the length of the test
concentration to be determined at different points along the whole section were also included to represent the flow profile in compact
test section. Consequently, these ranges of values allowed the form. Critical and subcritical conditions are represented when the
minimum, average, and maximum cuttings velocity and concen- minimum cuttings velocities are zero. Therefore, at some point
tration for the whole test section to be determined. along the length of the annulus, a stagnation point (as discussed
In these and other cases (of different mud rheologies and cut- earlier) will be encountered. Cuttings rate above the critical will
tings sizes), the simulator was run at different inlet cuttings rates lead to subcritical condition. Cuttings rate below the critical will
to determine the corresponding cuttings velocity profiles. This was lead to nonzero cuttings velocity.

Fig. 11—Comparison of experimental data and simulation results, horizontal borehole, Mud 1 drilling mud (␳f=1.00 g/cm3, µ=355
mPaⴢs, n=0.469, ␳s=2.40 g/cm3); mud rate=40 m3/hour; cuttings injection rate=0.013 and 0.018 L/min.

166 June 2003 SPE Journal


Fig. 12—Comparison of experimental data and simulation results, horizontal borehole, Mud 2 drilling mud (␳f=1.19 g/cm3, µ=216
mPaⴢs, n=0.610, ␳s=2.40 g/cm3); mud rate=40 m3/hour; cuttings injection rate=0.016, 0.250, 1.043, 1.696, and 2.148 L/min.

It is seen that for the cases simulated, the minimum cuttings envelope of minimum and maximum velocity observed over the
velocity tends to concave downward and approaches critical at an length of the test section at terminal simulation time followed the
inlet cuttings rate of approximately 1.4 l/min. What this means is same trend as that observed in Fig. 10. For the hydrodynamic con-
that for inlet cuttings rates above this critical value, a mud rate of ditions associated with Mud 1, the minimum cuttings velocity tends to
40 m3/hour would not be sufficient to remove cuttings effectively approach critical at an inlet cuttings rate of approximately 1.3 l/min.
from the system. A point must be made regarding the relatively poorer match
Similar simulation runs were made for non-Newtonian muds. between simulation results and experimental data of cuttings ve-
In Fig. 11, Mud 1 was used. Its flow rate was also 40 m3/hour. Two locity in this case (Mud 1) compared to the previous case (water).
different injected cuttings rates were used: 0.013 and 0.018 l/min, It was observed that simulation results deviated from experimental
respectively. The simulator calculated cuttings velocity values of data at extreme dilute cuttings rates (for the cases studied, less than
17.2 and 19.0 cm/s for the two runs, compared to their measured 0.05 l/min). In these cases, it is believed that the two-layer model
values of 21.5 and 10.8 cm/s, correspondingly. For this mud, the with a level interface does not represent the system adequately.

Fig. 13—Comparison of experimental data and simulation results, horizontal borehole, Mud 1 drilling mud (␳f=1.00 g/cm3, µ=355
mPaⴢs, n=0.469, ␳s=2.40 g/cm3); mud rate=40 m3/hour; cuttings injection rate=0.169 L/min.

June 2003 SPE Journal 167


This deviation of simulation results from experimental data was F␴ ⳱ surface force
also observed in the case of Mud 2 at extreme dilute cuttings rate g ⳱ gravitational acceleration
(as shown in Fig. 12). In the Mud 2 case, five different values of L12 ⳱ cross-sectional length of the level interface
injected cuttings rate were used, ranging from 0.016 l/min at the L1,inner ⳱ contact length between bed and the drillstem surface
low end to 2.146 l/min at the high end. Most of the values were
L2,inner ⳱ contact length between suspension and the drillstem
above 0.10 l/min. In these cases, the calculated cuttings velocities
matched very closely with measured values, as seen quite clearly in surface
Fig. 12. In the Mud 2 case, the minimum cuttings velocity tends to L1,outer ⳱ contact length between bed and the borehore surface
approach critical at an inlet cuttings rate of approximately 4.0 l/min. L2,outer ⳱ contact length between suspension and the borehore
In Fig. 13, Mud 1 was also used. In this case, the cuttings size surface
was 0.25 in. The mud flow rate was 40 m3/hour. Only one experi- m ⳱ hindered terminal settling correlation concentration
mental injected cuttings rate was used: 0.169 l/min. The cuttings exponent
velocity calculated by the model was 38.3 cm/s, which agreed to mENT ⳱ slope parameter of entrainment function
within 3.5% of the measured cuttings velocity of 37.0 cm/s. For
n ⳱ flow index
this case, the minimum cuttings velocity tends to approach critical
at an inlet cuttings rate of approximately 3.0 l/min. n ⳱ curvilinear borehole trace normal vector in (s, n)
p ⳱ pressure
Conclusions Qf,IN ⳱ fluid volumetric inflow rate
The following conclusions are offered from the study: Qs,IN ⳱ solid volumetric inflow rate
1. A transient, 1D numerical model has been developed to simulate R ⳱ local cross-sectional drillstring centered radial
the transport of drill cuttings in UBD. Its main features include coordinate in (r, ␪); or drillstring radius
two-layer formulation, interaction between the fluid phase Re1 ⳱ bed mixture Reynolds number
(mud) and the solid phase (drill cuttings) in the suspension layer, Re12 ⳱ bed-suspension interfacial shear Reynolds number
and interaction between the two layers (deposit bed and suspen- ReDEP ⳱ hindered terminal settling deposition Reynolds
sion layer). Its formulation enables, as shown by the results number
generated and presented in this paper, a wide range of transport
Ref2 ⳱ suspension fluid Reynolds number
phenomena to be simulated.
2. The model was run to generate results—including the unsteady- Rep ⳱ particle Reynolds number
state profiles of pressure distribution, drill cuttings and mud s0 ⳱ drill-bit inlet boundary location
velocity distributions, cuttings concentration distribution, distri- s ⳱ curvilinear borehole trace tangent vector in (s, n)
bution of the deposit bed velocity, and height—for different S ⳱ borehole axial coordinate
mud rheologies, mud rates, cuttings sizes, and cuttings rates. t ⳱ time
Explanations are offered to provide possible insights into dif- u1 ⳱ bed bulk velocity
ferent behaviors observed from consideration of physical fea- u12 ⳱ bed-suspension fluid shear velocity
tures of the model. u*12 ⳱ critical bed-suspension fluid shear velocity
3. In a majority of the cases studied, the numerical model success- uf2 ⳱ suspension fluid bulk velocity
fully matched experimental data. The calculated cuttings veloc-
us2 ⳱ suspension solid bulk velocity
ity typically agreed quantitatively with the measured cuttings
velocity. The match was, however, relatively poorer in the case vDEP ⳱ volumetric deposition transfer rate
of extremely dilute cuttings injection rates, as the two-layer vENT ⳱ volumetric entrainment transfer rate
model with a level interface does not adequately describe the vp ⳱ single-particle terminal settling velocity
interfacial phenomena. vDEP ⳱ deposition velocity
4. In addition to being used to match experimental data, the numerical vENT ⳱ entrainment velocity
model was also run for other conditions to establish envelopes of V ⳱ volume
minimum and maximum velocity for critical behavior. w ⳱ density related coefficients
x ⳱ Cartesian horizontal coordinate axis in (x, z)
Nomenclature Xⴕ ⳱ base unknown solution vector
A ⳱ total annulus cross-sectional area open to flow z ⳱ Cartesian vertical coordinate axis in (x, z)
A1 ⳱ bed cross-sectional area ⭸ ⳱ partial differential operator
A2 ⳱ suspension cross-sectional area ␣ ⳱ wellbore inclination angle
A12 ⳱ interfacial surface area between bed and suspension ␣1 ⳱ bed wall friction coefficient parameter
CD ⳱ drag coefficient ␣2 ⳱ suspension wall friction coefficient parameter
Cf1 ⳱ volumetric bed fluid concentration ␣D ⳱ drag factor
Cf2 ⳱ volumetric suspension fluid concentration ␤1 ⳱ bed wall friction coefficient parameter
Cs1 ⳱ volumetric bed solid concentration ␤2 ⳱ suspension wall friction coefficient parameter
Cs2 ⳱ volumetric suspension solid concentration ⌬ ⳱ difference operator
dp ⳱ particle diameter ␨ ⳱ cross-sectional drillstring centered Cartesian vertical
D1 ⳱ bed region equivalent hydraulic diameter coordinate axis in (␩, ␨); or bed-suspension interface
D2 ⳱ suspension region equivalent hydraulic diameter height
Douter ⳱ wellbore diameter ␨offset ⳱ vertical borehole center offset
f1 ⳱ bed mixture hydrodynamic wall friction coefficient ␩ ⳱ cross-sectional drillstring centered Cartesian
f12 ⳱ bed-suspension interfacial friction coefficient horizontal coordinate axis in (␩, ␨); or dry friction
ff2 ⳱ suspension fluid hydrodynamic wall friction coefficient
coefficient ␪ ⳱ cross-sectional drillstring centered polar coordinate in
fi ⳱ solid-fluid interaction function (Kirkwood et al.15) (r, ␪); or bed-suspension interface angle
F1 ⳱ dry bed friction force ␭1 ⳱ bed hydrodynamic wall shear stress function
FD ⳱ drag force ␭2 ⳱ suspension hydrodynamic wall shear stress function
Fg ⳱ gravitational force ␭12 ⳱ bed-suspension hydrodynamic shear stress function
Fsf ⳱ solid-fluid drag force ␭sf ⳱ solid-fluid drag function

168 June 2003 SPE Journal


␮1 ⳱ bed mixture viscosity 15. Kirkwood, J.G., Maun, E.K., and Adler, B.J.: “Radial Distribution
␮f ⳱ fluid viscosity Functions and the Equation of State for a Fluid Composed of Rigid
␳1 ⳱ bed bulk density Spherical Particles,” Jour. of Chemistry & Physics (1950) 18, 1040.
␳2 ⳱ suspension bulk density 16. Televantos, Y. et al.: “Flow of Slurries of Coarse Particles at High
␳s ⳱ solid bulk density Solids Concentrations,” Can. Jour. of Chemical Engg. (1979) 57, 255.
␳f ⳱ fluid bulk density 17. Masuda, Y. et al.: “Experimental Study to Determine Critical Flow
␶1 ⳱ bed wall shear stress Rate of Cuttings Transport During Underbalanced Drilling,” paper
␶2 ⳱ suspension wall shear stress IADC/SPE 62737 presented at the 2000 IADC/SPE Asia Pacific Drill-
ing Technology Conference, Kuala Lumpur, 11–13 September.
␶12 ⳱ bed suspension shear stress
Subscripts
Appendix A—Mass Conservation Equations
1 ⳱ bed region
Mass balances over differential volume elements of length ⌬s will
2 ⳱ suspension region
relate the mass accumulation over time to the net rate of mass
12 ⳱ interface between bed and suspension regions inflow across the bounding surface. For the moving bed mixture in
D ⳱ drag Region 1,
DEP ⳱ deposition
ENT ⳱ entrainment ⭸ ⭸
F ⳱ fluid 共A ␳ 兲⌬s + 共A1␳1u1兲⌬s
⭸t 1 1 ⭸s
G ⳱ gravitational
I ⳱ spatial grid index
inner ⳱ drillstring wall inner boundary
= ␳s + 冉
Cf1

␳ C v L ⌬s − ␳1vENT L12⌬s. . . . (A-1)
Cs1 f s2 DEP 12
offset ⳱ borehole center offset The mass conservation for the solids component and fluid com-
outer ⳱ borehole wall outer boundary ponent in the flowing suspension mixture in Region 2 is given by
p ⳱ particle
s ⳱ solid ⭸ ⭸
sf ⳱ solid-fluid 共A ␳ C 兲⌬s + 共A2␳sCs2us2兲⌬s
⭸t 2 s s2 ⭸s
␴ ⳱ surface = −␳sCs2vDEP L12⌬s + ␳sCs1vENT L12⌬s
+ ␳sQs,IN. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-2)
References
1. Rommetveit, R. et al.: “A Dynamic Model for Underbalanced Drilling ⭸ ⭸
共A ␳ C 兲⌬s + 共A2␳f Cf 2uf 2兲⌬s
With Coiled Tubing,” paper SPE/IADC 29363 presented at the 1995 ⭸t 2 f f 2 ⭸s
SPE/IADC Drilling Conference, Amsterdam, 28 February–2 March. Cf1
2. Liu, G. and Medley, G.H. Jr.: “Foam Computer Model Helps in Analy- = −␳f Cs2vDEP L12⌬s
Cs1
sis of Underbalanced Drilling,” Oil & Gas Journal (1 July 1996) 114. + ␳f Cf1vENT L12⌬s + ␳f Qf,IN. . . . . . . . . . . . . . . (A-3)
3. Wang, Z. et al.: “Underbalanced Drilling Requires Downhole Pressure
Management,” Oil & Gas Journal (16 June 1997) 54.
Momentum Conservation Equations
4. Wang, Z. et al.: “Underbalanced Drilling Model Simulates Dynamic
Well Bore Conditions,” Oil & Gas Journal (7 July 1997) 62. The rate of change in total linear momentum over a stationary
5. Langlinais, J.P., Bourgoyne, A.T., and Holden, W.R.: “Frictional Pres- volume element will have contributions from transient local ac-
sure Losses for Annular Flow of Drilling Mud and Mud-Gas Mixtures,” celerations, net convective acceleration and momentum transfer
Jour. of Energy Resources Technology (March 1985) 142. associated with mass deposition and entrainment between suspen-
6. Luo, Y., Bern, P.A., and Chambers, B.D.: “Flow Rate Predictions for sion and bed regions.
Cleaning Deviated Wells,” paper IADC/SPE 23884 presented at the Momentum balance for the moving bed in Region 1 is given by
1992 IADC/SPE Drilling Conference, New Orleans, 18–21 February.
7. Sharma, M.P. and Chowdhry, D.V.: “A Computational Model for ⭸ ⭸
共A ␳ u 兲⌬s + 共A1␳1u1u1兲⌬s
Drilled Cutting Transport in Air (or Gas) Drilling Operations,” Jour. of ⭸t 1 1 1 ⭸s
Energy Resources Technology (March 1986) 8. ⭸p
= −A1 ⌬s − A1⌬s␳1g sin ␣
8. Hemphill, T. and Larsen, T.I.: “Hole-Cleaning Capabilities of Oil- ⭸s
Based and Water-Based Drilling Fluids: A Comparative Experimental + ␭12共uf 2 − u1兲L12⌬s
Study,” paper SPE 26328 presented at the 1993 SPE Annual Technical − ␭1u1共L1,outer + L1,inner兲⌬s
Conference and Exhibition, Houston, 3–6 October. − u1vENT␳1L12⌬s + us2vDEPCs2␳s L12⌬s
9. Guo, B. and Rajtar, J.M.: “Volume Requirements for Aerated Mud Cf1
Drilling,” SPEDC (September 1995) 165. + uf 2vDEPCs2 ␳f L12⌬s − F1. . . . . . . . . . . . . (A-4)
Cs1
10. Gao, E. et al.: “A New MTV Computer Package for Hole Cleaning
Design and Analysis,” paper SPE 26217 presented at the 1993 SPE
Momentum balance for the suspended solids and fluid in Region 2
Computer Conference, New Orleans, 11–14 June.
are given by
11. Larsen, T.I., Pilehvari, A.A., and Azar, J.J.: “Development of a New
Cuttings Transport Model for High-Angle Wellbores Including Hori-
⭸ ⭸
zontal Wells,” SPEDC (June 1997) 129. 共A2␳sCs2us2兲⌬s + 共A2␳sCs2us2us2兲⌬s
12. Kamp, A.M. and Rivero, M.: “Layer Modeling for Cuttings Transport ⭸t ⭸s
⭸p

冤 冥
in Highly Inclined Wellbores,” paper SPE 53942 presented at the 1999
A2 ⌬s + A2⌬s␳s g sin ␣
SPE Latin American and Carribean Petroleum Engineering Confer- ⭸s
ence, Caracas, 21–23 April. = −Cs2 + ␭12共uf 2 − u1兲L12⌬s
13. Shook, C.A. and Roco, M.C.: Slurry Flow: Principles and Practice,
Butterworth-Heinemann, Burlington, Massachusetts (1991).
+ ␭2uf 2共L2,outer + L2,inner兲⌬s
14. Doron, P., Granica, D., and Barnea, D.: “Slurry Flow in Horizontal + u1vENTCs1␳sL12⌬s − us2vDEPCs2␳sL12⌬s
Pipes—Experimental and Modeling,” Intl. Jour. of Multiphase Flow
(1987) 13, No. 4, 535. + A2␭sf 共uf 2 − us2兲⌬s, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-5)

June 2003 SPE Journal 169


⭸ ⭸ e-mail: oguztoreli@ibm.net. Previously, he was with Real Op-
共A ␳ C u 兲⌬s + 共A2␳fCf2uf2uf2兲⌬s tions Group in Milan, Italy, as Director of Oil & Gas, Energy and
⭸t 2 f f2 f2 ⭸s
Natural Resources. He has also worked as a consultant spe-
⭸p

冤 冥
A2 ⌬s + A2⌬s␳f g sin ␣ cializing in mathematical modeling, numerical simulation, and
⭸s software development in the areas of engineering, manage-
= −Cf2 + ␭12共uf 2 − u1兲L12⌬s ment, and finance. Currently, he is developing stochastic dy-
namic programming decision-making tools to value optimal
+ ␭2uf 2共L2,outer + L2,inner兲⌬s operating policies and path-dependent action plans for peak-
ing power plants and gas reservoir storage assets with con-
Cf1 straints contingent on economic, physical, operational, and
+ u1vENTCf1␳fL12⌬s − uf2vDEP Cs2␳fL12⌬s contractual states. Oguztoreli holds a PhD degree in engineer-
Cs1
ing management, MBA and MSc degrees in petroleum engi-
− A2␭sf 共uf 2 − us2兲⌬s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-6) neering, and a BSc degree in physics. Yoshihiro Masuda is cur-
rently an associate professor in the Dept. of Geosystem Engi-
The sum of driving forces in Region 1 must overcome the dry neering at the U. of Tokyo, Japan. e-mail: masuda@geosys.t.
friction resistance force, F1, in direction opposing flow for the bed u-tokyo.ac.jp. His current research activities include hole
to move (Doron et al.14), as follows: cleaning in underbalanced drilling, in-situ polymer gelation in

再 冋冉 冊冉 冊册冎
sediments, and development of a numerical simulator predict-
2共␨1 + ␨兲 ␶
1 −1 ␪+ ing gas production behavior from methane-hydrate reservoirs.
F1 = ␩ 共␳ − ␳f兲gC1sDouter
2
Douter 2 He previously was a petroleum engineer with Japan Petroleum
2 s + cos␪ Exploration Co., Ltd. He holds a doctoral degree in engineer-
⌬s sgnu1cos␣ ing from the U. of Tokyo. Tetsuo Yonezawa is currently a Project
Manager of the Research Project Team, Methane Hydrate,
⭸p Technology Research Center of Japan National Oil Corp., Ja-
F1␴ = A1 ⌬s − ␶12L12⌬s + ␶1共L1,outer + Li,inner兲⌬s
⭸s pan. e-mail: yonezw-t@jnoc.go.jp. His current activities include
+ F1 and 兺
Fdriving ⬎ F1 for |u1| ⬎ 0, . . . . . . . . . . . . . . . (A-7)
drilling and completion R&D projects including, among others,
UBD and borehole stability. He holds a BA degree in engineer-
where ␩⳱the dry friction coefficient, and sgnu1 cos ␣ accounts for ing from Hokkaido U., Japan. Atsushi Kobayashi is currently a
the inclined flow direction. section manager in the Tokyo Technology Center of Telnite
Co., Ltd. e-mail: atsushi_kobayashi@telnite.co.jp. He previously
joined with JNOC/TRC for the Underbalanced Drilling Project in
Quang T. Doan is a reservoir engineer at APA Petroleum Engi- JNOC. Kobayashi holds a BSc degree in chemical engineering
neering Inc. in Calgary, Canada. e-mail: qdoan@apa- from Akita U., Japan. Shigemi Naganawa is currently a re-
inc.com. He was previously a senior engineer in the R & D search associate in the Dept. of Geosystem Engineering at the
Group of the Computer Modelling Group Ltd. in Calgary. Be- U. of Tokyo, Japan. e-mail: naganawa@kelly.t.u-tokyo.ac.jp.
fore joining CMG Ltd., he was a consultant involved in devel- His current research activities include cuttings transport in di-
oping simulators for multiphase flow in production and drilling rectional wells and modeling of drill-bit/strings dynamics. Na-
processes and undertaking a number of reservoir analysis and ganawa holds a Meng degree in mineral development engi-
well-performance studies. He has taught courses in production neering from the U. of Tokyo, Japan. Arjan M. Kamp was until
engineering, reservoir engineering, and simulation, as well as recently employed as researcher and project leader in PDVSA
supervised MSc and PhD theses on thermal recovery pro- Intevep in Venezuela. e-mail: caroyarjan@cantv.net. His re-
cesses, cold production of heavy oil, and multiphase flow search activities include multiphase flow applications in the oil
modeling. His interests include heavy oil recovery and mul- industry, varying from drill cuttings transport, underbalanced
tiphase flow phenomena. Doan holds BSc, MSc, and PhD de- drilling, flow in horizontal wells, and heavy oil flow in reservoirs.
grees in petroleum engineering, all from the U. of Alberta, Mr. Kamp is an applied physicist from the Eindhoven U. of Tech-
Canada. Mustafa Oguztoreli is currently Director of Quantita- nology in the Netherlands. He holds a PhD degree in fluid me-
tive Resources at Shell Trading Gas and Power in Houston. chanics from the Inst. Natl. Polytechnique de Toulouse, France.

170 June 2003 SPE Journal

Das könnte Ihnen auch gefallen