Вы находитесь на странице: 1из 13

SPE 124195

Simulation of Liquid Unloading From a Gas Well With Coiled Tubing Using
a Transient Software
P. Salim, and J. Li, BJ Services Company

Copyright 2009, Society of Petroleum Engineers


This paper was prepared for presentation at the 2009 SPE Annual Technical Conference and Exhibition held in New Orleans, Louisiana, USA, 47 October 2009.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
Unloading gas wells is one of the most common applications for coiled tubing (CT). Despite the large number of jobs
completed, fundamental questions about the optimum gas lift rate, run-in-hole speed and how much nitrogen is required
remain. This is because the unloading process is not steady state, and the commonly used CT computer simulations can only
model steady state flow.
This paper describes transient software that has been developed and used to determine the nitrogen volume and cleaning
time required when optimizing the process of liquid unloading from a gas well with CT.
Based on experimental test results with a full scale flow loop, a critical gas velocity model was developed. This model
determines how much liquid can be lifted for a given gas rate under varying operating conditions. It also determines the
minimum gas rate required for complete liquid unloading. Several examples are presented which illustrate the transient
characteristics of the liquid unloading process. A few field cases illustrate the benefits of using the transient model and some
problems with conventional design methods.
Introduction
Unloading gas wells is a common operation in the oil industry. After workover operations the liquid used to kill or stimulate
the well must be unloaded from the tubing string to return the well to normal operation. Some wells naturally liquid load and
require a gas lift to put them back on production. Two techniques are commonly used to unload wells. One method would
be to use gas lift mandrels in the completion and pressurized gas in the area. Another is to inject nitrogen in the completion
with CT. The CT method will be our focus in this paper.
There are a few steady state design methods which can be used to design the liquid unloading process. One method
involves the generation of a family of curves comparing bottom hole pressure (BHP) and gas injection rate for various
assumed produced liquid rates, while holding CT size and circulation depth constant. Based on these curves, the potential
liquid unloading rate can be estimated at different injection rates for a particular condition. In order to estimate the liquid
unloading rate at other conditions, similar curves can be generated. However, with such conventional steady state design
methods, the transient behavior of liquid unloading process, i.e. pressure/fluid rate changing at the surface or at the bottom of
hole, can not be captured and the unloading time is poorly estimated. It will be shown that simulating transient behavior of
liquid unloading process is very important for the design and execution of that process.
Process design includes the determination of CT size, tripping speeds, nitrogen rate and surface choke setting. A good
design will bring the well back to production in a timely manner with minimal nitrogen consumption but sometimes these
goals are compromised by surface fluid rates and dynamic BHP. Surface fluid rate can be crucial when there is limited
surface handling equipment as is common in many offshore operations. Dynamic BHP can be critical as some reservoirs are
very sensitive to fluid losses during the liquid unloading process.
This paper describes transient software that has been developed and used to optimize the process of liquid unloading in a
gas well with CT. The transient model is validated with Lages (2000) full scale well test data and field operational data. The
transient feature of the unloading process is clearly captured by the software. The model tracks how the whole system
dynamically responds to changes in major operational conditions such as the variation of pump rate, choke size, and
downhole conditions. A critical gas velocity model was developed and incorporated into the software.

SPE 124195

Experimental Setup and Data Collection


The flow loop, shown in figure 1, was used throughout this project. The loop consisted of a 20ft-long transparent lexan pipe
with a 5-inch inner diameter simulating the open hole and a 2-3/8-inch steel inner pipe simulating a gas lift pipe. The inner
pipe was positioned on the bottom of Lexan pipe to simulate the worst case for the liquid removal (eccentricity = 100%).
The loop was mounted on a rigid guide rail and could be inclined at any angle in the range of 0o-90o from vertical. A
schematic of the flow loop system is shown in figure 2. The annular test section is first filled with liquid after which the
centrifugal pump is shut down. Nitrogen is then flowed into the test section. After a steady state is established, pneumatic
actuating valves isolate the test section. The liquid and gas volumes in the annulus of the test section are then measured.
From this information, the liquid holdup and the average in-situ velocity for gas phase can be determined. This liquid holdup
was called the zero net liquid flow (ZNLF) holdup. Tests were conducted at increasing gas rates until we reached the critical
gas rate at which the liquid from the test section was completely removed.
The flow rate was measured by two mass flow meters. The mass flow meters were equipped with a densitometer feature
that could measure the density of gas and liquid phases. Water, two different densities of brines and two different biopolymer
fluids were used to conduct the tests. Results were used to investigate the effect of liquid density and viscosity on the ZNLF
holdup and the minimum critical velocity. For each liquid the tests were conducted at several different deviations. The
pressure inside of the test section could be controlled with a back choke. The maximum pressure inside of the test section was
150 psi. Most of tests were conducted with the choke fully open and a pressure of roughly 20 psi. A set of tests were also
conducted to investigate the effect of pressure on the critical velocity using the maximum test section pressure of 150 psi.
Test results showed that the ZNLF holdup increased with deviation angle reaching a maximum at about 55o, it then
decreased as the deviation angle increased. This was similar to the other investigators observations. For the same gas
velocity, the ZNLF holdup increased as the fluid density or viscosity increased. There was less ZNLF holdup when bottom
hole pressure was increased due to increased gas density with pressure. More than 300 test data points were collected and
compiled into figure 3. It is shown that the ZNLF holdup could be correlated with a single curve as a function of the velocity
ratio between the gas superficial velocity and the gas critical velocity. A generalized correlation was developed to predict the
critical gas velocity at which all the liquid can be removed completely from the wellbore. Based on this correlation, the liquid
holdup can be predicted when the gas velocity is less than the critical velocity.
Transient Model
The mathematical model of liquid unloading is based on the conservation of mass and a drift-flux model. One momentum
equation is solved and the slip velocity between gas and liquid phases is determined with steady state equations that consider
the flow regime. A steady-state isothermal temperature profile is assumed in the wellbore. The space domain is assumed to
be one dimension in axial direction and is discretised into N nodes.
The conservation of mass is divided into two parts: (1) the conservation of mass for each gas and (2) the conservation of
mass for each liquid. The conservation of mass for gas j in node i is given by:

g , ji g , jiVi
= Fg , ji
t

(1)

and the conservation of mass for liquid k in node i is given by:

l , ki l , kiVi
= Fl , ki
t
j = ng

k = nl

j =1

k =1

(2)

g , ji + l ,ki = 1
In Eqs. (1) to (3),

(3)

g, ji

fraction of gas j in node i,


variable, and Fg , ji and

l, ki

is the density of gas j in node i,

l, ki

Fl , ki

is the density of liquid k in node i,

g, ji

is the volume

is the volume fraction of liquid k in node i, Vi is the volume of node i, t is the time
are respectively defined as the mass transfer of gas j and liquid k in node i:

Fg , ji = ( g , ji u g , i g , ji Ai ) ( g , ji u g , i g , ji Ai )
in

out

S g , ji

Fl , ki = ( l , ki uli l , ki Ai )in ( l , ki ul , i l , ki Ai )out S l , ki

(4)
(5)

SPE 124195

ul , i is liquid velocity in node i, Ai is the flow area of node i, S g , ji is the mass


source of gas j entering node i, and S l , ki is the mass source of liquid k entering node i. u g , i and u l , i in Eqs. (4) and (5) are

where u g , i is gas velocity in node i,

computed from the drift-flux model shown in Eqs. (6) and (7):

u g , i = c m , i u m, i + u d , i
u g , i c m, i
ud ,i
ul , i =
u m, i +
=
, i cm
,i
,i
cm
cm

(6)
(7)

, i are distribution coefficients in node i, u d , i is drift velocity currently assumed as zero, and u m, i is
where c m, i and c m
the mixture velocity in node i and is computed by means of the Bernoulli equation. In single phase flow, both c m, i and

, i are set to unity. In gas-liquid flow, c m, i and c m


, i are estimated from a gas-liquid model (such as Taitel-Dukler
cm
model, Duns-Ros model, etc. depending on the flow regime), but for particular cases in which the superficial gas velocity
, i are estimated from the correlation generated from figure 3 to
becomes less than the critical gas velocity, c m, i and c m
account for the ZNLF holdup. After some mathematical manipulations for Eqs. (1) and (2), a Transient Pressure Equation is
generated and shown as Eq. (8) below:
j = n g F
k = nl
k = n l F
P
j =ng
g , ji
l , ki Vi
i

(8)
=
Vi g , ji K g , ji + l , ki K l , ki
+

g , ji k =1 l , ki t
t
j =1
=
1
=
1
k
j

where: K g , ji is the compressibility of gas j in node i, K l , ki is the compressibility of liquid k in node i, and Pi is pressure
in node i.
Eqs. (1), (2) and (8) are categorized as an initial value problem and are solved from an initial condition where the initial
masses of each phase and the initial pressure and temperature of each node are known. For the computation stability and
speed, the equations are solved by using the fully implicit or backward time numerical scheme, in which a set of simultaneous
linear equations are solved at each time step.
The initial condition was obtained from a steady-state simulation described by Craig (2003), Li et al. (2002), Misselbrook
et al. (1991), Nasr-El-Din et al. (2006), and Ovesen et al. (2003). The steady-state simulation runs in Windows on a standard
PC and is a powerful analytical tool that provides a complete evaluation for both flow and force analyses at each control
volume. The simulation has three main parts:
1. Input Parameters: Any well path, completion and CT combination can be described. Non-Newtonian and Newtonian
fluids in single phase and energized mixtures can be considered. Reservoir pressure is specified, its flow rate can aid the
cleanout or not at the users discretion. Inputs are run through hundreds of validations prior to being sent to the
calculation algorithms. Extensive hints help guide the user to fix validation errors.
2. Job Design: Typically one varies flow rates, fluid types, penetration rate, circulation and trip times to optimize the job.
3. Output Information: The simulation generates outputs of all pertinent variables in both tabular and graphical formats. The
output includes pressure distribution, velocities of liquid and gas, liquid holdup and flow regimes both inside the CT and
in the CT/completion annulus. Information on shear rates, effective viscosity, flow regime, friction gradients and
hydrostatic gradients are also provided. Numerous warnings and messages are also generated to alert and guide the user
about potential problems.
The modeling for both the steady-state and transient flow with single or two-phase fluid in the coiled tubing application,
pipeline application and other applications has been extensively validated with the full scale experimental data, field data, and
other data available in the public domain.
Simulation Results and Case Studies
Four cases related to the liquid unloading process are chosen for discussion in the following section. In case #1 the model is
compared to full scale well test data where a well is unloaded but continues to produce water. Case #2 discusses the effect of
main parameters (i.e. gas rate, the rate of penetration of CT, CT size, and wellhead pressure) on the unloading time and liquid
return rate. This case study also shows some typical features of the transient unloading a gas well and how to utilize the
software to optimize the unloading process. In case study #3 the transient software was used to help design the job.
Simulation and field results are compared. In case #4 we again compare field results to the transient simulation but this job
design did not benefit from the transient simulation and relied only on conventional steady state methods.

SPE 124195

Case Study # 1: Model validation with the Full-Scale Well Test Data
A full-scale investigation was executed by Lage (2000) in a 1278 m vertical well with a drill pipe of 3.5 OD x 2.764 ID
placed inside a casing of 6.276 ID (seen in figure 4). Detail of the well setup was included in Lage (2000). Initially the well
is full of water. Water is then injected into the drill pipe at the rate of 160 gpm. 13 minutes later, nitrogen is also injected at
the rate of 300 scfm through a parasite string located at 760 m. The drill pipe remains stationary at 1262 m. The downhole
memory gauges were attached to the drillstring to record annular pressures at 998 m, 605 m and 185 m. The end of the
drillstring was open, this enabled the pressure sensor in the logging tool to be used as a real time monitor of the conditions at
the bottom of the drillstring.
A transient simulation was conducted based on the Lages test condition. Choke size at wellhead was assumed to be fully
open and was set at 100 in./64. Figures 5 to 7 show the comparison between the simulation results and experimental data for
completion pressures at 1262 m, 605 m and 185 m. During the simulation, u-tubing phenomena (Kalessidis et al., 1994) was
encountered in the drill pipe after 15:50 minutes. Pressure oscillations shown in figures 5 to 7 are related to how the transient
simulation handled the u-tubing phenomena. In general, the transient model is shown to reproduce the experimental results.
Case Study #2: Optimizing the Unloading Process with Transient Model
We were requested to optimize a well kickoff operation for a 10600 ft. vertical well. Perforations were located at 10,000 ft,
BHT is 250 oF, and WHT is 80 oF.
0.7
p 2 p B2
f
Mscf/D; Gas gravity = 0.65; Oil-to-Gas Ratio = 100 bbl/MMscf;
Reservoir characteristics: QG = 5.0

1000

The well is filled with brine to 3200 ft and has a wellhead pressure of 150 psi (1 Mpa). During the liquid unloading
process, the coiled tubing is run into hole from 3000 ft to the bottom of the well (10600 ft) with a fixed rate of penetration
(ROP) and a fixed nitrogen pump rate. Once stable well production is achieved, nitrogen is stopped and the coiled tubing is
pulled out of hole. The affect of varying coiled tubing ODs (1.25, 1.5 and 1.75), coiled tubing ROPs (10 to 100 ft/min)
and nitrogen pump rates (100 to 1500 scfm) on the time required to achieve stable well production was recorded.
For this comparison, the start of stable well production was arbitrarily defined as the first time that the bottom hole
pressure falls below 2700 psi. To illustrate some of the challenges in this case study detailed results were generated for one
scenario. 1.25 OD coiled tubing is run into the well at 50 ft/min while pumping 700 scfm nitrogen, results are shown in
figures 8 to 11. The time to reach stable well production was recorded when the bottom hole pressure reached 2700 psi, this
occurred at approximately 71 minutes. At 71 minutes nitrogen pumping was stopped (shown in figure 8) and the coiled
tubing was pulled out of hole at the rate of 30 ft/min (shown in figure 9). Figure 8 also shows how transient delays can affect
the job design. N2 was halted at 71 minutes but it took another 50 minutes for the reservoir and surface gas rates to reach a
steady-state.
Figure 10 shows the pressure response versus time at key points in the well. You will notice that the initial pressure at the
perforations is 3300 psi, giving a 300 psi overbalance. Hence the reservoir is initially charged with kill fluid. Figure 11
shows reservoir and wellhead liquid rates versus time but excludes the initial charging rate. This was done to focus the
graph on some interesting details that occurred later. As the perforation pressure started to decline below pf (3,000 psi),
reservoir production increased as expected but then dropped around 59 minutes. This drop reflected the last of the kill fluid
and the fact that it could be produced back more easily than the produced fluids. A liquid slug around 90 minutes
corresponded to the gas break through at surface (comparing figures 8 and 11). It was encouraging to note that the simulation
predicted the liquid rate would spike before the gas rate on surface, this has been observed in many field operations. All the
parameters (i.e. pressures and surface rates of liquid and gas) approached a steady state 100 minutes after the initial unloading
process.
This scenario was then modified to test the affect of coiled tubing ODs (1.25, 1.5 and 1.75), coiled tubing ROPs (10 to
100 ft/min) and nitrogen rates (100 to 1500 scfm). The typical results for the runs are presented in figures 12 to 14.
Figure 12 shows the time to reach stable well production for coiled tubing 1.25 OD at various ROPs and nitrogen pump
rates. As observed in experiments the liquid unloading required a critical minimum gas velocity. Figure 12 shows that
nitrogen rates just above the critical gas velocity needed more time to reach stable well production. Also increasing the
nitrogen rate did not always result in reduced unloading times. Once an optimum N2 rate is achieved the unloading time is
roughly constant. The optimum value changes with ROP. At an ROP of 10 ft/min, a nitrogen rate of 300 scfm appeared to
be the optimum. At 30 ft/min, a nitrogen rate of 500 scfm was required. At 70 ft/min, the optimum nitrogen rate was 900. In
figure 12 we can see that the slope of the total consumed N2 volume changes around the optimum N2 rate. In addition,
increasing ROP from 10 to 30 ft/min showed remarkable reduction in liquid unloading time, but ROPs beyond 30 ft/min did
not.
Figure 13 plots the average liquid unloading rate for various gas rates and ROPs. For a given injection gas rate, a higher
ROP results in a higher liquid unloading rate at the surface. Similarly for a given ROP, the higher the injection gas rate is, the
higher the liquid unloading rate will be. Figure 13 indicates that the reduction in kick off times with increasing nitrogen rate
tends to flatten off in all ROP curves. This trend also indicates that there is an optimum gas rate for a given ROP curve.

SPE 124195

Figure 14 shows how the CT size affects the average unloading liquid rate at the surface. For a given injection gas rate, a
larger CT would result in a higher unloading liquid rate. However, it seems that the CT size would not affect the optimum gas
injection rate and for this scenario in figure 14, it is about 700 scfm. The optimum gas injection rate is mainly affected by the
RIH speed.
In general the surface equipment (separator, choke size and surface return line) is selected according to normal well
production rates. However unloading operations usually required significantly higher rates. If the surface equipment is
undersized, dramatically higher well head pressures (WHP) can result. Figure 15 shows how increasing WHP affects the time
to reach stable well production. Higher WHPs result in longer unloading times as the pressure increases gas density which in
turn reduces gas velocity and hence the efficiency of the whole process
After considering CT availability in the operating region and other logistical issues, it was recommended to use 1.5 CT,
a RIH speed of 50 ft/min and a nitrogen rate of 700 scfm.
Case Study #3: N2 Well Unloading and Cleanup Assistance
A subsea water injector in the Caspian Sea was completed with an expanded sand screen and downhole flow control devices
to inject water between two isolated sands in the formation. The well profile and wellbore information are plotted in figures
16 and 17, respectively. The well is 3924.2 m MD with a maximum deviation of 57o. After swapping the well bore fluid over
to base oil it was planned to assist the well flowback and cleanup using coiled tubing N2 circulation. While cleaning up, the
N2 rate is adjusted to generate the desired drawdown across the sand-face/screens.
The BHA included a down hole memory gauge which collected the down hole pressure and temperature near the
BHA/circulation point during the liquid unloading process. Downhole completion pressure was recorded with a transducer
installed in the well at 3648 m MD.
Job Details: Figure 18 shows operational details recorded in the field. The flowhead valves were opened and CT was RIH
circulating N2. The N2 rate used was the minimum critical gas rate determined by the transient software. At 210 m the CT was
stopped and it was confirmed that the fluid quantity recovered in well test surge tanks was equivalent to the landing string
volume. With the fluid unloaded from the landing string confirmed, CT continued to RIH and N2 was circulated at a rate
lower than the critical rate.
After 200 min the coil reached 850 m (30o wellbore deviation) and fluid returns were lost at surface. After shutting down
the N2 rate was again increased to its critical value and fluid returns were re-established. CT was then RIH as the well
started to flow. Real time downhole gauges confirmed reservoir drawdown.
With the CT parked at 2500 m N2 rate was changed in an attempt to control the drawdown on the well. BHP stabilized at
approximately 2500 psi and the decision was made to RIH with CT to 3000 m. It was hoped that running to 3000 m would
increase drawdown on the well.
At 3250 m the N2 supply was exhausted by circulating at 900 scfm for a further 14 hours. During this period, the desired
500 psi drawdown was generated downhole and controlled by adjusting the surface choke in the range of 54/64 60/64.
With this drawdown, the well flowed an average of 4500 bbl/day (3500 bbl/day low 5500 bbl/day high). CT pull tests were
conducted continually throughout the duration of the well flow period.
CT was pulled out of the well without circulating N2. The flowhead swab valve and surface choke were closed. The CT
reel was clamped and the well was monitored for 8 hours. Injectivity tests were conducted on both zones and a tracer
chemical was displaced into the formation. CT equipment and surface iron were rigged down.
Job Modeling: It was unknown if 1 CT could unload a 9-5/8 riser. It was believed that the well would flow with N2
assistance after being kicked off, but first the hydrostatic column had to be reduced enough to initiate reservoir flow.
Extensive transient modeling was carried out and the results indicated that liquid unloading was indeed possible. Intermittent
or slug flow was predicted at the start of the unloading process but this would change to continuous flow once the well
drawdown was established. Surface flows during the job confirmed this prediction.
After the job the actual data run with the transient software and the results are shown in figure 19. A reservoir PI of 9.4
bbl/day/psi was calculated from the well test data and was used in the modeling. A few observations:
Figure 19 indicates continuous flow is established after 250 minutes, this was confirmed by the well test package.
Figure 20 shows that predicted pressure trends (WHP, CTP, BHP and BHAP) match well with job data.
The model accurately predicted the critical gas velocity required for unloading the 9-5/8 riser. After running through the
riser and stopping at 210 m the volume of fluid returned was equal to the riser volume (60 bbls).
As seen in figure 19 the model predicts flow slugging after 150 min. This was noted on the job and after 200 min the CT
was halted at 850 m and the N2 was shut down. After some discussion the N2 rate was again increased beyond the
predicted critical gas rate. Both the simulation and job observations confirmed that continuous returns were re-established
and maintained.
A 10-20 minute delay between changing N2 rate and BHP change was predicted by the model and confirmed by job data.
In the end the well was flowed back for approximately 28 hours with 4600 bbls of oil being recovered from the reservoir.
The average drawdown generated downhole during the flowback was 500 psi. This method of cleanup resulted in far greater
fluid volumes being recovered from similar wells using well surge methods.

SPE 124195

Although not used here another approach to well clean up involves tripping to total depth (TD) first. Figure 21 shows the
surface return rates for both liquid and gas phase when N2 is not circulated during the CT RIH but is circulated at 1500 scfm
only when CT reaches TD (3250 m MD). The liquid return rate when N2 is not circulated during CT RIH is almost double
that when N2 is circulated during RIH (see figure 21). In this case the higher return rate would have caused a problem for the
surface separator. The model also predicted that the time to unload the well is less when circulating only on bottom and not
during RIH. Again the potential for a severe slug determined that unloading as the CT was RIH was the safest option.
Actual well performance parameters (BHP gauge data, CT BHA memory gauge data, CTP, WHP, N2 circulation rate,
well flow rates) were measured and compared closely with the results of the transient software. Given the model is the only
means of evaluating the feasibility of future well unloads, it was important to gain an understanding of actual versus
predicted well behavior. Having the real time downhole pressure gauge data displayed in the CT cab was very beneficial in
helping control the drawdown being placed on the well. The quickest way to control bottom hole pressure is by adjusting the
returns choke. Changing the N2 rate was much slower. There was a delay of 10-20 minutes between changing N2 rate and
noticing a change in the bottom hole pressure.
Case Study #4: N2 Lifting the Post Acidizing Liquid
A horizontal gas well with TMD of 6255 m and TVD of 4306 m was drilled in the Canadian Turner Valley formation. The
well profile and the detail wellbore information are plotted in figures 22 and 23, respectively. The BHP was about 1885 psi
and the BHT was roughly 110 oC. The well is completed with a 3-1/2 production tubing to 4329 m, 3-1/2 slotted tubing to
4674 m and left open hole in the horizontal section to 6255m MD. The objective of this operation was to displace the acid in
the open hole section with a 2 tapered CT. After displacement of the treatment fluids, N2 was used to unload the well.
Job details of the acid stimulation and cleanup are shown in figure 24. This paper will focus only on the unloading
process which started 4260 minutes into the job. At this time the CT is at 5000 m and acid displacement is finished. Acid is
assumed to occupy the last 2000 m of the well giving it a fluid top of 4250 m. The unloading process was modeled using
field data for CT movement and nitrogen rates.
The unloading stage consisted of the following steps.
1. Set choke size to 25 in/64.
2. CT RIH from 3000 m to 4928 m at the rate of 30.48 m/min, N2 rate small at 1 scm/min.
3. At TD the simulation was given 260 minutes to achieve a steady state. Equilibrium between acid in the reservoir
and that in the wellbore was achieved, but no acid was returned to surface.
4. Nitrogen pumping was varied as presented in figure 25.
5. CT was moved to different location as seen in figure 25.
Figure 25 shows how pressures at the BHA (pressure at the circulation point) varied in time with CT depth and N2 pump
rate during the liquid unloading process. It indicates that the computed BHAPs are in fair agreement with the measured
BHAPs. The pressure spikes at the BHA due to the changes nitrogen rate and the CT depth are clearly captured by the
transient simulation. The dynamics of liquid going in and out of reservoir are shown in figure 26, as are the surface liquid
oscillations that result. In figure 26 it can be seen that the predicted and measured reservoir flows didnt match. The
productivity index, bottom hole pressure and temperature and the choke schedule were all estimated or unknown. By
making changes to key bottom hole parameters theres a possibility that post simulation matching could lead to valuable
insights about the current reservoir state.
In this case the job design was conducted in the conventional steady state manner without the benefit of a transient
simulation. Both the actual job data and simulated results indicate that liquid was squeezed back into the reservoir during the
unloading process. Obviously this was inefficient, future designs conducted with the transient software should help avoid or
at least mitigate this behavior.
Conclusion
A sophisticated transient software for the CT application has been developed and used to study the transient behavior of
liquid unloading process. The following conclusions are drawn:
1. The simulation can accurately predict the transient behavior of the liquid unloading process with the proper input
information. The interaction between the reservoir and the wellbore could significantly affect the simulation results of the
transient behavior.
2. For a given wellbore condition, there is a minimum gas velocity above which all the liquid can be removed from the
wellbore. This critical velocity is a function of liquid properties, wellbore deviation angle and the downhole pressure.
3. For a given RIH speed and CT size, the time to reach the start of stable well production could not be reduced once a
certain N2 rate was achieved, this is defined as the optimum N2 rate.
4. For a given size of CT, a higher RIH speed would result in less time to unload the well with the same N2 rate. Larger CT
diameters give higher liquid return rates when holding N2 rate and RIH speed constant.
5. The transient simulation can be used to optimize the unloading process and generate a guide for the field engineer to
execute the job. Surface equipment, CT size, N2 pump rate and a RIH/POOH procedure all have a significant affect on the
outcome.

SPE 124195

Acknowledgments
The authors would like to express their appreciation to BJ Services Company, for the opportunity to present this paper. We
wish to further thank colleagues: Simon Smith and Bill Gavin to provide the field operation information; Manfred Sach, Bill
Aitken and Lance Portman for their valuable input and the time spent to edit the manuscript. The contribution of Dr. Marco
Teixeira, our previous colleague from 2000 to 2006, on the initial development of transient multi-phase simulation is highly
appreciated. Last but not least, the transient simulations user-interface would not have been as user-friendly without the great
contribution of Ee Ker who has initiated and maintained the user-interface codes since 2000.
Nomenclature
BHA = bottom hole assembly
BHAP = pressure at the BHA or the circulation point
BHP = bottom hole pressure
BHT = bottom hole temperature
BHAT = temperature at the BHA or the circulation point
CT = coiled tubing
CTP = pressure at CT injection point
gpm = US gallon/min
ID = internal diameter
MD = measured depth
OD = outside diameter
pB = bottom hole pressure
pf = formation/reservoir pressure
PI = productivity index
POOH = pull out of hole
QG = produced gas flow rate
RIH = run in hole
ROP = rate of penetration
scfm = standard cubic feet per minute
TD = target depth
TMD = total measured depth
TVD = true vertical depth
Vcrit = critical gas velocity below which the liquid can not completely removed
Vsg = superficial gas velocity
WHP = wellhead pressure
WHT = wellhead temperature
ZNLF =zero net liquid flow
SI Metric Conversion Factors
bbl x159
E +00 = liter
gpm x 3.7854
E +00 = liter
ft x 0.3048
E +00 = m
inch x 25.4
E 03 = m
psi x 6.895
E +03 = Pa
References
Craig, S.H.: A Multi-Well Review of Coiled Tubing Force Matching, SPE 81715 presented at the SPE/ICoTA Coiled
Tubing Conference held in Houston, TX, USA, April 8-9, 2003.
Kalessidis, V.C., Rafferty, R., Merlo, A., and Maglione, R.: Simulator Models U-Tubing to Improve Primary Cementing,
Oil & Gas Journal (March 1994), 72-80.
Lage, A.C.V.M., K.K. Fjelde and R.W. Time: Underbalanced Drilling Dynamics: Two-Phase Flow Modeling and
Experiments, IADC/SPE 62743 presentation at the 2000 IADC/SPE Asia Pacific Drilling Technology held in Kuala
Lumpur, Malaysia 1113 September 2000.
Li, J., Walker, S., and Aitken, B.: How to Efficiently Remove Sand From Deviated Wellbores with a Solid Transport
Simulator and A Coiled Tubing Cleanout Tool, SPE 77527 presented at the 2002 SPE Annual Technical Conference
held in San Antonio, TX, USA, September 29-October 2, 2002.
Misselbrook, J., Wilde, G., and Falk, K.: The Development and Use of a Coiled-Tubing Simulation for Horizontal
Applications, SPE 22822 presented at the 66th Annual Technical Conference and Exhibition of the Society of Petroleum
Engineers held in Dallas, TX, USA, October 6-9, 1991.

SPE 124195

Nasr-El-Din, H.A., Al-Anazi, M.A., Balto, A.A., Proctor, R.J., and Saleh, R.M.: Challenging Wellbore Cleanouts with
Coiled-Tubing Made Easy with Computer Modeling Technology, SPE 100129 presented at the 2006 SPE/ICoTA Coiled
Tubing and Well Intervention Conference and Exhibition held in the Woodlands, TX, USA, April 4-5, 2006.
Ovesen, M., Sach, M., Laun, L., Gill, G.E., Juel, H.: Efficient Sand Cleanouts in Larger Wellbores Using Coiled Tubing: A
New Approach Making An Old Problem Simple, SPE 81727 presented at the SPE/ICoTA Coiled Tubing Conference
held in Houston, TX, USA, April 8-9, 2003.

SPE 124195

Liquid N2
tank
N2 line
Mass flow
meter

Pressure
Relieve valve

valve

valve
N2 pump
& heating unit

Check
valve
Test section

Lexan pipe

Gate valve

valve

Inner pipe
Pressure
Relieve valve

valve

Check
valve

H2O bypass
valve

N2 bypass
valve

vent
Mass flow
meter
Separator
tank

valve

Mass flow
regulator
valve
Water tank

1 m3
Water line
valve

Figure 1 Photo of full scale flow loop

Centrifuge
pump

Figure 2 Schematic of the flow loop system

1
0.9
0.8

Liquid holdup

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0.01

0.1

Vsg/Vcrit

Figure 3 Data compilation of liquid holdup under


the zero net liquid flow condition

Figure 4 Well configuration for the full-scale tests in case #1


(Courtesy of Lage, 2000)

1000

1900

Experimental Data (Lage, 2000)

Experimental Data (Lage, 2000)


1800

Simulation Results

Simulation Results

900

1700

800
Pressure, psi

Pressure, psi

1600

1500

700

1400

600
1300

500
1200

1100
15:21

15:28

15:36

15:43

15:50

15:57

16:04

Time

Figure 5 Pressure at 1262m for case #1

16:12

400
15:21

15:28

15:36

15:43

15:50

15:57

16:04

Time

Figure 6 Pressure at 605m for case #1

16:12

10

SPE 124195

300
Experimental Data (Lage, 2000)

Wellhead Gas Rate

2500

Reservoir Gas Rate

Pumped Gas Rate

Simulation Results
250

2000

Pressure, psi

200

1500
150

1000
100

500
50

0
0
15:21

15:28

15:36

15:43

15:50

15:57

16:04

16:12

Time

Figure 7 Pressure at 185m for case #1

Reservoir Liquid Rate

100

150
Time [min]

200

250

300

Figure 8 Simulated gas rates for case #2

Figure 9 Position and speed of the BHA for case #2

Wellhead Liquid Rate

50

Figure 10 Simulated pressures for case #2

400

Pumped Liquid Rate

140

3.0
350
Consumed N2 with ROP=10 ft/min

2.0

1.5

1.0

120
Consumed N2 with ROP=30 ft/min

300
Consumed N2 with ROP=70 ft/min

100

250
80
200
60
Time with ROP=10 ft/min

150

40

Time with ROP=30 ft/min

100

Consumed N2, Mscf

Time to reach the stable production, min

2.5

Time with ROP=70 ft/min

0.5

20

50

0.0

0
0

50

100

150
Time [min]

200

250

Figure 11 Simulated liquid rates for case #2

300

200

400

600

800

1000

1200

1400

1600

Nitrogen pump rate, scfm

Figure 12 Effect of N2 rate and ROP on the unloading time


and N2 consumption for 1.25 CT for case #2

SPE 124195

11

0.6

0.5

0.45
1.25"CT @ ROP=50 ft/min

ROP=10 ft/min

Average liquid return rate at the surface , bbl/min

Average liquid return rate at the surface , bbl/min

0.5
0.4

ROP=30 ft/min
ROP=70 ft/min

0.35

0.3

0.25

0.2

0.15

0.1

1.5"CT @ ROP=50 ft/min

1.75"CT @ ROP=50 ft/min


0.4

0.3

0.2

0.1
0.05

0
0

200

400

600

800

1000

1200

1400

1600

200

400

600

800

1000

1200

1400

1600

Nitrogen pump rate, scfm

Nitrogen pump rate, scfm

Figure 13 Effect of N2 rate and ROP on the average


unloading liquid rate with 1.25 CT for case #2

Figure 14 Effect of N2 rate and CT size on the average


unloading liquid rate at 50 ft/min of ROP for case #2

Time to reach the incipient of stable well production, min

600

500

400

300

200

100

0
0

500

1000

1500

Well head pressure, psi

Figure 15 Effect of wellhead pressure on the liquid


unloading time for case #2

Figure 16 Wellbore profile for case #3

RIH unloading well with N2

70

3000

Circulate N2 to generate desired drawdown, move CT frequently

4000

60

Deviation angle

2500

3500
CT depth, m

Vertical depth

40
1500
30
1000
20

WHP (psi), CTP (psi), BHAP (psi), BHP (psi), CT Depth (m

Vertical depth(m)

2000

Deviation angle (o), Completion ID (in)

Bottom hole pressure (BHP), psi

50

3000

2500
Pressure at BHA (BHAP), psi

2000

1500

Pressure at CT injection point (CTP), psi

1000
Pressure at wellhead (WHP), psi

500

10
Completion ID

0
0

500

1000

1500

2000

2500

3000

3500

0
4000

500

0
0

100

200

300

400

500

600

700

800

900

1000

1100

Elapsed Time, min

Measurement depth(m)

Figure 17 Wellbore information for case #3

Figure 18 Field operation condition for case #3

12

SPE 124195

4000

Measured BHP
Predicted BHP
Measured BHAP
Predicted BHAP
Measured WHP
Predicted WHP
Measured CTP
Predicted CTP

3500

Bottom hole pressure (BHP), psi

WHP (psi), CTP (psi), BHAP (psi), BHP (psi)

3000

2500

Pressure at circulation point (BHAP), psi


2000

1500

Pressure at CT injection point (CTP), psi


1000

Pressure at wellhead (WHP), psi


500

0
0

100

200

300

400

500

600

700

800

900

1000

1100

Elapsed Time, min

Elapsed time [min]

Figure 19 Simulation results for case #3

Figure 20 Comparison of simulation results and field data


for case #3

2500

Wellhead Liquid Rate

2000

Wellhead Gas Rate


6

5
1500

1000
3

Gas Rate [scf/min]

Liquid Rate [bbl/min]

Pumped Gas Rate

2
500
1

0
0

200

400

600

800

1000

1200

Time, min

5000

100

4500

90

4000

80

3500

70

Figure 22 Wellbore profile for case #4

RIH/POOH and acidizing

2500

50
Deviation angle

2000

40

1500

30

1000

20

500

10

Completion ID

0
0

1000

2000

3000

4000

5000

6000

0
7000

Measurement depth (m)

Figure 23 Wellbore information for case #4

WHP(psi), CTP(psi), BHAP(psi), CT depth(m)

60

600

6000

Deviation angle (o), Completion ID (in)

Vertical depth (m)

Vertical depth

3000

Unloading well with N2

7000

500
CT depth, m

5000
400

4000
Liquid rate, LPM
300
Pressure at the circulation point(BHAP), psi
3000
Pressure at CT injection point (CTP), psi
o

Temperature at the circulation point(BHAT), C

200

2000

100

1000

WHP, psi

N2 rate, m /min

0
1000

0
2000

3000

4000

5000

6000

7000

8000

Elapsed time, min

Figure 24 Field operation condition for case #4

9000

N2 rate(m3/min), Liquid rate(LPM),BHAT(oC)

Figure 21 Simulation results of fluids return at the surface


for case #3 without gas lifting during RIH period

SPE 124195

13

18

Field Data - BHAP


Field Data - CT Depth

40

Simulated CT Depth

5000

Field Data - N2 Pump Rate

35

30

25
3000
20

2000

15

10
1000

0
4000

4500

5000

5500

6000

6500

7000

7500

8000

8500

Time, min

Figure 25 Comparison of simulation results


and field data for case #4

N2 Pump Rate (scm/min)

Simulated N2 Pump Rate


4000

Simulated or measured liquid return rate at surface (gpm)

16

Simulated BHAP

BHAP (psi), CT Depth (m)

100
Simulated reservoir liquid rate

45

50
14

12
Simulated surface liquid rate
10

-50
Measured surface liquid rate

-100

4
-150

0
9000

-200

4000

4500

5000

5500

6000

6500

7000

7500

8000

8500

Time, min

Figure 26 Simulated liquid rates for case #4

9000

Simulated liquid rate at reservoir (gpm)

6000

Вам также может понравиться