You are on page 1of 11

SPE 152975

Using Dynamic Simulations to Optimize the Start-up Procedure of a Lazy


Horizontal Oil Well
Rahel Yusuf, SPE, Scandpower Petroleum Technology Sdn Bhd

Copyright 2012, Society of Petroleum Engineers


This paper was prepared for presentation at the SPE Oil and Gas India Conference and Exhibition held in Mumbai, India, 2830 March 2012.
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
During a well start-up, fluid distribution within the well shifts from a static segregated phase (gas-oil-water) column to a
flowing three-phase mixture. For this transition to be successful, the well has to overcome the additional hydrostatic head
associated with the phase segregation. This process can be problematic for wells with limited pressure support as an improper
bean-up procedure can lead to a situation where the pressure driving force across the well is unable to overcome the
hydrostatic head imposed by the liquid column during start-up thus resulting in a failed start-up. The failed start-up may add
extra liquid present within the wellbore thus making subsequent start-up attempts even more difficult. For such wells, correct
bean-up strategy is important to ensure successful start-up.
In the current work, dynamic simulations are conducted with a commercial transient-multiphase-simulator to understand the
start-up flow and pressure transients and optimize the start-up procedure of a lazy horizontal oil well which has experienced
start-up problems after a resting period.
Simulations of the failed bean-up procedure showed that a water-lock formation in the tubing hampers a successful well startup as the hydrostatic head imposed by the water-lock does not allow the well to flow. An alternate bean-up procedure was then
virtually tested by simulation where the wellhead choke was quickly opened. This led to a successful start-up as a gas pocket
originally trapped in the horizontal section could escape to the vertical tubing only under quick bean-up thus lowering the
hydrostatic pressure below the pressure driving force across the well. Dynamic simulations were also used to devise a strategy
for the operator as to how long the well be rested before any bean-up is attempted. The proposed strategy enabled the operator
to use the readings from downhole pressure gauge as a guide in deciding the resting period and bean-up procedure.
This study brings forth the significance of dynamic simulations in optimizing the start-up of lazy wells where failed bean-ups
can lead to higher down times thus adversely affecting production. Further, dynamic simulations also help in arriving at
guidelines for the operators on how to avoid failed start-ups.
Introduction
Well start-up is a transient phenomenon characterized by temporal variation of fluid distribution from a static segregated gasoil-water column to a flowing three mixture as shown in Figure 1. As a start-up is initiated by opening up the well, the gas
column escapes and the tubing is progressively filled with oil, water and the incoming reservoir fluid. Pressure transients
during this period can be significant as the static liquid column moves inside the wellbore, especially in horizontal wells where
the hydrostatic head can increase and reach a maximum as the liquid in the horizontal part moves into the vertical section of
the wellbore (Figure 2). For a start-up to be successful, it is thus important that the pressure driving force (Reservoir pressure
minus the outlet boundary pressure) across the well is able to overcome the maximum hydrostatic head encountered during
start-up. This issue can become critical for start-up of wells having limited pressure support from the reservoir (lazy wells) as
an improper bean-up procedure can lead to a situation where the pressure driving force across the well can be slightly below
the maximum hydrostatic head imposed by the liquid column during start-up thus resulting in a failed start-up. The failed startup may result in increased liquid content within the wellbore thus making subsequent start-up attempts even more prone to
failure. This will result in reduced well uptime thus adversely affecting production. For such wells, correct bean-up strategy is
important to ensure successful start-up (Calvert and Davis[1]).
Keeping in view the risks of failure associated with trying different bean-up procedures in the field, virtually testing the
different bean-up procedures through simulations is an attractive option that is much cheaper and risk free. The transient nature

SPE-152975

of well start-up virtually rules out using a steady-state simulation approach like Nodal analysis and a transient multiphase
modeling approach is thus required to simulate well start-up. In the present work, a comprehensive transient multiphase
simulator OLGA is used to simulate well shut-in and start-up in order to optimize the start-up of a lazy horizontal oil well
which fails to flow after a resting period.

[2]

Figure 1 Fluid distribution in a well before and after a successful start-up (Adapted from Mantecon

et al., 2004)

Figure 2 Hypothetic temporal variation of hydrostatic pressure in a horizontal oil well during start-up

Transient multiphase flow modelling


The multiphase flow model incorporated in OLGA is based on first principles whereby the equations of mass, momentum and
energy are solved in time and space in order to arrive at temporal and spatial variation of phase fractions, velocities, pressure
and temperature. The model is closed by appropriate closure laws defining interfacial and wall friction, deposition and
entrainment of droplets and bubble entrainment. Finer details of the model are avaialable in Bendiksen[3] et al. The model is
suited to well simulation applications as it can simulate complex trajectories, smart completions, gas lift valves and the
transient heat transfer in wells by accounting for all the applicable modes of heat transfer (conduction, convection and
radiation) between the tubing, casing and formation.
System Modelling
The well chosen in this study is a drytree horizontal oil producer. The trajectory of the well is shown in Figure 3 where the
depths are with reference to rig table (RT). RT elevation above mean sea level (MSL) and sea bed depth are 158 ft and 291 ft
respectively. The wellhead is located at 80 ft MDRT while the true depth (TD) of the well is 16723 ft MDRT. The well
produces through perforations in eight zones in the horizontal completion. The flow from the perforated zones comes into the
production tubing through SSDs which remain fully open. The true reservoir pressure is 3900 psi at 7400 ft TVD and the

SPE-152975

reservoir temperature is 90.5 oC. However, the the long term reservoir support to the well is estimated to be limited to 3300
psi. The reason for this phenomenon is rooted in the low effective permeability of the well which is close to 0.1mD
A schematic of the modeled system is shown in Figure 4. The modeled system includes the well and a chokeline between
the wellhead and the separator. The size and length of the chokeline are 5 inch and 30 ft respectively. The key inputs required
for model building in OLGA are discussed in the following sections.

Figure 3 Trajectory of candidate well

Figure 4 Schematic of the modeled system

Geometry
The system geometry input in OLGA requires elevation, length, I.D. and roughness of the pipes. The elevation and length
are calculated from the true vertical depths (TVDs) and measured depths (MDs) from Figure 3. The production tubing I.D. is
3.958 inch in the top completion and 2.922 inch in the eight zones in the horizontal completion. A tubing roughness of 5.5e-05
m is assumed in this work.
Heat Transfer
Heat transfer to the ambient environment is modeled in detail rather than assuming a constant overall heat transfer
coefficient. Depending on the system component, heat losses to the ambient can occur through the tubing material, annular
fluid, casing material and the formation which are modeled as series of resistances to the heat flow in the radial direction. The
ambient temperature varies from 90.5 oC at the reservoir to 4 oC at the seabed and 10 oC at the sea level. A vertical

SPE-152975

interpolation is used to obtain the ambient temperature at any given section between the bottomhole and sea level. A constant
ambient temperature of 10 oC is assumed above the sea level.
Inflow Source
The inflow from the perforated zones is modeled by a linear relationship as

q = PI * ( Pres Pwf

Where q is the volumetric liquid flow rate, PI is the productivity index, Pwf is the flowing bottomhole pressure, and Pres is the
reservoir pressure. The total liquid phase (oil plus water) PI of the well is 1.6 Stb/d/psi as indicated from the welltest data. It is
assumed that all the zones contribure equally and the PI for each perforation in the model is hence 0.2 Stb/d/psi.
Outflow Boundary and Topside Choke
The outflow boundary is modeled as a pressure outlet representing the backpressure at the rig. The topside choke is placed
at the end of the chokeline. The variation of choke flow coefficient (CV) with equivalent bean size (1/64) is summarized in
Table 1. The table is incorporated in OLGA in terms of the variation of CV with choke opening. The choke opening represents
the fraction of total area available to flow. The area is calculated from the equivalent bean size and the opening is obtained by
dividing the area at a given bean size by the area corresponding to the maximum bean size (53/64). The calculated opening is
also summarized in Table 1.
Table 1 Variation of choke CV with beansize and calculated opening
Equivalent
Opening
CV value
beansize
0
0
0
1
17
0.10
2
20
0.14
3
23
0.19
4
28
0.28
5
30
0.32
6
33
0.39
7
37
0.49
8
39
0.54
9
42
0.63
10
44
0.69
11
46
0.75
12
47
0.79
13
49
0.85
14
51
0.92
15
53
1
Reservoir Fluid Composition
The fluid composition as summarized in Table 2 is used to generate three phase (gas/oil/water) fluid property tables in
PVTsim with a watercut of 32%.
Table 2 Reservoir fluid composition upto C7+ Components
Mol wt
Liquid Density
Component
Mol %
g/gmol
g/cm
N2
0.140
28.01
CO2
0.67
44.01
C1
53.18
16.04
C2
7.32
30.07
C3
5.33
44.09
iC4
0.86
58.12
nC4
2.92
58.12
iC5
1.10
72.15
nC5
1.83
72.15
C6
3.00
86.17
0.6640
C7+
23.65
212.40
0.8640

SPE-152975

Simulation Scenarios
The following scenarios are investigated

Match the predicted production of gas, oil and water against the measured values for three welltest data
Run shut-in simulations to match the predicted bottomhole pressure against the measured values
Run start-up simulation according to the chosen bean-up procedure and come up with a different start-up procedure if
the well is difficult to start-up according to the chosen start-up procedure

Results and Discussion


This section presents and discusses the results obtained from the simulation scenarios outlined in the previous section. It
should be noted that unless otherwise specified, the reservoir pressure is set to 3300 psi in all the aforementioned scenarios. As
indicated earlier, the value of 3300 psi is the estimated long term pressure support from the reservoir.
History matching the production during normal operations
In this scenario, three different cases are chosen for history matching the predicted production against measured data.
Three different welltest cases are chosen and summarized in Table 3. All the values indicated in the table are averaged over
three days.
Table 3 Summary of welltest cases chosen for normal production history matching
Case #
Chokesize
GOR
WC
backpressure
[1/64]
[Scf/Stb]
[-]
[psi]
1
15.3
1478
0.326
522
2
24
1478
0.326
537
3
15.3
1415
0.320
522
The three cases are simulated using the GOR, WC and backpressure shown in Table 3 and the chokesize is tuned to obtain
a match (within 10%) between the predicted and measured flowrates and bottomhole pressure. The bottomhole gauge is
located at 7610 ft MDRT. A comparison of the predicted and measured values is shown in Table 4.

% error

Predicted

Measured

% error*

BHP
[psig]

Measured

1
1073
1082
-0.8
726
732
-0.8
2
1046
1082
-3.3
708
732
-3.3
3
1148
1172
-2.0
809
828
-2.3
*error is calculated for pressure drop (BHP-backpressure)

Water rate
Stb/d]

Predicted

% error

Measured

Predicted

% error

Measured

Predicted

Table 4 Summary of normal production history matching


Case #
Gas rate
Oil rate
[Mscf/d]
[Stb/d]

351
342
381

354
354
389

-0.9
-3.4
-2

2290
2306
2218

2179
2286
2182

6.6
1.1
2.1

Shut-in simulations
Simulations are conducted for Case # 2 to match the predicted bottomhole pressure against the measured value for a three
day shut-in period. The predicted shut-in pressure at the bottomhole gauge position is 2816 psig which compares very well
with the measured value of 2902 psig.
Shut-in simulations provide detailed insight into fluid segregation during the shut-in period. Details of fluid segregation are
useful to understand the difficulties that can be encountered during start-up. Figure 5 shows the profile of gas, oil and water
volume fraction along the tubing after three days of shut-in. The tubing geometry is also plotted in Figure 5 in terms of
elevation vs. length. The length axis begins at the wellhead and ends at the first bullplug. It can be seen that the gas collects at
the top of the inclined section due to its lower density. The gas is followed by oil and water in the inclined section of the well.
It is worth noting that due to a high point in the horizontal section (between 10000 and 14000 ft), another fluid segregation
occurs in the horizontal section where a gas pocket is trapped between two oil layers

SPE-152975

Figure 5 Fluid segregation along the tubing after 3 days of shut-in

Start-up simulations
Start-up simulations are run as restart simulations from the shut-in scenario. The choke bean-up procedure during well
start-up is summarized in Table 5.
Figure 6 plots the temporal variation of oil, water and gas flowrates during the first 15 hours of start-up after shut-in. It is
seen from Figure 6 that the gas trapped at the top during the shut-in escapes first. The gas is followed by oil and finally a big
slug of water arrives at the surface at around 2.5 hours after start-up. The flowrates slowly stabilize thereafter and a steady
state is attained 10 hours after start-up.
Table 5 Bean-up sequence during well start-up
Start-up Scenario
Choke bean-up procedure
Start-up after 3 days of shut-in

14/64 for first 60 min


20/64 next 30 min
24/64 thereafter

Contrary to field observations of difficulties in well start-up, simulations show that the well restarts after shut-in. The
GOR, flowrates, watercut etc. are known with a degree of certainty but the there could be uncertainty in the reservoir pressure
at start-up. It could be possible that due to limited pressure support from the reservoir, the reservoir pressure after the shut-in is
lower than 3300 psi, as originally assumed.
Start-up simulations are then conducted for different reservoir pressures to observe the effect of reduction in reservoir
pressure on start-up. The GOR, watercut, backpressure and bean-up procedure are kept same as for the original start-up
scenario. Trial simulations with lower reservoir pressure reveal that the well flows after start-up up to a reservoir pressure of
3050 psi. However, when the reservoir pressure is lowered to 3000 psi, the well doesnt flow on start-up as seen in Figure 7
where the water flowrate at surface is zero.

SPE-152975

Figure 6 Temporal variation of standard volumetric flowrates for oil, water and gas phase during the first 15 hours of start-up after
shut-in (Pres= 3300 psi)

Figure 7 Temporal variation of standard volumetric flowrates of oil and water during the first 10 hours of start-up after shut-in (Pres=
3000 psi)

In order to ascertain the reason for the failure of the well to restart at 3000 psi reservoir pressure, profiles of phase volume
fractions are plotted at different time instants during the start-up (Figure 8). It is seen from Figure 8 that the gas at the top of
the tubing escapes initially and some oil also follows suit. However, the gas pocket in the horizontal part of the well is unable
to escape and it nearly stagnates after four hours. A water lock is eventually created (see plot for t=3 days in Figure 8) whereby
the water from the horizontal part moves into the inclined part of the well.
In order to understand the role of water lock in failed well start-up, Figure 9 plots the temporal variation of hydrostatic
pressure inside the tubing during the unsuccessful start-up. It can be seen from Figure 9 that the hydrostatic pressure imposed
by the water lock exceeds the available pressure driving force across the well (Reservoir pressure minus the rig back pressure)
thus not allowing the well to flow.

Figure 8 Profile plots of phase volume fractions at different time instants during the unsuccessful start-up (Pres= 3000 psi)

Figure 9 Temporal variation of hydrostatic head inside the tubing during the unsuccessful start-up

SPE-152975

SPE-152975

It is now seen that the well is unable to restart at a reservoir pressure of 3000 psi if the bean-up procedure outlined in Table
5 is adopted as the maximum hydrostatic pressure encountered during the well start-up slightly exceeds the available pressure
driving force. Calvert and Davis[1] have suggested that in such borderline scenarios, there exists an optimum bean-up
procedure that can minimize the hydrostatic head to be overcome to attain a successful start-up. A new bean-up procedure is
virtually tested through simulations in order to achieve a successful start-up. In the new bean-up procedure, the choke is fully
opened over the first two minutes after start-up. The choke then remains fully open for the next seven hours and finally beaned
back to 24/64 after seven hours.
Figure 10 plots the temporal variation of oil, water and gas flowrates during the first 30 hours after start-up with the new
bean-up procedure. Initially some slugging is observed but the well flows stably after 20 hours.
The profile of phase volume fractions at different time instants is plotted in Figure 11. It is seen that due to the fully open
choke, the gas pocket in the horizontal section of the well is able to escape thus paving the way for the successful start-up of
the well. Figure 12 plots the temporal variation of hydrostatic pressure inside the tubing during the successful start-up which
clearly indicates that the faster bean-up minimizes the hydrostatic head to a value below the available pressure driving force
thus leading to a successful well start-up.

Figure 10 Temporal variation of standard volumetric flowrates for gas, oil and water phase during the first 30 hours of start-up with
the new bean-up procedure (Pres= 3000 psi)

It should be noted that the proposed new bean-up procedure too has its limits and trial simulations showed that the well
fails to flow at a reservoir pressure of 2900 psi even with the new bean-up procedure.
From an operators viewpoint, it is important to know how the limits of the reservoir pressure corresponding to different
bean-up procedures correlate with the shut-in bottomhole gauge pressure. Figure 13 plots the bottomhole gauge pressure (at
7610 ft MDRT) against the bottomhole pressure near the perforation (which is close to the reservoir pressure at shut-in). Stage
1 in the plot shows the bottomhole gauge pressure which corresponds to 2900 psi reservoir pressure when both the original and
the proposed bean-up procedures fail. The well should be shut-in till bottomhole gauge pressure reaches a value that
corresponds with 3000 psi reservoir pressure as depicted by Stage 2 in the plot. At this juncture, the operator can first try the
original bean-up procedure and if that fails, the new bean-up procedure can be tried. If the well is shut-in for long enough so
that the bottomhole gauge pressure reaches Stage 3 in the plot, the well can be safely restarted by using the original bean-up
procedure as the reservoir pressure is now 3300 psi.

10

Figure 11 Profile of phase volume fractions at different time instants during the successful start-up (Pres= 3000 psi)

Figure 12 Temporal variation of hydrostatic head inside the tubing during the successful start-up

SPE-152975

SPE-152975

11

2850

2800

Bottomholegaugepressure,[psig]

2750

2700

2650

2600

2550

2500

2450

2400
2800

2850

2900

2950

3000

3050

3100

3150

3200

3250

3300

Bottomholepressure,[psig]

Figure 13 Limits of different bean-up procedures with respect to bottomhole gauge pressure available to the operator

Conclusions
Transient simulations of a lazy, horizontal oil well are conducted with OLGA. The simulations cover normal production, shutin and start-up scenarios. The predicted flowrates and bottomhole pressure during normal production compare favourably
(within 10%) with the measured values. Good match is obtained between the predicted and the measured bottomhole gauge
pressure for the shut-in scenario. Start-up simulations after the shut-in scenario indicate that the well is able to flow with the
bean-up procedure practiced in the field. The successful start-up is in contrast to field observations where difficulties are
encountered in restarting the well. Owing to uncertainty in reservoir pressure, the assumed reservoir pressure of 3300 psi in the
present study could be higher than the actual value. The well is unable to start-up when the reservoir pressure is lowered to
3000 psi. The hydrostatic head imposed by a water lock formation is seen as the possible cause of unsuccessful start-up at the
lower reservoir pressure. An alternative bean-up procedure, where the choke is fully opened quickly over a period of two
minutes, helps to successfully restart the well as the maximum hydrostatic head experienced in this bean-up procedure is
below the available pressure driving force.
Nomenclature
GOR
= gas oil ratio
IPR
= inflow performance relationship
MD
= measured depth
PI
= productivity index
= reservoir pressure
Pres
= flowing bottomhole pressure
Pwf
RT
= rig table
SSD
= sliding sleeve device
TD
= true depth
TVD
= true vertical depth
WC
= watercut
References
1.

2.
3.

Calvert, P. and Davis, J.: A Dynamic Business Needs Dynamic Solutions; How Field of the Future Has Turned BP into a Smooth
Operator, SPE paper 128682 presented at the SPE Intelligent Energy Conference and Exhibition, Utrecht, The Netherlands, 23-25
March 2010
Mantecon, J.C., Andersen, I. and Freeman, D.: Impact of Dynamic Simulation on Establishing Watercut Limits for Well Kickoff, Presented at the SPE Asia Pacific Oil and Gas Conference and Exhibition, Perth, Australia, 1820 October 2004
Bendiksen, K.H., Malnes, D., Moe, R. and Nuland, S.: The Dynamic Two-Fluid Model OLGA: Theory and Application, SPEPE,
171, May 1991.

You might also like