Académique Documents
Professionnel Documents
Culture Documents
subsea tieback
S R Davies, T Lockett, R G Chapman, G Cristofoli, O Smith
BP Exploration Operating Company Ltd., UK
There was a desire to optimise the minimum acceptable flow rate for an existing HPHT
gas-condensate tieback and to reduce the volume of hydrate inhibitor that is used during
start-ups or restarts. Transient multiphase flow simulations were benchmarked against
current operating data and were then re-run for the proposed future operating state to
demonstrate that the system could be maintained outside the hydrate region for steady
state and shut-in conditions. Compositional tracking simulations were performed to
optimise the required inhibitor injection requirement (rate and duration) for restarts.
1
INTRODUCTION
133
The tieback consists of three HPHT wells that are connected to a subsea manifold via
three well flowlines, each 6.5 inches internal diameter and between 2 and 3 km in length.
The manifold is connected to the processing platform via a 14-inch internal diameter
flowline that is 45 km in length. The separator on the processing platform was
represented by a constant pressure boundary condition. All of the flowlines use pipe-inpipe insulation to conserve heat during steady-state operation and for short duration shutins; the thermal behaviour of the flowlines was represented using a series of wall layers
which followed the flowline design. A wall roughness of 110-5 m was also taken from
the original design. Short bare steel pipes known as ambient spools connect the wellhead
choke valves to the well flowlines; the spools are between 50 and 90 meters in length.
The ambient spools have a high heat transfer coefficient to ensure that the fluid from the
wellhead warms up above the minimum acceptable operating temperature during restart.
The length of the ambient spool is critical: too short and the flowline temperature would
drop below the minimum design temperature during restart, too long and the excessive
cooling would negate the intention of the pipe-in-pipe to keep the fluid outside of the
hydrate region during steady state flow. To regulate the flow, each wellhead is fitted with
a choke valve with a flow coefficient (Cv) of 241. The local water depth is approximately
120 meters and the reservoir has a temperature of approximately 150C, a pressure of
570 barg and is located at a total vertical depth of approximately 4.6 km below the
seabed. The ambient temperature of the subsurface was assumed to vary linearly with
elevation between the reservoir and the seabed. The ambient temperature of the sea was
set to 8C based on the minimum recorded seasonal temperature over 5 years of operation
and the topsides were assumed to be at the minimum expected ambient temperature of 5C.
The well performance was modelled using the reservoir conditions and then applying
simple Productivity Indices (PIs). PIs are rarely use for compressible fluids; this type of
model is rarely used for gas systems but was applied due to the limited available data.
There are also issues associated with applying PI models to transient scenarios as
discussed in Section 4. The Productivity Indices (PIs) for the well models were adjusted
to provide the currently observed steady state gas flow rates. The predicted temperatures
and pressures for two of the wells upstream and downstream of the choke valve are
shown in Table 1 along with the measured data. The agreement between the models and
field data was excellent for these conditions. Unfortunately no measurements of the
bottom-hole pressure and temperature were available for comparison. The arrival
temperature at the host platform was predicted to be 50.1C at a pressure of 33 barg
which was also in agreement with field measurements.
Table 1: A Comparison of the Steady State Model Predictions
to Field Data for Steady State Operation
Reported
3.1
Well 1
Predicted
Error
Reported
3.4
Well 2
Predicted
Error
98
104.0
6%
142
144.9
2%
33
33.2
1%
66
63.3
4%
Upstream T (C)
102
98.9
3%
102
104.0
2%
Downstream T (C)
90
87.5
3%
101
95.0
6%
Upstream P (barg)
111
122.0
10%
133
140.9
6%
Downstream P (barg)
107
118.0
10%
127
129.8
2%
134
512
Cubic
sI
2/cell
6/cell
512
51264
Cubic
sII
16/cell
435663
51268
Hexagonal
+
3/cell
136 H2O
8/cell
512
sH
46 H2O
2/cell
34 H2O
1/cell
135
500
500
450
30 mol% in water
30 mol% in water
40 mol% in water
40 mol% in water
Prediction - 5.6 mol%
400
400
Prediction - 30 mol%
350
Prediction - 40 mol%
300
Pressure (bara)
Pressure (bara)
350
250
200
Prediction - 40 mol%
300
250
200
150
150
100
100
50
50
0
-30
-20
-10
10
20
-30
-20
Temperature ( C)
-10
0
Temperature (oC)
10
20
45
45
40
40
35
35
30
30
Pressure (bara)
Pressure (bara)
Figure 2: A Comparison of sI Hydrate Equilibrium Data for Methane and Water [2]
against Predictions from the First Model (left) and the Second Model (right)
25
20
0 wt% in water
20 wt% in water
40 wt% in water
60 wt% in water
80 wt% in water
Prediction 0
Prediction 20
Prediction 40
Prediction 60
Prediction 80
15
10
25
20
0 wt% in water
20 wt% in water
40 wt% in water
60 wt% in water
80 wt% in water
Prediction 0
Prediction 20
Prediction 40
Prediction 60
Prediction 80
15
10
0
-30
-20
-10
10
20
-30
-20
Temperature (oC)
-10
10
Temperature (oC)
136
20
RESULTS
80
Pressure (barg)
70
60
Ambient
Seawater
50
40
33.2
33.1
33.6
30
20
12.1
12.2
8.1
8.2
12.6
10
0
-40
-30
-20
-10
10
8.6
20
30
40
50
60
Temperature ( C)
Figure 4: Comparison of the Steady State Arrival Conditions to the Hydrate Phase
Boundary; Numbers show Arrival Pressure (barg) and Gas Flow Rate (MSm3/d)
Table 2: Predicted Cool-Down Times for the Pipe-in-Pipe for Unplanned (ESD)
Shut-ins
Arrival Pressure (barg)
Gas Flow Rate (MSm3/d)
Cool Down Time (days)
8
1
8
2
8
6
5
12
1
12
2
12
6
4.9
33
1
2.8
33
2
7.6
33
6
4.5
137
Pressure (bara)
70
60
50
40
30
Downstream of Choke
Start
End
No Inhibitor
1 kg IMS per kg Water
2 kg IMS per kg Water
3 kg IMS per kg Water
5 kg IMS per kg Water
7 kg IMS per kg Water
9 kg IMS per kg Water
11 kg IMS per kg Water
13 kg IMS per kg Water
15 kg IMS per kg Water
20
10
0
-100
-80
-60
-40
-20
20
40
60
Temperature (oC)
138
strategy, the position of the choke valve was matched to the measured opening position
from the field data using the Cv curve of the choke valve. The trim of the choke valve
had recently been replaced so this curve was considered to be reliable.
The simulation results from Strategy 1 are compared to the field data in Figures 6, 7 and
8 for the gas flow rate, the temperature and the pressure upstream and downstream of the
choke valve. The initial well restart occurred at 31.25 hours. Operational issues led to
large deviations in the flow rate, pressure and temperature from 55 hours onwards (24
hours after starting flow).
3.5
Data
Acquisition
Error
Field Data
OLGA Prediction
2.5
2
1.5
1
0.5
0
30
35
40
45
50
55
Time (hours)
60
65
70
75
Temperature ( C)
100
80
60
40
20
0
-20
-40
-60
30
35
40
45
50
55
60
65
70
75
Time (hours)
Figure 7: Comparison of Model Predictions from Strategy 1 to Field Data for the
Temperature Upstream and Downstream of the Choke Valve during a Recent
Well 2 Restart
139
600
Pressure (barg)
500
400
300
200
100
0
30
35
40
45
50
55
Time (hours)
60
65
70
75
Figure 8: Comparison of the Model Predictions from Strategy 1 to Field Data for
Pressure Upstream and Downstream of the Choke Valve during Recent Restart
of Well 2
A good agreement was obtained between the predictions and the observations: the
pressure and temperatures agreed well, but the flow rate was slightly higher than the
model prediction. It is believed that this discrepancy is due to the treatment of the well
Productivity Index (PI) within the simulator as discussed in Section 2; the PI relates the
flow rate from the well to the pressure difference between the bottom of the wellbore and
the reservoir. At steady-state operation where the PIs are valid, a significant pressure
drop occurs within the formation itself. For restart operations however, the pressure drop
within the formation is initially much lower so the well can provide a higher flow rate
than would be predicted by a PI model.
For Strategy 2, the predicted gas flow rate is shown in Figure 9. The simulation
significantly under predicted the gas flow rate. The Cv curve for the choke valve is
considered valid since the trim had only recently been replaced; this discrepancy might
therefore indicate an error in the recorded stem position of the valve or could indicate a
problem with the choke model in OLGA.
The fluctuations in pressure, temperature and flow rate from 55 hours onwards (24 hours
after starting flow) were not part of the typical restart procedure and are not considered
further. The well restart procedure stated that the choke should initially be set to an
opening corresponding to a gas flow rate of 1.53 MSm3/d. In practice the gas flow rate
took several hours to reach this value; prolonging the time that the downstream
temperature was below the hydrate equilibrium line. The procedure required an initial
IMS dose rate of 5.2 m3/hr that was to be reduced to 1.2 m3/hr after the well had been
flowing for 1 hour, or when the wellhead temperature reached 20C. IMS injection was to
continue at the reduced rate for a further 11 hours. The actual IMS injection rate is
compared to the IMS injection rate from the procedure in Figure 10. Time values less
than 31.25 hours are before the well was restarted. IMS injection during this time
corresponds to pre-dosing. In the actual restart, several instances of IMS injection were
recorded prior to the well being opened, possibly due to operational issues on the
receiving platform. The flow rate was slightly above the prescribed IMS flow rate of 5.2
m3/hr when the well was opened and the rate was decreased to 1.2 m3/hr when the well
140
head temperature reached 20C, as per procedure. The IMS injection rate then dropped
off almost linearly over time and fell significantly below the prescribed flow rate of 1.2
m3/hr.
3.5
Field Data
OLGA Predictions
2.5
2
1.5
1
0.5
0
30
35
40
45
50
55
Time (hours)
60
65
70
75
120
Recorded IMS Flowrate
80
Temperature ( C)
100
Wellhead Temperature
60
40
2
20
1
0
0
0
10
20
30
40
50
Time (hrs)
Figure 10: Comparison of the Measured IMS Rate for the Restart of Well 2 to that
Specified in the Procedure; the Measured Well head Temperature is Also Shown
Simulations predicted the water flow rate and the IMS concentration downstream of the
choke valve during the restart as shown in Figure 11. The model predicted that the IMS
that had been injected into the line prior to the restart was produced almost immediately
as a slug of IMS through the choke valve. The IMS flow rate had been reduced 45
minutes after the restart. The water flow rate through the choke valve increased at
approximately 2 hours after the restart and this rapidly diluted the remaining IMS
concentration to 25 wt%. The IMS concentration continued to fall, reaching 5 wt% at 40
141
hours. It was of interest to determine if sufficient IMS was present to inhibit hydrate
formation during this period. The hydrate equilibrium temperature is a function of both
pressure and the concentration of IMS in the water phase. Pressure-temperature curves
were produced using the Multiflash CPA model for a range of IMS concentrations and
the hydrate equilibrium temperature was found by interpolating between the adjacent
pressures and compositions at each time step and location.
35
IMS Slug
30
25
0.7
0.6
20
0.5
0.4
15
0.3
10
0.2
Water
Breakthrough
0.1
0.8
0.9
0
30
32
34
36
38
40
Time (hrs)
Figure 11: Predicted Water Flow Rate and IMS Concentration as a Function of
Time Downstream of the Choke Valve for the Model Simulation of the Restart
The fluids temperature downstream of the choke valve is compared to the local hydrate
equilibrium temperature in Figure 12. The system was predicted to enter hydrate region
for approximately 3 hours, at a simulation time of 33 hours (1.75 hrs after starting flow
from the well). This coincided with an increase in the water flow rate from the well. The
100
80
Startup
Water Breakthrough
40
Temperature ( C)
60
20
0
-20
-40
-60
30
32
34
36
38
Time (hrs)
40
42
44
142
46
IMS flow rate had already been reduced at approximately 32 hours once the wellhead
temperature reached 20C as per the procedure. An alternative procedure might be to
flow the IMS until the temperature downstream of the choke valve increases above the
hydrate temperature at the flowline pressure during restart (50 barg) of 12C.
4.3
Optimisation of the restart procedure
It is of interest to try to minimise the amount of IMS required for a successful restart
while still protecting the flowline from hydrate formation. Several potential restart
procedures were considered. For simplicity only the last two variations are described in
this paper and are referred to as Option 1 and Option 2. The hydrate temperature at the
riser base is 8.7C at the desired operating pressure of the inlet separator of 33 bar. It is
necessary to protect this section of the flowline from hydrates until the cold fluid from
the restart has left the flowline and the temperature at the receiving platform increases
above 8.7C. Simulations suggested that it could take approximately 35 hours from
initiating the flow before the cold fluid would leave the flowline. It was found to be
necessary to dose the system with a small volume of IMS for 24 hours after the initial
restart in order to avoid the hydrate formation region during this period.
Option 1 of the restart procedure for Well 2 for a separator pressure of 33 bar is as
follows: set the separator pressure to 12 bar; inject IMS at the wellhead at 5.2 m3/h; start
the well at 1.53 MSm3/d; continue dosing the IMS until the temperature downstream of
the choke exceeds 12C; continue injecting IMS at a rate of 0.15 m3/h for 24 hours; keep
the separator pressure at 10 bar until the temperature downstream of the choke exceeds
50C; pack the line to 33 bar; increase the flow rate from the well once the line has been
flowing with IMS injection for 24 hours. The results simulating Option 1 are shown in
Figures 13 and 14. The predicted fluid temperature downstream of the choke is compared
to the hydrate equilibrium temperature in Figure 13. The fluid temperature at the riser
base is compared to the hydrate equilibrium temperature in Figure 14. Option 1 of the
restart procedure avoided the hydrate region at both locations for the duration of the
restart with a desired arrival pressure of 33 bar.
70
30
Temperature ( C)
50
10
-10
Fluid Temperature
-30
Hydrate Equilibrium
Temperature
-50
0
0.5
1.5
2.5
3.5
4.5
Time (hrs)
143
20
15
IMS Stops
Arriving
Temperature ( C)
10
5
0
-5
Fluid Temperature
Hydrate Equilibrium Temperature
Line Packing
-10
10
15
20
25
30
35
40
45
Time (hrs)
Temperature ( C)
10
5
0
-5
Fluid Temperature
-10
Hydrate Temperature
-15
-20
0
10
20
30
40
50
Figure 15: Comparison of the Fluid Temperature in the Flowline to the Hydrate
Equilibrium Temperature 3 Hours after the Start of the First Well with Option 1
144
In order to prevent the system entering the hydrate region, it was necessary to dose the
flowline with a reduced IMS flow rate of 1.2 m3/h for a further 8 hours after the initial
5.2 m3/h. If the separator pressure is to be increased to 33 bar then IMS would also be
required at a rate of 0.15 m3/h in order to inhibit the riser base for a further 16 hours as
before. This second revision to the restart procedures is termed Option 2 and the
corresponding plot of fluid temperature versus the hydrate temperature is shown in
Figure 16.
20
15
Temperature ( C)
10
5
0
-5
Fluid Temperature
-10
Hydrate Temperature
-15
-20
0
10
20
30
40
50
Distance (km)
Figure 16: A Profile Plot Comparing the Fluid Temperature in the Flowline to the
Hydrate Equilibrium Temperature 3 Hours after the Start of the First Well with
Option 2 of the IMS Injection Strategy
In the previous simulations, the pressure in the inlet separator was increased from 10 to
33 bar 7 hours after the well had been started. It was of interest to see if this criterion
could be reduced since the gas would have to be flared at arrival pressures below 33 bar.
Figure 17 shows the predicted fluid temperature and pressure profiles in the flowline as
the line is packed between 7 hours and 8 hours into the simulation. The pressure and
temperature path is compared to the hydrate region. As the line is packed, both the
pressure and temperature of the fluid increase and the inhibition requirements of the fluid
are similar at each stage of the line packing process. This suggests that packing the line
should not incur any additional risk of hydrate formation.
145
50
45
Pressure (bara)
40
35
30
25
20
15
7 hrs
10
8 hrs
9 hrs
0
-20
-10
10
20
30
40
50
Temperature (oC)
Figure 17: A Profile Plot Comparing the Fluid Temperature and Pressure in the
Flowline during the Line Packing Process to the Hydrate Equilibrium
Temperature at Various Times
The profile plots of the pipeline were compared to the local hydrate equilibrium
temperatures at various times into the simulation and it was confirmed that the flowline
would not enter the hydrate formation region using the revised IMS injection strategy,
even if the flowline was pressurised 1 hour after the restart. Figure 18 confirms that the
fluid temperature at the riser base is outside the hydrate formation region if the line is
packed after only 1 hour instead of 7 hours.
20
Fluid Temperature
Temperature ( C)
15
Hydrate Temperature
10
-5
0
10
20
30
40
50
60
70
Time (hrs)
146
80
CONCLUSIONS
This paper has investigated the IMS requirements for hydrate mitigation during the
restart of the wells on a HPHT tieback to a receiving platform. It was of interest to try to
minimise the amount of IMS required for a successful restart in order to reduce the
problems caused for the downstream process. It was suspected that the volume of IMS
that was specified in the original procedure could be reduced if the lower shut-in well
head pressure and the warmer sea bed temperature were considered.
Recent experimental data was used to validate the available predictive tools. One model
was found to best predict structure I hydrate equilibrium data and a different model was
found to best predict structure II hydrate equilibrium data.
In the original restart procedures, the IMS injection requirements had been based on the
well head temperature measurements and the time after restart. Recent field data
indicated that the well had been started-up slower than was expected in the operating
procedure. The model predicted that this slower restart extended the duration for which
the system was subcooled and caused the piping downstream of the choke valve to enter
the hydrate region some time after the IMS injection rate had been reduced by the
operators. Fortunately no hydrate plug had formed during the restart.
Subsequent simulations allowed the IMS requirements to be optimised; larger quantities
of IMS were required initially but the injection rates could be reduced to much lower
flow rates than those specified in the original procedures while still protecting the
flowline from hydrate formation. Revised procedures were proposed in which the IMS
injection rates were based on the temperatures downstream of the choke valve. These
procedures were intended to be more robust and less sensitive to the rates at which the
wells are started-up. The revised procedures would require approximately 8% less IMS
than the original procedures provided that the wells were started up quickly.
ACKNOWLEDGEMENTS
The authors would like to thank BP for permission to publish this data.
REFERENCES
1.
2.
3.
Sloan E.D., Koh C.A. Clathrate Hydrates of Natural Gases Third Edition, CRC
Press, 2008.
Anderson R. et al. Proceedings of the 6th International Conference on Gas Hydrates,
Vancouver, July 2008.
Private Communication, Centre For Gas Hydrate Research, Hydrate
Thermodynamic Measurements, Heriot-Watt University, July 2005.
147