1. Performance analysis of gas-lifted subsea wells from combined well testsby
Wim der Kinderen
Consultant Production Technologist
Shell UK Exploration and Production
Aberdeen
API/ASME Gas Lift Workshop, Houston, February 2001

2. Subsea gas lift
why is gas lift optimisation of subsea wells so difficult?
what are the consequences of limited testing and surveillance?
how can we improve testing and surveillance in a cost-effective way?

3. typical GL performance curve (fixed FTHP)
net or gross (m3/d)
2000
1500
1000
500 50
100
150
200
injection rate (*1000m3/d)

4. influence of wellhead choke or flowline
500
1000
1500
2000 50
100
150
200
injection rate (*1000m3/d)
gross (m3/d)

5. Problems of subsea gas lift
wells usually share flowline to platform:
FTHP cannot be considered constant
over-injecting lift gas causes oil deferment
flowline/riser system is prone to severe slugging:
limited validity of steady-state models
difficult well kick off (risk of platform trip)
subsea wells are hardly ever surveyed (expensive access)
wells are sporadically tested (oil deferment)
downhole gauges/flowmeters are lacking/ malfunctioning

6. Well testing
long flowline -> several hours stabilisation time (typ. > 8 hrs after GL rate change)
slugging -> long test times (typ. >6 hours)
difficult to test at normal operating conditions
cumbersome playing with chokes to match normal FTHP
multi-rate testing of one well takes days

7. Limited survey and test data
usually just one P/Q datapoint at one lift gas rate available
lift point, reservoir pressure, and IPR are uncertain:
too many degrees of freedom to match well and flowline models
performance analysis is very time-consuming
models are useless for lift gas allocation and well routing

8. How to improve surveillance in a cost-effective way?
use pressure drop across subsea oil chokes for production trending:
install dP transmitters (FTHP and manifold pressure gauges are too inaccurate at small dP)
install distributed temperature sensors (DTS) in the wellbore
apply modified ‘piggy-back’ well test method:
less production deferment
reduced slugging problems
multi-rate test data available

9. Gannet D field - central North Sea6”
4” (gas lift)
MPM
Bulk Sep
R31
R32
Gannet A
Gannet G
Test Sep
GD-01
GD-02
GD-03
GD-04
GD-06
Andrew Tay
reservoirs
subsea manifold
15.5 km from platform
New well
4 well test campaigns cost 2 MM£/yr

10. Testing two subsea wells simultaneously
standard ‘piggy-back’ testing is not possible:
the FTHP of well A changes when well B is added to the test flowline:
therefore, the production of well A is no longer known unless its PQ curve is first established:
this requires a multi-rate well test, or a calibrated well model
combined test data can be unravelled:
by assuming that the PQ and lift performance curves of a well can be linearised around an operating point

11. Linearised PQ curves Well model, no GL Linearised, no GL10 20 30 40 50 60 70 80 90
100
200
300
400
500
600
700
Well model, no GL
Linearised, no GL
model, GL 20 Km3/d
Linearised
model, GL 40 Km3/d
THP (bar)
800
900
Gross flowrate (m3/d)

12. Linearised GL curves Gross flowrate (m3/d) model, THP 50 bar
350 50
100
150
300
250
200
Gross flowrate (m3/d)
model, THP 50 bar
model, 55 bar
model, 60 bar
Linearised, 50 bar
Linearised, 55 bar
Linearised, 60 bar 10 20 30 40 50 60 70 80 90
Lift gas injection rate (Km3/d)

13. Linearised equations
The production of a well can be written as a function of
the wellhead pressure (THP) and the lift gas rate (Qgl)
where the constant c includes the reference Qgl and THP
Around an operating point for well 1:
For a second well:
Adding both wells:
6 unknowns (a, b, c, d, e, f) ---->
6 independent equations to be solved

14. Dual well test Procedure: 1. Select two wells for combined testing
2. Test most prolific well at one THP and GL rate
3. Add second well and select GL combinations
4. Use choke to change THPs if required
5. Solve inverse matrix to find a, b, c etc. (Excel macro)
Matrix of measured THPs and GL rates

15. Dual well test - input data set
Example:

16. Input data plot THP (barg) test 3, GD01 (40 K) + GD04 (40 K)80 70 60 50
THP (barg) 40
test 1, GD4 only, no GL 30
test 2, GD4 only, 40 K GL
test 3, GD01 (40 K) + GD04 (40 K) 20
test 4, GD01 (40 K) + GD04 (40 K)
test 5, GD01 (45 K) + GD04 (no GL) 10
test 6, GD01 (20 K) + GD04 (no GL)
200
400
600
800
1000
1200
Total gross flowrate (m3/d)

17. Split after matrix inversion10 20 30 40 50 60 70 80
200
400
600
800
1000
1200
Gross flowrate (m3/d)
THP (barg)
test 1 GD04, no GL
test 2 GD04, 40 K GL
split test 3
split test 4
split test 5
split test 6
GD04, no GL
GD04, 40 K GL
GD01, 20 K GL
GD01, 37.5 K GL
GD01, 45 K GL
GD04: sm3/d/bar
GD01: 6.04 sm3/d/bar

18. How to unravel BSW and GOR?
Single well test of well 1 reveals BSW1 and GOR1
BSW2 = (Qwater,combined - BSW1* Qgross,1)/Qgross,2
Assumption: BSW and GOR are not rate-dependent:
GOR2 = (Qgas,combined - GOR1* Qnet,1)/Qnet,2
Similar for the GOR of well 2:
Requires accurate well test measurements!

19. Conclusions Deriving PQ curves from dual well tests is feasible
Method works for gross, net and produced gas
Method requires just one single well test (and 5 combined tests)
Adequate accuracy for most purposes, but depends on input data quality (a.o. THP, liftgas rate)
Provides valuable information for system model calibration (-> to generate actual P/Q and GL curves)
Significant cost savings (1-2 MM£ for Gannet D)
Now applied in other subsea fields of Shell Expro

20. Enhancements
use multivariate analysis when more well tests are available
use downhole gauge data where available
use oil and/or water composition from samples to improve production allocation
extend method to multiple well testing