Professional Documents
Culture Documents
Basically, these tests were carried out to test the productivity Hence, following quality control, blowout rates were calculated
(and sand production) of the wells, but for several reasons, they offer and compared with the actually measured rates.
an excellent opportunity to verify the blowout-rate calculations on
subsea wells of the in-house blowout-control modeling tool: Test-Data Quality Control
• The configuration is very similar to the configuration encoun- The well was tested at a number of rates, varying from 0 (closed in)
tered in many blowouts, with annular flow between the drillstring to 1.3×106 std m3/d, at flowing wellhead pressures (FWHPs) ranging
and casings/open hole when control is lost over the well (e.g., from 252 (closed in) to 67 bar. For each flow rate, the flowing tub-
after a kick is taken). inghead pressure and the flowing bottomhole pressure (FBHP) were
• Subsea wells blow out against the pressure at seabed. Dur- measured. This allows a distinction between the inflow performance
ing the tests, wellhead pressures between 70 and 220 bar were (the relation between the flow rate and the FBHP) and the vertical
observed, which is representative for subsea blowouts in water lift performance (the relation between the pressure drop over the
depths between 680 and 2100 m. tubing from top to bottom and the flow rate). Both elements of the
overall well performance will be examined separately.
where pres is reservoir pressure, bar; pwf is FBHP, bar; and qgas is
9.625 in.
inCsg
casing 3.500 in.
inDPdrillpipe gas rate, 1000 std m3/d.
53.50lbm/ft
53.50 lbm/ft 10.20lbm/ft
10.20 lbm/ft
A, F, and pres can be obtained basically by plotting the square of the
3340 m FBHP as a function of gas rate and fitting a second-order polynomial
3400 m
to the measurements. For the test points, however, this procedure
failed because the fit gives a negative value for A. Because A is the
Darcy-flow term, which is related to the pressure drop in the porous
permeable rock of the reservoir, this term cannot be negative because
this would imply that in the absence of turbulence, pressure would
increase in the direction of flow. This is physically impossible.
Hence, it was decided to suppress the non-Darcy F term, by
7.000 in.
inCsg
casing forcing the term to vanish. The result is shown in Fig. 3.
8.500 in.
in P 32.00lbm/ft
32.00 lbm/ft The fit gives the following result where pres = 291 bar and
4774 m
3401 m A = 26.3 bar2/(1000 std m3/d). Given the true vertical depth of the
formation (2840 m), this means that the reservoir is normally pres-
sured (0.102 bar/m). The value for A indicates that the formation
has a low permeability, in the millidarcy range.
It is noted that there is considerable scatter in the test points
(the coefficient of determination, R2, indicating the quality of the
fit, is only 0.77), particularly at rates below 800,000 std m3/d. This
can have several causes:
• Poor measurements of the FBHP and/or the rates
• Slugging in (part of) the well at low rates
Fig. 2—Well-test configuration. • Poor stabilization of the tests
2 , bar2
50000
pwf 40000
30000
20000
10000
0
0 500 1000 1500
qgas, 1000 std m3/d
It is difficult to distinguish between these causes. Slugging because these would have introduced a linear dependence on gas
would explain that the scatter is found mainly at the low-rate points rate—gives the following results: B = 1.39 (dimensionless) and
(loading conditions), but this could also be attributed to flow-rate C = 0.0235 bar2/(1000 std m3/d)2.
measurements close to the lower limit of the dynamic range of It is noted that the scatter observed in the inflow-performance
the orifice meter. plot is completely absent in the vertical-lift-performance plot. This
Alternatively, the consistency of the measurements can be indicates that the rates and pressures have basically been measured
checked by examination of the vertical lift performance. correctly and accurately and rules out slugging as a possible cause
for the scatter because this would certainly have given rise to a
Vertical Lift Performance. For “dry” gas wells, the vertical lift deviation from the (pseudohomogeneous) Cullender-Smith model.
performance can be described with the Cullender-Smith B- and Stabilization is not an issue for the tubing flow in view of the
C-factor relation (Katz 1959): limited volume of the tubing and the short residence time of the
gas in the tubing (approximately 200 seconds).
2
pwf = B ⋅ ptf2 + C ⋅ qgas
2
, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2) This leaves poor reservoir stabilization as cause for the scatter
in the inflow data [i.e., (pseudo) steady-state reservoir flow has
where ptf is flowing wellhead pressure, bar; pwf is FBHP, bar; and not been reached for the low flow rates during the tests]. This is
qgas is gas rate, 1000 std m3/d. typical for tight gas reservoirs, which take a long time to stabilize
The B factor accounts for the hydrostatic pressure drop in the (Dake 2001). Because the scatter is not systematic, the test points
well and depends on gas density, compressibility, and temperature. can be used to verify the blowout-rate calculations, but the scatter
The C factor gives the frictional pressure drop and depends on gas will be reflected in the results of these calculations.
density, compressibility, temperature, tubing size, and hydraulic
roughness of the conduits. By plotting (pwf /ptf)2 vs. (qgas/ptf)2 and Comparison With Calculated Blowout Rates
fitting a straight line, B can be derived from the intercept of the line By comparing measured and calculated rates, two modules of the
and C can be derived from the slope. This is shown in Fig. 4. blowout simulator (Oudeman et al. 1993) can be investigated:
The excellent fit—the coefficient of determination, R2, equals 1. The solution routines that, given a surface outflow pressure,
0.99, which also implies that accelerational terms play no role here search for the matching blowout rate and downhole pressure. These
12
10
8
(pwf /ptf )2
0
0 50 100 150 200 250 300 350 400
(qgas/ptf )2 (1000 std m3/d.bar)2
1.8 1.8
0.2 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
0
Measured rate, million std m3/d
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Measured rate, million std m3/d Fig. 6—Comparison of measured and calculated rates, using
an adapted hydraulic roughness.
Fig. 5—Comparison of measured and calculated rates, using
the default roughness.
actually occurring in the well are not taken into account. This
routines are based on the same principles as nodal analysis, with increases the calculated blowout rates.
thorough modeling of sonic conditions in the well (if present), usu-
ally at wellhead (this makes the intake pressure curve, or “J-curve,” Calculations With Adapted Hydraulic Parameters. As men-
deviate from the standard curve because the wellhead pressure will tioned, the hydraulic roughness used in the calculations is 0.001
not be constant but dependent on rate for critical rates), and model- in. (25 m), a value determined by Gray (1974) to give the best
ing of along-well temperatures, given the considerable expansion prediction of tubing pressure drop in wet gas wells.
of the gas and the associated Joule-Thomson cooling. In this case, however, the inside of the 7-in. casing and the
2. The vertical lift calculations in the annular geometry (3.5-in. outside of the 3.5-in. drillpipe have been exposed to drilling fluids
drillpipe × 7-in. casing) that were recently improved by including and cuttings and mechanical damage over a long period. Hence, it
an improved description of hydraulic diameter for annular flow is reasonable to assume that the hydraulic roughness of the steel
(Oudeman 2007). will be closer to the roughness of rusted steel [i.e., on the order
Correspondence between calculated and measured rates can be of 0.01 to 0.02 in. (250 to 500 m)]. Furthermore, the presence
expected only if both modules perform correctly in the simulator. of crossovers and couplings will add to the effective hydraulic
roughness of the blowout conduit.
Calculations With Default Hydraulic Parameters. To calculate Fig. 6 shows the result of calculations with a roughness of 0.01
frictional pressure drop in high-rate gas/liquid flow, a value for in. For the highest rates, an excellent match with measurements is
hydraulic roughness, also known as sand-grain roughness, is obtained using this value for hydraulic roughness. On average, the
required because at high Reynolds numbers, this will completely rates are underpredicted by only 2% (i.e., well within the range of
determine frictional pressure drop (Nikuradse 1933). Default val- accuracy of the data). The scatter in the lower-rate points remains
ues for hydraulic roughness for various materials have been because, as pointed out before, this is related to poor reservoir stabi-
listed, but actual values may differ considerably because of the lization and not to the selected value for roughness. This also applies
manufacturing process and condition of the material. For gas-well to the two points for which no solution for the rate could be found.
tubing, Gray (1974) determined a value of 25 m (0.001 in.) to
give the best fit to field data, and this value is used as a default Preliminary Conclusion. From the preceding sensitivity analysis
in calculations with the in-house modeling tools, although other on the effect of hydraulic roughness, it can be concluded that the
authors determined values as low as 15 m (0.0006 in.) in gas standard value for hydraulic roughness in cases such as that being
wells (Smith et al. 1954). studied gives at best a worst-case estimate for the blowout rates and
A comparison between the measured and calculated rates for that, using an appropriate value for roughness, blowout rates can be
the wellhead pressures measured is shown in Fig. 5. calculated with high accuracy. This conclusion, however, is based
It appears from this figure that on this single case and leaves the possibility that, for example, an
• For two wellhead pressures, 224 and 227 bar, a zero outflow erroneous calculation of the pressure loss is corrected by adapting
rate is calculated (i.e., it is predicted that, under steady-state con- the hydraulic roughness. To exclude this possibility and confirm
ditions, the well will not flow against these pressures because of the preliminary conclusion, the blowout-rate calculations were
downhole liquid accumulation). However, measured flow rates also tested against the results of an earlier test. This is discussed
were 0.5 and 0.6 million std m3/d. Probably, this can be attrib- in the next section.
uted to poor stabilization, as pointed out in the Test-Data Quality
Control section. The closed-in wellhead pressure is 251 bar, so Testing the Preliminary Conclusion
if the measurements were taken soon after opening up the well, In 1994, a dedicated field test was set up to study the dynamics of
the measured wellhead pressures are not representative for stable the hydraulic kill of a blowing well by injecting kill fluid down-
flowing conditions. hole in a 5.5-in. well through 1.75-in. coiled tubing. This test in a
• The scatter in the low-rate points, which was observed in the Groningen gas well is described in detail in Oudeman and Mason
inflow-performance analysis, is indeed reflected in the calculated (1998). The flow configuration is shown in Fig. 7.
rates. The configuration is somewhat less representative for an actual
• The blowout rates are overpredicted almost systematically by blowout situation, where the inner pipe would usually be a 3.5- or 5-
approximately 11% on average. This suggests that pressure losses in. drillstring. The gas/liquid ratios (GLRs) during the tests ranged
2 , bar2
26550
pwf
26500
26450
26400
26350
26300
0 200 400 600 800 1000 1200 1400
qgas, 1000 std m3/d
1.5
3.5×7 in.
1 1.75×5.5 in.
0.5
0
0 0.5 1 1.5 2
Measured rate, million std m3/d