Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
the inelastic near/far ratio as a function of water, oil, and gas constraints. To implement the solution, the problem has been
holdup. These equations assume that the carbon contribution from reformulated such that the solver tries to minimize the following
the gas scales with gas density. This assumption appears to be function.
S D S D S D
reasonably robust if the gas is hydrocarbon. (The equation must be
reformulated if the gas is nitrogen, carbon dioxide, or steam.) RN 2 R0N 2
RF 2 R0F 2
RI 2 R0I 2
z5 1 1 , . . . . . . . . . . (7)
Inherent to these equations are the assumptions that the borehole sR N sR F sR I
holdups sum to unity and that the formation oil and water satura-
tions also sum to unity (no gas in the formation), i.e., where the superscript 0 on the R terms indicates the reconstructed
1 5 yO 1 yW 1 yG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (4) ratio from Eqs. 1 through 3, and the nonsuperscripted value indi-
cates the measured value from logging. The sx values in the
and 1 5 SO 1 SW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (5) denominator are the standard deviations of the various measured
parameters. If none of the constraints of Eq. 6 are violated, then the
The determination of the N, F, and G coefficients depends on the linear programming solution will be identical to the algebraic
formation properties and the borehole completion. The G coeffi- solution.
cients are calculated from a regression to either modeled or labo-
ratory data. The derivation of the N and F coefficients is discussed
in the literature.8 For the rest of this section, we will assume that Field Tests
all these parameters are known. The field tests were performed with the PL Flagship tool string,
Given the above equations and data, there are five unknowns (yO, which incorporates measurements of oil velocity, water velocity
yW, yG, SO, and SW) and five equations; this means that they can all (by two independent methods), and water holdup in horizontal
be solved for directly. If adequate statistics are available in the wells through its various sensors. In addition, the tool string usually
measurement, this can be done straightforwardly; however, with contains spinners and pressure and temperature sensors. Field
limited statistics, the poor orthogonality of the borehole and for- testing the TPHL measurement with these other measurements
mation response causes some problems in the solution, making it provides a way of at least partially validating the measurement
appear very noisy. The dominating factor is the determination of the based on other independent measurements.
formation oil and water saturations. Large error fluctuations in The FloView Plus measurement, which can measure the water
these parameters, because of statistics, result in large perturbations holdup in horizontal wells,10, 11 is of specific interest for these field
in the holdup answers. One way of addressing this is to set the tests. This measurement is made by using eight local electrical
saturations to some fixed value (say 50%) and solve for the holdups probes mounted on arms of a spring-centralizer system. The arms
directly. This dramatically improves the precision of the determi- open automatically in the wellbore. The probes are sensitive to the
nation, but it also results in a noticeable and unpredictable bias in passage of oil and water. A global estimate of the water holdup can
the answer. By letting the saturations float, the accuracy is im- be made and an image of the distribution of the wellbore fluids
proved, but the precision is degraded. produced by combining the information from the eight local and
The approach used to date is based on solving for all parameters spatially separated probes. This measurement is a water holdup
simultaneously. If the precision on formation saturations is poor, measurement only because the electrical probes cannot differentiate
they are just ignored and not reported. To make the mathematical between oil and gas. In other words, they differentiate water from
solution more stable in poor statistical conditions, a constrained hydrocarbons.
solution has been used that limits the range of values for the
unknowns to reasonable values. The constraints imposed are Example 1. Gas Well. The first example is from a gas well in the
Middle East. The well was completed with two lateral sections. The
0 # yO # 1, 0 # yG # 1, 0 # yW # 1,
overall purpose of the logging program was to better understand the
0 # fSO # f, and 0 # fSW # f. . . . . . . . . . . . . . . . . . (6) reservoir and inflow behavior before starting a stimulation pro-
gram. This was to be accomplished by logging one of the laterals
Given Eqs. 1 through 5 and the constraints of Eq. 6, the problem for a production profile and logging above both laterals to obtain
can be solved by use of a linear programming technique with the production split. This was especially important because there
were no production data available on the well. It was decided to run temperatures, the gas and condensate densities were 0.20 and 0.58
the PL Flagship tool string to obtain redundancy in the measure- g/mL, respectively. The well was logged at a relatively high logging
ments because it was feared that debris in the well might cause speed of 1,000 ft/hr because the primary parameter of interest was
problems for some of the tool string sensors, such as the spinner. gas holdup. (This meant that oil/water differentiation would be
The TPHL measurement was added to the logging plan because of highly statistical.)
the need to measure gas holdup. Fig. 1 shows the inputs from the RST tool that are used for the
The well was a 6-in. borehole with a barefoot completion in a 8 TPHL processing. The inelastic near/far ratio is running around 1.1
porosity unit (p.u.) limestone formation. At downhole pressures and for most of the well with a few positive excursions to higher values.
This ratio was formed for both the capture and inelastic elemental
Example 2. Oil/Water/Gas. When logged, this well was a newly yields and these two ratios were then averaged (Fig. 5). Because of
drilled horizontal well producing 95% water and had twice the the high logging speed, the data were heavily depth averaged;
gas/oil ratio of adjacent wells. This indicated that the well may have however, the results clearly indicate that the lower part of the well
intersected a suspected fault that was connected to a gas cap; has carbonate concentrations up to '50%. For TPHL processing,
therefore, the well was logged to determine entries of unwanted a mixture of 50% sandstone and 50% carbonate was assumed for
fluids. The TPHL measurement was one of the measurements used the lower section of the well. (However, this proportion is probably
to evaluate this well in addition to the entire PL Flagship tool string. not constant over this section of the well.) This lithological analysis
The well was drilled with a 61⁄8-in. drillbit and completed with a was accepted by the client. If the well had been logged at a slower
41⁄2-in., 17 kg/m casing that was cemented in place. The well was speed with the RST tool, then the SpectroLith processing could
perforated at seven different intervals. The formation was reported have been applied to the data to give a more detailed picture of the
to be a 14 p.u. sandstone. At downhole pressures and temperatures, formation lithology.12
the gas and oil densities were 0.19 and 0.74 g/mL, respectively. For The interpretation was straightforward when the data were pro-
the TPHL measurement, the logging program included station cessed assuming the better estimate of lithology. Figs. 6 and 7 show
measurements between the perforation intervals (to coincide with the final interpreted log including both FloView Plus and TPHL
fluid velocity measurements) and a continuous pass at 600 ft/hr results. Fig. 6 shows the data in a standard interpretation format.
over one zone of interest. This includes information on the well trajectory, useful in analyzing
Fig. 3—Final TPHL results for Example 1. FloView Plus data are plotted on top of the TPHL data and show
good agreement with the TPHL water holdup measurement. The FloView Plus tool is insensitive to oil vs.
gas.
Fig. 5—Average of the capture and inelastic lithology ratio for Example 2. Because of the high logging
speed, a heavy depth filter was applied to the data for this processing. This result shows the large
change in lithology in this well.
the results from the client perspective. Fig. 7 shows a portion of the perforations at this depth. This log conclusively shows that the zone
data where the FloView Plus and TPHL holdup data are plotted on of gas production is from the upper set of perforations. If gas had
top of each other (11-level depth smoothing). This helps increase been produced from lower perforations, then it would have accu-
confidence in the results. The FloView Plus measurement differ- mulated at the high point of the trajectory at about 880 ft.
entiates only water and hydrocarbon and is insensitive to changes
from oil to gas. Even with these poor statistics of oil/water holdup,
the agreement between FloView Plus and TPHL water holdups is Summary and Conclusions
reasonable, despite the fact that the lithology has been very roughly The ability of the RST tool to perform a three-phase holdup
estimated. If a better measure of the changing lithology had been measurement in horizontal wells has been demonstrated. The TPHL
available for this well, the results from the two methods would measurement uses the RST tool’s near/far inelastic ratio with the
probably be in even better agreement. It should also be noted that near and far C/O ratios as inputs to a linear programming solution
the stationary TPHL measurements, even though not shown, agree to obtain quantitative oil, water, and gas holdups in horizontal
well with the continuous TPHL measurement and the FloView Plus wells. Because the tool responds to both the formation and borehole
results. (even when centered), the analysis must correct for the formation
The gas entry at 730 ft coincides with the change of lithology for response (lithology and formation fluid) to obtain accurate borehole
sandstone to a mix of sandstone and carbonate and with a set of holdups. If this information is not included in the analysis, then
errors on the calculated holdups can vary from a couple of percent difficult to obtain a statistically significant formation oil volume;
up to about 20% depending on the lithology and porosity. For this however, as long as the formation oil volume is constrained to
reason, the formation lithology and oil volume are included in the realistic values in the analysis, statistically significant and accurate
analysis. At high logging speeds, insufficient statistics can make it values of borehole holdup can be obtained.
Tool response calibration is derived with a combination of additional useful information to the interpretation of production
laboratory data and Monte Carlo data. This approach is necessary problems.
to account properly for oil and gas density variations under down- Finally, it is worth noting that a holdup measurement by itself is
hole conditions. This has been demonstrated from earlier work and not usually sufficient to diagnose production problems in horizontal
from the field examples cited in this report. wells. Only under some specific conditions and/or for some specific
The field examples shown here demonstrate the TPHL measure- problems will stand-alone holdup provide an adequate answer.
ment under two extremely different sets of conditions, a gas well Therefore, holdup must usually be combined with velocity infor-
with a little liquid and a liquid well with a little gas. Both of these mation to understand fully the production environment in horizon-
are important conditions for a production logging tool to work tal wells.
properly. In both examples, the TPHL water holdup results com-
pared favorably with the FloView Plus results, which is a totally Nomenclature
independent water holdup technique. Unfortunately, there are no
data available with which to validate the gas vs. oil holdup results CF 5 far carbon yield
quantitatively; however, other qualitative measures indicate that the CN 5 near carbon yield
results are quite reasonable. Fx 5 far detector sensitivity factors
Example 2 showed the importance of understanding the forma- Gx 5 near/far gas sensitivity factors
tion to interpret properly a holdup log with a pulsed-neutron tool. Nx 5 near detector sensitivity factors
In this example, the lithology change caused a large perturbation to OF 5 far oxygen yield
the amount of carbon in the formation, severely affecting the holdup ON 5 near oxygen yield
calculation. Fortunately, the RST tool can also provide lithological RF 5 far carbon to oxygen yield ratio
information to identify and correct for these effects. Because of the RI 5 inelastic near to far yield ratio
high logging speed in this example (600 ft/hr), the quality of the RN 5 near carbon to oxygen yield ratio
lithology information was marginal as a result of the heavy depth SO 5 formation oil saturation
smoothing required. If the log had been performed at a lower SW 5 formation water saturation
logging speed, the RST lithology information could have dramat- yG 5 gas holdup
ically improved, and much less depth smoothing would have been yO 5 oil holdup
required. Not only would this have improved the lithology analysis, yW 5 water holdup
it would also have improved the accuracy and precision of the f 5 porosity
borehole holdups. rG 5 downhole gas density
The examples shown here were for continuous logging passes at rO 5 downhole oil density
relatively high logging speeds. These high logging speeds (600 and
1,000 ft/hr) could be used for these wells because the primary Acknowledgments
parameter of interest was the gas vs. liquid production in these I thank Chris Lenn, Peter Hook, Khalid Azim, Syed Shakil Ahsan,
wells. This is why the oil and water holdups appear fairly noisy. To Bruce Cassell, and Kazuo Yuassa of the Schlumberger field orga-
separate water from oil with the TPHL measurement, better sta- nization for assistance in obtaining the data and releases for this
tistics, and hence slower logging speeds, are required than for study.
separating liquid from gas. For this reason, the FloView Plus and
TPHL tool combination is useful in augmenting one another’s
interpretation. In addition, the redundant measurements of water References
holdup give added confidence in the results. Another helpful feature 1. Theron, B.E. and Unwin, T.: “Stratified Flow Model and Interpretation
of this tool combination is in slotted liner completions, because the in Horizontal Wells,” paper SPE 36560 presented at the 1996 SPE
FloView Plus tool sees only what is inside the liner, whereas the Annual Technical Conference and Exhibition, Denver, Colorado, 6–9
TPHL tool sees both inside and outside the liner. This can add October.