FUNDAMENTALS OF PRODUCTION LOGGING

PART 2
COURTESY SCHLUMBERGER
COLIN WHITTAKER

1
 4. Spinner Velocity Tools
• Although turbine, or spinner, flowmeters are used extensively within industry with small to negligible errors, the situation is
very different in oilfield applications.
• The fullbore spinner does not, as its name suggests, cover the entire pipe cross section (Fig. 4-1).
• Typically a fullbore spinner sweeps only about 40% of the casing’s cross-sectional area.
• In addition, the blades do not have a progressive pitch (as on a ship’s propellers and in gas turbines) because the requirement
to collapse down to a diameter of 111⁄16 in precludes anything more complicated than a flat spinner blade.
• The continuous, or tubing, spinner has a progressive- pitch spinner, which is more effective at extracting energy from the well
fluids (and therefore reducing the spinner threshold, which is the minimum velocity needed to start the spinner turning) (Fig.
4-2).
• Unfortunately, the much reduced spinner diameter more than negates the effect of the improved blade profile, and the
threshold velocity of a standard tubing spinner is about 3 times higher than that of a fullbore spinner.
• The Flow Scanner minispinner (Fig. 4-3) is used in an array of five minispinners recording velocities on the vertical axis of the
pipe, from the bottom to the top.
• This arrangement is discussed in detail for the Flow Scanner horizontal and deviated well production logging system (see the
“Flow Scanner Interpretation” chapter).
• A change of bearing technology together with the progressive pitch of a tubing spinner delivers a spinner threshold
comparable to that of a fullbore spinner.

2
Figure 4-1. Fullbore spinner schematic.
3
Figure 4-2. Continuous, or tubing, spinner.

4
Figure 4-3. Flow Scanner minispinner.
5
 Theoretical spinner response model
• Consider the response of an ideal spinner flowmeter (Fig. 4-4).
• The spinner speed in revolutions per second (rps) is directly proportional to the fluid velocity passing through the spinner.
• The slope of the response curve, measured in rps/ft/min [rps/m/min], comes directly from the spinner pitch measured in
inches [centimeters].
• Once friction in the bearings is included, the response becomes a little more complicated, with two response lines, one for
positive spinner readings and a second for negative spinner readings.
• There is now a range of low fluid velocities where the spinner does not turn because the spinner torque is smaller than the
bearing friction.
• Close inspection of the response lines in Fig. 4-4 shows a small curve at low spinner speeds owing to the action of static friction
(stiction), viscosity effects, or both.
• To avoid the complications arising from nonlinear spinner response, near-zero spinner readings are discarded if they look at all
suspicious.
• Increasing the viscosity of the fluid passing through the spinner produces some unusual results.
• The threshold first increases and then decreases while the spinner slope changes by about a factor of 5 or more.
• Figures 4-5 and 4-6 show the results of previously unpublished experimental data from the Schlumberger Gould Research
Center.
• Changing the density of the fluid passing through the spinner (e.g., from liquid to gas) also increases the size of the low-velocity
region where the spinner does not turn but should not significantly change the spinner response slope (Fig. 4-7).
6
Figure 4-4. Spinner response with friction and stiction. 7
Figure 4-5. Spinner response to increasing viscosity.
8
Figure 4-6. Spinner response to increasing viscosity, amplified scale. 9
Figure 4-7. Spinner response to decreasing density. 10
Figure 4-8. Fluid turbulence and the fullbore spinner.

11
• However, changing the fluid from liquid to gas causes a big change in the pseudo-Reynolds number (created using the tool
velocity) and makes the creation of turbulence and vortices much easier, thus leading to the situation of Fig. 4-8:

• On the left of Fig. 4-8 is an upward-moving tool (equivalent to a negative fluid velocity) generating vortices that travel down
and confuse the spinner.
• Behavior like this creates asymmetric spinner slopes and thresholds.
• The effect can be expected to be bigger in gas wells, but it is still present in water and the lighter oils.
• In the middle and on the right of Fig. 4-8 is the effect of the spinner cage stirring up the flow.
• To a first-order approximation the effects seem to be equal for flow from above and flow from below.

12
 Practical spinner response model

• The theoretical spinner response model described in the previous section is too complicated for everyday use. Instead, some
approximations are introduced until the model of Fig. 4-9 is reached.
• This shows a spinner in an unknown fluid as having a positive and negative spinner slope and a positive and negative spinner
threshold.
• Because these four parameters can change with velocity, fluid density, fluid viscosity, casing diameter, and other conditions,
the model needs judicious updating over an interpretation interval.

13
Figure 4-9. Approximation to a spinner response.
14
 Spinner interpretation—Initial laboratory characterization
• The first approach (Meunier et al., 1971) to calibrating the spinner involved laboratory characterization and the creation of
interpretation nomograms (Fig. 4-10).
• The laboratory-determined spinner threshold and spinner pitch were combined with a stationary spinner reading, velocity-
profile viscosity model, and the pipe internal diameter to deliver a downhole flow rate.
• Unfortunately, the log analyst did not often have reliable downhole viscosity information, so measurements of the spinner
threshold in the laboratory were rarely representative of the field, and logs versus depth could not easily be processed.

15
Figure 4-10. Spinner interpretation nomogram (Meunier et al., 1971). 16
17
 In situ spinner calibration
• Fortunately, the in situ spinner calibration technique was developed. The earliest reference to this technique dates back to
Peebler (1982).
• Because it is impractical to vary the downhole fluid velocity while measuring the spinner rps, the problem of spinner
characterization (usually known as calibration) is approached from a different direction.
• Instead of varying the fluid velocity, the tool is moved up and down at different speeds and the spinner rps plotted versus tool
speed.

18
• In a zero-flow environment it is possible to determine the positive and negative spinner slopes and the positive and negative spinner
thresholds (Fig. 4-11).
• But where does a zero-flow environment exist? Shutting in the well at surface does not guarantee zero flow because there may be
crossflow between zones.
• Above the top perforation, after the well has been shut in for many hours, there should be no flow, but in the time available for
production logging there may be wellbore storage (unlikely) or liquid fallback from the tubing that extends to surface.
• Although it is safer to assume that there is no flow below the deepest perforation (in the absence of casing plugs isolating deeper
zones), the fluid density and viscosity below the deepest perforation are often unrepresentative of the fluid density and viscosity
flowing in the well (the fluid for which the spinner calibration is intended).
• Usually the best way to identify a zero-flow region is to refer to the temperature log and identify what appears to be a geothermal
gradient because geothermal gradients are incompatible with fluid movement.
• The spinner calibration in Fig. 4-12 corresponds to a small positive velocity (downflow) but this analysis can be delivered with
confidence only after inspecting the temperature log in Fig. 4-13.
• The log in Fig. 4-13 shows cable velocity, depth, spinner rps, the extent of the perforated intervals in red, spinner calibration zone in
yellow, rate calibration zones in gray, pressure, temperature, and density (indicating a water- filled borehole).
• The steep change in temperature at about 3,900 ft shows where the cooler crossflowing water is lost into a perforation (indicated in
red) and the geothermal gradient is recovered.

19
 20
Figure 4-11. Zero-flow spinner calibration.
Figure 4-12. Ambiguous near-zero spinner calibration.
21
Figure 4-13. Near-zero spinner calibration plot data.

22
• Will the positive and negative spinner thresholds be of equal magnitude?
• To a first-order approximation, the answer is yes, but more precisely the answer is generally no.
• It is normal to see a positive threshold that is between 40% and 60% of the distance from the negative intercept to the positive
intercept, but a range of 30% to 70% is still acceptable.
• Only when values of less than 30% or more than 70% are encountered should the validity of the spinner calibration or spinner
measurements be questioned.
• When the well is flowing, the same spinner response is translated to the left, with the measured velocity found by subtracting
the positive spinner threshold from the positive intercept (Fig. 4-14).
• Zero spinner readings have an associated uncertainty on the cable velocity axis no matter how accurately the cable velocity is
measured.
• Consequently a zero (or near zero, say less than 0.5 rps) spinner reading should never be used to calculate a spinner slope or
fluid velocity.
• Stationary spinner readings, which are spinner readings with a cable velocity of zero, can be added to the crossplot, where
they complement the pass data.
• Note that the data quality of the flowing crossplot in Fig. 4-14 occurs only in stable monophasic flowing conditions.
• Multiphase flow has more scatter on the data points and more ambiguity in choosing the appropriate spinner slope and
threshold.
• Where the fluid velocity is about 60 ft/min [20 m/min] or faster, it is not be possible to determine the negative spinner
intercept with any degree of confidence if at all.
• Therefore, the positive spinner threshold has to be assumed from lower speed measurements deeper in the well. 23

• There are errors in this assumption of the order of 10% to 20% of the positive spinner threshold.
Figure 4-14. Spinner calibration in a flowing well.
24
• Taking a typical fullbore spinner threshold of 5 ft/min, the expected error in the assumed threshold is about 0.5 to 1.0 ft/min.
• However, compared with a fluid velocity of 60 ft/min or higher, an error of 0.5 to 1.0 ft/min is not significant.
• In deviated wells with multiphase flow it is not uncommon to see the positive and negative intercepts move closer to each
other and finally move past each other, resulting in an overlapping intercept (Fig. 4-15).
• In this case there are no physical models to allow any correction to be applied; therefore, averaging the positive and negative
intercepts and calling this the spinner velocity is a simple and not too bad way of creating a fluid velocity.
• The log in Fig. 4-16 is from a deviated well under shut-in conditions and possibly with a little bit of water fallback and oil
bubbling upward.
• Track 1 shows the cable velocity followed by the depth, fullbore spinner rps, perforation interval in red, spinner calibration
zones in yellow, rate calibration zones in gray, pressure, temperature, mixture density, X caliper, Y caliper, cable tension, and
formation gamma ray.
• Above 2,815 m there is 95⁄8-in casing and below 2,817 m there is a 7-in liner.
• The wellbore oil/water contact is at 2,820.5 m. The temperature response at 2,821.5 m is oppo- site a perforation and may
indicate some inflow.
• Three spinner calibration zones were chosen.
• The top zone is in the 95⁄8-in casing in the oil column that may have some water falling back from the tubing when the well
was shut in.
• The second zone is in similar conditions except for a reduction in the casing size.
• The third zone is in the water leg in the 7-in liner.

25
Figure 4-15. Crossed intercepts on a spinner calibration.
26
Figure 4-16. Effect of changing diameter, changing density,
and recirculation.

27
• In Fig. 4-17 it can be observed that the top two zones have crossed intercepts. Each calibration zone has a different slope as
the spinner responds to changes in mixture density and casing ID.
• No one slope is correct, and the log analyst must choose where the slopes (and thresholds) are calculated and where the
interpolation between zones should occur.
On occasion it is impossible to calibrate the spinner downhole with a series of up and down passes at different speeds.
Perhaps the tool failed after the first logging pass or well conditions changed.
• In these cases a default spinner slope and threshold must be applied.
• The table in the Appendix to this book contains the theoretical spinner slopes (and pitches) and reason- able estimates of
spinner thresholds for all current and most historical Schlumberger spinners.
• Because it is normal to see some changes in slope and threshold over the logged interval, the use of a default slope and
threshold is deprecated as it often results in an inferior interpretation answer.

28
Figure 4-17. Spinner calibration from data of Fig. 4-16.
29
 Spinner velocity to mixture velocity
• The spinner measures a fluid velocity but this is not the same as an average mixture velocity.
• Across the pipe there are faster moving fluids in the center and slower moving fluids closer to the pipe wall.
• The centralized spinner reads some kind of average of the faster moving fluid velocities swept by the spinner blade (Fig. 4-18).
• A correction factor is required to reduce the spinner velocity to the average velocity.
• The first approach (Nicolas and Meunier, 1970) to this problem was entirely empirical, based on a survey of 45 wells where
surface separator rates could be compared with downhole velocity measurements (Fig. 4-19).
• The spinner diameter used for these surveys was probably about 1.5 in [3.8 cm], thus requiring a smaller correction coefficient
than a more modern spinner of 21⁄2-in or 31⁄2-in [6.4- to 8.9-cm] diameter.
• Despite the lack of scientific rigor (the correction factor should increase from the bottom to the top of the well as the velocity
increases and the velocity profile becomes flatter), the correction factor of 0.83 is still widely used today and does not
introduce unmanageable errors.

30
Figure 4-18. Effect of velocity profile on apparent spinner velocity. 31
Figure 4-19. Empirical study of 45 wells to determine a spinner correction factor (Nicolas and Meunier, 1970).

32
• Most modern computer-based interpretation techniques use a model for the velocity profile across the pipe and then
average the velocity that would be seen by the swept area of the spinner.
• For a spinner (Fig. 4-20) with blades of external radius R2, internal radius R1, and in laminar flow, the average velocity seen
by the spinner is

33
• Figure 4-20. Spinner blade dimensions.
34
• Most modern computer-based interpretation techniques use a model for the velocity profile across the pipe and then
average the velocity that would be seen by the swept area of the spinner.
• For a spinner (Fig. 4-20) with blades of external radius R2, internal radius R1, and in laminar flow, the average velocity seen by
the spinner is
• In turbulent flow and using the Prandtl relationship,

• In the case of two- or three-phase flow, the same correction algorithms developed for monophasic flow are used.
• Where the multiphase velocity profile looks similar to a monophasic velocity (in vertical wells or high-velocity deviated wells)
this approximation is acceptable.
• However, in deviated wells with low-velocity multiphase-flow, recirculation occurs.

35
 Recirculation and the spinner
• In anywhere from 5° to 75° deviation, a low mixture velocity combined with a high slip velocity produces gravity segregation of
the phases, giving rise to large high-side velocities, significant shear between light and
• heavy phases in the pipe, and finally a downflow of the heavy phase on the low side of the pipe (Fig. 4-21).
• A centered spinner tries to average the mixture of velocities passing through the swept area of the blades but generally
returns a velocity that is too low if not actually negative.
• A poorly centralized spinner, lying closer to the low side of the pipe, returns an answer still more heavily weighted for heavy-
phase recirculation.
• Increasing the diameter of the spinner blades reduces the size of the errors and is the simplest way of improv- ing data quality
under these conditions.
• The data in Fig. 4-22 shows the effect of increasing the spinner swept area (however, the use of heavy dead oil has greatly
reduced the slip velocity and the recirculation effect).
• While recirculation normally occurs with gas or oil bubbling up through water, the inverse situation, of water falling back
through oil or gas, is often seen when a well has just been shut in.
• In these cases the spinner sees an erroneous uphole flow rate.

36
Figure 4-21. Three-dimensional axial velocity
distribution in recirculation.

37
Figure 4-22. Effect of increasing spinner deviation at a constant oil–water flow rate on recirculation errors.

38
 Diverter flowmeters
• Another approach to handling the problems of recirculation is use of the diverter flowmeter.
• Because recirculation requires a low mixture velocity, the diverter, or petal basket (Fig. 4-23), flowmeter funnels the flow of
the well through a 1- to 11⁄2-in diameter tube in which a spinner velocity is recorded.
• If run in a 7-in completion the diverter flowmeter increases the velocity by about 20 times and therefore reduces the poorly
defined thresh- old for recirculation effects to one twentieth.
• However, once the diverter is deployed the diverter flowmeter tool can only be used to make stationary measurements.
• This means that the accelerated flow passing through the diverter must exceed the spinner threshold (which is of the order of
15 ft/min for the small spinner used).
• In addition, the in situ spinner calibration of the slope and threshold must be made with the flowmeter closed and in a
different flow regime from that being used for the station measurements.
• Nevertheless, the diverter flow- meter is currently the least bad approach to measuring velocity in the presence of
recirculation.

39
Figure 4-23. Diverter, or petal basket, flowmeter.

40
 Graphical interpretation techniques
• In the past, graphical techniques for interpreting spinner data were very popular. One elegant approach is still seen in
production log interpretation manuals.
• In Fig. 4-24, after the various stable interval responses are plotted together with a zero-flow response, the reading line for
the zero tool velocity spinner is projected right, across to the zero-flow line, from which a vertical line is dropped down to the
horizontal axis, which delivers the fluid velocity.
• Increasing the slope of the zero-flow line by the reciprocal of the velocity profile correction factor automatically corrects for
the velocity profile across the pipe.
• This technique assumes that the zero-flow spinner response slope and threshold are applicable to all the stable intervals
above.
• In practice this is true only for well-behaved monophasic wells with a clean, uncontaminated zero-flow region at the bottom.
• This plotting technique is no longer used commercially.

41
Figure 4-24. Graphical techniques for computing the mixture velocity (Production Log Interpretation, 1973).
42
 Computer processing algorithms
• Computer-based interpretation techniques tend to use one of two approaches for delivering a velocity to the flow rate
interpretation engine.
• In the first approach, a spinner calibration is per- formed above and below every production zone of interest.
• The velocity computed by the spinner calibration is delivered directly to the interpretation engine. No flow rates are
computed outside of the spinner calibration zones.
• (This technique is used by BorFlow software.)
• In the second, a spinner calibration is performed every time the spinner calibration slope or threshold is suspected of
changing.
• In a monophasic well this requires a single spinner calibration, whereas in a multiphase well with changing holdup and
velocity a series of spinner calibrations are made.

43
• From each calibration zone only the slopes and thresholds are used (the computed spinner velocity is ignored).
• Each spinner pass is converted into a spinner velocity pass using the following equation for positive spinner readings:

• Different spinner calibration zones have different values of slopes and thresholds.
• For depths between spinner calibration zones, the slopes and thresholds are normally interpolated from the adjacent
calibration zones.
• A survey of three up and three down passes delivers six velocity curves.
• It is a good quality control practice to inspect the multiple velocity curves for repeatability and either discard bad velocity
sections or revisit the spinner calibration and correct the slopes and thresholds before averaging the resulting velocities and
loading them into the interpretation solver. 44
 Spinner response in the presence of local slip
• In bubble flow the spinner encounters the continuous phase, typically water traveling slowly, and the discontinuous bubble
phase, typically oil traveling quicker.
• The traditional spinner interpretation approach is based on the following equations, to which subscripts identifying
monophasic and multiphase have been added for clarity.
• This approach is known as volumetric averaging.

45
• Volumetric averaging has been the default model used in production log interpretation for many years and it works
reasonably well where there is only a small density difference between the phases present or the slip velocity is quite small.
• Where a large density difference exists between two flowing phases, the velocity of the denser phase has significantly more
weight in the average velocity the spinner reads.
• (An extreme example is a paddle steamer that can move only because the paddle in the water creates more thrust than the
paddle in the air.)
• To correct for the effects of differing density, Whittaker et al. (2006) propose using the mass-fraction average:

• In the presence of a high density contrast between the oil and water and a high slip velocity between the oil and water there
is a large difference in the resulting spinner velocity depending on which mixing model is applied.
• For gas and water the differences are still larger, but in many cases there is no bubble flow and the mass-fraction model is
applicable only to bubble flow.

46
 Minimum velocities registered by multipass spinner surveys
• Most of the turbine flowmeters in the world are fixed within the pipe work and have a lower threshold velocity below which
they do not turn.
• Because this threshold velocity is an important quality measurement for any turbine or flowmeter, we also measure the
thresholds for our production logging flowmeters.
• However, although the threshold describes the lower velocity limit for a fixed spinner, a multispeed in situ calibration of a
spinner takes the quantifiable velocity limit significantly lower.
• Figure 4-25 shows the modeled spinner response in a 7-in liner flowing a total of 1,000 bbl/d of water.
• A typical spinner with a slope of 0.091 rps/ft/min and a threshold of 6 ft/min was simulated at a number of logging speeds
(ft/min) and also as station measurements.
• 10 8 6 4 2
• Of the five station measurements, only the last two stations, in 800- and 1,000-bbl/d flow, had sufficient water velocity to
start turning the spinner.
• This indicates that a survey comprising just station measurements would be unsatisfactory.
• A lower spinner threshold would be required to monitor the smaller flow rates.
• However, if a series of logging passes were recorded, the spinner would be rotating over the entire interval to indicate the
complete flow profile.
• It therefore follows that the spinner threshold cannot be used to indicate the minimum flow velocity detectable by a
production logging spinner.
• If the spinner threshold does not indicate the mini- mum mixture velocity that a spinner can quantify, then what does?
47 The
following factors affect the minimum detectable fluid velocity measurable by a spinner.
Figure 4-25. Effect of a threshold on a spinner response.

48
■ A bigger spinner diameter is able to detect lower fluid velocities.
■ Spinner pitch has an impact, as does the blade profile (is it flat or with a progressive pitch?).
■ The logging speed must be fast enough to make the spinner turn but not so fast that the blades deform.
■ High-density, low-viscosity fluids improve the sensitivity.
■ Shallow vertical wells with smooth tool movement allow smaller velocities to be detected than in deeper, more highly deviated
wells.
■ The flowmeter bearing design and the maintenance the bearing has received are also important.
■ Single-phase flow multiphase flow.
• The complexity is such that models must be abandoned and reliance put on empiricism and experience. For the log analyst
lacking experience, the following observations may help.
o A modern well-designed fullbore flowmeter should be able to resolve down to about 3 ft/min in a water-filled borehole
of 45° deviation and 9,000-ft depth.
o In a shallow vertical well filled with water it is possible to resolve down to about 1.5 ft/min.

49
 What is the upper viscosity limit for spinner operations?
Highly viscous crude oils may not allow the spinner flowmeter to turn.
Work in the Schlumberger Gould Research Center shows that a local Reynolds number can be used to determine the cutoff point:

• Because fullbore spinners collapse or deform under the stresses experienced in highly viscous fluids, this computation should
be limited to rigid turbine spinners and the minispinners.
• Unfortunately this method cannot be used for the newer highly viscous non-Newtonian polymers, for which the spinner
situation may be worse (Zheng and Liu, 2006).

50
 References

• Meunier, D., Tixier, M.P., and Bonnet, J.L.: “The Production Combination Tool—A New System for Production
Monitoring,” Journal of Petroleum Technology (May 1971), 603–613.
• Nicolas, Y., and Meunier, D.: “From Memos by Y. Nicholas [sic], EPS, and D. Meunier, STS,” Technical Review
(October 1970), 18, No. 5, 21–25.
• Peebler, B.: Multipass Interpretation of the Full Bore Spinner, Houston, Texas, USA, Schlumberger Well
Services (1982).
• Production Log Interpretation, Houston, Texas, USA, Schlumberger Well Surveying Corporation (1973).
• Whittaker, A.C., Lenn, C.P., and Hammond, P.: “Improving Multiphase Production Logging Answers with the
Mass- Fraction Spinner Response Model for Gas-Liquid Flows,” Petrophysics (April 2006) 47, No. 2, 120–128;
originally presented at the SPWLA 46th Annual Logging Symposium, New Orleans, Louisiana, USA (June 26–
29, 2005), paper NN.
• Zheng, H., and Liu, X.-B.: “Injection Profiling in Polymer Injectors in Daqing Oilfield,” paper SPE 103703
presented at the 2006 SPE International Oil & Gas Conference and Exhibition, Beijing, China (December 5–7,
2006).

51
THANK YOU