1

PRODUCTION LOGGING MEASUREMENTS AND

ANALYSIS

I. Introduction

Production logging encompasses a number of cased hole logging measurements

usually run in completed injection or production wells to evaluate the performance of the
well itself or of the reservoir. Although the most common application of production
logging is to obtain the well flow profile, a measure of the distribution of fluid flow into
or out of the wellbore, there are numerous other applications, such as detecting channels
or leaks. In .this article, the primary production logging measurements and interpretation
procedures will be reviewed; the application of production logging to well and reservoir
behavior diagnosis is treated in Chapter 14.
The chapter is organized according to the logging environment or logging
objective. First, the use of temperature, radioactive tracer, and spinner flowmeter logs for
measuring flow profiles in single-phase flow is described. Then, the more difficult
application of logging in multiphase flow is treated, introducing the basket flowmeter,
density, and capacitance logs that are used in this environment. Finally, production
logging techniques for completion evaluation are considered. For a more complete
treatment of the logging methods presented here, the reader is referred to Hill [1990].

II. Production Logging Measurements in Single-Phase Flow

2.1 Temperature Logging

The temperature log is probably the simplest, most accurate, and most widely
applicable production log. A temperature log is a measurement of the temperature in the
well as a function of depth -a typical temperature log from a gas production well is shown
in Figure 1 [Atlas Wireline Services, 1982.]
 2

Fig. 1 Typical temperature log displays

Notice that two curves are displayed, with one being the temperature versus depth
(usually called the gradient curve) and the other being the derivative of temperature with
depth (the differential curve). The differential curve is recorded to accentuate changes in
temperature behavior. Often, temperature logs will be run both with the well flowing and
with the well shut-in, since a shut-in period will sometimes result in larger temperature
anomalies at injection or production intervals.
The temperature in a well depends on many factors, including the temperature of
the surrounding formations, the wellbore flow conditions, the heat transfer characteristics
of the completion, and fluid movements near the wellbore. The natural temperature
distribution in the earth is called the geothermal temperature profile. Because of the heat
transfer from the earth’s interior to the atmosphere, the temperature in the earth’s crust
increases with depth, leading to a geothermal temperature profile such as that shown in
Figure 2 [Connolly, 1965] from a region in Canada.
 3

Fig. 2 Geothermal temperature profile, Leduc area, Canada

The geothermal temperature profile varies significantly from area to area, and the
slope of the geothermal temperature (the geothermal gradient) varies from formation to
formation. Knowledge of the geothermal temperature profile is often necessary for
temperature log interpretation. Also, the geothermal gradient is generally assumed to be
constant when interpreting temperature logs. When significant variations in lithology
 4

occur, as in Fig. 2, the variation in geothermal gradient should be considered.

When fluids are injected or produced through a wellbore, the temperature is
perturbed from the geothermal temperature. In general, fluid injection will cool the
wellbore relative to the surrounding formations, while fluid production leads to wellbore
temperatures that are higher than the surroundings, as illustrated in Fig. 3 [Hill, 1990.]

Fig. 3 Temperature behavior in injection or production wells

Ramey [1962] presented a model for the temperature in a well away from any
injection or production intervals as follows:

dT T − TG
= (1)
dz AR

where

[ ]
AR = mCPf λ + f (t )r1U
2Πλr1U
(2)

In these equations, T is the wellbore temperature, TG is the geothermal

 5

temperature, z is depth, m is the mass flow rate, Cpf is the heat capacity of the wellbore
fluid, λ is the thermal conductivity of the formation, r1 is the radius of the wellbore flow
path (the tubing inside radius, or the casing inside radius, if no tubing is present), U is the
overall heat transfer coefficient of the completion, and f(t) is a time function. If the
geothermal temperature varies linearly with depth (constant geothermal gradient), the
solution to Equation 1 is

T = gGz + Ts − gGAR + [Tο (t ) + gGAR − Ts ]e − z / AR (3)

where Ts is the formation temperature at z=0, gG is the geothermal gradient, and To(t) is
the wellbore fluid temperature at z=0. The function f(t) in Eq. 2 depends on the form of
the boundary condition assumed and is given graphically in Fig. 4.

Fig. 4 Time function in the Ramey equation

For later times (log (αt/r22 > 2.5), all solutions converge to the line-source solution, and
f(t) is given approximately by

r2
f (t ) = − ln − 0.290 (4)
2 αt
 6

where α is the formation thermal diffusivity, equal to λ/ρCp, and r2 is the casing outside
radius. Equation 4 can be used to calculate the wellbore temperature as a function of
depth above the zones where fluid is entering or leaving the wellbore while Eq. 1 can be
used directly to interpret temperature logs.

Example 1 Calculation of Wellbore Temperature Profile

Calculate the wellbore temperature as a function of depth from the surface to

10,000 ft after 60 days of injection of 500 b/d of fresh water. The following conditions
are known:

casing OD = 7 inch tubing ID = 2.5 inch

λ = 5 Btu/hr-ft-◦F α = 0.06 ft2 /hr U = 300 Btu/day-ft2◦F
geothermal temperature = 68◦F + (0.01 ◦F/ft) (z)
water temperature at the surface = 80◦F

Solution
Using Eq. 3, the wellbore temperature at a number of depth locations can be
calculated to determine the wellbore temperature profile. First, f(t) is determined by
calculating log(αt/r22):

 ft 2   24hr  
   0 . 06 (60days ) 
 αt 
  hr   day  
log   = log   = 3.01 (5)
r 2   3 . 5 
2

    ft
2
 2   12  


If log (αt/r22) is greater than about 2.5, f(t) can be calculated with Eq. 4:

  3.5  
   
  12 
f(t) = -ln   − 0.290 = 3.86 (6)
 2 (0.06 )(60 )(24 ) 
 
Next, AR is calculated with Eq. 2.
 7

 ft 3  lb  day  lb
m = (500b / d ) 5.615  62.4 m3   = 7,300 m (7)
 bbl  ft  24hr  hr

 lb  Btu   
+ (3.86 )(0.104 ft )12.5
Btu Btu
 7,300 m 1 5 
  lbm − ° F  hr − ft − ° F − − °
2
hr  hr ft F 
AR = = 1,788 ft
 Btu  1.25  Btu 
2π  5  ft 12.5 
 hr − ft − ° F  12  hr − ft 2 − ° F 
(8)
Substituting in Eq. 3, the temperature as a function of depth is

T = 0.01z + 68 − (0.01)(1,788) + (80 + (0.01)(1,788) − 68)e − z / 1788 (9)

The calculated wellbore temperature profile is shown in Figure 5.

Fig. 5 Temperature profile calculated with the Ramey equation

The Ramey solution (Eq. 1 or 3) does not account for convective heat transfer
occurring with fluid movement in the formation and thus cannot be applied opposite
injection or production zones. No simple solution exists for the temperature behavior in
such regions, and qualitative methods are generally applied to interpret temperature logs.
In injection wells, the injected fluid will usually be cooler than the reservoir. The
injection of large volumes of cool fluid into the reservoir creates a cool region around the
 8

wellbore that can be detected with temperature logs. When the well is shut-in, the
wellbore will begin to warm towards the geothermal temperature, but in the injection
zones, this warming is slower than elsewhere because of the bank of cool fluid around the
wellbore. Fig. 6 illustrates injection zones can be identified as cool anomalies on a shut-in
temperature log. Although the base of the injection interval is easily identified on both
the flowing and shut-in temperature profiles, the top of the injection interval can usually
only be identified with the shut-in log.

Fig. 6 Flowing and shut-in logs for an injection well

In production wells, production zones may or may not be clearly identifiable on a

temperature log, depending on fluid properties and fluid entry rates. When a free gas
phase is flowing in the reservoir, significant cooling of the gas may occur in the near-
wellbore vicinity due to Joule-Thomson cooling. When this occurs, gas entries are
identified as the locations of cool anomalies on a temperature log (Fig. 7 [Kunz and
Tixier, 1955]). Joule-Thomson cooling will be more significant for relatively low
bottomhole pressures, high production rates, and high drawdown.
 9

Fig. 7. Gas-entry locations identified by cool regions on a temperature log

In oil production wells, the temperature of the oil entering the wellbore will be the
geothermal temperature at the depth or slightly higher because of frictional heating. The
temperature will vary little across individual production intervals; however, there may be
significant changes between production zones. An estimate of the flow profile can be
obtained from these changes by using the Ramey equation (Eq. 1) or by applying an
energy balance.

The Romero-Juarez Method

Romero-Juarez [1969] pointed out that Eq. 1 can be rearranged to estimate the
 10

flow rate from a temperature log. If the fluid properties and heat transfer characteristics
of all zones are the same, substituting 2 into 1 and applying the result at different
wellbore locations, we have

 
T −T 
 G 
 dT 
 
q i  dz  i
=
qo  
T −T 
 G 
 dT 
 
 dz  o
(10)
where i denotes any wellbore position and o denotes a location above the uppermost
production or injection zone. Since the Ramey model does not apply in or near a
production or injection interval, the locations for log interpretation must be selected far
enough from these zones to avoid their influence.

The Mixing Method

When fluid enters the wellbore and mixes with the fluid stream from below that
location, and if the heat capacities of all the fluids are the same, an energy balance on the
mixing zone yields

mi T A − TB
=
mo Ti − TB
(11)
where mi is the mass flow rate from the Zone i, mo is the mass flow rate above the mixing
zone, TA and TB are the temperatures above and below the mixing zone, respectively, and
Ti is the temperature of the fluid room Zone i. TA and TB can be read from the
temperature log; Ti is the geothermal temperature of Zone i if no significant Joule-
Thomson cooling or frictional heating is occurring.

Example 2 Interpreting the Flow Profile with a Temperature Log

The temperature log is shown in Fig. 8 was obtained in a well producing oil and
water. Estimate the fraction of flow from each perforated zone (the flow profile) using
the Romero-Juarez Method and the Mixing Method.
 11

Fig. 8 Example 2 temperature log

Solution
Romero-Juarez Method
The analysis consists of the following steps:
1. Choose stations for interpretation. These should be approximately midway
between perforated zones.
2. Measure the slope of the temperature log, dT/dz at each station.
 12

3. Measure the difference between the wellbore temperature and the geothermal
temperature, T-TG, at each station.
4. Calculate the fraction of flow at each station using Eq. 10.

For the example log, the middle set of perforations did not appear to be producing
any fluids, so stations were chosen at 5690 ft, above the top perforations, and at 5740 ft,
approximately midway between the top and bottom perforations. In this well, the
geothermal temperature could be extrapolated from the temperature log response in the
rathole below the lowest perforations. The values of dT/dz and T-TG for each station were
then obtained as shown in Fig. 9 and are summarized in Table 1.

Fig. 9 Romero-Juarez interpretation of Example 2 log

Table 1
 13

Romero-Juarez Temperature Log Interpretation (Example 2)

Station Depth dT/dz T-TG AR % total flow

◦ ◦
(ft) F/ft F ft
A 5690 0.0033 1.1 333 100
B 5740 0.008 0.4 50 15
The interpretation shows about 85% of the total flow coming from the top zone
and about 15% from the bottom zone.

Mixing Method
Applying the mixing method at the bottom of Zone 1, TA and TB are read from
the temperature log and Ti determined from the estimated geothermal profile, as shown in
Fig. 10. The fraction of flow from Zone 1 is then calculated with Eq. 11. Zone 1 is
estimated to contribute 80% of the total flow, compared with 85% found with the
Romero-Juarez method.

Fig. 10 Mixing method interpretation of Example 2

 14

2.2 Radioactive Tracer Logging

Radioactive tracer logs are most commonly used to measure flow profiles in water
injection wells, though they are also sometimes used in steam or CO2 injection wells and
in production wells. A radioactive tracer log is run by ejecting from the tool a small
amount of radioactive material that is miscible with the wellbore fluid. The tracer mixes
with the wellbore fluid and moves down the wellbore while its movement is monitored
by one or more γ-ray detectors. The two most common logging methods are the tracer
loss method and the velocity shot method.

Tracer Loss Method

In the tracer loss method, a relatively large tracer slug is ejected above all injection zones
and its movement tracked by passing the γ-ray detectors repeatedly through the slug.
Thus, a tracer loss log will consist of a number of recordings of high γ-ray intensity that
follow the path of the tracer down the wellbore fluid, the area under each γ-ray intensity
curve is proportional to the wellbore flow rate at that location. The interpretation consists
of measuring the areas under the γ-ray intensity curves; then

qi A
= i
q o Ao
(12)
where qi is the flow rate at the depth of the γ-ray curve peak, qo is the flow rate above all
injection zones, and Ai and Ao are the areas under the γ-ray intensity curves
corresponding to these depth locations. Since all calculations are normalized with the
area under a γ-ray intensity curve above all injection zones, it is important that a good
measurement of the tracer slug (preferably two or more) is made above the injection
intervals.

Velocity Shot Method

To run a velocity shot log, the radioactive tracer logging tool is positioned stationary at a
number of depth locations; at each location, a small shot of tracer is ejected and its travel
time between two detectors is measured. The travel time is usually measured either as the
difference between the arrival times of the first detectable tracer at each detector (the
leading edge travel time) or as the difference between the arrival times of the γ-ray
intensity peak at each detector (the peak to peak travel time). If the flow is turbulent, the
 15

leading edge travel time measures the maximum velocity of the flow stream, while the
peak to peak travel time corresponds to the average velocity. For a discussion of tracer
logging in laminar flow, see Hill [1990].
The volumetric flow rate is calculated from the travel time as

π (D 2 − Dt2 )L
q=
4∆t pp
(13)
when the peak to peak travel time is measured, or by

π (D 2 − Dt2 )L  u 
q=  
4∆t le u 
 max 
(14)
when the leading edge travel time is used. In these equations, D is the casing ID,
Dt is the logging tool OD, and L is the spacing between detectors. The ration, u / u max , in
Equation 13-14 ranges from 0.75 to 0.86 and is usually taken to be 0.83 in production log
interpretation. If the wellbore cross-sectional area is constant and the flow is turbulent at
all velocity shot locations, the flow at any depth location is related to the flow rate above
all injection zones by

qi ∆t o
=
q o ∆t i
(15)
The flow rates calculated with velocity shot logs should generally be assigned to
the depth midway between the two detectors.

Example 3 Radioactive Tracer Log Interpretation

Determine the flow profile from the tracer loss log shown in Fig. 11 and from the
velocity shot data obtained from the same well, given in Table 3. For the velocity shot
log, the detector spacing is 5 ft and the travel times given are leading edge travel times.
The injection rate into the well is 460 b/d, the tool OD is 1-3/8 in., and the ID of the 4-1/2
in. casing is 4.09 in.
 16

Fig. 11 Tracer loss log

Solution
Tracer loss log
To determine the flow profile, the area under each γ-ray curve must be calculated.
Notice that a base γ-ray log was recorded – this background radiation level should be
subtracted from the area computed. The areas obtained by planimetry are shown in Table
2. The tracer slug was logged three times before it reached the first perforations, so the
first three γ-ray traces can be averaged to yield Ao. The fraction of total flow at each
tracer peak depth is then calculated with Eq. 12, with the results shown in Table 2. The
interpreted flow profile is sketched in Figure 12.
 17

Fig. 12 Flow profiles interpreted from radioactive tracer logs

Table 2
Tracer Loss Log Results (Example 3)
depth area of γ-ray curve % of total flow (qi/qox100)
3735 21.32 100.0
3750 22.24 average = 23.16 100.0
3765 25.92 100.0
3797 23.76 103.0
3826 23.84 103.0
3842 17.88 75.0
3860 16.20 68.0
3883 13.56 56.5
3898 10.00 41.5
3912 9.68 40.0
3922 10.00 41.5
3936 6.36 26.5
 18

3944 0.16 0.01

Velocity shot log

Since the wellbore cross-sectional area is presumably constant throughout
the logged interval, the fraction of flow at each velocity shot location can be calculated
with Eq. 15; the results are given in Table 3. The flow profile is plotted in Fig. 12. It is
good practice to calculate the volumetric flow rate from a velocity shot measurement
above the perforations so that it can be compared with the surface rate (a large
discrepancy can indicate a tubing leak). For station 1, using Eq. 14 since the travel times
are leading edge travel times,

 
π (4.09 2 − 1.375 2 in 2 )(5 ft )
ft 2
 2
q=  144in
 (0.85) = 392b / d
 day  ft 
3
4(13.5 sec )  5.615 
 86,400 sec  bbl 
(16)
This compares fairly well with the surface rate of 460 b/d.

In general, the velocity shot log is more accurate than the tracer loss log, primarily
because of its better depth resolution.

Table 3
Velocity Shot Data and Analysis Results (Example 3)

Shot No. Depth-top detector Δt (sec) Δto/Δti qi/qo

1 3715 13.50 1.00 1.00
2 3745 14.25 0.95 0.95
3 3777 13.50 1.00 1.00
4 3795 13.50 1.00 1.00
5 3807 15.00 0.90 0.90
6 3827 13.50 1.00 1.00
7 3830 15.90 0.85 0.85
8 3835 16.50 0.82 0.82
9 3840 18.30 0.74 0.74
 19

10 3853 22.20 0.61 0.61

11 3865 21.00 0.64 0.64
12 3877 21.00 0.64 0.64
13 3933 36.00 0.38 0.38
14 3957 234.0 0.06 0.04
15 3979 294.0 0.05 0.04
16 3999 ∞ 0.00 0.00

2.3 Spinner Flowmeter Logging

Another commonly used device for measuring flow profiles, both in injection and
production wells, is the spinner flowmeter. A spinner flowmeter is an impeller that is
placed in the well to measure fluid velocity in the same matter that a turbine meter
measures flow rate in a pipeline. Like a turbine meter, the force of the moving fluid
causes the spinner to rotate. The rotational velocity of the spinner is assumed linearly
proportional to fluid velocity and some electronic means are incorporated in the tool to
monitor rotational velocity and sometimes direction.
The preferred method of running a spinner flowmeter log involves making several
passes of the spinner at different tool speeds, moving both upwards and downwards
through the region to be logged. It is also helpful to make a series of measurements with
the tool held stationary at different depth locations. With multiple passes of the spinner
through the well at different cable speeds, an in-situ calibration of the spinner response is
obtained; this method of running and interpreting a spinner flowmeter log is called the
multipass method.
The multipass interpretation is based on the linear spinner response,

f = m(u f + u t − uT ) (17)

where f is the spinner response, usually recorded in rps, m is the slope of the spinner
response versus cable speed curve, uf is fluid velocity, ut is tool velocity (cable speed),
and uT is the threshold velocity. Solving for the fluid velocity when the tool velocity is
zero yields

fo
uf = + uT (18)
m
 20

where fo is the intercept of the spinner response curve at ut = 0. With the spinner
centralized, as it always should be, the velocity measured is near the maximum velocity.
The volumetric flow rate is then

 u 
q = Au f   (19)

 u max 

where A is the wellbore cross-sectional area. Thus, to interpret multiple passes of a

spinner flowmeter log using Eqs. 18 and 19, we need the slope and intercept of the
spinner response curve and an estimate of the threshold velocity.
The threshold velocity is the velocity needed to force the spinner to begin turning
if the response is perfectly linear and is obtained from a plot of spinner response versus
cable speed. If the spinner has been run at a variety of cable speeds, both in the direction
of and opposite the direction of flow, a calibration such as that shown in Fig. 13 can be
obtained.

Fig. 13 Spinner flowmeter in-situ calibration plot

 21

The negative responses shown occur when the spinner is rotating in the opposite
direction from the runs that yield positive responses. For example, in an injection well,
when the spinner is being pulled up against the flow, the spinner will rotate in one
direction throughout the well (assume these are clockwise rotations). When the spinner is
moved downwards in the direction of flow, the spinner will experience a net downward
force and continue to rotate clockwise unless the fluid velocity is less than the tool
velocity. When the tool velocity exceeds the fluid velocity, the net force is upwards and
the tool will rotate counterclockwise. Since there is a threshold velocity for each direction
of rotation, when all responses are plotted as in Fig. 13, the difference in the calibration
curve intercepts on the cable speed axis is approximately twice the threshold velocity, or

u tp − u tn
uT = (20)
2

where utp and utn are the intercepts of the positive and negative spinner response curves
with the f=0 axis, respectively.
The threshold velocity is usually obtained in the lower part of the well where the
lower fluid velocity makes it easier to obtain data with both clockwise and
counterclockwise rotations of the spinner. It is not good practice to measure the threshold
velocity in the rathole, since the rathole fluid may be different from the produced fluid.
The threshold velocity can also be measured with the well shut-in. In this case, the
threshold velocity should be measured above all injection or production intervals to avoid
the effects of crossflow that may occur between zones in a shut-in well.
The steps in the multipass method are summarized below.
1. Choose the stations (depth locations) at which the flow rates will be
calculated. As a minimum, a station should be chosen between each
perforated interval.
2. Read the spinner response at different cable speeds at each station.
3. For each station, plot spinner response versus tool velocity (cable speed).
4. Calculate the slope, mp or mn, for each response line.
5. For each station where positive and negative spinner responses occur,
determine the threshold velocity, uT, from Eq. 20.
6. Calculate the fluid velocity at each station from the intercept of the response
line at ut = 0, applying Eq. 18.
7. Convert the fluid velocities to volumetric flow rates with Eq. 19.
 22

Example 4 Spinner Flowmeter Interpretation with the Multipass

Method

The spinner flowmeter log shown in Fig. 14 was obtained in a water injection
well with 6 in. ID casing using a Schlumberger Fullbore Flowmeter. Using the multipass
method, calculate the volumetric flow rates at 5780, 5800, 5820, 5845, and 5880 ft.

Fig. 14 Example 4 spinner flowmeter log

Solution
For the stations given, first the spinner responses at each cable speed are read.
 23

Since the fullbore spinner does not sense direction of rotation, this must be determined
from the character of the log so that positive or negative values can be assigned to the
spinner responses. Since the spinner will rotate in the same direction throughout the well
when it is pulled against the flow direction, it is convenient to assign positive values to
the spinner responses on those passes. For this injection well, this means positive values
are assigned to all spinner responses from the up passes.
For the down passes, the spinner may be rotating in either direction. Notice that
on the down run at a cable speed of 40 ft/min, the spinner response is low in the upper
part of the well, drops to zero, then begins to increase in the lower part of the well. This
behavior indicates that the spinner has reversed directions. In the upper part of the well,
the fluid velocity is greater than the tool velocity and the spinner is rotating in the same
direction as in the up passes; in the lower part of the well, the spinner is rotating in the
opposite direction and negative values are assigned to the responses. At the higher cable
speed in the down direction, the spinner response is negative throughout the well. The
spinner responses are summarized in Table 4.

Table 4
Spinner Flowmeter Responses (Example 4)
Cable Speed
Station up 30 ft/min up 60 ft/min down 40 ft/min down 80 ft/min
5780 2.85 4.10 0.08 -0.88
5800 2.50 3.68 0.00 -1.16
5820 2.08 3.26 -0.08 -1.68
5845 1.42 2.77 -0.50 -2.32
5880 0.85 2.10 -1.28 -2.85
 24

Fig. 15 Multipass method in-situ calibration plot

Next, the tool responses are plotted against cable speed for each station as in Fig.
15. From the station at 5880 ft the threshold velocity is found with Eq. 20 as

ft  ft 
9.6 −  − 7.4 
min  min  ft
uT = = 8.5 (21)
2 min

For each station, the slope, mp, and the cable-speed axis intercept are determined and the
fluid velocity and volumetric flow rate are calculated from Eqs. 18 and 19. For example,
at 5780 feet, mp = 0.040 rps/ft/min, fo = 1.67 rps, and

1.67 rps ft ft
uf = + 8.5 = 50.2
rps min min
0.04
ft / min
(22)

ft  π  6    bbl  1,440 min 

2

q =  50.2   ft  (0.83) 
3 
 = 2,100b / d
 min  4  12    5.615 ft  day 
 25

(23)
The results are summarized in Table 5.

Table 5
Multipass Interpretation Results (Example 4)

station mp mn fo vf q
rps/ft/min rps/ft/min rps ft/min b/d
5780 0.040 -- 1.67 50.2 2100
5800 0.039 -- 1.32 42.1 1800
5820 0.039 0.040 0.90 31.4 1300
5845 0.045 0.046 0.07 10.1 400
5880 0.042 0.039 -0.40 0 0

This log had the minimum number of passes needed to apply the multipass
method- two in each direction. Preferably, three or more passes are run in each direction.
Notice that there was some variation in the spinner response curve slopes, with mp
ranging from 0.039 to 0.045. This variation is caused by the tool itself or unsteady flow
conditions in the well and is an indication of some inaccuracy in the log. In the
interpretation, the slope obtained at each station was used to calculate the flow rate at that
station. However, often an average value of mp is used for all stations. Finally, at 5880 ft,
the intercept fo, was negative. This indicates that the fluid velocity at that depth is below
the threshold velocity and Eq. 18 no longer applies. The fluid velocity is assumed to be
approximately zero in this instance.

III. Production Logging Measurements in Multiphase Flow

In production wells, most of the time two or more phases flow simultaneously.
Water is produced in many wells and free gas flows in the wellbore whenever the
wellbore pressure is below the bubble point. The goal of production logging in
production wells is to determine the locations and rates of oil, water, and gas entries to
the wellbore, a much more difficult task than the measurement of a single-phase flow
profile. In multiphase flow, in addition to measuring velocity or total flow rate, it is
necessary to make measurements of the amount of each fluid present in different
 26

locations in the wellbore. These measurements are interpreted using models of two-phase
flow behavior to estimate the flow distribution into the wellbore.

3.1 Fluid Velocity Measurement

The same radioactive tracer and spinner flowmeter logging techniques used in
single-phase flow are applied in multiphase flow. A third device, a flow-concentrating
flowmeter, is also used to measure flow velocity in production wells.
Radioactive tracer logging (the velocity shot method) can be used to measure
velocity in production wells. Because radioactive material is produced to the surface,
these techniques are seldom used in production wells. However, by taking precautions
such as storing the produced fluid until the level of radioactivity has decayed to a safe
level, radioactive tracer logs can be used in production wells. In a multiphase stream, a
radioactive tracer will most likely move with the continuous phase. Thus, a radioactive
tracer log will provide a measure of the in-situ average velocity of the continuous phase,
not the average velocity of the total flow stream.
The same spinner flowmeters used in single-phase flow logging are employed to
measure velocity in multiphase flow. The spinner flowmeter’s efficacy is greatly
diminished in multiphase flow, however, because of the complexity of the flow itself.
One major difficulty is that a spinner flowmeter measures a localized velocity that does
not necessarily represent the average flow velocity. Since the velocity profile in
multiphase flow is much less predictable than that in single-phase flow, the small
sampling area of a spinner flowmeter can be a serious detriment. Particularly when the
well is inclined, the velocity distribution in multiphase flow can be highly non-uniform,
causing serious interpretation problems with spinner flowmeters. Velocity fluctuations in
multiphase flow also contribute to the inaccuracy of spinner flowmeters.
Spinner flowmeters will yield better results in multiphase flow the more
homogeneous the flow stream; e.g., when relatively small bubbles of one phase are
distributed relatively uniformly in another phase, a spinner flowmeter can be expected to
perform well. In general, spinner flowmeters will be more reliable in high velocity flow
streams in multiphase flow. No absolute guidelines are available at present to predict
what flow rates are sufficient for accurate use of spinner flowmeters. When running a
spinner flowmeter in a multiphase well, the standard single-phase interpretation methods
are often unreliable. It is often best to first make a few passes with the spinner flowmeter
in the well to get a general picture of the flow behavior, then make a series of stationary
measurements. With the spinner stationary, the response can be averaged over time,
 27

eliminating some of the error caused by flow fluctuations.

The best devices currently available to measure velocity in multiphase flow when
the volumetric flow rates are relatively low are flow-concentrating flowmeters, tools that
force all flow to pass through a small chamber containing a turbine flowmeter. By forcing
the flow into a much smaller diameter flow path inside the tool, the fluid velocities are
increased enough to make the flow stream relatively homogeneous and thus the turbine
meter responds linearly to total velocity. Two types of flow concentrating flowmeters
have been used – the packer flowmeter and the basket flowmeter. Currently, the basket
type flowmeter is the most commonly used flow concentrating flowmeter.

3.2 Fluid Density Measurement

In two-phase flow, the in-situ average density, ρ is related to the holdup by

ρ = y1ρ 1 + (1 − y1)ρg (24)

for gas-liquid flow, or

ρ = ywpw + (1 − yw )ρo (25)

for oil-water flow. (See Chapter 7 for a discussion of holdup and other general features of
two-phase flow). Thus, a measurement of the average density of the flow stream provides
a measure of the holdup of each phase in a two-phase flow. Two types of tools are used
to measure fluid density-those based on γ-ray absorption and those based on a pressure
drop measurement.
The most common type of fluid density tool is the γ-ray densitometer. This tool is
based on the fact that γ-ray absorbance is inversely proportional to the density of the
material through which the γ-rays are passing. The device consists of a γ-ray source (a
collimated beam), an open space through which wellbore fluid can pass, and a γ-ray
detector. A collimated beam is used to prevent backscatter from the casing wall. The tool
response decreases logarithmically as the fluid density increases so that

log K = a − b ρ (26)

where K is the count rate (HZ), a and b are constants, and ρ is the fluid density.
A γ-ray densitometer has several inherent limitations, including the statistical
 28

nature of the measurement, small sampling size, and low sensitivity in oil-water flow.
Like any nuclear measurement, there will be statistical fluctuations in γ-ray densitometer
readings because the γ-ray source does not emit radiation at a constant rate. This effect
can be minimized by using a strong γ-ray source and by averaging the tool response over
a finite time period. Thus, stationary measurements are advantageous with a γ-ray
densitometer so that the response can be averaged to eliminate statistical fluctuations.
Another major problem with a γ-ray tool is the small sampling size. As with a
spinner flowmeter, a γ-ray densitometer is sensing density of only a small portion of the
flow stream and thus the density measured may not represent the average fluid density. If
the flow regime is highly nonuniform, as in inclined flow, a γ-ray densitometer may yield
a reading that has no relation to average density.
As with any density tool, sensitivity is low in an oil-water stream in which there is
a small difference between the densities of the oil and the water. A density tool will most
clearly distinguish between gas and liquids.
The primary density tool based on a measurement of pressure drop is the
Schlumberger GradiomanometerTM .The Gradiomanometer determines average fluid
density by measuring the pressure differential across two feet of wellbore. Assuming that
ΔpKE, the pressure gradient due to kinetic energy change is negligible, and that the
frictional pressure gradient, ΔpF, is small compared to the hydrostatic head term, then

∆p = g ρ cos θ (27)

where θ is the wellbore deviation from vertical. Thus, for these conditions, the average
fluid density, ρ , can be determined from a measurement of pressure gradient in the
wellbore.
If the frictional pressure drop is significant compared to the gravitational term, the
fluid density calculated from Eq. 27 will, of course, be in error. Using ρGR to represent
the apparent density from the gradiomanometer reading,

gρGR = g ρ + ∆pF (28)

written in practice as

ρGR
ρ= (29)
1+ F
 29

where F is a correction factor to account for frictional pressure losses. Schlumberger has
empirically determined values of F for flow rates greater than 2000 b/d.

3.3 Fluid Capacitance Measurements

To overcome the problems inherent in fluid density tools in distinguishing

between oil and water, another class of tools has been developed to measure the water
fraction in multiphase flow more accurately. These devices are based on a measurement
of electrical capacitance and are sometimes referred to as holdup meters or watercut
meters.
Capacitance tools are essentially coaxial capacitors. By applying a voltage
potential between a central electrode and the outside of the logging tool, the capacitance
of the device is determined. Since the capacitance measured is a function of the dielectric
constant of the fluids in the sample chamber, the capacitance tool provides a
measurement of the dielectric constant. Liquid hydrocarbons have dielectric constants on
the order of 2 to 6 while water has a dielectric constant of about 80 and thus a dielectric
constant measurement distinguishes well between hydrocarbons and water. Since the
dielectric constants of gases are near 1, the capacitance log does not discriminate well
between oil and gas.
The advantage, then, of a capacitance tool measurement is the wider separation
between the responses to oil and water, thus providing a better means of measuring water
holdup. Furthermore, in three-phase flow, a capacitance measurement in conjunction with
a density measurement can provide an estimate of the holdup of each phase.
Capacitance tools do not respond linearly to water fraction. When water is the
continuous phase in a multiphase flow, a capacitance tool has little sensitivity since a
continuous electrical path exists through the water. A typical calibration curve for a
capacitance tool is shown in Fig. 16 [Carlson and Roesner, 1982]. Notice that when the
input water fraction is greater than 0.5, the tool sensitivity to water fraction is greatly
diminished compared with the behavior at lower water fractions.
 30

Fig. 16 Capacitance tool flow loop calibration

The capacitance tool shares the problem of small sampling size with the density
and spinner flowmeter logs. Particularly in inclined wells a capacitance log may indicate
100% water even though a significant oil stream is present because the oil is flowing
along the upper side of the pipe and is undetected by the capacitance tool.

3.4 Production Log Interpretation in Multiphase Flow

Log interpretation in multiphase flow is generally based upon the measurement of the
average total velocity or mixture velocity, um, and the holdup, y, of the phases. When
three phases are present (oil, water, and gas), we assume that there is no slip between the
liquids, so that the oil and water can be treated as a single liquid phase. Thus, the analysis
is always for two-phase flow of gas and liquid or oil and water. Letting α denote the less
dense phase and β denote the denser phase, the volumetric flow rates of the phases are
obtained from

qα = A(1 − y β )(u s y β + u m ) (30)

 31

qt = Au m (31)
q β = qt − qα (32)

where qα, qβ, and qt are the volumetric flow rates of phase α, phase β, and the total flow
stream and us is the slip velocity. Applying these equations to oil-water flow, α is the oil
phase and β is the water phase, while in gas-liquid flow, α denotes the gas and β denotes
the liquid.
The slip velocity must be obtained independently from the log measurements.
Several methods of estimating the slip velocity are commonly used [Hill, 1990]. Two
methods are presented here – the correlation of Nicolas and Witterholt [1972] and the
method suggested by Curtis [1967].
The Nicolas-Witterholt correlation is based on laboratory measurements of slip
velocity in bubble flow and is given by

u s = ( y w ) u ∞ with1 / 2〈 n〈 2
n
(33)
where u∞, the terminal bubble rise velocity is

1
 gσ∆ρ  4
u ∞ = c  2  with1.53〈 c〈1.61 (34)
 ρ β 

In these equations, c is usually taken to be 1.53, n is usually assumed equal to one,

g is the acceleration of gravity, σ is the interfacial tension, Δρ is the difference between
the phase densities, and ρβ is the density of the denser phase. Hill [1992] found this
correlation to work particularly well for vertical oil-water flow, though it appears to
overestimate u∞.
The Curtis method is based on the knowledge of the flow rates of both phases
above the production zones from the surface flow rates. For example, in oil-water flow,

q o = Bo q os (35)
and
q w = Bw q ws (36)
where qo and qw are the downhole volumetric flow rates of oil and water above all
production zones, qos and qws are the phase volumetric flow rates at the surface, and Bo
and Bw are the formation volume factors of the oil and water. With a measurement of yw
above the perforations, Eqs. 30 through 32 can be solved for the slip velocity to yield
 32

qo q
us = − w (37)
A(1 − y w ) Ay w

The slip velocity calculated for the location above the perforations is then used to
interpret the log responses at all other locations; i.e., the slip velocity is assumed to be
constant throughout the well.

Example 5: Determination of the Flow Profile in a Two-Phase Production Well

A schematic of a well producing oil and gas and the production log data obtained
at five stations with a basket flowmeter and a gradiomanometer are shown in Fig. 17. The
volumetric flow rates at station 1 above all the perforations are determined from surface
conditions to be 2000 b/d oil and 10,000 ft 3 / d gas. Assume that all the gas enters the
wellbore as free gas. At bottomhole conditions, the oil density is 0.85 g/cc, the gas
density is 0.10 g/cc and the interfacial tension is 30 dynes/cm. Calculate the flow profile
for this well, first using the Nicolas-Witterholt correlation, then using the Curtis method.

Fig. 17 Example 5 schematic and log values

Solution
Nicolas-Witterholt
 33

Eqs. 30 through 32 apply. First yo and um are determined at each station from the log data.
Rearranging Eq. 24 to solve for holdup and applying the result for station 1,

ρ − ρg 0.64 − 0.1
yo = = = 0.72 (38)
p o − ρ g 0.75 − 0.1
Similar calculations are performed for the other stations.

Basket flowmeters are usually assumed to respond linearly to total average

velocity (um) with negligible threshold velocity. Thus, the mixture velocity at any station
can be obtained from the known mixture velocity at a station above the perforations. For
station 1,

  ft 3  ft 3 
 (2,000b / d )
 5 . 615  + 10, 000 
qo + q g   bbl  day  1day  ft
u m1 = = 2   = 73.7
A  0.2 ft  1,440 min  min
 
 
(39)
At any other station i,

fi
u mi = u m1 (40)
f1

where fi and f1 are the flowmeter responses at stations i and 1, respectively.

To calculate the slip velocity, u∞ is calculated with Eq. 34:

1
  lbm  4
  
  32.2
ft  dynes  sec 2 (0.85 − 0.1) 62.4 lbm  
 30  0 . 0022
 sec 2  cm  dyne  
 ft 3  
   
 cm  ft ft
u ∞ = 1.53 2  = 0.66 = 39.8
  lbm   sec min
  (0 . 85 )(62 . 4 )  

 ft 3 
 
 
 
(41)
Then the slip velocity at each station is calculated with Eq. 33. Finally, the volumetric
 34

flow rates are calculated with Eqs. 30 through 32. For example, for station 1,

 ft 
u s = (0.72 ) 39.8
ft
 = 28.7 (42)
 min  min
 ft  1,440 min 
( () ) ( )
q g = 0.2 ft 1 − 0.72  0.72  28.7
2 ft 
 + 74
min 

min  day 
 = 7,600
ft 3
(43)
  day
ft  1,440 min 
( 
)
qT = 0.2 ft 2  74   = 21,200
ft 3
(44)
 min  day  day
ft 3 b
q o = 21,200 − 7,600 = 13,600 = 2,400 (45)
day d

The results for all stations are given in Table 6.

Table 6

Nicolas-Witterholt Interpretation Results (Example 5)

station flowmeter density yo um us qT qg qo qo

(rps) (g/cc) (ft/min) (ft/min) (ft3/d) (ft /d) (ft3/d)
3
(b/d)
1 85 0.64 0.72 74 28.7 21200 7600 13600 2400
2 62 0.52 0.56 54 22.3 15500 8400 7100 1300
3 46 0.62 0.69 40 27.6 11500 5200 6300 1100
4 23 0.80 0.93 20 37.1 5700 1000 4700 800
5 0 1.05 1+ 0 - 0 0 0 0

Curtis Method
With this method, the slip velocity above the perforations (station 1) is calculated
from the known rates at this location using Eq. 37. This slip velocity is then used at all
other stations to calculate the volumetric flow rates of the phases. The total average
velocity, um, and hence, the total volumetric flow rate will be the same for this method as
with the Nicolas-Witterholt method. Thus,

 ft 3  ft 3  
10,000 (2, 000b / d )
 5 . 615 
1  day  bbl   day ft
us = 2 
−  = 69.8
0.2 ft  (1 − 0.72 ) 0.72  1,440 min min
 

(46)
 35

Using this value of us, qg, qo, and qT are calculated with Eqs. 30 through 32; the results
are shown in Table 7.

Table 7
Curtis Method Interpretation Results (Example 5)

station qT qg qo qo
(ft3/d) (ft3/d) (ft3/d) (b/d)
1 21200 10000 11200 2000
2 15500 11800 3700 700
3 11500 7800 3700 650
4 5700 1700 4000 700
5 0 0 0 0

The flow profiles obtained are plotted in Fig. 18. The profile is presented as a plot
of total flow rate and a plot of oil flow rate, with the difference between these values
being the gas flow rate. A few points about these results are worth noting. First, at station
5, the flowmeter reads 0 rps and the density tool shows a density higher than the oil
density. These results show this to be the rathole region, indicating no oil or gas is
flowing at this depth.
 36

Fig. 18 Example 5 two-phase log interpreted flow profile

Comparing the results obtained with the Nicolas-Witterholt correlation with those
obtained with the Curtis method, qualitatively they are similar, with both showing the
majority of the oil production coming from Zones A and D. Both show the primary gas
production from Zones B and C. However, notice that with the Curtis method, Zones B
and C are interpreted to produce essentially no oil, while with the Nicolas-Witterholt
interpretation, over 400 b/d of oil is found to be produced by these zones. Unfortunately,
it is not possible to know which is more correct, as both methods are empirical
correlations of complex two-phase flow phenomena.

IV. Completion Evaluation with Production Logs

Production logs are often used to inspect the well completion. Locating tubing,
casing, or packer leaks, detecting channels behind pipe, and evaluating the condition of
the cement are all common applications. In this section, we will review methods for
locating channels or leaks followed by a discussion of cement evaluation.

Channel and Leak Detection

 37

Channels through the cemented region behind the casing can be located with
temperature, radioactive tracer, and noise logs while leaks in casing or tubing are
routinely detected with noise and radioactive tracer logs. Sometimes, a comparison of
two or more production logs is the best means of locating leaks or channels.
A temperature log can sometimes indirectly detect channeling in an injection well
by locating a cool anomaly away from the intended injection intervals. If there is no
means for fluid to exit the wellbore at the location of the anomaly (such as through a
casing leak), the fluid cooling the formation apparently traveled to the cool region
through a channel.

Example 6 Channel Detection with a Temperature Log

Figure 19 [Hill, 1990] is an 18 hour shut-in temperature log from a well with a
suspected channel. Looking first at the perforated interval, no cool anomaly has
developed in this region, so it appears that little of the injected fluid is entering the
formation opposite the perforations. The cool anomaly from 4790 ft to 4844 ft suggests
that fluid is entering the formation over this interval after channeling up from the
perforations beginning at 4848 ft. The cool anomaly at 4924 ft-4950 ft is likely due to
injection into the openhole section of the well at this depth. On this particular log, the
flowing temperature log indicates where fluid leaves the wellbore as the regions where
temperature is increasing more rapidly with depth, such as from 4855 to 4865 ft.
However, in most injection wells, the flowing temperature log will show little definition
across the overall injection interval.
 38

Fig. 19 Temperature log showing channeling behind casing

Radioactive tracer logs can also be used to find leaks and channels. Since the
gamma-rays generated by the radioactive tracer can penetrate 1-2 ft, it is possible to
observe fluid movement outside of the casing with tracer loss logs. If the fluid movement
can be clearly identified as being outside the casing, the tracer loss log can give a positive
indication of channeling.
Channeling is identified on a tracer loss log by the development of a secondary
peak of tracer concentration (gamma-ray intensity). Figure 20 illustrates this technique
[Schlumberger, 1973]. After tracer enters the perforations at Sand #3, a secondary peak
develops which moves back up the well (f, j, n, v). This movement indicates fluid
channeling up the casing-formation annulus to Sand #4. Also, tracer is detected moving
below the lowest perforated interval at Sand #2 (1, p). Presumably, this movement is due
to channeling down to Sand #1. Finally, the secondary tracer peak remaining stationary at
the packer is attributed to tracer caught in the hardware and in turbulent eddies and is not
an indication of channeling.
 39

Fig. 20 Hypothetical behavior of a tracer loss log in a well with channeling behind pipe

The noise log is one of the most positive means of detecting leaks or channels. A
noise log is simply a record of a passive measure of the audible sound detected by a
sensitive hydrophone at a number of locations in the wellbore. Since sound is generated
by fluid turbulence, high noise amplitudes indicate locations where the flow path is such
that additional turbulence is developed. Fluid moving through restricted channels, leaks,
flow from perforations, and flow past the logging sonde are among the phenomena that
can produce characteristic sounds in the wellbore and thus may be detected with a noise
log. Analysis of the frequency characteristics of the measured noise can distinguish
between the various possible sources of high sound amplitudes.
The noise log has been used primarily as a qualitative indicator of channeling
behind pipe [McKinley et al., 1973]. Flow in a channel is indicated on a noise log by the
presence of high amplitude noise at places where restrictions in the channel cause
 40

throttling of the fluid, as shown in Fig. 21 [Atlas Wireline Services, 1982]. Similarly,
flow through a leak results in a pressure drop that generates detectable noise.
Fig. 21 Noise

Fig. 21 Noise log

Example 7 Detection of a Leak with a Noise Log

Figure 22 [Hill, 1990] shows a noise log run in a well with a suspected packer
leak allowing flow into the annulus. The log shown in Fig. 22 was run with 1200 psi
surface pressure on the annulus; with this backpressure, no leak is evident as no noise
anomalies occur at the packer location. Pressure was then bled off the annulus and the
noise log repeated (Fig. 23 [Hill, 1990]). With low pressure in the annulus, a leak at the
packer was clearly indicated by the noise amplitude peaks at the packer location.
 41

Fig. 22 Noise log to detect a packer leak – pressure on the annulus

 42

Fig. 23 packer leak indicated on noise log with annulus pressure bled off

Sometimes, a combination of logs is needed to clearly distinguish between leaks

and channeling behind pipe. Consider, for example, the log responses that would be
obtained in an injection well with a channel downward from the lowest perforations and
another injection well with a casing leak below the lowest perforations [Hill, 1990]. The
temperature log responses that would be expected in these situations are shown in Figs.
24 and 25.
 43

Fig. 24 Temperature logs for injection well with a downward channel

 44

Fig. 25 Temperature logs for injection well with a casing leak below the bottom zone

The temperature logs are identical, as the temperature responds primarily to where
the fluid enters the formation. Spinner and radioactive tracer logs for these wells are
shown in Figs. 26 and 27. Now differences are seen between the logs. When channeling
is occurring, the spinner and velocity shot logs detect no flow below the perforations
(Fig. 26), while a casing leak results in flow in the wellbore past the bottom of the
perforations, as detected by the spinner or velocity shot log. In the case of channeling, the
spinner or velocity shot log alone cannot find the anomalous well behavior. However,
when they are compared with a temperature or tracer loss log, channeling is conclusively
identified.

Fig. 26 Radioactive tracer logs for injection well with a downward channel
 45

Fig. 27 Radioactive tracer logs for injection well with a casing leak below the bottom
zone

4.2 Cement Evaluation

Acoustic logging techniques, primarily the cement bond log, have been used for many
years to try to directly measure the quality of the cement between the casing and the formation.
More recently, ultrasonic pulse-echo techniques have been developed in an attempt to eliminate
some of the deficiencies of the cement bond log for cement evaluation.
A primary function of the cement is to prevent fluid movement between the various zones
in a reservoir and between the reservoir and other zones up or down the hole. Thus, cement
quality logging is aimed at determining whether the cement is of sufficient strength and is
sufficiently distributed to prevent fluid communication between zones. Ideally, a cement quality
log should indicate whether the cement is bonded to the pipe, if the cement is bonded to the
formation, and if the channels are present in the cement. A cement quality log does not directly
measure the capability of the cement to prevent fluid communication – this is inferred from the
 46

degree of acoustic coupling of the cement to the pipe and the formation as measured by the logs.
For this reason, cement quality logs are not an absolute measure of the hydraulic integrity of the
cement; however, when run and interpreted properly, they have been shown to be generally
reliable predictors of cement placement.
The two primary logs for evaluating cement quality, the cement bond log and the
ultrasonic cement evaluation log, are both acoustic logs that differ primarily in the path taken by
the sound waves between the transmitter and the detector. With the cement bond log, sound
travels axially down the casing and through the cement and formation to detectors (usually two)
located below the sound source on the logging tool (Fig. 28 [Hill, 1990]). An ultrasonic cement
tool, on the other hand, has an array of transducers, or rotating transducers, that serve as both
transmitters and receivers of sound energy, so that the sound path is radial to and from the
transducers – Schlumberger’s Cement Evaluation Tool is pictured in Fig. 29 [Froelich et al.,
1982]. Alternatively, the ultrasonic logging devises have transducers mounted on a rotating
section of the tool . so that continuous acoustic scans of the cement conditions around the
borehole circumference can be made. Thus, the two types of logs are fundamentally different
measurements and must be treated separately.

Fig. 28 Cement bond log tool

 47

Fig. 29 Ultrasonic pulse-echo log tool

Cement Bond Log

A cement bond log usually records three separate measurements of the acoustic
energy received: the transit time, a measure of the time from sound transmission to the
first arrival of sound energy at the near detector; the amplitude, the amplitude of the first
wave arriving at the near detector; and the full wave train, a presentation of all the
acoustic energy received by the far detector for a short time period. The full wave train is
often displayed as a variable density log, constructed by rectifying the wave train and
assigning varying shades of gray to the waves, based on their amplitude. These logs yield
information about the acoustic coupling between the cement and the pipe and between
the cement and the formation. Following are a few examples of cement bond log
interpretation for different bonding conditions.
 48

Free pipe – In uncemented casing, the amplitude log shows high amplitude and
the transit time corresponds to the casing arrival time (the time required for sound waves
to pass through the wellbore fluid and the casing). The variable density log shows
strongly contrasting parallel vertical lines with no indication of formation signals. Casing
collars show up distinctively on a cement bond log in free pipe. Collar reflections result
in chevrons (capital W’s on their sides) on the variable density log, a decrease in
amplitude, and an increase in transit time. A cement bond log in free pipe is shown in
Fig. 30 [Hill, 1990]. It is important to log in an area of free pipe if possible when running
a cement bond log. Any deviation from the expected response in free pipe indicates a
malfunctioning or improperly centralized tool. This calibrates the tool in a known
environment under logging conditions.

Fig. 30 Cement bond log in free pipe

Good bond to formation and casing –With good bonding, the amplitude is low.
The full wave train display shows weak or no casing signals and strong formation arrivals
unless the formation attenuation is high such as would be observed for an unconsolidated
 49

gas sand, weak shales, or other low velocity formations. Comparison of the cement bond
log with an open hole sonic log can help identify regions of high attenuation in the
formation. An example of good bonding to the pipe and the formation is presented in Fig.
31 [Hill, 1990].

Fig. 31 Cement bond log with good bonding to the pipe and the formation

Good casing bond but poor formation bond – This is characterized by weak
casing arrivals as indicated by low amplitude and low contrast on the variable density log
at casing arrival times and weak formation signals on a full wave train display.
Unfortunately, these same characteristics can be caused by other factors including high
 50

formation acoustic attenuation and tool eccentricity. Good bonding to the pipe but not to
the formation can easily occur opposite permeable zones where a mud cake is built up
that is not displaced by cement. Figure 32 from Bigelow [1985] presents such a case
where both casing and formation amplitudes are low. Interpretation of the bonding from
the amplitude curve alone would give an erroneous picture of cement integrity – the lack
of acoustic coupling to the formation indicates poor cementing even though the pipe
amplitude is quite low. However, this behavior alone is not sufficient to prove a lack of
hydraulic seal, as mud occupying the space between the cement and the formation may
be immobile.

Fig. 32 Cement bond showing good bond to the pipe but poor bonding to the formation

Ultrasonic pulse-echo logs

Ultrasonic pulse-echo techniques have been developed in an attempt to overcome some
of the deficiencies of traditional cement bond logs. The primary advantage of the ultrasonic
devices is that they provide a circumferential picture of cement quality by utilizing multiple
 51

transducers arrayed around the tool, or by rotating the transducer(s) to give continuous
measurement of cement conditions around the well circumference. The ultrasonic
measurements are less sensitive, however, to acoustic coupling to the formation.
Ultrasonic pulse-echo tool originally consisted of an array of eight ultrasonic transducers
spaced around the body of the tool such as shown for the Schlumberger Cement Evaluation Tool
in Fig. 29. A ninth transducer is aligned axially and aimed at an acoustic mirror so that an in-situ
measure of travel time in the wellbore fluid can be made. Pulse-echo tools operate within the
resonance frequency of steel pipe, so that the casing will resonate if it is not well bonded by
cement. These tools measure the bonding to the casing by measuring the rate of decay of casing
vibration. The output from the eight transducer is presented as a map of the bonding conditions
around the casing.
Newer ultrasonic tools have replaced the eight fixed transducers with rotatable
transducers that continually sweep around the borehole (Fig. 33, Morris et al.,2007). The angled
transducers measure flextural attenuation of the acoustic energy.

Fig. 33 Rotating ultrasonic transducers on cement imaging tool

Example 8 Channel Detection with an Ultrasonic Pulse-Echo Log

 52

One of the primary advantages of the ultrasonic pulse-echo log is that it can
identify unsupported sections of the pipe circumference since it measures bonding
conditions at eight positions circumferentially around the pipe. Figures 34 from Catala et
al. [1984] shows a typical response to a channel, with a few of the tracks showing poor
bonding, while good bonding is indicated around the rest of the pipe. The channel
appears to be spiraling around the pipe; however, the relative bearing recording indicates
that the tool was slowly rotating as the log was run – the channel is consistently on one
side of the pipe.

Fig. 34 Ultrasonic pulse-echo log

V. Production Logging Tools and Methods for Inclined or Horizontal

Wells
 53

5.1 Introduction
The recognition of the asymmetric phase distributions that occur in multiphase
flow in inclined or horizontal wellbores, and the rapid implementation of horizontal wells
beginning in the 80’s has led to the gradual development of a new class of production
logging tools and techniques. When the phases tend to segregate with the lighter phase
concentrated on the upper side of the wellbore and the denser phase primarily occupying
the lower side, it is clear that single point measurements with small tools such as
traditional spinner flowmeters could not possibly measure average flow properties.
The advent of horizontal wells has had other profound impacts on production
logging measurements. Fluid density measurements based on a Δp measurement are not
possible in a horizontal well because there is no potential energy pressure drop along a
horizontal wellbore. Temperature log interpretations of the well flow profile which are
based on fluids from different zones having different geothermal temperatures do not
work for horizontal wells.
These difficulties, particularly the tendency for phases to segregate in a two-or-
three- phase flow in horizontal wells, has led to the development of an entire new class of
production logging tools and methodologies. For the flow regimes that occur in
horizontal two-phase flow [Brill and Beggs, 1978] (Fig. 35), to measure the volumetric
flow rates of all phases present requires one of three approaches: (1) make measurements
that interrogate the full cross-sectional area of flow and properly average the properties of
interest; (2) use an array of sensors, make measurements at multiple locations in the
wellbore cross-section; or (3) make a measurement that targets a specific phase. All three
of these approaches are being applied in the new class of production logging instruments.
In this section, we will describe the following tools and logging methods – capacitance
probes, optical probes, spinner flowmeter arrays, tracer methods targeting specific
phases, and unfocused γ-ray density tools. The section also includes a discussion of
combinations of these measurements.
 54

Fig. 35 Two-phase Flow Regimes in Horizontal Wells

5.2 Arrays of Capacitance or Optical Probes

A simple deployment of capacitance probes to get some measure of the phase
distribution is to attach capacitance probes to centralizer arms that are usually included
with production logging strings as shown in Fig. 36 [Mas et al., 2001].
 55

Fig. 36 Array of Capacitance Probes Attached to Centralizer Bands

Each probe will send signals like those in Fig. 37 that yield the average length of
time that the probe is immersed in a particular phase [Theron et al., 2000].
 56

Fig. 37 Output from a Probe Sensitive to the Phase Contacting the

Probe

Capacitance probes distinguish between hydrocarbons and water, while the

optical probes are most sensitive in identifying gas versus liquid. These local holdup
measurements provide an approximate map of the average distribution of the phases at a
location in the well. This type measurement is best made with the tool stationary (or
moving very slowly) so that a representative temporal average is obtained. The
measurement relies on discrete bubbles being measurable by the probe, which requires
the bubbles to be the size of the probe or larger. This can lead to underestimation of the
holdup of the dispersed phase when bubbles are very small, as occurs in high velocity
flows.
A more elaborate use of an array of capacitance probes is used on the tool shown
in Fig. 38 [Chase et al., 2000]. Capacitance probes are distributed in 8 locations along
the extended arm of the tool, giving 8 measurements of local holdup from the high to the
low side of the wellbore. In addition, in six of these locations, there are 4 probes aligned
axially with the tool body. The purpose of these aligned probes is to measure the velocity
of bubbles passing the probes. By timing the arrivals of the signal changes that occur
when the phase contacting the probes changes, a time of flight velocity measurement is
obtained. This approach depends on discrete bubbles of one phase being dispersed in the
other phase.
 57

Fig. 38 Capacitance Arrays on Extendable Arms to Measure Local

Holdup and Velocity Distributions

Optical probes are used to measure the local holdup in gas-liquid flows [Mas et
al., 2001, Jackson et al., 2001] and can also be used to distinguish between oil and water.
Each optical probe on this tool measures the refractive index of the fluid in which it is
immersed by sending light from an LED to a sapphire tip on the probe. When immersed
in gas, all of the light reaching the tip is reflected, while only part of the light is reflected
when immersed in oil or water [Mas et al., 2001].

5.3 Arrays of Spinner Flowmeters

A tool developed to measure the velocity profile using an array of spinner

flowmeters is shown in Fig. 39 [Dandaray et al., 2005], while Fig. 40 depicts such a tool
in a gas-oil-water flow [www.slb.com, 2009]. As with the capacitance array tool, this
tool uses an extendable arm that folds into the tool housing to place the multiple spinners
across the flow stream. By averaging the output of each spinner over a short time period
with the tool held stationary, the velocity profile can be constructed. Combining this with
an array of capacitance or optical probes allows for the mapping of the flow distribution
of multiple phases. For example, Fig. 41 [Dandaray et al., 2005] shows a velocity profile
measured with a spinner array in a deviated well. High upwards velocities were
measured in the upper part of the pipe, while a reverse flow was measured with the
bottom spinner. This type of fallback of a denser phase is common in inclined wellbores
[Hill and Oolman, 1982].
 58

Fig. 39 An Array of Spinner Flowmeters

 59

Fig. 40 Multiple Spinners in Complex Multiphase Flow

Fig. 41 Velocity Profile Measured with an Array of Spinner Flowmeters

 60

5.4 Tracer Methods in Multiphase Flow

Tracers have the potential to directly measure the velocity of only one of the
multiple phases present in a complex multiphase flow if the tracer can be presumed to
reside in only one phase. The two tracer techniques that are currently applied in
multiphase flow are the activation of water with neutron pulses to create a short-lived
isotope to track water movement, and chemical tracers, either oil or water soluble, that
can be introduced into the flow and tracked with a pulsed neutron log.
A pulsed neutron logging instrument can be used to activate water molecules, and
track the movement of the resulting isotopes traveling in the water in the wellbore as they
move past multiple γ-ray detectors. Fig. 42 [Chase et al., 2000] illustrates this technique.
The neutron pulses activate the oxygen molecules in the water to N17, which immediately
begins decaying back to O16, emitting γ-radiation in the process. This radiation is
detected by the γ-ray detectors located downstream of the neutron source. By timing the
passage of a cloud of decaying N17, the velocity of the water is obtained. If this velocity
is combined with a valid overall holdup measurement, the volumetric flow rate of water
is directly obtained as

qw = vw y w A (5.1)

The wellbore cross-sectional area is the annular area between the casing inner diameter
and the tool diameter.

Fig. 42 Oxygen Activation Method to Measure Water Velocity

In theory, radioactive tracers, such as solutions of Iodine131 could be injected into

the flow stream using a traditional radioactive tracer logging tool to measure phase
velocities in multiphase production wells. In practice, this is not done because of the
 61

hazards of producing the radioactive tracer material to the surface. Instead, a chemical
marker method has been developed [Roscoe et al., 1997] in which an element,
gadolinium, having a very high macroscopic capture cross-section, is injected into the
well upstream of the neutron source of a pulsed neutron instrument (Fig. 43). With its
very high capture cross-section (1000 times that of chlorine), the gadolinium creates a
great deal of γ-radiation after being activated by the neutron pulse. Both oil and water
soluble gadolinium tracers were developed so that the velocity of either phase can be
measured. In order to measure the velocity of a dispersed phase with this technique, the
tracer must be able to pass through the continuous phase and mix with the dispersed
phase before reaching the neutron source. Whether this is possible will likely depend on
the flow regime.

Fig. 43 Chemical Markers to Measure Individual Phase Velocities

5.5 Unfocused γ-ray Fluid Density Measurements

The traditional in-line γ-ray densitometers that are used to measure wellbore fluid
density do not measure the average flow stream density when the phases are non-
uniformly distributed, as is so often the case in inclined or horizontal wellbores because
the fluid passing between the γ-ray source and detector is not in the same proportion of
phases as the overall average. To overcome this problem, an unfocused γ-ray
densitometer has been developed (Fig. 44 [Kessler and Frisch, 1995]). This tool has a
radiation shield between the γ-ray source and detector so that the radiation detected is
from back-scattered radiation from all around the pipe cross-section. Laboratory tests
with this tool have shown good linearity between the overall holdup of liquid in a gas-
liquid flow, even with segregated flow regimes.
 62

Fig. 44 Unfocused γ-ray Density Tool

5.6 Interpretations and Log Displays with Distributed Measurements

With measurements of local holdup and/or velocity at multiple locations around

the wellbore cross-section, we can now map the distribution of these properties to gain
more insight into the flow conditions in the wellbore. Integration of the product of
velocity and local holdup yields an estimate of the volumetric flow rates of multiple
phases at any position along the wellbore. Following are examples of such
interpretations and log displays.

Stagnant Water along a Horizontal Well

Figures 45 and 46 are logs from a capacitance array tool run during two tests of a
well being initially brought on production [Chandran et al., 2005]. Producing with a
smaller choke setting, and hence, lower total flow rate, the log shows large water holdups
in the lower elevation (sump) regions along the well (Fig. 45), in spite of the low
produced water cut of 7-10%. With the choke size increased, approximately doubling the
total production rate from the well, the amount of stagnant water in the sumps has been
 63

significantly reduced (Fig. 46). The velocity profiles plotted in the middle tracks of these
logs shows that the velocity in the water in the sumps is near zero, with the oil flowing
over the standing water. This type profile is typical of nominally horizontal wells with
slight undulations in trajectory.

Fig. 45 Horizontal Well Flow Profile with Small Choke Size

 64

Fig. 46 Horizontal Well Flow Profile with Large Choke Size

Analysis of Water Shutoff Procedure

 65

Figs. 47 and 48 show production logs run with optical and electrical probe arrays
to measure holdup profiles before and after a casing patch was run to shut off a zone
producing excessive water [Jackson et al., 2001]. The log before the workover (Fig.
5.13) shows the well was producing large volumes of water from 6,494 to 6,460 feet.
After a casing patch was set across this interval, the subsequent log showed that the water
entry had been eliminated, though there was now some water production from just above
the patch location.

Fig. 47 Production Logs Showing Water Inflow Before Casing Patch

Workover
 66

Fig. 48 Production Logs Showing Altered Water Inflow After

Placement of Casing Patch

Effect of Small Inclination Changes on Holdup Behavior

In nominally horizontal wells, small changes in well inclination can change the
flow regime and the holdup in the well significantly. With traditional, single point
production logging measurements, such changes often lead to misinterpretations of the
well profile and the phase inflows. Distributed holdup measurements make these two-
phase flow effects readily identifiable. Fig. 49 [Fitz et al., 2006] shows production logs
run with arrays of spinners and holdup probes in a nominally horizontal well with very
slight variations in inclination. In this gas-liquid producing well, in regions of the well
towards the toe where the inclination is slightly downward, the liquid holdup is relatively
small, reflecting the segregated flow regime that occurs in a downward flow. In the
slightly upwardly inclined section near the heel of the well, almost of the pipe is liquid
filled, with gas flowing along the upwards part of the pipe as a bubbly flow.
 67

Fig. 49 Effect of Slight Inclination Changes on Phase Distribution in a

Nominally Horizontal Well

5.7 Downhole Video as an Alternative to Production Logs

Advances in downhole video equipment now offer this measurement as an

alternative to traditional or the new class of production logging measurements. In most
production wells, it is possible to clearly see oil or gas entries into the well, because
almost all production wells contain water through which the hydrocarbons are passing.
Thus, a downhole video log is a means of directly measuring the locations of oil and gas
entries into wellbore. High rate water entries can also be detected from the image
distortion caused by high levels of turbulence.
Oil entries in a production well typically are clearly visible as dark bubbles of oil
entering the wellbore from perforations as shown in Fig. 50 [Whittaker, 2009]. With a
video image, the oil entry rate can be estimated by measuring the rate of bubble entry and
estimating the bubble sizes (the diameter being approximately the diameter of
perforations.) Gas entries are often not as clearly distinguishable as oil entries because of
gas-water miscibility and small bubble sizes, but they are still usually detectable. The
location of both water and gas entries with downhole video in a horizontal well was
illustrated by Sask et al. [2007]. Gas inflows into a water-filled wellbore were seen as
streams of gas bubbles (Fig. 51). The location of the gas-liquid interface (Fig. 52) in
different sections of the well indicated water entry locations.
 68

Fig. 50 Video Image of Oil Entry

Fig. 51 Video Image of Gas Entry

Fig. 52 Video Image of Gas-Water Interface in

Horizontal Gas-Producing Well
 69

References
Bigelow, E. L.: “A Practical Approach to the Interpretation of Cement Bond Logs,” JPT,
(July 1985) 1285-1294.

Brill, J. P. and Beggs, H. D., Two-Phase Flow in Pipes, University of Tulsa, Tulsa, OK,
1978.

Carlson, N. R. and Roesner, R. E.: “Water-Oil Flow Surveys with Basket Fluid
Capacitance Tool,” Proceedings, SPWLA Annual Logging Symposium, Corpus Christi,
Texas (July 6-9, 1982) paper F.

Catala, G. N., Stowe, I. D., and Henry, D. J.: “A Combination of Acoustic Measurements
to Evaluate Cementations,” paper SPE 13139, 1984.

Chandran, T., Talabani, S., Jehad, A., Al-Anzi, E., Clark, jr., R., and Wells, J.C.:
“Solutions to Challenges in Production Logging of Horizontal Wells Using New Tool,”
SPE 10248 presented at the International Petroleum Technology Conference, Doha,
Qatar, Nov. 21-23, 2005.

Chase, David, Wang, J., Mirzwinski, R., Maxit, J., and Trcka, D.,: “Applications of a
New Multiple Sensor Production Logging System for Horizontal and Highly-Deviated
Multiphase Producers,” SPE 63141 presented at the SPE ATCE, Oct. 1-4, 2000, Dallas,
Texas.

Connolly, E. T.: “Resume and Current Status of the Use of Logs in Production,”
Proceedings, 1965 SPWLA Annual Logging Symposium, Dallas, Texas, May 4-7, 1965.

Curtis, R. M.: “Flow Analysis in Producing Wells,” paper SPE 1908, 1967.

Dandaray, H., Al-Mutairi, A., Abdouche, G., Khedr, O., and Elkadi, A.: “A New
Production Logging Tool Allows a Superior Mapping of the Fluid Velocities and
Holdups Inside the Wellbore,” SPE 93556 presented at 14th SPE Middle East Oil and Gas
Show and Conference, Bahrain, March 12-15, 2005.

Fitz, D. E., Guzman-Garcia, Angel, Sunder, Ram, Billingham, Matt, and Smolensky,
Vitaly: “Pushing the Envelope for Production Logging in Extended Reach Horizontal
 70

Wells in Chayvo Field, Sakhalin, Russia – New Conveyance and Flow Profiling
Approach,” SPE 103589 presented at the SPE Russian Oil and Gas Technical Conference
and Exhibition, Moscow, Russia, Oct. 3-6, 2006.

Froelich, B., Pittman, D., and Seeman, B.: “Cement Evaluation Tool: A New Approach
to Cement Evaluation,” JPT (Aug. 1982) p. 1835-1841.

Interpretive Methods for Production Well Logs, 2nd Edition, Atlas Wireline Services,
Western Atlas Inc., Houston (1982).

Jackson, R. R., Ayan, C., and Wakefield, J.: “Flow Diagnosis and Production Evaluation
in High Flowrate Oil-Water Producers Using Optical-Fibre Holdup Sensors,” SPE 71727
presented at the SPE ATCE, New Orleans, LA., Sept. 30 – Oct. 3, 2001.

Kessler, Calvin and Frisch, Gary: “New Fullbore Logging Sensor Improves the
Evaluation of Production in Deviated and Horizontal Wells,” SPE 29815 presented at the
1995 Middle East Oil Technical Conference and Exhibition, Bahrain, March 11-14, 1995.

Mas, C., Walterhouse, B., Van Vliet, J. P. M., and Rahman, H. A.: “Evaluations of Sub
Horizontal Well Performance with Optical and Electrical Probes,” SPE 72150 presented
at the SPE Asia Pacific Improved Recovery Conference, Kuala Lumpur, Malaysia, Oct.
8-9, 2001.

Morris, C., Sabbagh, L., Wydrinski, R., Hupp, J., Van Kuijk, R., and Froelich, B.:
“Application of Enhanced Ultrasonic Measurements for Cement and Casing Evaluation,”
SPE 105648 presented at the SPE/IADC Drilling Conference, Amsterdam, The
Netherlands, Feb. 20-22, 2007.

Sask, D., Hundt, C. D., Slade, J. M., and Daly, P.: “Production and Video Logging in
Horizontal Low-Permeability Gas Wells,” SPE 108084 presented at the SPE Rocky
Mountain Oil and Gas Technology Symposium, Denver, CO., April 16-18, 2007.

Theron, B., Vuhoang, D., Rezgui, F., Catala, G., McKeon, D., and Silipigno, L.:
“Improved Determination of Gas Holdup Using Optical Fibre Sensors,” presented at the
SPWLA Annual Symposium, Dallas, Texas, June 4-7, 2000. SPWLA (cited in SPE
71727).
 71

Whittaker, Jeff, Expro Group, personal communication, 2009.