SPE 101078

A Novel Cased Hole Density-Neutron LogCharacteristics and Interpretation

P.A.S. Elkington, SPE, C.A. Pereira, and J.R. Samworth, Weatherford

Copyright 2006, Society of Petroleum Engineers

This paper was prepared for presentation at the 2006 SPE Asia Pacific Oil & Gas Conference
and Exhibition held in Adelaide, Australia, 1113 September 2006.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to an abstract of not more than
300 words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract
We have characterized the response of a compact 2-inch
(57mm) diameter open-hole formation density logging tool for
cased hole environments. Data are processed with an
established open hole transform in which the casing effect
appears as a simple attenuation term in the count rate domain,
and variations in cement thickness are compensated using a
classical dual-detector spine-and-ribs approach applied in the
density domain. The combination of through-casing density
and casing-corrected neutron porosity has been applied to the
evaluation of by-passed pay and shallow gas, and to the
evaluation of wells where open hole acquisition has not been
feasible for operational, hole quality, or economic reasons.
The tool-specific neutron porosity excavation effect has been
characterized for gas-bearing sands.
Case history and model results suggest accurate formation
densities are achievable for casing thicknesses up to about
0.35 inches (9mm) if cement is less than about 1 inch (25mm)
thick, albeit with loss of precision relative to open hole.
Formation sensitivity declines with increased cement
thickness until practically all is lost for casing standoffs in
excess of 1.5 inches (38mm). For modest thicknesses,
however, cased hole density-neutron gas evaluation has
advantages relative to the neutron-dipole sonic method; in
particular it does not rely on good cement bond, it has
generally superior spatial resolution, and (optionally) data can
be acquired in memory mode on slickline in operating
environments that do not favour conventional wireline units.
Introduction
The present work was stimulated by the search for shallow
gas, and has since found broader application in the evaluation
of by-passed pay (including light oil), in the monitoring of
fluid contacts and saturation changes over time, and in the
general evaluation of intervals that, for a variety of reasons,
may not have been logged open hole.

Gas behind casing can be inferred from pulsed neutron tool

count rate logs, and the 111/16 inch (43mm) diameter variants
may be the sole evaluation option in wells with small diameter
tubing restrictions. A common alternative in other cases (and
where pulsed neutron tools are precluded by other operational
or cost issues) is a dual porosity approach with neutron and
sonic curves scaled empirically to overlay in clean waterbearing zones. Sonic porosity measurements have the
advantage of being insensitive to hole enlargement behind
casing, but quality control of waveform processing is timeconsuming, and the velocity-porosity transform may be
uncertain. In poorly bonded casing the formation arrival may
be lost altogether. A more technically challenging third option
is the density-neutron combination. This is feasible where
casing thickness and standoff are modest, but is demanding
because of the need to deal with variable casing standoff.
Density tools respond to casing as though it were heavy
mudcake, except measured density values are typically high
and beyond the range for which tools are normally
characterized. When this is compounded with unknown casing
standoff, standard processing yields inappropriate values of
delta-rho and inaccurate compensated densities.
One approach to cased hole density correction is to
normalize the compensated density log in zones of known
porosity, or align to neutron porosity logs in clean waterbearing intervals (Cigni and Magrassi, 1987). This can be
adequate for delineating zones of interest, but is unlikely to
furnish good accuracy over a broad range of density values,
and provides little or no information to allow variations in
casing standoff to be flagged or accounted for. A better
approach is to correct individual apparent density
measurements (Near and Far curves in the case of a dual
detector device), but it is important to recognise that measured
and actual densities may not be linearly related outside the
range of densities for which the tool is normally calibrated.
Among the factors affecting measured density in cased
wells are casing thickness, casing density, cement thickness
and cement density, as well as formation density. To account
fully for all these variables implies a requirement for multiple
independent density measurements with different depths of
investigation. Laboratory results from an experimental fourdetector device have been published (Moake, 1998), but there
is no evidence of the tool being run commercially. Field
experiences with a conventionally-sized density tool processed
with laboratory-derived cased hole calibrations have also been
reported (Ellis et.al., 2004). The present work relates to a
commercial dual-detector tool, and is an extension of a

SPE 101078

methodology that assumes casing properties and cement

density are constant in the interval analysed (Samworth and
Calvert, 2002).
The tool is 2-inch (57mm) in diameter, developed for
open holes but capable of being deployed into casing and
production tubing. It is run on wireline for real-time surface
readout, but in battery-memory mode it can be conveyed on
slickline, drillpipe or coiled tubing (Elkington et. al., 2000). It
is often run in combination with other 2-inch diameter tools
from the same family, including neutron porosity and sonic.
The ability to acquire open and cased data with the same tools
has operational and evaluation advantages, particularly in
monitoring applications that use open hole logs as the
baseline.
The skid that contains the source and detectors is typically
faced with an 8-inch (203mm) curvature wear plate, but for
this study we used a 4-inch (102mm) profile plate developed
originally for slim open holes, and which also provides a
better fit to casing and tubing with internal diameters less than
6-inches (152mm). The skid is able to move in and out of a
carrier, and it is reasonable to assume that it is in intimate
contact with the casing at all times i.e. tool to casing standoff
is assumed zero. The residual effects of well fluid density and
casing diameter are small, and accounted for within the
processing.
The cased hole density analysis starts with the single
detector count rate transform. We derive the casing attenuation
factors before moving to a dual detector compensation
scheme. These factors were initially derived empirically, and a
limited database of values was established. Monte Carlo
modelling subsequently supported the empirical observations,
and has helped to quantify sensitivities to key variables, and
extend the range of application. We also show how cased hole
density and neutron logs can be combined to provide
quantitative porosity and saturation estimates in gas-bearing
sands, and give examples of its application.
Cased Hole Density Response
We derive formation density in two stages: transformation of
count rates into apparent density values for each detector (the
single detector response), followed by computation of
formation density from a combination of apparent densities
(the dual detector response). The process is identical to that
employed in open hole, the only difference being the values
assigned to certain key variables.

Figure 1 shows count rate vs. density characteristics when

the borehole term is zero. It is similar to the commonly used
linear log-normal characteristic over the range of typical
reservoir rock densities, but is a more realistic representation
of the actual response at high apparent densities (such as low
porosity formations in cased wells), as well as at low apparent
densities (such as coals in open holes). It also conforms to the
boundary condition (violated by the linear log-normal form) of
zero non-tool counts for density values of zero and infinity.
The low density response is largely irrelevant in cased wells,
but a good description of the response at high apparent
densities simplifies the casing correction.

Fig. 1 Count rate vs. density response.

The count rate transform was investigated with the MCNP

Monte Carlo modelling code. We looked at casing thicknesses
of 0.10, 0.14, 0.20, 0.25, 0.35, 0.5 and 0.75 inches (2.5, 3.5,
5.0, 6.4, 8.9, 12.7 and 19mm) with zero cement thickness and
formation densities of 2.65, 2.40, 2.16, and 1.66 g/cc. The
formation part of equation (1) can be made to fit the count rate
distribution for each casing thickness simply by modifying the
E term (while making no change to k). This implies that the
casing is attenuating the count rate in a straightforward way.
The degree of attenuation is shown in Figure 2; it shows that
the logarithm of the scaling factor applied to E varies linearly
with casing thickness, reducing the Near and Far spacing
values to 38% and 43% of their respective open hole levels for
a casing thickness of 0.25 inches (6.4mm).

Single Detector Response

In the single detector response model (Samworth, 1992),
count rate, I, is related to formation density, , by:

I = Ee-k + F(c-t)(me-km - e-k) + G

(1)

where m is borehole fluid density, c and t are hole and tool

profile diameters respectively, and E, F, G and k are constants.
E is a transmission term, and F and G relate to contributions
from the borehole and tool respectively. F is of minor
significance in open hole, being about four orders of
magnitude less than the nominal value of E, but becomes more
significant as casing thickness increases (and count rates
decrease). G arises principally from internal low-activity 137Cs
sources used in the tools gain stabilization system.

Fig. 2 Attenuation factor vs. casing thickness.

SPE 101078

The leading term in equation (1) now becomes:

I = E e-mt e-k

(2)

where t is casing thickness and m the slope of the line in figure 2.

Dual Detector Response
Standoff effects for dual detector tools in open hole are
commonly dealt with by the well-known spine-and-ribs
technique. The method is valid even when the density of the
material in front of the tool exceeds that of the formation
itself, so might therefore be expected to work in cased wells.
In practice, however, implementations are tuned to work with
realistic mudcake densities and thicknesses, not high density
steel casing (approximately 8 g/cc). Apparent densities
measured by each detector are influenced by casing, cement
and formation densities, and are typically above the range for
which most tools are characterized. Therefore we do not
expect standard open hole processing to succeed in cased
wells.
There is another important consideration in respect spineand-ribs processing that relates to precision and repeatability.
In the limiting case of large standoff, ribs re-join the spine at a
point corresponding to the density of the material between tool
and formation figure 3. The formation density of any point
along a rib is found by projecting the point back along the rib
to the spine at the zero standoff condition, but small variations
in the log density values at extreme standoffs have a large
lever effect on the projection, causing the compensated result
to oscillate wildly. For this reason, real-world implementations
of spine-and-ribs constrain the projection i.e. they sacrifice
some accuracy for improved precision. This is a particularly
important consideration in cased holes where reduced count
rates reduce statistical precision.

Fig 3 Uncertainty in compensated density at large standoffs, no

casing corrections applied. As Near and Far log values approach
the high apparent density point, uncertainty in the projected
compensated density value increases.

The single detector casing corrections remove casing

effects on the individual Near and Far measurements so that
the points are both closer to the spine and within the normal
range of density values for which the tool is calibrated.
Having determined the E attenuation factor for each
detector, it is assumed that any residual differences between
Near and Far apparent density values are due to variations in
cement thickness and/or cement density. In general there are

insufficient degrees of freedom with a two detector system to

solve uniquely for both cement parameters, but if cement
density is known independently we can still derive formation
density using spine-and-ribs processing (after casing
correction the ribs meet the spine at formation and cement
density values).
In many cases, however, we can avoid the need to know
(or solve for) cement density by using a linear rib assumption.
Samworth (1992) has shown that within the classical spineand-ribs framework, the density of the material between tool
(or casing in this instance) and formation has no influence on
the compensated density over the linear part of a rib (that part
close to the spine).
Although real ribs are curved (because the rates at which
formation sensitivity declines as standoff increases are
different for each detector), the linear rib assumption is
reasonable for modest standoffs, and for reasons outlined
previously, this also has an advantage in respect of precision.
With these insights, a casing corrected density is readily
computed by substituting equation (2) into the formation part
of equation (1). The remaining process is identical to that for
open hole, producing a compensated density, C, and degree of
compensation, . The latter is computed in the usual way,
namely:
= C F

(3)

where F is the corrected Far density. reflects the degree of

casing standoff (cement thickness), and may be larger than in
open hole where standoff due to mudcake is typically less.
The method does not correct for the additional steel
thickness at casing collars, and it is important to review results
alongside a CCL log. The C curve tends to read light opposite
collars because they cause the Near to read heavier than the
Far, making negative. Collar effects can be mitigated by
suppressing the Near contribution over collar intervals.
Support from Empirical Evidence
The model-derived attenuation factors applied to the Near
and Far E values in cased wells can be verified empirically by
examining distributions of Near and Far density values in real
well environments.
Figure 4 is a crossplot of values obtained with standard
open hole processing in a sand-shale-coal sequence in 4.5-inch
(114mm), 9.5 lb/ft (14 kg/m) casing in a 6.25-inch (159mm)
well. Migrating this data back to the spine using open hole ribs
produces grossly inaccurate compensated densities. If we have
a-priori knowledge of at least one formation density value,
however, the data can be moved to the position it would
occupy on the crossplot if there were no casing (nominal
casing standoff needs to be assumed). We would then expect
to see all non-collar points distributed above the spine in a
manner that reflects cement standoff only.
Linear shifts are unsatisfactory (because high apparent
density values may not be linearly related to actual densities),
but using equations (1) and (2) it is possible to determine the
attenuation factors that move the data to match the density of
the known point. Figure 5 shows the same data corrected in
this way. Note that the data cluster approaches the spine at
two points: at high formation densities, and at the density of
the cement (about 2 g/cc). The lowest density values are coals,
and these appear below the spine after correction. Attenuation

SPE 101078

factors inferred in this way are consistent with those from

model results.
Efficacy of the cement thickness compensation may be
judged by comparing passes acquired with the density tool in
different azimuthal orientations, and processed in two
different ways (cement thickness is assumed to vary around
the casing except where it is centralized in a circular section
hole). Figure 6 shows two such passes, one processed by
shifting Near and Far linearly, the other using the count rate
model and non-linear shift. The latter has the better repeatability
between main and repeat passes. The greatest difference is in
the interval 555m - 562m where differs between runs
(indicating the tool azimuth was different between runs);
linear processing does not correct the difference, whereas
count rate processing does.
Density logs processed in this way are consistent with
offset open hole data and other porosity logs where casing and
cement thicknesses are modest. Quality reduces as thicknesses
increase, with practical limits being in the region of 0.5 inches
(13mm) casing thickness and/or about 1.5 inches (38mm)
cement thickness. This is consistent with model results that
show, for example, that for a casing thickness of 0.35 inches
(9mm) the Far density has usable sensitivity to a cement
thickness of about 2 inches (51mm), but for the Near the limit
is about 1 inch (25mm).

Fig. 4 Near and Far density values before cased hole processing.

Fig. 5 Near and Far density values after cased hole processing.

Fig 6 Overlay of main and repeat pass cased hole density logs.
Upper log Near and Far shifted linearly. Lower log Near and
Far processed via the count rate model and non-linear shift.
Density scales are 1.9 to 2.9 g/cc; -0.05 to 0.45 g/cc over the
rightmost 5 grid lines; gamma ray logs and CCL in the left track.
Note the interval 555m - 562m where main and repeat passes
agree after processing with the count rate model.

SPE 101078

Cased Hole Neutron Response

Gas bearing formations are classically delineated in open hole
using a density-neutron combination, and with appropriate
corrections, the same can be true of cased holes. Indeed,
formations that have produced gas after casing may have a
stronger gas signature relative to the open hole situation (in
which the gas signal has been suppressed by fluids invasion).
It has also been suggested that count rate ratios (from which
porosity is derived) are more accurate predictors of gas
saturation than sigma derived from pulsed source tools
(Cowan and Wright, 1999). In this section we examine the
neutron cased hole response, and in particular we look at the
effects of gas saturation on the density and neutron responses
behind casing.

Fig. 8 Neutron porosity comparison. Green was acquired in

open hole, red in cased hole. Scale is 60 pu (left) to 0 pu (right).
Left track shows open and cased gamma ray(uncorrected), and
CCL. Casing is 7-inches, 21.5 lb/ft in an 8.5-inch bit size well
Fig.7 Neutron porosity cased hole corrections.

Cased hole corrections for the 2-inch (57mm) neutron

tool have been available for some time figure 7. They are
defined by three parameters - casing ID, casing thickness, and
cement thickness. These replace the borehole size and standoff
corrections required in open hole (the casing itself does not
have a large effect on the neutron flux around the tool, cement
having the greater influence). Corrections determined for each
parameter sum to the net correction. In the case shown,
apparent neutron porosity in 0.5-inch (13mm) thick, 5-inch
(127mm) ID casing with 1.25-inch (32mm) thick cement is 28
pu, the individual corrections are 3.9, -7.5 and -6.6 pu, and
corrected porosity is 17.8 pu. Efficacy is illustrated in figure 8,
which compares the same interval logged open then cased
hole.

Batzle and Wang (1992). Results are referred back to the

standard condition of 8-inch diameter open hole.
Apparent log densities were computed from formation
electron densities using the inverse Z/A relationship for this
tool (Samworth, 1992). In order to allow application to field
porosity logs computed assuming full water saturation and a
fixed value of filtrate density, we have converted measured
density values to apparent density porosities using a nominal
filtrate density value of 1.0 g/cc. Results for 2,000 psi/50C
(13.8 MPa/122F) are plotted on a conventional densityneutron crossplot in figure 9. Constant porosity lines are
extensions of the limestone to sandstone equi-porosity lines,
and lines of constant saturation are very approximately
orthogonal to them. Plotted on a dual-porosity crossplot, the
constant porosity and saturation lines are linear. Results for
7,000 psi and 120C (48.3 MPa / 248F) are plotted in Figure 10.

Gas Effect
Gas in the formation reduces apparent neutron porosity. This
so-called excavation effect has been described for an older
tool (Segasman and Liu, 1971). We used MCNP to model the
effects of gas saturation on the compact tool neutron response,
taking into account typical temperature/pressure gradients.
Gas composition was assumed to be methane, and water was
assumed fresh; temperature and pressure effects on water and
gas densities were computed from equations developed by

Example
Figure 11 is a through casing density-neutron porosity and
gas saturation analysis from a high-value horizontal well in
which no open hole density-neutron data were acquired due to
failure of an LWD string. Conveying wireline tools on coiled
tubing was considered, but it was not possible to use a coil
with integral wireline due to weight restrictions. The solution
was to deploy the compact density-neutron in battery-memory

SPE 101078

Fig. 9 Gas saturation on density-neutron crossplot.

Fig. 10 Gas saturation on dual porosity crossplot.

Fig. 11 Porosity and gas saturation analysis.

SPE 101078

mode on the wireless coil. Nominal casing and cement

thicknesses are 0.27 (7mm) and 0.75 inches (19mm) respectively.
Correcting both the density and neutron for casing reveals
that gas is confined to numerous thin intervals. Corrected
porosity plots between the density and neutron where there is
gas crossover, and saturation is shown in track 1. Separation
of Near and Far neutron count rates provide corroboratory
evidence for gas.
Comparison with Sonic-Neutron for Gas Detection
The processing has been applied to dozens of cased wells, and
interpretations compared with those from the sonic-neutron
method. Figure 12 is an overlay of through casing sonic and
density in a sand-shale-coal sequence. The sonic comes from
semblance processing, and even though cement bond is good
in this interval, it lacks the resolution of the density.
Alternative processing can recover some resolution, but it is
the density whose character is most similar to that of the
neutron porosity.

effects of variable casing standoff. As standoff increases,

small variations in measured density translate to increasingly
large changes in compensated density, but precision can be
maintained at the cost of some reduction in accuracy if linear
ribs are assumed. Linear ribs have the additional property of
not requiring explicit knowledge of cement density.
The combination of density and neutron through casing
provides an attractive gas indicator. Modelling has shown that
gas saturation has a straightforward effect on the neutron
response, and that porosity and saturation can be computed
from the dual porosity combination using simple linear
equations. This is a useful alternative to the sonic-neutron dual
porosity approach.
Acknowledgements
The authors are grateful to Weatherford Evaluation, Drilling
and Intervention Services for permission to publish this paper.
References
Batzle, M. and Wang, Z. (1992). Seismic properties of pore fluids.
Geophysics v 57, p 1396 - 1408
Cowan, P. and Wright, G. A., 1999. Investigations into improved
methods of saturation determination using pulsed neutron capture
tools. SPWLA 41st Annual Logging Symposium, paper P.
Cigni, M., and Magrassi, M., 1987. Gas detection from formation
density and compensated neutron logs in cased hole, SPWLA
28th Annual Logging Symposium, paper W.
Elkington, P. A. S., 2000. The role of open hole memory logging and
wireless conveyance systems in the evaluation of horizontal
wells, paper SPE 65461, presented at the SPE/Petroleum
Society of CIM International Conference on Horizontal Well
Technology, 6-8 November 2000.
Ellis, D., Lling, M., G., Markley, M., E., Moss, L., Neumann, S.,
Pilot, G., and Stowe, I., 2004. Cased hole formation density
logging some field experiences, SPWLA 45th Annual
Logging Symposium, paper G.
Moake, G.L., 1998. Design of a cased hole logging tool using
laboratory measurements, paper SPE 49226, presented at the
SPE 73rd Annual Technical Conference.
Samworth, J. R., 1992. The Dual Spaced Density Log
Characteristics, Calibration and Compensation, The Log
Analyst, vol. 33, n.1, p. 42-49
Samworth, J.R., and Calvert, S., 2002. Differentiation of reservoir
fluids through casing an application of the compact memory
logging technique, SPWLA European Formation Evaluation
Symposium, London, September 5th - 6th.
Segasman, F., and Liu, O., 1971. The excavation effect, SPWLA
12th Annual Logging Symposium, paper N.

Fig. 12 Cased hole density and sonic porosity overlay. Track 1:

Gamma ray (green) and CCL (blue). Track 2: Density (red), sonic
(blue) 60 to 0 pu.

Summary and Conclusions

The key to deriving formation density through casing from a
dual detector tool is appropriate individual processing of the
Near and Far curves to remove the effect of casing thickness;
spine-and-ribs compensation can then used to correct for the

SI Metric Conversion Factors

in. x 2.54*
E+00 = mm
psi x 6.894 757 E+03 = MPa
(F*- 32)/1.8
E+00 = C
*Conversion factor is exact.