Joint Inversion of MWD and Wireline Measurements

Raghu K. Chunduru*, Alberto G. Mezzatesta, Hal W. Meyer, Zhiyi Zhang and Rainer Busch, Baker Hughes
Incorporated, and Tom Maher, Shell Offshore Inc
Summary
Traditionally, measurement-while-drilling (MWD) data are
used primarily for geosteering purposes and drilling
decisions such as monitoring of hole direction, deviation,
and delineation of abnormally pressured zones. Wireline
resistivity measurements, galvanic and induction, play a
fundamental role in identifying and delineating oil- and
gas-bearing formations. The availability of both MWD and
wireline data not only provides the interpreter with
abundant information about subsurface formations but also
poses a new challenge to generate a unique model(s) that
better explains both data sets. Generally, MWD and
wireline data are interpreted independently to estimate
formation resistivities that may result in inconsistent earth
models. In this study, we performed a joint inversion of
MWD Multiple propagation resistivity (MPR), and wireline
High Definition Induction log (HDIL) data to come up with
an earth model that best explains bot the data sets. An
inversion strategy using a dual earth model, that describes
the appropriate logging conditions of both wireline and
MWD was also used in theinversion. Finally, the proposed
algorithm was implemented on synthetic and the Gulf of
Mexico data examples, and the results were compared with
conventional MPR and HDIL processing results.
Introduction
MWD and wireline data obtain responses from similar
subsurface formations at different times, reflecting different
borehole and invasion conditions. Each of these data offers
distinct advantages when compared to the other. In general,
MWD data are not affected by invasion, allowing better
interpretation of formation resistivities. On the other hand,
wireline data allow us to characterize the invasion profile,
resulting in identification of permeable and impermeable
zones, and thereby facilitating the evaluation of movable
and residual hydrocarbons.
Electromagnetic wave resistivity tools, such as MPR and
the HDIL tool provides a multitude of measurements
associated with different depths of investigation and
vertical resolution. Typically MPR tool measures both
attenuation and phase difference at two frequencies of 2
MHz and 400 kHz, and at two transmitter-receiver
spacings, providing eight different depths of investigation
measurements. The wireline tool measures data at a
frequency range of 10 kHz to 150 kHz for all seven
receiver arrays, resulting in fifty-six measurements.
Conventionally, MWD and wireline data are interpreted
independently to estimate formation resistivities that might

SEG 1999 Expanded Abstracts

result in inconsistent earth models. Inversion techniques

allow for simultaneous processing of two data sets, leading
to a consistent earth model that describes both data sets. In
this study, a dual earth model that describes the appropriate
logging conditions of both wireline and MWD is proposed
and implemented on synthetic and a real data example from
Gulf of Mexico. We also compare the resolution matched
curves of the MPR (Meyer et al., 1994) and HDIL
measurements (Zhou et al., 1994).
Inversion Strategy
Prior information, model parameterization, cost function,
efficient forward modeling code, and inversion strategy
play a key role in geophysical inversion. In this study we
propose an inversion strategy to combine MPR and HDIL
measurements. Primarily, joint inversion is performed in
two steps. In the first step, we estimate invasion zone
resistivity and fine tune bed boundary positions that are
derived from the inflection points using the shallow
measurements. If the data generated from the estimated
model matches all the shallow and deep measurements then
the model is treated as the final model of the inversion. In
regions where the data match is moderate, inversion is
performed by introducing invasion into the model
parameterization. In the following sections, the
applicability of the approach is demonstrated on synthetic
and field data examples.
Inversion Examples
Synthetic Oklahoma Benchmark Model Example Data
The synthetic data (HDIL and MPR) for this study were
simulated for a modified Oklahoma formation benchmark
model are shown in Figure 1. Track 1, of Figure 1 contains
2 MHz long-spacing and 400 kHz short- spacing phase and
attenuation MPR data. Track 3 contains 10, 30, 90, and 150
kHz curves for the third subarray of HDIL data.
Parameterization- Dual Model
The dual earth model consists of a 2-D earth model
characterized by a sequence of layers each identified by its
thickness, invasion zone length (Lxo), invasion zone
resistivity (Rxo), and virgin zone resistivities (Rt). The
common parameters for MPR 1-D model and 2-D HDIL
model are layers and Rt.
Forward Modeling and Inversion Algorithm
A 2-D finite difference and a 1-D analytical modeling code
are used to generate the responses for HDIL and MPR data,
respectively. A Levenberg-Marquardt least-squares
algorithm is used as a basis for the optimization process.

Joint Inversion of MWD and Wireline Measurements

Joint MPR and HDIL inversion
We performed joint inversion using 14 HDIL data curves
(7 subarrays and for frequencies 30 kHz and 90 kHz) and 8
MPR resistivity data curves generated from modified
Oklahoma bench mark model. The benck mark model
consists of a large range of layer thickness and resistivities,
with conductive invasion in resistive layers (Chunduru et.
al., (1997)). In Figure 1, track 1 shows Rt, track 2
represents the invasion zone length of the model, and in
track 3, Rxo and Rt are presented as dashed and solid lines,
respectively. In the inversion, HDIL data were generated
using a 2-D modeling code and the MPR data were
generated using a 1-D modeling code. The results obtained
from the joint inversion are presented in Figure 2, along
with the inversion results obtained using MPR and HDIL
data only. Comparison of the invasion zone lengths, true
Lxo (Lxo), HDIL inversion (LxoHDIL), and joint inversion
(LxoJOINT) are shown in track 1. In track 2, Rxo and Rt
corresponding to the true model, MPR (RtMPR), HDIL
(RxoHDIL, and RtHDIL), and joint (RxoJOINT, and RtJOINT)
inversion results are shown. There is good agreement
between the synthetic and final model data, and, also
among the results obtained from the three inversion
schemes, but the model obtained from the joint inversion is
closest to the true model.

from the uninvaded run could match all the raw HDIL and
MPR data, confirming the presence of no invasion that is
observed with the processed curves. The results obtained
from the joint inversion (RtJOINT) are shown in track 1 of
Figure 4. There is an excellent match between raw and
theoretical data generated from joint inversion results. The
RtJOINT obtained from the joint inversion result coincides
with the 60 inch MPR and 120 inch HDIL curves in regions
where they overlap each other. In regions where the two
curves differ, the joint inversion provides a resistivity value
that is consistent with both HDIL and MPR data.

Gulf of Mexico Field Data Inversion

Finally, we apply our joint inversion scheme to MPR and
HDIL data acquired in a well from the Gulf of Mexico over
the interval x290-x400 ft. In track 1 of Figure 3, we show
gamma ray (gr) and caliper (cal) logs, and in track 2 we
show density (den) and neutron porosity (por) and density
corrections (den.cor) logs. The resolution matched 35 and
60 inch MPR curves and 30 and 120 inch HDIL curves are
shown in track 3. The processed curves clearly show the
presence of hydrocarbons in the interval from x320-x365 ft.
In general, the MPR and HDIL curves stack on top of one
another. Lack of separation among the individual processed
curves, e.g., 30 and 120 inch, indicate little or no invasion.
Although there is a good resolution match between the
MPR and HDIL curves, MPR curves tend to show higher
apparent resistivities than the HDIL curves. The differences
between these two data sets can be attributed to different
processing
schemes,
e.g.,
algorithms,
different
assumptions, different borehole conditions, etc. By
performing a joint inversion we try to reconcile the
differences between these data and to obtain a unique
model that can explain both data sets.

Acknowledgements

Joint inversion of MPR and HDIL data is performed using

resistivity and phase MPR data, and 30kHz and 90kHz raw
HDIL data. The layers for the earth model are derived from
processed MPR and HDIL data. The initial guess for Rt is
obtained from the 400 kHz long-spacing phase curve.
Inversion was performed using the proposed strategy. In
the first step, we used an uninvaded model to invert
thickness and resistivity of the layers. The model obtained

SEG 1999 Expanded Abstracts

Conclusions
An inversion strategy to combine HDIL and MPR
measurements that assumes different subsurface borehole
conditions at the time of data acquisition was successfully
implemented on synthetic and field data example from Gulf
of Mexico. The dual model concept can be extended to any
combination of MWD and wireline data responses. The
advantages of performing joint inversion are to: a) validate
the results obtained from different measurement physics at
different times; b) reduce the uncertainty in the estimation
of formation parameters; and c) demonstrate the
consistency among various inversion results.

The authors would like to thank Baker Atlas for the

opportunity to publish this work. Authors would also like to
thank Steve McWilliams, Dan Georgi, Rod Hotz and Chris
Barnette for their support and helpful discussions.
References
Chunduru, R.K., Busch, R., Jervis, M.A., Frenkel, M.,
Mezzatesta, A., Strack, K-M., and Kennedy, D.W, 1997,
Russian resistivity log interpretation using inversion, 3rd
International Congress of the Brazilian Geophysical
Society.
Meyer, W.H., Maher, T., and Mc Lean, D.J., 1996, New
Methods Improve Interpretation of Propagation Resistivity
Data: Presented at the 37th SPWLA Annual Logging
Symposium, paper O.
Zhou, Q., Beard, D., and Tabarovsky, L., 1994, Induction
tool radial resolution, paper presented at the 1994
International Symposium on Well Logging Technology,
Xian, China.

Joint Inversion of MWD and Wireline Measurements

Figure 1. MPR and HDIL data simulated for a modified Oklahoma bench- mark models are shown in
tracks 1and 3, respectively. The Lxo, Rxo, and Rt of the bench mark model are shown in tracks 2 and 3,
respectively.

Figure 2. Comparison of MPR, HDIL and joint inversion results (tracks 1 and 2). True (-) and
estimated (--) data for C2F30, C4F30, RP2ML and RA4KL are shown in tracks 3, 4, 5 and 6
respectively.

SEG 1999 Expanded Abstracts

Joint Inversion of MWD and Wireline Measurements

X300
X350
X400
Figure 3. MPR and HDIL data from Gulf of Mexico. The resolution matched curves of 30 and 120 inch for HDIL, and 35
and 60 inch for MPR data are displayed in track 3. The interval x325-x375 shows the presence of hydrocarbons.

x300
x350
x400
Figure 4. MPR and HDIL joint inversion result (track 1). True (-) and estimated (--) data for C2F30, C4F30, RP2ML and
RA4KL are shown in tracks 3, 4, 5, and 6 respectively.

SEG 1999 Expanded Abstracts