You are on page 1of 4

In search of the well tie: What if I dont have a sonic log?

By STEVE ADCOCK
Mitchell Energy and Development Corporation
The Woodlands, Texas

E ffective seismic interpretation requires correlating seismic


data with well data. This is usually accomplished by visually
correlating a synthetic seismogram, generated from sonic and
density logs, with nearby seismic data. If a sonic log is not
available but a density log is, then Gardners relationship is
generally used to estimate sonic data. For those situations
where synthetic seismograms are derived but sonic or density
log data are not available, techniques have evolved to estimate the sonic traveltime from other well log data, primarily
the resistivity curve.
Sometimes multilinear regression between sonic data and
other log data in similar wells is used to obtain coefficients
for equations that can then be applied to logs in the well where
the sonic estimate is desired. More commonly, geophysicists
rely on Faust or Smiths equations (Table 1) to estimate the
traveltime from the resistivity. Frequently, an interpreter also
has access to either a gamma ray or an SP log, so I have
included in Table 1 an alternate method which can easily be
implemented with todays computers; it is a time average
equation
AT = f R ATsaturant + (l - f R) ATmatrix
where AT is estimated traveltime in ps/ft, fR is fractional
Porosity> wm.lrant is the estimated sonic traveltime of the rock
fluids, ATm&-ix is the estimated sonic traveltime of the rock.
Making rough estimates of porosity and lithology for entire
logs has become relatively easv using workstation and PC

software. For example, estimates of porosity may be made


from resistivity, neutron, or density data and estimates of
shale volume may be made from gamma ray or SP data.
Figure 1 shows a comparison of results in estimating
traveltimes using the equations in Table 1 over deeper log
intervals in three different Texas basins. In these better consolidated rocks, the Faust equation tends to overestimate
traveltime about 11 ps/ft, while the Smith equation tends to
underestimate it about 14 @ft. At 10 000 ft/s thats 11 percent
and 14 percent velocity error, respectively. That means a
Faust synthetic will appear about 11 percent longer in time
and a Smith synthetic about 14 percent shorter in time than
one made from a sonic in the same well. Figure 1 also implies
that a time average equation will provide a better estimate of
average velocity than can be expected from Faust or Smith,
at least in better consolidated rocks. The time average approach also provides a better average velocity estimate than
might be expected from NMO corrections; the Faust and
Smith equations do not.
In Figures 2-6, actual sonic logs are the basis for comparison of the three methods. In each figure, the gamma ray or
SP, the resistivity, and the estimated lithology are shown
Beside these are plots of sonic logs estimated by the three

Table 1. Three equations for estimating sonic traveltime


Equations
(AT in ps/ft)

Variables

Constants

Z = depth in ft
R = apparent
resistivity in Q-m

a =513.3
b =-1/6
c = -1/6

R = apparent
resistivity in Q-m

a = 94.2
b = -0.15

Faust:
AT=aZVC

Smith:
AT=a Rb

Time-average (two-lithology):
AT = ATF .fR + [hT,h fsh + ATsd t1 - .fsh)l (1 --j-R>
AT F = fluid traveltime
ATsh = shale traveltime
ATsd = sand traveltime
= fractional shale volume
fsh
= fractional porosity
fR

Figure 1. Errors in traveltime estimates using the three


methods in Table 1. Data were 90 000+ depth sampled
values in 22 wells at depths 22000 ft in the Forth Worth
Basin, SO00 ft in the East Texas Basin, and 27000 ft in
the Texas Gulf Coast.

DECEMBER 1993 THE LEADING EDGE 1161

Figure 2. Log data from the Texas Gulf Coast, Lavaca County: Gamma ray, resistivity, roughly estimated lithology,
estimated sonic logs (blue) using the time average Faust and Smith equations overlain on the actual sonic data (magenta).
Differences between sonic log data (DT) and estimated values (Tavg, Faust, Smith) are shown beside these overlays. The
zero line (orange) in these difference plots means zero error in estnnating the sonic data.
methods from Table 1 (blue) overlain on the actual sonic data
(magenta). Differences between sonic log data and estimated
values (DT minus the estimated sonic data) are shown beside
these overlays. The zero line (red) in these difference plots
means zero error in estimating the sonic data. Figure 6 is an
expanded view of a portion of Figure 5.
Errors in estimating traveltime using resistivity are especially large where average sonic traveltimes exceed about 100
ps/ft (usually the shallower well data). This is well demonstrated by the large residual errors left by all three methods
for shallow intervals of log data. The depth term, z-l/6, in the
Faust equation does help a little in reducing the transit time
error in the shallower log intervals (Figure 2) which is not
particularly surprising, since all Fausts data were over very
shallow log depths by todays standards. Fausts depth term
acts as a low-frequency correction that I have found can be
more accurately defined by checkshot correcting the estimated
sonic logs with seismic stacking velocities near the well.
Basically, the resistivity (porosity) and lithology data provide
the high frequency information, while the checkshot supplies
the low-frequency trend in the estimation of the sonic log.

ppendix. The two-lithology matrix form of the time average equation I used in these examples is

1162 THE LEADING EDGE DECEMBER 1993

AT= AT,fR + v&l fsh + AT, (1 - f,h>l (1 - fR)


where AT is traveltime, f R is fractional porosity (the decimal
equivalent of the percentage of total volume) calculated from
deep induction resistivity, AT, is traveltime representing the
shale portion of the matrix, AT, is traveltime representing the
sandy portion of the matrix, ATF is traveltime representing
the pore fluid, and fsh is the fractional shale volume calculated
from gamma ray data. The examples in Figures 2-6 have had
traveltime specifically estimated using
AT= 189f, + [9Of,, + ml - f&l (1 - fR)
I calculated fractional porosity, f R, from resistivity (to make
the comparisons more meaningful) at each log sample, using
Archies equations with n = 1, a = 2, m = 0.81, and assuming
100 percent saltwater saturation with a constant resistivity
(R,,,) = 0.045 Q-m. The specific equation I used here was
fR = O.l9/?R
where R is the deep induction resistivity values without
correction.
In practice, I choose constants on the basis of the nearest
available data and occasionally make corrections for Rw as a
function of depth. Matrix (in this case sand and shale) transit

Figure 3. Log data from offshore Texas (sequence same as Figure 2).

Figure 4. Log data from the Fort Worth basin (sequence as in Figure 3).

times can be estimated from the nearest available sonic data


over wet zones without washout problems at about the same
depths/pressures of interest. Porosity can also be estimated
from whatever source is available without reference to a
resistivity log.
I calculated fractional shale volume using
fssh = 0.33(22G - 1)

where G is the gamma ray index, ( GR observed - GR sand )/


(GR,hale - G&and)a GR&served is the gamma ray log value,
GLd is the estimated sand baseline value and G&hle is the
estimate shale baseline value. Calculating this can be simple
on a computer and is often click and point level software
in the workstation environment. Where a simplistic approach
(as in these examples) is taken, I have found that younger rocks

DECEMBER 1993 THE LEADING EDGE 1163

Figure 5. Log data from East Texas basin.

Figure 6. Expanded view of a portion of the log data in Figure 5.

require a faster shale baseline value or, as shown here, an


unreasonably high sand content in the lithology estimate.
The ability to control meaningful lithological parameters
when estimating the sonic response makes this a very useful
tool for stratigraphic modeling. Over intervals of a few hundred feet, the assumption used in these examples will often
provide a good fit to sonic data after application of a small
bulk shift.

1164

THE LEADING EDGE

DECEMBER 1993

Steve Adcock earned BS degrees in


physics (Louisiana State University)
and in geology (Centenary College),
and on MS in geophysics (University of
Houston). He starred in the oil business
as a geophysicist with Texacos Hourton ofice and now works for Mitchell
Energy as a senior staff geophysicist.

You might also like