Page 182 - The Petroleum System From Source to Trap

174 Deming

south-north cross-section. Heat flow estimates are also the following empirical correction, calibrated from the
shown, along with error bars. The heat flow estimates data of Birch and Clark (1940):
were derived from thermal conductivity measurements
along with high-resolution temperature logs made in the k(O) = k(25)[1.007 + 25(0.0037 - 0.0074/k(25))] (7)
upper few hundred meters of the same boreholes in
which the BHTs were measured (Deming et al., 1992). To k(D = k(O) [ 1.007 + T(0.0036 - 0.0072/k(O))] (8)
/
obtain accurate measurements of temperature in this
manner, it was necessary to compensate for drilling where k(25) is the thermal conductivity measured in the
disturbances by making repeated temperature logs over laboratory at 25°C, k(O) is the inferred thermal conduc­
a period of about 7 years and extrapolating to equilib­ tivity at oac, and kCD is the estimated thermal conduc­
rium conditions (Lachenbruch et al., 1987, 1988). The tivity at in situ temperature T. The thermal conductivity
temperature field estimated from BHTs and heat flow of water can be calculated as a known function of
estimates were thus derived from independent data sets. temperature (Touloukian et al., 1970):
However, both show a dramatic change in thermal state
across the basin; heat flow and thermal gradients are kw = a + bT + c'fl. (9)
both much higher in the north than in the south. The
utility of heat flow estimates is the demonstration that where kw is the thermal conductivity of pure water in
the change in thermal gradients cannot be ascribed W /m K (watts per meter degrees Kelvin). For 0 :;;; T :;;;
merely to changes in thermal conductivity related to 137°C, = a 5.65 x 1G-1, b = 1.88 x 1G-3, c = -7.23 x 10"-6; for
changes in lithology. Deming et al. (1992) considered a 137 :;;; T :;;; 300°C, a = 6.02 x 1Q-1, b = 1.31 x 1Q-3, and c =
number of hypotheses to explain the observed variation -5.14 x 10-6. Compared to the rock matrix, the thermal
in thermal state, but they concluded the only reasonable conductivity of water is relatively low. At 0°C, kw = 0.56
explanation involved a basin-wide groundwater flow W /m K, at 100°C, kw = 0.68 W /m K. Thus, the bulk
system driven by topography in the Brooks Range and thermal conductivity of a porous rock is inversely corre­
its foothills. Further speculation suggested that this same lated to porosity and temperature.
flow system may have been the mechanism by which Comprehensive compendia of thermal conductivity
hydrocarbons migrated to Prudhoe Bay. The North data are given by Clark (1966), Kappelmeyer and Haenel
Slope basin is thus an outstanding example of integrating (1974), and Roy et al. (1981). The in situ thermal conduc­
information from traditional heat flow studies with tivity of most sedimentary rocks is in the range of about
analysis of BHTs measured in oil and gas wells. The 1 . 0 -4.5 W /m K, although some lithologies fall outside of
resulting estimates of temperature and heat flow allow this range. Most coals are probably less than 1.0 W /m K
important inferences to be made concerning the and can be as low as 0.25 W /m K. In contrast, halite and
mechanism of oil migration in the basin and also consti­ quartzite are about 5-7 W /m K In general, the thermal
tute a present-day boundary condition on thermal conductivity of most clastic sedimentary rocks is
history and maturation studies. inversely correlated to their clay content. Most shales are
probably less than 1 . 5 W /m K (Blackwell and Steele,
Thermal Conductivity 1989), while clean sandstones commonly have thermal
conductivities of about 3-4.5 W /m K. Carbonates tend to
The thermal conductivity of rocks and sediments is an mostly fall in the range of 2-3 WI m K. These are all
intrinsic physical property that is determined by miner­ rough approximates that have common exceptions. It is
alogy, porosity, and temperature. Most sedimentary therefore risky to estimate (guess) thermal conductivity
rocks are an aggregate of minerals with pore spaces on the basis of lithology alone. The average error in such
saturated with saline water. Their bulk thermal conduc­ estimates is usually at least ±30-40%, and the maximum
tivity depends on both the solid rock component and the error may exceed 100% (e.g., see Issler and Beaumont,
pore fluid. A number of different mixing models have 1986).
been proposed to relate the thermal conductivity of an Due to the unacceptably large error involved in
aggregate to its individual components (Woodside and estimates based on lithology, it is necessary to measure
Messmer, 1 9 6 1 ) . The most common of these is the the thermal conductivity of representative samples in the
geometric mean model: laboratory if accuracy levels greater than ±30-40% are
desired. Measurements are commonly made on cores or
(6) drill chips using a divided bar apparatus (Sass et al.,
1971, 1984b; Galson et al., 1987) or a needle probe (line
where <jl is the fractional porosity, k r is the thermal source) (Von Herzen and Maxwell, 1959; Sass et al.,
p
conductivity of a porous rock, km is the thermal conduc­ 1984a; Lee, 1989). The absolute error of most measuring
tivity of the matrix, and kw is the thermal conductivity of devices is usually about ±5%; however, the corrections
the pore fluid (usually water). needed for in situ porosity and temperature commonly
Over the range of temperatures found in sedimentary raise this number to about ±10%. Additional uncertain­
basins, matrix thermal conductivity tends to decrease ties concerning anisotropy and sample bias, particularly
with increasing temperature. Most measurements are important when working with drill chips, can lead to
made in the laboratory at -25°C and then corrected for uncertainties of ±10-20% (e.g., Sass et al., 1992).
estimated in situ temperatures. Sass et al. (1992) suggest Figure 9.7 shows a profile of thermal conductivity