Synthetic samples of different grain size fractions of quartz were prepared and the thermal conductivities measured with a divided bar apparatus. The measurements were carried out on dried and water-saturated samples both perpendicular and parallel to the main stress direction during compaction of the samples. A clear correlation was observed between the porosity and the measured thermal conductivities. For water-saturated samples the thermal conductivity decreases more with increasing porosity than for ideal isotropic material as predicted by the geometric mean model. A clear positive correlation was also found between the thermal conductivity and the grain sizes of the samples. When the expected effect of porosity on the thermal conductivity was excluded, a linear correlation between the logarithm of the grain size and the measured thermal conductivities of the water-saturated samples was found. An anisotropy effect (a=k 0 /k ⊥ ) in thermal conductivity was observed for the samples, being most prominent for the coarsest ones. The measured anisotropy is assumed to be mainly due to the thermal anisotropic nature of the quartz crystal.

In spite of the fact that thermal conductivity is a key factor in basin modelling, knowledge regarding the thermal conductivity of sedimentary rocks is scarce. In particular, hardly any information exists about claystones and mudstones which can make up 70-80% of a sedimentary basin. For the Upper Jurassic sediments of a section across the northern North Sea, basin modelling programs, using the geometric mean model, gave a deviation of 50~ for the present-day temperature, simply because the matrix conductivity was estimated by two different models. These uncertainties in the determination of thermal conductivities are partly due to problems in measuring, and partly to difficulties in modelling the thermal conductivity. Future work should concentrate on comparative studies and tests on all types of sedimentary rocks, directed towards improved measurement methods. This is of particular importance for clays, mudstones and shales, as measurements on these are associated with the greatest uncertainties. Standardized procedures that include detailed descriptions of sampling, sample preparation and measurement techniques are important to ensure that the thermal conductivity measured is unaffected by factors related to the measurement method. In thermal conductivity modelling it is our experience that the influence of sediment texture is underestimated by most models currently applied. Estimates based on the geometric mean model seem to be most reliable when the determination of matrix conductivity is restricted to the sediment type, mineralogy and texture of the sediments.

No abstract

A parallel reaction model is developed for describing the conversion of smectite to illite. Each reaction represents a group of similar smectite layers that require the same activation energy and have the same iUitization rate. The model considers that the rate-determining reactant is smectite itself which follows first-order Arrhenius kinetics. By modelling the data from hydrothermal illitization experiments and from a Gulf Coast well, the activation energies are found to be distributed in the range of 11-24 kcal/mol with a maximum reaction at 18 kcal/mol, which involves 65% of reactive smectite. A frequency factor in the order of 10-3-10-4/s, obtained from the data fitting, appears to be adequate for modelling natural diagenesis in sedimentary basins. The distribution pattern of activation energies is considered to be controlled by the degree of heterogeneity of the initial smectite and the degree of electrostatic interactions between smectite layers and the newly formed illite layers during reaction.

The claystones and mudstones investigated are London Clay, Fullers Earth, Oxford Clay and Kimmeridge Clay. The thermal conductivities were measured using a divided bar apparatus and the values measured perpendicular to layering ranged from 0.68 to 0.97 W/mK. Comparative measurements of thermal conductivities were carried out by the needle probe method and Middleton's method. Deviations of up to 50% were obtained between the needle probe and the divided bar method. The thermal conductivities estimated from the geometric mean model based on mineralogy and water content ranged from 0.87 to 2.01 W/mK, considerably higher than the measured values. A correlation was found between the grain size distributions of the samples and the measured thermal conductivities. This textural effect on the thermal conductivity is assumed to be the main reason for the low measured values and the lack of correlation between the measured and the calculated values.