As the dielectric constant and conductivity of petroleum products are different from those of the pore water in soil, the electrical resistivity characteristics of oil-contaminated soil will be changed by the corresponding oil type and content. The contaminated soil specimens were manually prepared by static compaction method in the laboratory with commercial kaolin clay and diesel oil. The water content and dry density of the first group of soil specimens were controlled at 10 % and 1.58 g/cm(3). Corresponding electrical resistivities of the contaminated specimens were measured at the curing periods of 7, 14, and 28 and 90, 120, and 210 days on a modified oedometer cell with an LCR meter. Then, the electrical resistivity characteristics of diesel oil-contaminated kaolin clay were discussed. In order to realize a resistivity-based oil detection method, the other group of oil-contaminated kaolin clay specimens was also made and tested, but the initial water content, oil content, and dry density were controlled at 0~18 %, 0~18 %, 1.30~1.95 g/cm(3), respectively. Based on the test data, a resistivity-based artificial neural network (ANN) was developed. It was found that the electrical resistivity of kaolin clay decreased with the increase of oil content. Moreover, there was a good nonlinear relationship between electrical resistivity and corresponding oil content when the water content and dry density were kept constant. The decreasing velocity of the electrical resistivity of oil-contaminated kaolin clay was higher before the oil content of 12 % than after 12 %, which indicated a transition of the soil from pore water-controlled into oil-controlled electrical resistivity characteristics. Through microstructural analysis, the decrease of electrical resistivity could be explained by the increase of saturation degree together with the collapse of the electrical double layer. Environmental scanning electron microscopy (ESEM) photos indicated that the diesel oil in kaolin clay normally had three kinds of effects including oil filling, coating, and bridging. Finally, a resistivity-based ANN model was established based on the database collected from the experiment data. The performance of the model was proved to be reasonably accepted, which puts forward a possible simple, economic, and effective tool to detect the oil content in contaminated clayey soils just with four basic parameters: wet density, dry density, measured moisture content, and electrical resistivity.
