Influencing factors of rock thermal conductivity and applicability evaluation of its mixing law predictive models

•Lithology is the main reason for the difference of rock thermal conductivity, where special components should also be considered.•Thermal conductivity of dolomite is positively correlated with mineral grain size.•The commonly used temperature correction formula for thermal conductivity is not applicable to all rocks and can be measured in practice when available.•The geometric mean model, together with the voigt-ruess-hill average model, provided the best fit. The thermal conductivity (TC) of rocks is a very important parameter in geothermal theory and applied research, its value is related to rock composition, mineral grains characteristics, rock porosity, and pore fluid etc., and simultaneously is affected by formation thermodynamic environment (temperature, pressure,etc.). Statistical analysis results of the thermophysical property parameters of borehole samples from Jizhong Depression indicated that the TC is mainly controlled macroscopically by lithology, compared with sandstone, mudstone, conglomerate, and gneiss, dolomite has a greater TC. In the dry state, the TC of dolomite samples showed a decrease with increasing temperature, the values of TC decreased by 12.3–27.5% in the temperature interval of 30–130 °C. Moreover, under the same temperature conditions, excluding the effects of water content and mineral fraction, the TC of medium–coarse crystalline dolomite, medium–fine crystalline dolomite, microcrystalline dolomite, and mud-powder crystalline dolomite decreased successively, reflecting the influence of the mineral grain size on TC. The validity and applicability of six commonly used mixing law predictive models for TC were evaluated based on X-ray diffraction (XRD) data from 70 borehole samples. The results showed that Geometric mean model displayed the best fit (goodness of fit, R2 = 0.878) for rocks, and Voigt-Ruess-Hill average also showed a good fit, with an R2 of 0.868, and had the best prediction for mudstone (AME = 10.16, RMSE = 0.25). In cases where the calculated values deviate significantly from the measured values, it may reflect the limitation that the mixing models consider only the composition of minerals but ignores the structural characteristics of rocks.