煤矿高温火区发展蔓延过程具有明显的介尺度特性。 基于第一性原理，讨论煤矿高温火区 发展过程中，介尺度对微观传热机理与宏观能量迁移现象的关联作用，有助于推动煤自燃及热动力 灾害领域的深入研究。 首先，采用离散元方法(DEM)模拟煤颗粒下落、堆积及压实过程，建立数字 堆积结构;基于单煤颗粒尺度与采空区尺度之间的介尺度II，建立煤颗粒堆积体内流动及换热过程 的计算模型，经过实验验证该模型温度场预测平均误差小于 1.77%，具有较高精度。 其次，以所建 立的颗粒堆积结构及相关数值计算为基础，发展了适用于煤矿隐蔽火源发展蔓延规律研究的有效 导热系数尺度上联方法。 此方法以温度场数值计算结果为基础，以最小二乘优化算法为核心，有效 导热系数计算结果可以准确预测真实温度分布，有效地实现了单煤颗粒尺度到采空区尺度的信息 传递。 再次，计算了有效导热系数在颗粒雷诺数(Rep)介于 3~6 000 区间内的变化规律，结果表明 有效导热系数在该范围内与Rep成线性关系。 最后，提出了采空区内有效导热系数尺度上联经验 公式。 在其适用范围内(3<Rep <6 000)，该公式平均误差为 7.17%;针对采空区内实际情况(Rep < 200)，该公式平均误差为 4.50%。

Mesoscale characteristics are the intrinsic features of the development of fire zones in coal mines.To investi⁃ gate the effect of mesoscale on the connections between the microcosmic mechanism of heat transfer and the macro⁃ scopic phenomenon of energy variation during the development of fire zones is of great significance for the studies on coal spontaneous combustion and coal mine thermodynamic disaster.Firstly，the Discrete Element Method (DEM) was employed to simulate the falling，packing and compacting of coal particles，which generated the packing structure of coal particles.Based on the mesoscale II between the coal⁃particle scale and the gob scale，the mathematic model was built for the fluid flow and heat transfer in coal particles packing.The high⁃accuracy of the model was validated with the mean error of 1.77%.Secondly，on the basis of the generated packing and corresponding simulations，an up⁃ scaling method，i.e.the inverse operation method of effective heat conductivity，was developed specializing in the devel⁃ opment of fire zones.The proposed method utilized the numerical temperature solutions and nonlinear least⁃square fit⁃ ting method to calculate the effective heat conductivity that could bridge the gap between two boundary scales and pre⁃ dict the heat transfer process in the gob accurately. Thirdly， the effective heat conductivities were calculated according to the Rep ranging from 3 to 6 000.The results showed that the effective heat conductivity and Rep presented a linear relation.Finally，an empirical equation for up⁃scaling was developed.For 3<Rep <6 000，the mean error of the equation is 7.17%;for the practical gob (Rep <200)，the mean error of the equation is 4.50%.