The importance of pore-scale models deals with their ability to estimate macroscopic parameters such as capillary pressure, absolute and relative permeabilities that are important input data of any reservoir model while studying the microscopic structure of the given samples.

The current paper aims to discuss practical aspects of two widely applied techniques for absolute permeability estimation, lattice Boltzmann method (LBM) and pore network modeling. Both approaches analyzed three sandstone and one carbonate samples with different pore structures from the open database of Imperial College London. First, a direct pore-scale modeling via LBM approach was built based on the single relaxation time concept. Later on, pore networks were extracted by using the fast and efficient SNOW algorithm which could be considered as an improved version of a well-known watershed segmentation approach. The paper does not only compare the accuracy of two methods, but rather summarize in which cases a direct pore-scale modeling can be substituted by the simplified pore network approach. It also highlights importance of new insights, for instance, consideration of image resolution and representative elementary volume of the samples as a part of their analysis.

Results demonstrate that direct pore-scale modeling allowed to reach a reasonable accuracy in absolute permeability estimation for three studied sandstone samples, although this method will not properly work for bigger pore volume while running simulations on CPU. At the same time, pore networks were able to find values of absolute permeability with reasonably low errors for all samples by varying the filtering parameters while preparing images.

孔隙尺度模型的重要性在于它们能够在研究给定样品的微观结构时估计宏观参数，例如毛细管压力、绝对和相对渗透率，这些参数是任何储层模型的重要输入数据。

目前的论文旨在讨论两种广泛应用的绝对渗透率估计技术的实际方面，即格子 Boltzmann 方法 (LBM) 和孔隙网络建模。这两种方法都分析了伦敦帝国理工学院开放数据库中具有不同孔隙结构的三种砂岩和一种碳酸盐样品。首先，基于单一弛豫时间概念，通过 LBM 方法建立了直接孔隙尺度建模。后来，通过使用快速有效的 SNOW 算法提取孔隙网络，该算法可以被认为是众所周知的分水岭分割方法的改进版本。本文不仅比较了两种方法的准确性，还总结了在哪些情况下可以用简化的孔隙网络方法代替直接孔隙尺度建模。它还强调了新见解的重要性，

结果表明，直接孔隙尺度建模允许在三个研究的砂岩样品的绝对渗透率估计中达到合理的精度，尽管在 CPU 上运行模拟时，这种方法不适用于更大的孔隙体积。同时，通过在准备图像时改变过滤参数，孔隙网络能够以合理的低误差找到所有样品的绝对渗透率值。