HU is of great importance for the reservoir characterization, which could indicate the fluid behavior and petrography features of the reservoir. During the detailed development of any oil-gas field, the reliable estimation of HU in the reservoir makes an important impact on improving the efficiency of production and predicting the distribution of remaining oil. In this study, a novel method for HU classification and prediction was investigated. Firstly, the clustering analysis was used to classify the HUs of 723 cored samples in the S-95 block, Damintun Sag (China). Since two logging parameters, e.g. true formation resistivity (RT) and shale volume (V_sh), displayed a considerable relationship with flow zone indicator (FZI) through the Spearman's rank correlation analysis, RT and V_sh were used to establish a planar grid of logging bins, also called 2D-crossplot. Notably, the cumulative probability method and the exponential increase method were first applied to construct the intersection bin of logging parameters in the 2D-crossplot. Then the Bayes inference was applied to write the prediction program for HU, and the posterior probability was used to identify the HU for each bin in the crossplot. The crossplot case of V_sh & RT-A reached the desirable overall accuracy rates under 5 HU types condition, ranging from 88.01% to 93.26%, which was higher than that of the artificial neural network (83.29%). It indicated that the HU classification could effectively reflect the reservoir characteristics. Thus, V_sh & RT-A was used to predict HU of un-cored intervals/wells. In addition, a geological body composed of HU was established by petrel software platform based on sequential instruction simulation. During the static verification, the geological body well matched the lithology of the reservoir. R35 of cored samples was consistent with the prediction of HU, and the productivity performance of each group well matched the flow characteristics of the reservoir. And the numerical simulation results showed that the spatial distribution of the hydraulic unit determined the remaining oil distribution. HU#3 has general physical properties but a wide range of residual oil distribution, which is the focus of further development. In general, the system for predicting HU in this study can meet the needs of describing the low permeability reservoir in the Liaohe oilfield to the great extent, and had an applicable value for the promotion in similar sandstone reservoirs.