Geometric Similarity Recognition And Modeling Reliability Evaluation Of 3D Digital Core Model

The advanced technology that uses Computing Tomography(CT)scanning to build highdefinition digital core models and numerically analyzes the physical properties(i.e.electrical and mechanical)has been widely used in rock physics,oil/gas resource evaluation and pavement subgrade detection,etc.,due to its advantages over traditional laboratory measuring,such as high efficiency,sufficient data,as well as convenient analysis and storage.However,building 3D digital core models involves complex procedure and various image processing algorithms that may have multiple effects on characterizing the model geometries.As a consequence,significant errors may exist in the following analysis for the physical properties of digital core models,therefore cause misjudgment in the production and application.In order to fully understand the impact of image processing algorithms on the geometric and physical characterization during 3D digital core modeling,this dissertation investigates various deviation caused by modelling algorithms,and puts forward a method to recognize the geometric similarity between the 3D gray-scale models and 3D binary models.Based on the method,an evaluation criteria for digital core modeling reliability can be established.The criteria can be used to identify the digital core model with the geometric structure closest to the original core sample,thus ensuring modeling accuracy.The research serves well as a valuable reference for engineering applications.The research conducts as follows:(1)Use CT scanning to obtain multiple sets of rock images and build gray-scale digital core models.Then,several image processing algorithms are used to obtain massive binary models.Two digital cores with pore and grain,respectively,are obtained and used for core property analysis.(2)Analyze the effects of image processing algorithms on the geometric and physical properties during the process of digital core binary modeling.First,using various image processing algorithms and parameters to characterize the geometric properties of digital core and finite element method(FEM)is used to calculate the electrical property of digital core.Then,the effects of multiple threshold segmentation and morphological algorithms on different digital core models are analyzed.(3)An evaluation criterion for digital core reliability based on the geometric similarity is proposed and can be used to quantify errors in the modeling process.The criteria can be used to identify the geometric property of gray-scale digital core models and binary models,calculate the similarity between the two models,and provide quantified evaluation for the effects of digital core modeling.By comparing the simulated data with the measuring data,the results show that the method proposed in this paper can obtain the optimal digital core model with the best evaluation results,and the model has minimum deviation from the measuring data.The method can be used to recognize digital core modeling reliability and reduce errors in the process of modeling,thus providing guidance for industry production.