Modeling the complex and long term swelling behavior of argillaceous rocks

• Published in: Mining science and technology (China) 2011-09, Vol.21 (5), p.655-659
• Main Authors: Doostmohammadi, R., Mutschler, Th, Osan, C.
• Format: Article
• Language: English
• Subjects: Argillaceous rock; Artificial neural network; Artificial neural networks; Correlation; Learning theory; Long term swelling potential; Mathematical models; Neural networks; Rocks; Short term swelling potential; Swelling; 人工神经网络; 压力测试; 复杂性; 时间依赖; 泥质岩; 溶胀; 电动工具; 行为建模
• ISSN: 1674-5264; 2212-6066
• DOI: 10.1016/j.mstc.2011.10.007

The swelling behavior of argillaceous rocks is a complex phenomenon and has been determined using a lot of indexes in the literature. Determining the required modeling indexes that need to be performed requires expensive tests and extensive time in different laboratories. In some of the cases, it is too difficult to find a relation between the effective variables and swelling potential. This paper suggests a method for modeling the time dependent swelling pressure of argillaceous rocks. The trend of short term swelling potential during the first 3 days of the swelling pressure testing is used for modeling the long term swelling pressure of mudstone that is recorded during months. The artificial neural network （ANN） as a power tool is used for modeling this nonlinear and complex behavior. This method enables predicting the swelling potential of argillaceous rocks when the required indexes and also correlation between them is unattainable. This method facilitates the model of all studied samples under a unique formulation.