Determination of priorities and importance of some geotechnical factors in grouting of karst in large dam sites with decision tree model (Cart). The case study of salman farsi dam, fars province, southern iran

Abstract

As a consequence the leakage from karstic dam foundations and abutments, reservoir storage will be reduced. Due to the complexity of dam sites in karst, estimation of the cement to be used in the injecting boreholes and operating the sealing membrane cannot be readily done. Data mining with decision tree algorithm and searching of databases can help us to find the effective parameters in cement take. In this study, Salman Farsi Dam in Fars province in southern Iran is selected. In this article as a case study, in the left gallery at level of 853 a.m.s.l, 6 boreholes were selected. Important information such as the depth of the borehole drilling, shaft angle, and cement take at different depth intervals, hydro-mechanical behaviors (ground water flow type) gathered with Lugeon tests and were analyzed. 47.76% of the test data indicates Lugeon ranges in impermeable to very low permeability (3 < Lu ? 10). Application of the technique of decision tree showed that when the angles of boreholes lie between 29 and 31.5°, depth of the borehole is less than 32.5 m, Lugeon greater than 15.5 and Hydro-mechanical behavior belongs to dilatation, wash out or void filling, cement take will follow the empirical Ewert (1985) theory. If the angle of the hole is higher than 31.5 degrees, flow type is void filling or linear, and Lugeon greater than 3.5, cement take and luegon number has not complied. Thus we can order the role of effective parameters on cement take as follows: angle of shaft, hydromechanical flow type, average Lugeon, and finally borehole depth. Eventually, the validity of the model is tested.