Embankment dam seepage evaluation from resistivity monitoring data

Abstract

ABSTRACTMethods for monitoring seepage are important for the safety of embankment dams. Increased seepage may be associated with internal erosion in the dam and internal erosion is one of the main reasons for dam failures. Internal erosion progresses inside the dam and is difficult to detect by conventional methods. Therefore there is a need for new or improved methods. The resistivity method is a nondestructive method that may accomplish this task. It has been tried in an on‐going research programme in Sweden. Daily resistivity measurements are carried out on permanent installations on two Swedish embankment dams. In this paper the installations on the Sädva embankment dam are described and selected parts of the results are presented. In addition, a method for evaluating the seepage from resistivity monitoring data is theoretically described and tested for four selected areas in the foundation of the Sädva dam. Seasonal resistivity variations are apparent in the reservoir as well as inside the dam. Most parts of the dam have a homogeneous resistivity distribution with consistent variations. The overall status of the dam is satisfactory. However part of the foundation demonstrates a slightly different behaviour pattern with regard to the seasonal variation. The four selected areas represent localities with low, intermediate and high variations in seasonal resistivity. The areas are compared qualitatively and thereby permeable zones within the dam may be identified. Quantitative assessment of the seepage flow is also carried out as an initial test of the described method. It is concluded that the experiences from the Sädva dam are valuable with regard to the use of the resistivity method on embankment dams. Resistivity monitoring data may be used to qualitatively assess the seepage through the dam. For quantitative assessment, the method is promising and the data from the Sädva dam constitute an interesting initial approach. However, many assumptions and simplifications are made and more work on refining the method is needed.