Effect of hydraulic functions in transient seepage analysis through the earth dam and its foundation (Case study: Derik earth dam)

Seepage through earth dam is one of the major problems that threat the stability and the safety of the earth dam. Therefore, this subject is interesting for researchers and engineers meanwhile numerous research and investigation are conducting all over the world. One of the main issues refer to the problem is definition of material properties in right manner and one of the main porous media properties that affect seepage and water flow in dam body is hydraulic conductivity. For homogenies earth dams the effect of hydraulic conductivity in the terms of constant value and the function of matric suction has been explained in literature. In current manuscript, the role of permeability of zoned earth dam has been investigated numerically and the results validated in a zoned earth dam.
In this research, effect of hydraulic conductivity on the unsteady seepage through zoned earth dams was studied by GeoStudio software and SEEP/W module. Hydraulic conductivity was considered in two different states as a constant value and also as function related to soil suction. In order to investigate and validate the accuracy of the achieved results Derik Dam as a zoned embankment dam was selected as a case study structure. The mentioned dam is located in the western region of Salmas city in the West-Azerbaijan, Iran. In this study data, collocated during 4 years of operation, were used to comprise with numerical results. Furthermore, two statistical parameters as RMSE and NSE (Residual Mean Square Efficiency and Nash- Sutcliffe Efficiency) were employed to quantitative comparison between numerical and observation results. In this research, data collected from 7 different piezometers (include 3 in dam core and 4 in dam foundation) were studied. Therefore, numerical results were compared with observed data from monitoring instruments installed in Derik earth dam embankment. The time depending numerical simulation of seepage for two different conditions: hydraulic conductivity as a function and a constant value equal to saturated permeability started based on the same initial conditions. For the initial condition the water level at the reservoir was assumed at the normal level. Then, the transient analysis for each state (variable and constant hydraulic conductivity value) conducted by employing real four years fluctuation of water level in reservoir, as upper boundary condition. The time steps involved in transient analysis has about 30 day’s intervals. However for some cases the comparison among simulated and observed data conducted around 90 day’s intervals as seasonally.
Results of this study revealed that pore water pressure and phreatic line through the embankment would be determined correctly while variable hydraulic conductivity was considered in seepage analysis, the error values were reduced up to 10%. Comparison between results of two conditions of applying hydraulic conductivity -as a constant value instead of a function- showed that the pore pressure was estimated higher, in some case higher than 200 times, while permeability to be considered as function of suction whereas lower seepage flow rate was calculated. Evaluation of seepage flow through embankment and foundation of a dam confirmed advantages of using hydraulic conductivity function instead of a constant value. Investigation about statistical parameters of NSE and RMSE confirmed that the model could not estimate the pore pressure in the core very well but the accuracy of model was better while the hydraulic conductivity was defined as function instead of constant saturated value. The model could estimate correctly the pore pressure in foundation with fairy high accuracy in the both state of employing permeability as matric suction and/or constant. Of course, the results showed upper estimating of pore pressure in down side of the cut-off wall. This perhaps is because of well water proofing of cut-off wall in breaking and reducing of pore water pressure, even better than expected values. However, by the results, for saturated regions or for the confined media employing of constant value for hydraulic conductivity would have the same results while hydraulic conductivity is used as a function. Of course, the experience of model running revealed that the convergence in case of constant permeability was better and the time of modeling was shorter. Therefore, in order to fast performance of the numerical model, considering hydraulic conductivity as constant value for fully saturated zones of dam is highly recommended.