An Approximate Finite Element Analysis for Unconfined Flow through Layered Porous Media

Abstract

When the effect of seepage control and the safety of banking structure (for example, embankment dams, river embankments) and natural foundations are examined, the seepage analysis is very important. Most natural foundations consist of several layers which have mutually different physical properties and it must be considered that many banking structures consist of several layer in the light of constructional properties, too.
For free surface seepage, most of the previous finite element procedures involve modifications of the finite element mesh below the free surface with successive iterations. But, it is difficult to analyze free surface seepage through the layered porous media by these procedures because of the modification of the finite element mesh.
There are some finite element procedures based on the invariant finite element mesh concept which were investigated by Desai, Bathe & Khoshgoftaar and Neumann. However, under the present condition the.procedure investigated by Neuman (N procedure) is impractical because it has physical constants which are difficult to determine without knowing the unsaturated seepage.
In this paper, the suitability of the finite element procedures investigated by Bathe and Koshgoftaar, and Desai (B-K homogeneous procedure, and D procedure) only for analyzing the free surface seepage through homogeneous porous media were studied for analyzing the free surface seepage through layered porous media along with their respective characteristics.
It was concluded that the two approximate finite element procedures are proper for analyzing free surface seepage through layered porous media.
Following are the characteristics of the two procedures:
1. The free surface locations by B-K procedure are affected locally by the finite element mesh.
2. The free surface location by D procedure is affected wholly by the finite element mesh.
3. Sometimes, because of the characteristics 1. and 2. above, the free surfaces located by the two procedures have hardly any difference. Since both the procedures utilize almost the same input data, the same model should be analyzed to check the free surface location by cne procedure with the other.
4. The discharge by B-K procedure has generally bigger reliability than the one by D procedure.
5. With B-K procedure, approximate discharge is calculated fairly easily because only small computation error arises even if free-surface is not taken into consideration.