In-place measurement of permeability in heterogeneous media: 2. Experimental and computational considerations

The computational problem pointed out in part 1 of this sequence of papers is evaluated in some detail. The problem arises from the need for compromise between the accuracy of calculated permeability and the ease of performing the calculation. Finite difference methods were expected to provide such a compromise. However, imprecision in numerical differentiation prevented use of the difference method. A study of the accuracy required for measuring permeability indicates that 7.5 times as much variation is allowable in the permeability as in the potential pattern. Since potential patterns are the tools needed to treat practical problems, errors in permeability measurements as great as 37 to 75 per cent may be tolerated and still provide answers within the common engineering accuracy range of 5 to 10 per cent. Linear flow approximations are found to give rather reliable permeability results for lateral flow systems. The geometrical shape of net elements in a properly constructed flow net for heterogeneous soils is utilized in the method. An expression for the approximation error is presented for application to any problem being analyzed.