This paper presents the findings of an investigation regarding the effects of procedural and equipmental factors on the laboratory measurement of permeability of a bentonite-sand liner. In the experimental program, permeability tests were performed on compacted specimens of a bentonite-sand mixture to study the effect of (1) permeameter type (rigid- versus flexible-wall permeameters), (2) back pressure application, and (3) hydraulic gradient magnitude on the measured permeabilities. The liner material used in the tests was a mixture of 90% P20–R30 Ottawa sand and 10% bentonite clay. Water was used to prepare the specimens and as the permeant. Specimens were prepared by compaction at moisture contents exceeding optimum using the standard proctor method. A pressure of 380 kPa was used in the tests involving back pressure. The low and high hydraulic gradients used were nominally 29 and 290, respectively. Prior to the termination of a test, a red indicator was passed through the permeameters under the high gradient. The specimens were dissected and inspected after the test for the identification of flow areas and channels.
The test results indicated that specimens continue to hydrate during permeation. Unless wetter specimens are used, this continuing hydration interferes with the inflow-outflow balance, depriving the tester of an important check for leaks. Differential hydration throughout the specimen results in different soil structures and zones of flow. For instance, most of the flow may take place in an annular area surrounding a less hydrated core in the center. Since the total cross-sectional area is used in computing the coefficient of permeability, the values may be underestimated. Back pressure, often used to enhance saturation during testing, appears to have a detrimental effect when applied in the rigid-wall permeameters by increasing the potential for formation of channels and side flow. Test results are affected by hydraulic gradient in different ways depending on the type of permeameter. While gradients as high as 360 did not induce piping in the gap-graded liner material tested, the application of very high gradients to accelerate testing is not desirable for a number of other effects observed.