 |
 |
A Modified Way of Testing Filters for Clay Cores of Embankment Dams
Professor of Civil Engineering, Middle East Technical University, 06531 Ankara, Turkey Tanfer Gürler Gama Industrial Plants Manufacturing Co., Atatürk Bulvari, 211/7, Kavaklidere, 06680 Ankara, Turkey. |
Different forms of testing filters to retain cracked clay cores of dams have previously been proposed. The earliest of these, the perfect filter (PF) test was aimed at designing a filter that would retain the finest particles that would be contained in the eroded material, namely clay flocs. This test has subsequently been found to be too conservative, and an alternative procedure, the no erosion filter (NEF) test, was devised in which a layer of the clay core material is compacted above the filter to be tested, and erosion is induced in a 1 mm to 1.5 mm diameter pinhole in the clay layer by applying a sufficiently high hydraulic gradient. In the present study, the PF test has been modified to test a representative sample of the actual clay core material by continuously stirring the suspension above the filter to be tested, to prevent the deposition of the coarser material on the filter surface. Three different clay core materials have been tested, and the results found to be in close conformity with those of the NEF test. The modified form of the PF test seems suitable for application as an independent check of the NEF test results.
KEYWORDS: clay cores; embankment dams; filters; laboratory tests; perfect filter test; no erosion filter test.
Abstract
Introduction
Experimental study
Discussion
Conclusion
Acknowledgement
References
Different forms of testing filters for protecting the cracked cores of embankment dams have previously been proposed; some of these tests have been summarized by Foster and Fell (2001). The earliest of such tests, the perfect filter (PF) test, was devised by Vaughan and Soares (1982) for testing sand filters aimed at retaining the finest particles, claimed to be clay flocs, that would be carried away by erosion if the clay core of an embankment dam cracks. This procedure was however found to be rather conservative (e.g., Khor and Woo, 1989), as the coarser particles that are also carried along during erosion aid the filtering action. A more realistic way of testing such filters, the no erosion filter (NEF) test (Sherard and Dunningan, 1989), also called the pinhole filter test (Khor and Woo, 1989), was proposed. In the NEF test, a 25 mm thick layer of the core material (the base soil) is compacted in a 100 mm diameter steel mould, above the layer of the sand filter to be tested (the test filter). The base soil layer contains a 1 mm diameter pinhole. Erosion is induced in this pinhole by applying a very high hydraulic gradient to it (of the order of 1600), and the turbidity of the water passing through the filter is examined for five to ten minutes. If the filter retains the eroded soil, water coming through it is clear, and there is no visible sign of erosion of the pinhole; if the filter fails, this water is turbid, and erosion of the pinhole is clearly visible; even in tests in which the discharge water becomes clear within three to five minutes, the filter is taken as unsuccessful. The maximum size of the smallest 15% of the filter (D15) was found to be the dominant property governing the success or failure of a uniform filter, and the filter boundary (D15B) was defined as the value of D15 dividing successful and unsuccessful filters.
Foster and Fell (2001) quote evidence showing that design criteria based on the NEF test are widely accepted by major dam authorities around the world. So, regarding this test, it is desirable to eliminate any doubts as to how representative of the base soil, the periphery of a 1 mm diameter pinhole is, especially when erosion practically stops in a test using a successful filter. The aim of the present study was to see whether the PF test could be modified to enable its use as an independent check of the NEF test results. For this, the PF test was applied using, instead of clay flocs, the same concentration of the actual core material in the soil suspension, and constantly stirring the suspension above the test filter to prevent the coarser particles from settling on the filter surface, inducing self-filtration. The results of this modified perfect filter (MPF) test were compared with those of the NEF test on three different base soils, as well as those of PF tests on clay flocs extracted from the same soils.
Three samples of clay core material, passing the 2.38 mm test sieve, from the Ayvali dam site, some 30 km southeast of Kahramanmaras (37o E, 37.5o N) were chosen as the base soils. The index properties of these are given in Table 1, and the gradation curves, obtained by the hydrometer test using a dispersant shown on the left of Fig. 1. The filters to be tested were made up by mixing sand and fine gravel of different size ranges in appropriate proportions to give gradation curves similar to those shown on the right of Fig. 1, with the D15 sizes ranging between 0.074mm and 2.0mm.
Table 1. Index properties of the base soils tested
| ||||||
| 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| Base soil | Liquid limit (%) |
Plastic limit (%) |
Clay fraction (<2mm) (%) |
Activity | D50 of clay flocs (mm) |
Unified soil classification |
| ||||||
| BS1 | 55 | 23 | 35 | 0.91 | 6 | CH |
| BS2 | 70 | 22 | 54 | 0.89 | 8 | CH |
| BS3 | 87 | 28 | 44 | 1.34 | 2.4 | CH |
| ||||||
Modified perfect filter (MPF) tests
The full lines in Fig. 2 represent the setup essentially used for the perfect filter (PF) test by Vaughan and Soares (1982); the only modification to the apparatus in the MPF test is represented by the dashed lines: a 32mm diameter propeller P driven by an electric motor M, movable along a vertical supporting post. This propeller was introduced into the suspension containing not only clay flocs as in the PF test, but actual core material at the same concentration (25 g/litre) as in the PF test, and run at such a height (180 mm for the minimum speed of 1000 rpm of the available motor) above the top of the test filter TF that the material tending to settle on top of the filter was instantly carried into the suspension without disturbing the filter. The lack of disturbance to the filter was verified by permeability tests, circulating water alone in the system; no change was observed in the permeability of the filter when the water above it was stirred. The 75mm thick test filter was protected by a 25mm thick coarse filter CF (D15 = 2 mm), itself protected by a 25mm thick layer of coarse base filter CBF of material retained on the 4.76mm test sieve.
Plastic tubing of suitable length, 8mm in internal diameter, were attached to the nozzles marked I, OV, and BO, and two standpipes to S1 and S2 (Fig. 2). As a few pilot tests had indicated no difference in the results of tests using ordinary tap water and river water from the dam site, tap water was used throughout this study. Initially, water was pumped into the 500mm long 45mm internal diameter glass cylinder (cf. the 50mm diameter acrylic plastic cylinder used by Vaughan and Soares, 1982), supported on a metallic base MB, through the inlet I by means of an aquarium pump operating in a small plastic container; the excess water ran back into the plastic container through the overflow OV, maintaining a constant head of 325 mm above the test filter. The water passing through the test filter also discharged into the plastic container through the bottom outlet BO. The total amount of water circulating in the system was 3 litres. The permeability k of the test filter was determined.
To apply the MPF test, 75 g of the base soil was first mixed in 200 ml of water for 15 minutes; the mixture was then increased to 800 ml by adding water; 800 ml of clear water was extracted from the continuous circulation system through the outlets OV and BO; the 800 ml soil suspension was thoroughly shaken, poured into the plastic container, and stirred at about half-minute intervals while being pumped into the test cylinder through the inlet I. At about one-minute intervals after the soil suspension reached the test filter, the k value of the filter was determined, and the turbidity of the water passing through the filter examined. As in the NEF test, this examination was done visually. Turbid water discharged through the filter initially, becoming progressively clearer within at most 20 minutes for the successful filters, depending on the D15 size of the test filter.Fig. 3
shows an example of the variation, with the D15 size of the test filter, of the time required for the water discharging through the filter to become clear for the base soil BS2 (Fig. 1). Test filters for which the discharging water remained turbid, or became progressively clearer but was still turbid after 20 minutes were considered as unsuccessful. Fig 4 shows the variation, with the D15 size of the test filter, of k when clear water first discharged and the determination of the filter boundary D15B for the same base soil. The gradation of the boundary filters for the base soils BS1 to BS3 have been denoted as MPF1 to MPF3 respectively in Fig. 1.
The apparatus used and the procedure, outlined earlier, followed in the NEF tests were essentially the same as those used by Sherard and Dunningan (1989) and Khor and Woo (1989). The comparison of the various dimensions and the water pressure used by these authors and those used in the present study is given in Table 2. Sherard and Dunningan (1989) report that test results are independent of the dimensions of the laboratory apparatus and the water pressure used, provided this is high enough to cause the initial erosion on the bottom of the base soil.
Table 2. Comparison of some values used
in the NEF tests with those in earlier applications
| ||||||
| Item | Values used by Khor & Woo (1989) |
Values used by Sherard & Dunningan (1989) |
present study |
|||
| ||||||
| Diameter of cylinder (mm): | 170 | |||||
| Diameter of pinhole (mm): | 1.5 | 1 | 1.5 | |||
| Thickness of base soil (mm): | 25 | 25 | 25 | |||
| Thickness of test filter (mm): | 130 | ~ 80 | ||||
| Tap water pressure (kPa): | 392 | 413 | 250 | |||
| ||||||
The gradation of the boundary filters found by the NEF tests for the base soils BS1 to BS3 have been denoted as NEF1 to NEF3 respectively in Fig. 1. The D15B values of the filters obtained by the MPF tests and by the NEF tests are compared in Fig. 5.
For purposes of comparison, also shown in Fig. 5 are the D15B values obtained from PF tests carried out on clay flocs, estimated between 2 mm and 15 mm in size, with the average size D50 given in Table 1, column 6, extracted from the base soils, following the procedures given by Vaughan and Soares (1982). Stirring the suspension during the PF tests did not have any effect on the test results. By contrast, not stirring the suspension in the MPF tests resulted in D15B values of 0.98 mm, 0.73 mm, and 1.43 mm for base soils BS1 to BS3 respectively, instead of the 0.83 mm, 0.54 mm, and 1.43 mm when stirring was applied. No change in the D15B value for BS3 may be explained by the fact that this soil behaved as a naturally dispersed clay, there being very little difference in the gradation shown in Fig. 1 and that obtained from a hydrometer test without adding a dispersant.
Fig. 5 shows that for two of the three base soils tested the D15B sizes determined by the MPF test are of the same order as those obtained from the NEF tests; for the third base soil, behaving as a naturally dispersed soil, the MPF test gave a less conservative result for the test filter than the NEF test. All MPF test results are clearly larger than the PF test values. The limited number of base soils tested so far indicate that the MPF test may be used as an independent check of the NEF test results for clay core materials with about 80 % finer than 74 mm.
The no erosion filter test is a sound way of simulating the conditions in a cracked clay core, and is widely used for the design of filters for such conditions. One query that may arise regarding this test is how representative of the base soil, the material eroding from around the periphery of a 1-mm pinhole is. On the other hand, there can be little doubt about the material passing through the test filter in the case of the perfect filter test using clay flocs, but this test gives unduly conservative results. In this study, by using actual clay core material instead of clay flocs in the perfect filter test, and by continuously stirring the suspension above the test filter, results have been obtained for three base soils that are either of the same order as those given by the no erosion filter test, or less conservative than the latter. On the basis of the limited number of base soils tested so far, the modified perfect filter test seems a good way of verifying the results of the no erosion filter test for clay core materials with about 80 % finer than 74 mm.
This study was carried out at the Middle East Technical University, and the authors are grateful to Messrs A. Bal, S. Murat, and M. Yildirim for help with the testing, and to Mr. M. Ekinci for the tracings, and to Mr. O. Pekcan for having these scanned.
 | |
| © 2004 ejge | |
