 |
 |
Geotechnical Evaluation of a Morning Glory Spillway Failure Founded on Expansive and Dispersive Clay
Associate Professor, Civil Engineering Department, and Dean Civil Engineering MCE, |
Abstract
Surla dam, an irrigation/storage dam was constructed in 1985 across a small river in Chakwal area of Potohar Plateau of Pakistan. A morning glory spillway structure was constructed on right abutment of the dam for handling excess/flood water.The morning glory lip and upper concrete slabs of glacis and stilling basin are resting on shale type foundation material whereas the lower concrete slabs ofglacis rest on sand stone.
KEYWORDS: Spillway failure, swelling clays, erodable soils, foundation heave
INTRODUCTION
Surla Dam (See Dam.mpg Please see editor's note on movies) is a 25 m high, 250 m long, homogeneous earthen dam resulting in a reservoir with gross storage of 2 million m3 ( Surla Dam Project Report, 1976). A typical cross section of Surla dam is presented in Figure 1a.
Figure 1 (a). Typical dam cross section
Records show that detailed investigations pertaining to site geology, construction materials, hydrology, irrigation requirements etc., were carried out prior to initiation of the project.
A concrete spillway structure has been provided to cater for floodwater.
Figure 1 (b) Plan view of the spillway
A stilling basin has been provided at the down stream end of the spillway to induce hydraulic jump for proper dissipation of energy. Two bore holes were put down in the spillway area and it was concluded that the rock units in the spillway area are the same as those at the dam site i.e., alternating beds of sandstone and clay are present in the spillway area as well. The alignment of the spillway structure is across the strike direction. The concrete spillway structure rests mostly on clay beds except for the last two rows of concrete panels and stilling basin that rest on sandstone bed (See Bed2.mpg Please see editor's note on movies).
Figure 1 (c). Cross section of the spillway
During first event of spill the spillway structure experienced swear distress. Most of the concrete slabs resting on clay were washed down and expensive repairs were required to salvage the project. This paper reports findings of the investigations undertaken to determine causes ofspillway failure.
FIELD & LABORATORY TESTING RESULTS
Three locations at site, two on the upstream of the semi circular crest and one on theright abutment just next to the stilling basin were selected for the collectionof undisturbed block samples and in situ testing. The samples were taken out from the shale type foundation material that has not been subjected to moisture change due to ponding. The field tests were also conducted on the same material. The index tests yielded a liquid limit of 40, Plastic limit of 18 and shrinkage limit of 13. The hydrometric analysis indicated that the soil containsapproximately 25 % clay. The activity of the Surla shale therefore comes out to be 1.10. The soil classifies as Clay of low plasticity.
Standard penetration tests conducted on the foundation material yielded average blow count of 40 for 3 m depth. The corresponding allowable bearing capacity is 4.8 kg/cm2 under undisturbed state. The same soil yielded a low blow count of 7 when the foundation was saturated yielding allowable bearing capacity of 0.7 kg/cm2. Equation proposed by Peck et al., (1974) for 25 mm settlement was used for these computations. Plate Load test conducted in the field in dry conditions on the U/S of the spillway crest yielded ultimatebearing capacity of 11.4 kg/cm2 corresponding to 25mm settlement (Figure 2). UnconfinedCompression test was conducted on nine undisturbed specimens carefully carved out of block sample. These tests yielded average unconfined compression strength of 12 kg/cm2 which is inconformance with the results obtained from SPT and plate load test.Unconsolidated Undrained Triaxial Tests were conducted on two specimens. These specimens were saturated prior to shearing and yielded undrained strength of 0.9 kg/cm2. Thisindicates that saturation of this material can cause ten times reduction in compressive strength. Free swell test (Katzir & David, 1968, Das, B.M, 1990) was conducted in the odometer cell under a small surcharge of 0.07 kg/cm2. The test was conducted on two specimens (Figure 3) and for each the specimen swelled from original height of 20 mm to 22 mm. The swell is, therefore, 2 mm and the value of free swell comes out to be 10%.
The swell pressure of Surla shale was determined throughthe restrained swell test. The specimen was placed in an odometer under a small surcharge of about 0.07 kg/cm2. The load on the specimen was increased periodically after it was inundated so that the height of the specimen remained constant. The vertical stress necessary tomaintain Zero volume change is the swelling pressure. When this test was conducted on Surla clay it yielded a swell pressure of 1kg/cm2. The final moisture content of the specimen was14.4 %.
Figure 2. Results of plate load test (right abutment)
Figure 3. Results of unrestrained swell test
A number of unrestrained swell tests were also conducted on Surla clay. An initial pressure of 0.1 kg/cm 2 was applied over each specimen to simulate field loading from the concrete slab. The specimen was inundated and allowed to swell. Once the swelling seized the specimen was loaded. The pressure required to bring back the specimen to the original height is the swelling pressure. Typicaltest results reported in Fig. 4 and Fig. 5 indicate that the average swell pressure is 2 kg/cm2.
Figure 4. Results of unrestrained swell and compression test in Oedometer
Figure 5. Results of unrestrained swell and compression test in oedometer
ANALYSIS OF LAB & FIELD TEST RESULTS
Collapsible soils occur frequently in this region and it was speculated that failure ofspillway occurred due to soil collapse upon saturation.
Figure 6. Liquid limit and insitu dry density of Surla clay
plotted on the curve proposed by Mitchell and Gardener (1975)
Figure 7. Identification of potentially swelling soils using
activity and % clay (after Seed et al, 1952)
These results confirm that the tables and curves developed by various agencies and researchersare reasonably accurate for predicting swelling potential of a soil based onsimple index tests.
ANALYSIS OF THE SPILLWAY FAILURE
As stated earlier the morning glory spillway structure at Surla dam is resting on alternate beds of shale and sand stone. The concrete spillway structure was constructed after excavating a deep trench in the right abutment of the main dam. The concrete structure rests on shale in its upper portion where as the last two rows of concrete panels for sloping glacis and stilling basin rest on sand stone and weakly cemented conglomerate beds. At the top the thickness of the concrete slab is only 0.5 m that increases gradually down the slope of glacis. The downward pressure exerted by the self weight of slab, therefore comes out to be around 0.1 kg/cm2. As stated above the foundation is safe against bearing capacity failure. Oedometer tests (Figures 4 and 5) conducted on specimen prepared from undisturbed block specimens yielded most important results that seem to provide insight into failure mechanism. Results of all specimens collected from U/S, center and D/S portions of the spillway yielded similar results i.e., the soil expanded when it was exposed to water. The free swell of specimens was in the order of 10 %. Another set of Odometer tests indicated that the swelling pressure for the foundation material of Surla dam spillway is 1.0 kg/cm2. This swell pressure is roughly ten times the downward pressure being exerted by the slab itself. This explains the lifting of slabs upon first filling of the reservoir for it is well known that swelling can initiate failure (Hudson, J.A and Harrison, J.D, 1997; Katzir & David 1968 ; Franklin, J.A and Dusseault, M.B 1989). Swelling of foundation shale upon exposure to moisture definitely took place. The heavy cracking of the slabs and possible opening up of joints can easily be visualized if the process of first ponding is taken into consideration. As water was stored in the reservoir for the first time, a waterfront within the foundation soil started to move towards the spillway structure. One can conceive that the waterfront started to move from U/S of the spillway towards its D/S end. Since the permeability of the foundation clay is reasonably low, in the order of 10-5 cm/s , it is speculated that the movement of this front was slow. As soon as the foundation material under the concreted U/S portion of the spillway received water it expanded or swelled. The swelling clay exerted uplift pressure on the slab whereas the slab tried to contain this expansion. The D/S portion of foundation soil, however, was not exposed to water due to slow movement of the waterfront and thus remained unaffected. This resulted in a pressure distribution on the slab whereby the upper portion of the concrete structure was subjected to uplift forces and the down stream portion experienced only down ward pressure due to self-weight. At some point along the structure the overturning forces were sufficiently high to cause cracking of the concrete slab reinforced with temperature steel only. This process continued as waterfront moved downstream and all slabs resting on shale experienced substantial heaving and subsequent cracking. These conclusions are further augmented by the fact that at site the concrete slabs resting on the sand stone and conglomerate beds have not experienced any distress. The foundation of these slabs is stable and therefore these slabs did not experience any damage. The situation was further aggravated by the dispersive nature of foundation soil (See Erosion.mpg Please see editor's note on movies). As already stated the soil is highly erodable. During first flood as soon as the high velocity water passed over the morning glory crest and sloping glacis, it entered the crevices and started eroding the foundation. Dispersive clay did not offer any resistance to scouring. Subsequently large caverns were formed right under the concrete slabs. Since the slabs have been provided with only temperature steel on the top, they were damaged heavily after removal of the support from the ground. Some of these were washed down the glacis into the stilling basin. The presence of natural deep cavities and under ground caverns in close vicinity of the spillway structure confirms the dispersive nature of Surla shale and testifies the findings of this study. Expensive repairs to spillway structure were subsequently required to prevent complete destruction of spillway and failure of project.
CONCLUSIONS
Based on site investigations, lab testing and subsequent analysis of data it is concluded that following are the causes of Surla dam morning glory spillway failure. Such causes can result into severe damage/failure of such structures on similar soils.
References
 | |
| © 2004 ejge | |