There are around 2500 large, raised reservoirs in the UK and according to data from 1975-2000, there has been an estimated 1600 dam incidents recorded for every failure during this period. Sixty percent of these are associated with seepage-induced internal erosion of soil. Two serious erosion incidents also occurred each year.
The problem, as Elisabeth Bowman, a reader in geomechanics from the University of Sheffield, explains, is when you do an assessment of a dam with some kind of seepage distress that could be manifested as excessive seepage or settlement, it’s often inconclusive as to what the current cause is. And one of the issues is that UK embankment dams tend to be more clay dominated and/or of more mixed soils (sand, gravel, clay and silt). Whereas most dams globally tend to be more granular dominated so that the rules of internal erosion are a bit more cut and dried.
If looking at internal erosion in general, it manifests more evidently in non-plastic or granular soils as different types of internal erosion such as suffusion, backward erosion, contact erosion, and concentrated leak erosion. These have all been well documented and can lead to excessive leakage and sink holes etc, depending on grain size, distribution of the soil, hydraulic load, and total stresses applied.
Dispersion on the other hand, tends to manifest in plastic or clayey soils. Most Pennine dams have puddle clay cores with mixed fill around them. So the question is, is seepage induced due to stress, due to internal erosion, due to dispersion, or some combination of this? Or something entirely different, Bowman asked.
To give an idea of what can go wrong with dispersive soils in a dam in quite a catastrophic way, she gave the example of Tinaztepe Dam in Turkey. It was first filled in 1993 and developed sinkholes in 2000 due to use of dispersive soil in the core. We hopefully won’t have problems of this magnitude in UK, Bowman said, adding she just wanted to illustrate the point.
What is dispersion?
The issue is if you have a clay soil significantly rich in sodium ions, they can rapidly deflocculate in water with a slow salt concentration, even when the water is not flowing. And that’s the key point, Bowman says. It doesn’t require the passage of water to cause that actual erosion or dispersion of the material. Deflocculation occurs as an inter-particle repulsion. If there is flow, the clay can be removed as part of the fluid and eventually over time a crack can develop or larger particles detach, allowing internal erosion to occur.
Examples of tests for dispersion include the crumb test. This is a useful and quick test you carry out in the field and where you get dirty, Bowman said. Here a crumb of soil is placed in a beaker of water, and you wait about ten minutes up to a few hours, to see if the solution becomes cloudy (the soil disperses). There can be variations with different states of the initial crumb and fluid. It’s a nice and easy test but depending on the type of water you put it in, you can get different outcomes, Bowman warns.
In comparison, the double hydrometer test compares the percentage of clay in a sample that has been artificially dispersed, to that of another sample which has no artificial dispersing agent.
Classic case
Bowman then went on to discuss a case study called reservoir X. It was formed by a typical Pennine type embankment dam constructed in the 1870s. With an embankment crest length over 600m, maximum height of 20m, it is zoned with a central puddle clay core carried down into a cut-off trench up to 18m deep. Clayey material was placed in inner zones on other side of the clay core with thicker layers of more stoney material in the outer zones.
The geology consists of bedrock of the millstone grit group (ie grits and sandstones), interbedded with siltstones, mudstones, marine shales and thin coal seams. Superficial deposits are absent across the reservoir; however, peat is shown across the wider valley area.
The issue at the dam is that a variation in embankment settlement is evident despite a small difference in crest height. There’s been a maximum settlement of 117mm over past 30 or so years. The surface of the downstream shoulder is irregular compared to nearby dams of a similar age. Recorded variable drainage flows indicate that leaks could initiate, develop and enlarge through internal erosion.
Bowman describes this as a pretty classic case, suggesting something isn’t happy in that dam.
The importance of pretreatment in determining the potential for dispersion was also highlighted. Soil tested with pretreatment using hydrogen peroxide to remove organic matter is compared to that without pretreatment. The results also demonstrated that the amount of soil used in the hydrometer test should be carefully considered to avoid both hindered settling (using too much soil) at one extreme and poor hydrometer response (using too little soil) at the
other end.
Recommendations
In conclusion, Bowman explains that the average age of a UK dam is 125 years. They’re some of oldest in the world and variable soils have been used in older dams, particularly the Pennine types.
The concern is that internal erosion is a time dependent process and “we’re somewhere down the line there”, Bowman adds, explaining there could be pockets of internal erosion going on at a slow rate, with dispersion assisting it, and developing over time.
To assess the risk it is recommended that double hydrometer tests are undertaken with routine treatment for organics, even if considered to be low as “they’re almost never not there in UK dams”, Bowman said.
Reference
Note: Rialynn Davy is a Senior Principal Geotechnical Engineer at Stantec, and some of her recent PHD research has focused on understanding the mechanical pathways of seepage distress in UK embankment dams, focusing on Pennine-type dams. Her results were presented by Elisabeth Bowman at this seminar.
