Large-scale instability caused by groundwater

Inadequately controlled groundwater can cause large-scale stability prob-lems in a variety of ways, the mechanisms likely to be prevalent at a given site largely being controlled by ground conditions. These large-scale mecha-nisms can be explained in terms of effective stress and pore water pressures.

It is worth considering two classic cases of ground conditions. First, an excavation with battered side slopes made into a sandy soil forming an unconfined aquifer. Second, an excavation made through a low permeabil-ity stratum overlying a confined aquifer. The mechanisms of potential insta-bility will be quite different in each case.

4.3.1 Unconfined aquifers

Consider a hole with sloping sides dug in a bed of silty fine sand below the standing groundwater level. If the inflow water is pumped from a sump within the excavation, the sides will slump in when a depth of about 0.5–1.0 m below original standing water level is reached. As digging proceeds the situation will get progressively worse, and the edges of the excavation will recede. The bottom will soon fill with a sand slurry in an almost liquid con-dition which will be constantly renewed by material slumping from the side slopes. The collapse of the side slopes results from the presence of positive

pore water and seepage pressures which are developed in the ground by the flow of water to the pumping sump of the open excavation (Figure 4.1).

The mechanisms causing this unstable condition can be explained in terms of effective stress.

Above the water table, assuming that the soil is dry (i.e. has zero pore water pressure) then it can stand in stable slopes of up to ␾⬘ to the horizon-tal. In reality the pore water pressure above the water table may not be zero;

in fine-grained soils such as silty sands negative pore water pressures may exist temporarily due to capillary effects. As a result, stable slopes even steeper than ␾⬘ may temporarily be possible. Figure 9.12 shows a near verti-cal excavation face in a silty fine sand where groundwater has been lowered by wellpoints. However, in time the negative pore water pressure will decay as the capillary water exposed on and near the face dries out. After some finite period of time (which may be no more than a few hours) the over-steep-ened slope will eventually crumble to that of the long-term stable slope angle.

Where seepage emerges from the slope there will be positive pore water pressures. Positive pore water pressures will reduce effective stress, the shear strength of the soil will reduce in turn and the soil will not stand at slopes as steep as in dry soils. This is the mechanism leading to the ‘slumping’ effect seen when excavating below the groundwater level.

Considering Figure 4.1 in detail, flow line ‘A’ represents the flow line formed by the water table or phreatic surface. Seepage into the excavation will cause a slight lowering of the water table so flow line ‘A’ curves downward and emerges almost parallel to the surface of the excavation. Immediately

Figure 4.1 Instability due to seepage into an excavation in an unconfined aquifer.

below its point of emergence the positive pore water pressures generated by the seepage mean that the soil can no longer support a slope of ␾⬘. Below the emergence of the seepage line the soil slope will slump to form shallower angles. At the point of emergence of the almost horizontal flow line ‘B’ the sand will stand at ␾⬘ or less. Where there is upward seepage into the exca-vation base (flow line ‘C’) the effective stress may approach zero and the soil cannot sustain any slope at all. In these circumstances the soil may ‘boil’

or ‘fluidize’ and lose its ability to support anything placed on it – this is the so-called ‘quicksand’ case.

‘Running sand’ is another term used to describe conditions when a gran-ular soil becomes so weak that it cannot support any slope or cut face, and becomes an almost liquid slurry. The term is often used as if it were a prop-erty of the sand itself. Actually, it is the flow of groundwater through the soil, and the resultant low values of effective stress which cause this con-dition. Effective groundwater lowering can change ‘running sand’ into a stable and workable material.

In addition to the loss of strength, seepage of groundwater through slopes may cause erosion and undermining of the excavation slopes. This is espe-cially a problem in fine-grained sandy soils, and is discussed in Section 4.4.

Any solution to stability problems in unconfined aquifers will need to sta-bilize both the sides and base of the excavation. The most commonly used expedient is to install a system of dewatering wells (see Section 5.4) around the excavation to lower the groundwater level to below the base of the excavation, and to ensure that seepage does not emerge from the side slopes. A typical target is to lower the groundwater level a short distance (say 0.5 m) below the deepest excavation formation level.

If the excavation is only penetrating a short distance (say less than 1.5 m) below the original groundwater level, a sump pumping system (Chapter 8) used in combination with slope drainage (see Section 4.4) might also be effective. Extreme care must be taken when using sump pumping in this way, to avoid destabilizing seepages into the excavation; the risk of this increases for excavations further below original groundwater level.

If the sides of the excavation are supported by physical cut-off walls (see Section 5.3) then it is possible to sump pump from within the excavation, without the risk of instability of side slopes. The risk of fluidization of the base due to upward seepage remains (Figure 4.2). Fluidization will theoreti-cally occur when the upward hydraulic gradient exceeds a critical value i_crit

i_crit: (4.3)

approached. In design it is normal to limit the predicted hydraulic gradients to be less than i_crit/F, where F is a factor of safety of 1.2 or greater.

The hydraulic gradient for upward seepage is defined as the head differ-ence dh divided by the flow path length dl (see Figure 4.2). The hydraulic gradient can be controlled in two ways.

(a) By increasing dl. This can be achieved by ensuring the cut-off walls pen-etrate a sufficient depth below formation level. This is an important part of the design of cut-off walls in unconfined aquifers (see Williams and Waite 1995).

(b) By reducing dh. This requires reducing the groundwater head difference between the inside and the outside of the excavation enclosed within the cut-off walls. The most obvious way of doing this is to use a system of dewatering wells outside the excavation to lower the external groundwater level and so reduce the head difference to an acceptable level, while still sump pumping from within the excavation. This approach might be taken a step further and the external wells be used to lower the groundwater level to below formation level, and avoid the need for sump pumping altogether. An alternative approach to reduc-ing dh is to keep the excavation partly or fully topped up with water and carry out excavation underwater (e.g. using grabs) and then place a tremie concrete plug in the base. This method is sometimes used for the construction of shafts and cofferdams.

Figure 4.2 Fluidization of excavation base due to upward seepage in unconfined aquifer.

Upward hydraulic gradient below excavation floor:dh/dl.

4.3.2 Confined aquifers

The classic confined aquifer case is shown in Figure 4.3. It shows an exca-vation through a clay stratum, with the sides of the excaexca-vation supported by walls of some sort. The formation level of the excavation is in a very low permeability clay stratum that forms a confining layer above a permeable confined aquifer. The piezometric level in the aquifer is considerably above the base of the clay layer.

When assessing the stability of this case, the critical horizon is the inter-face between the confined aquifer and the underside of the overlying con-fining bed. Consider the two separate sets of pressures acting at this level.

The excavation will be stable and safe provided the downward pressure from the weight of the residual ‘plug’ of unexcavated clay is sufficiently greater than the upward pressure of pore water confined in the aquifer. This is assuming that the confining stratum is consistently of very low perme-ability, and is competent and unpunctured (e.g. it is not penetrated by any poorly sealed investigation boreholes, which could form water pathways).

If excavation within the cofferdam is sunk further into the clay without any reduction of pore water pressures in the sand aquifer, there will come a time when the upward water pressure in the aquifer will exceed the weight of the clay plug. There will then be an upward movement or ‘heave’ of the excavation formation. If the clay plug heaves sufficiently the clay may rupture, allowing an uprush of water and sand (sometimes known as a ‘blow’) from

Figure 4.3 Excavation in very low permeability soil overlying a confined aquifer – stable con-dition. For a relatively shallow excavation the weight of the unexcavated plug of clay is greater than the upward water pressure from the confined aquifer.

Figure 4.4 Excavation in very low permeability soil overlying a confined aquifer – unstable condition. For a deeper excavation the upward water pressure from the confined aquifer exceeds the weight of the clay plug, leading to base heave and, ultimately a ‘blow’.

the aquifer which may even lead to the collapse of the excavation (see Figure 4.4).

The ‘heave’ situation can be avoided either by:

(a) Adequately reducing the pore water pressure in the confined aquifer by pumping from suitable dewatering wells inside the cofferdam.

Alternatively, the wells may be sited outside the cofferdam but will be less hydraulically efficient. The wells should lower the aquifer water pressure so that the downward forces exceed the upward pressures by a suitable factor of safety.

(b) Increasing the depth of the physical cut-off wall sufficient to penetrate below the base of the confined aquifer. Provided the cut-off walls are water tight this will prevent further recharge, thus leaving only the water pressure contained in the aquifer within the cut-off walls to be dealt with. This can be effected by installing relief wells (see Section 11.3) prior to excavation. The economics of this alternative will depend primarily on the depth needed to secure a seal and the effectiveness of that seal.

(c) Increasing the downward pressure on the base of the excavation by keeping it partly topped up with water during the deeper stages of work. Excavation is made underwater, and a tremie concrete plug used to seal the base on completion.

The above description describes the principal features of the ‘base heave’

failure in confined aquifers. For relatively narrow excavations the shear resistance between the clay plug and surrounding ground may be significant and should be considered when estimating factors of safety. Hartwell and Nisbet (1987) discuss the problem further.

Base heave can also sometimes occur in nominally unconfined aquifers that contain discrete (but perhaps very thin) very low permeability clay layers.

Figure 4.5 shows a case where dewatering wells are used to lower ground-water levels around an excavation, but do not penetrate the clay layer below the excavation. The dewatering system has lowered water pressures above the clay layer, but the original high pressures remain beneath. These high pres-sures can cause base heave. This mechanism is sometimes known as ‘bed sep-aration’, because as the clay layer moves upward a reservoir of water will develop beneath. This form of base heave can be avoided by using deeper dewatering wells to control pore water pressures at depth.