The efficientcy of drainage blankets in surcharging of soft clays can profoundly affect the success of the vertical drain treatment. Tony Barry and Linda van der Bend report.
Vertical drain treatment of soft clay deposits relies on the functionality of the horizontal drainage blanket placed at the surface.
A number of factors affect the efficiency of the blanket, including rate of discharge from the drains, the plan extent of the blanket and initial perched water from hydraulic filling and rainfall.
When designing vertical drain schemes it is generally assumed that the water pressure at the original ground surface, or at the drainage blanket level, is zero.
Much effort is spent in assessing the drain capacity and smear effects which in fact are generally of little consequence (Barry et al, 2001), whereas the capability of the drainage blanket can profoundly affect the success of the treatment.
It is helpful to look first at the effects of a poorly designed, limited functionality or inadequate drainage blanket on the behaviour of soil installed with vertical drains.
Effect of impeded drainage Figure 1(a) shows a thick layer of compressible clay, with the water table at the surface, and hydrostatic conditions, U h which is typical of the situation requiring a prefabricated vertical drain (PVD) solution.
Figure 1(b) shows the application of a surface load from filling and the initial increase in pore pressure to some higher value, U 0, throughout the clay layer. It is assumed that the fill placed is free draining.
After vertical drains are installed, pore pressures dissipate with time to a . nal hydrostatic condition, U h, as shown on Figure 1(c), ignoring vertical drainage near the surface, which is not significant where closely spaced drains are installed.
This is Case
However, if for any reason the water level in the . ll is higher than the original ground surface, then the situation is as shown on Figure 2.
Since the vertical drains are in hydraulic continuity with the fill, then the pore pressure in the drain will be hydrostatic from the top water level in the fill.
If the water level in the fill remains static (Case 2), then consolidation of the clay will be less than the previous condition.
If the water level in the fill drops with time (Case 3: impeded or delayed drainage of the fill) then the consolidation of the clay will be delayed compared with the previous condition.
These three cases are shown in Figure 3; a case study showing this effect is described later in this note.
Causes of impeded or delayed drainage
Three significant sources of delayed drainage have been encountered:
Drain discharge. The total settlement of the clay layer requires the discharge of an equivalent amount of water. Thus, for example, a 20m layer of clay could settle by 2m under the load, which will generate 2m 3/m 2 of water over the period of drainage.
The significance of this discharge is highly dependent on the drain spacings; close drain spacings will result in faster discharge and the potential for a higher water table in the fill. Quantification of this effect is dealt with later.
Rainfall. Where the fill is relatively permeable, for instance where it is all sand, then in high rainfall areas such as the tropics, infiltration can be significant.
Hydraulic fill retained water. For larger reclamations, hydraulic filling is often used. It is normally assumed that the sand is free draining and therefore the water level in the sand will rapidly fall to original ground level, sea level or some other longterm static level.
In fact, for large excavations, even with relatively permeable sand, the time to drain the water from the reclamation can be very significant in relation to PVD performance.
Quantification of the effect Figure 4 shows a common situation where a road embankment is constructed over PVD treated ground.
A drainage layer is placed, followed by relatively low permeability fill. In this case the drainage layer needs to dispose of only the water arising from the drains.
The settlement of the soil is predicted in the normal manner, as shown on Figure 5. The drain discharge can then be assessed, by calculating the rate of settlement by drawing chords on the settlement graph, and thus developing a water discharge rate graph as shown on Figure 5.
For normal drain spacings and soil types a clean coarse sand provides an adequate drainage layer with only a small head being generated in the layer at the beginning of loading.
However, where the site is to be reclaimed over a large area using hydraulic fill , other factors also become important.
An example is a recent project in which an area of about 400m by 300m of soft clay with a depth in excess of 20m was to be treated with drains at 1m centres and reclaimed using hydraulically filled coarse to medium sand.
The sand had an average D10 size of 0.3mm to 0.4mm which Hazen's equation predicts would have a permeability of 9 x 10 -4 m/s to 16 x 10 -4 m/s.
A single constant head permeability test on a representative sample indicated a permeability of 4.5 x 10 -4 m/s.
Finite element programs such as SEEP/W and Plaxfl w allow the rate of drainage to be predicted, taking into account the surface rainfall infi ltration and the recharge from the vertical drains.
Figure 6(a) (see previous page) shows the predicted and actual water levels in the fill at the centre of the reclamation for a 20m clay thickness, after an adjustment of the two-dimensional analysis to allow for flow in the third dimension and with free drainage at the boundary.
The first line shows the predicted behaviour when only the water from the reclamation is considered.
For this case, the actual fill level in the final lift of filling was more than 7m, but site inspections, trial pits and standpipes in the fill indicated that at this level some drainage occurred from the face of the fil during placement, and on completion the initial water level was about 5.5m above datum.
This was not modelled in the initial analysis, which assumed an initial water level of +7.5m.
The second line shows the additional effect of the water exiting from the drains, modelled as recharge at the base of the sand.
The amount of discharge is not as high as could possibly be the case, as in this study the fi al layer of fill was placed after a staged delay and the continuing settlement after final layer placement was less than half the total.
The third line shows the further additional effect of rainfall, assuming 100% infi tration and a constant 200mm/month rainfall.
In this case the fill comes to equilibrium with a mound of water in the reclamation and a level of 3m above datum at its centre.
The analysis did not take into account the reduction in level of the base of the sand due to the settlement of the underlying clay.
However this is offset by the probably lower permeability of the lower part of the sand where it has become mixed with the underlying clay during placement.
Finally, Figure 6(a) shows the actual measured water level in the centre of the reclamation.
Figure 6(b) shows the effect of assuming an initial water level at the lower and more realistic level of +5.5m.
There is a substantial initial rise in level as rapid early settlement occurs, and the close agreement with the prediction model should be noted.
This may be somewhat fortuitous, since further analysis of the finite element model showed the results to be very sensitive to the mesh shape assumed; this is understandable as the thickness of the zone of water flow is extremely small in relation to its length.
For substantial reclamations, drainage of water from the drainage layer can take a very significant time.
This will in turn affect the efficiency of vertical drains, and thus the rate of settlement of the underlying soft clay. Where a rapid completion of surcharging is required the effect of this delayed drainage needs to be assessed. Approaches such as a suitably increased surcharge height, or the installation of pipe drains or geotextile within the drainage layer may be required to mitigate its effect.
Tony Barry is an independent consultant, Kuala Lumpur, Malaysia and Linda van der Bend is a geotechnical engineer at van Oord.
Barry AJ, Satriyo B, Daud S and Febrijanto R (2001). The Kaliwungu trial embankments in north Java, Symposium on soft ground improvement and geosynthetic applications, Asian Institute of Technology, Bangkok, Thailand.