It was 50 years ago that I undertook my first pumping test. As a fairly raw graduate, it was a rapid education into the ways of itinerant drillers whose pleasures in life included educating young engineers who got too close to the action by covering them in wet slurry and arisings from piezometer holes they were drilling. This education was near Braintree, in Essex, where I was working for Essex River Authority.

Observation holes were drilled, including one in the clay overlying the aquifer. During the test the person making the manual measurements in this piezometer noted that the water level was rising after the start of pumping. This caused some consternation and a little mockery. However, it turned out that this was a phenomenon that had been observed previously and was known as the Noordbergum effect.

Before the pumping test, the overburden pressure and upthrust (on the underside of the low permeability clay layer) were in balance. The pumping test then reduced the upthrust, resulting in a rise in the pore pressure within the overlying clay.

I was hooked on the mysteries of groundwater.

As an illustration of what has changed, back then, at the same time as the pumping test a colleague was writing a report to justify why the Authority should invest in calculators that had a ‘memory’. Not a memory chip but merely a ‘M+’ key so that conversions were simplified. This was the time of metrication.

A few years later, I moved to Soil Mechanics Ltd and it’s consultancy unit, ERCON. There was a wealth of highly experienced staff and many opportunities to get involved in challenging projects. Over the next few years I got involved with a number of them, including some that had groundwater issues with deep shafts and pumping stations. I wrote a paper with Bob Nisbet (Proc. IX ECSMFE, Dublin, 1987, pp 691–694) entitled ‘Groundwater Problems associated with the construction of large pumping stations’, which discussed 10 projects anonymously but still gets quoted.

On rereading the paper, I realise there were two particularly relevant projects (both in London – Beckton Pumping Station, Durand’s Wharf) that are relevant to more recent projects.

Beckton Pumping Station

In 1975/6 Mowlem were constructing a large pumping station in east London. The site is now in a roundabout adjacent to Beckton DLR station. The design was novel being two circular diaphragm walls that intersected one another; hence being efficient structurally and in terms of space.

Ground conditions were difficult being:

  • 9m Made Ground and Alluvium, over
  • 8m of Gravel, over
  • more than 8m silty sand (Thanet Sands), over
  • Chalk

The excavation was to be 16m-deep and the toe of the diaphragm wall (D-wall) was at 19.2m depth. The larger shaft internal diameter was 34.5m.

The plan was to drill six deep wells into the Chalk to create a cone of depression in the groundwater table and drain the overlying Thanet Sands and so reduce the pore pressure to an acceptable level, to stabilise the sands. It was a system that had been used successfully elsewhere in London.

During drilling of the wells, an efficient site engineer, Mac Dobson, noted that the base of the sand was quite clayey and raised the issue, as he realised it might form a layer that would prevent downward drainage as planned. The design was modified to include two wells pumping from above the clay layer. The young engineer had been correct, drainage was impeded by the clayey layer at the base of the sand and was much slower than planned. The critical piezometer took several months to reach the required level. Despite geology not being uniform, the excavation was completed safely.

Durand’s Wharf

Some years later, around 1995/6, construction of the Jubilee Line Extension was being undertaken into south east London. Traditionally this was challenging tunnelling territory due to difficult ground conditions (Brunel would have agreed) and perhaps explains why the Underground (‘Tube’) network had not previously been extended in this direction.

A critical shaft was to be sunk at Durand’s Wharf to launch tunnelling machines. Ground conditions were similar to those at Beckton and a similar dewatering/ depressurisation strategy was proposed. The shaft sinking method was to be by caisson. As excavation proceeded it was realised that the pumping from the Chalk was not reducing the critical pore pressures in the Thanet Sands.

Detailed investigations were implemented and the problem of the low permeability layer at the base of the Thanet Sands was again identified. Time was becoming critical with no certainty that additional dewatering would help.

The Contractor proposed applying compressed air and completing the excavation in what some considered the old fashioned way; the German partner in the JV had extensive experience. I recall a meeting at which the Client’s Engineer asked if the Contractor had done this before to which the German Engineer produced a promotional leaflet showing the 50 projects they had completed in this way – end of discussion (for more details, see T&T Jan 1997, pp 31–33).

In conclusion, beware of low permeability/clay layers that might confine groundwater flow and possibly create artesian pressures. For those readers unfamiliar with such problems Figure 1 should be considered as an image, from another project, that shows the fountain upflow of water from some artesian conditions.

Ringsend Pumping Station, Dublin

A further project in the 1987 paper, and one of those I had direct experience, was a major structure being constructed as the main lift pumping station for drainage of south Dublin.

The ground conditions were variable but typically:

  • 4m Fill and Estuarine Silt, over
  • 7m Sands and Gravels, over
  • 10m Stiff Silty Clay, over
  • Limestone

The excavation was to be 14m-deep and a sheet pile wall was installed to 18m depth. The shaft’s internal diameter was 31m.

The initial assumption was that the sheet piles would form a cut off but as excavation approached the target level a blow occurred at one point. Despite loading the area with gravel the flow increased and the excavation was flooded. There was considerable debate (and still is) as to the cause of the failure. Possible causes were uplift from the Limestone, declutching of a pile joint or a piping failure fed from the Gravels possibly aggravated by a boulder pushed down causing a flow path or a bent toe of a pile. I believe this last to be the most likely cause.

Remedial works included pressure relief wells and driving the sheet piles deeper. Unfortunately, one area of piles could not be accessed and a second blow occurred. This involved mobilising the Dublin fire brigade to pump water into the shaft!

The final solution was to construct a slurry trench cut off outside the sheet piles to cut off the sands and gravels and any possible sand layers within the clays (see Little et al. Proc. IX ECSMFE, Dublin, 1987, pp 183–188). In my opinion, a failure mechanism involving piping around the toe of a cut off is the most common cause of such failures; particularly where they can be fed by high permeability gravels from a source such as the river. For completeness, I describe below two such projects.

Olympic Cable Tunnels, London

When London was awarded the 2012 Olympics, in 2006, there was a very urgent need to remove the overhead high voltage (HV) power lines that crossed the Olympic site. The programme of diverting power lines into tunnels was critical. Two shafts south of the Olympic site were excavated to launch tunnel boring machines using jacked caisson techniques.

The ground conditions comprised Alluvium over Thames Gravels over Lambeth Group (mainly clays) over Thanet Sands over Chalk. The 15m-diameter shaft was to be 39m-deep, i.e. it would just toe into the Chalk.

Excavation encountered various problems primarily due to the very dense deposits. At around 30m depth there was a major failure – water (presumably from the Thanet Sands) led to erosion of the sands and loss of the bentonite lubricant behind the caisson. The sad situation can be seen in Figure 2. It should be noted that whilst the sands are very dense they do readily erode.

In order to complete the shaft sinking into the base of the Thanet Sands, additional measures were required. A system of inclined well points radiating from the shaft into the lower sands was developed, (see Figure 3), and proved effective and enabled the shaft to be completed.

Pumping Station, Preston, UK

In 2010, during the construction of two adjacent shafts for a stormwater pumping system, in Preston, problems were encountered.

The geology comprised fluvio glacial deposits (sand and gravels with some clay layers) up to 15m deep overlying Sherwood Sandstone. The strategy was to construct D-walls to 18.5m below the ground surface to provide a cut-off and toe into competent sandstones.

As the excavation approached the toe of the D-wall an inflow of water and then sand occurred; the shaft was flooded to balance water pressures. Subsequently, there were a number of similar events in both shafts.

The original concept that water could flow through the Sherwood Sandstone and be collected in sumps was clearly not valid. The problem appeared to be that the ‘sandstone’ was a marginal material in places, i.e., the cores recovered appeared as weak rock when drained but, under high hydraulic gradients, in the excavation they behaved as an erodible soil.

The project was recovered using an outer ring of jet grouted columns and an extensive programme of dewatering, including angled drains from inside the shaft as at the Olympic cable tunnels. The drains are shown in Figure 4. There was considerable debate about water pressures and flow directions, in part due to a number of standpipes with very long response lengths. It was still with some trepidation that excavation proceeded below the toe of D-wall. Happily, the project was completed but debate still continued about the nature of the top of the sandstones.

OTHER ISSUES – PROBLEMS WITH ROCK

The above cases all involve erodible material and a water source, with the most predominant problems being the short flow path around the toe of a cut-off.

This continues to be the most common issue and is consistent with the highest hydraulic gradients being at the toe of the wall, which is readily apparent from a simple flow net analysis.

In cases where the ground is not erodible, for example in rock, problems may still occasionally be encountered.

Empingham Reservoir

The 1987 paper, cited above, also refers to a deep shaft at the Empingham reservoir site (now called Rutland Water), near Oakham, in Leicestershire, England. Ground conditions comprised 40m of Lias Clay over a Marlstone rock band.

At 26m depth, the base plug heaved and significant inflows of groundwater occurred and the shaft was flooded. The situation was recovered after the installation of 13 deep wells into the rock band, pumping 230 l/s in total.

Such highly permeable bands of very stiff rock are sometimes encountered, such as in the Chalk aquifer in south east England, the Melbourn Rock being a band in the very stiff water bearing Chalk long sought by water supply engineers.

Prudhoe Treatment Works

Another example of failure in rock was the deep shaft for an innovative sewage treatment process, at Prudhoe, on the banks of the River Tyne, in England.

Cautiously, two boreholes were drilled rather than the usual one shaft centreline borehole. Unfortunately, the boreholes found coal but missed the old coal workings which were abandoned and flooded from the surrounding hills, resulting in a significant artesian head. The inevitable failure led to abandonment of the Prudhoe project.

These stories are told with the knowledge that some were supressed at the time but the author feels that in the interest of future knowledge the stories should not be forgotten.