Preparations for load tests and problems with the ground water

1111 First excavation

The investigations in the test field at Vatthammar were performed during the summer of 1995. During this period, the pore pressures in the piezometers at Vagverket were read off occasionally. Preparations for the load tests were also made by drilling holes for the four ground anchors down to about 18 m depth. The ground anchors consisted of expander bodies of the Swellex type, which were lowered down into the pre-drilled holes and connected to an injection tube and tie rods. The holes were made using Geobor S equipment and the holes were filled with a thick suspension of bentonite slurry to stabilise them and to seal them when the drilling tubes were withdrawn. The expander bodies were then lowered down to the bottom of the holes and should later be expanded by injection of cement grout under high pressure, Fig. 4.2.15.

Due to other urgent tasks for the field crews, the test programme was delayed and could not start before the beginning of September. The programme was to start with an excavation to 1.8 m depth, instrumentation and casting of the square concrete test plates which were to be loaded. During the occasional checks of the pore pressures during the summer, the free ground water level had been found to be located more than 2 metres below the ground surface. A final thorough inspection on 28th August confirmed these conditions.

The following week was spent in Linkoping, 350 km away from the test field, making preparations for instrumentation, prefabrication of moulds, excavator hire etc. The equipment was transported to the site on Monday 4th September and arrived in the evening. On Tuesday morning, the excavation was started.

Unknown to the test team, it had been raining heavily in the Borlange area over the weekend. The rain started on Friday and was unusually heavy on Sunday. It then continued with lesser intensity on Monday and it still drizzled on Tuesday morning when the excavation started.

When the excavation reached a depth of about 1.5 m, water started to seep in and the walls started to cave in. The excavation work was then stopped immediately, but the bottom was observed to soften up because of the water seepage. It was also observed that boreholes from the previous investigations opened up and the flow of water in these holes was strong enough to bring eroded soil up to the surface.

Soon, the excavated area was ruined with respect to any plate loading tests which

SGI Report No 54 72

Drilling the holes for the anchors.

Attaching the tie rods.

Lowering the expander body and the injection tubes.

Fig. 4.2.15 Installation of ground anchors.

Investigations and load tets in silty soils 73

were intended as references for predicted bearing capacity and settlements, Fig. 4.2.16.

The excavation remained open for a few days, during which the water level was observed and attempts were made to seal the open boreholes that were found by injecting bentonite slurry. The largest boreholes for the pressuremeter tests had been filled with bentonite slurry all the time. Other boreholes are normally not sealed in this type of soil but are expected to collapse and be self-sealing. The water level kept rising all the time and was close to I metre below the ground surface when the excavation was filled in. It also continued to rain for the rest of that week.

Fortunately, the ground anchors had not been fastened and two of the expander bodies could be taken up and moved to two new holes on the opposite side of the remaining pair of anchors. The location for the plate load tests was thus moved about 15 metres away from the intended location. A check to ensure uniform soil conditions was made by an additional CPT test and new piezometers were installed around the new test location. The main activities were then transferred to the test field at Vatthammar while a decision was made on how to deal with the ground water at Viigverket. The current experiences and the subsequent studies of how the ground water situation varied with the climatic conditions showed that the ground water level throughout the profile reacted strongly and rapidly on rainfall. Having entered the autumn, which is the rainy season in Sweden, there was little hope of a return to a low and stable ground water table. Furthermore, with occasional heavy rains and thunderstorms also during the summer period and the rapid reaction in the soil to such rains, there was no guarantee for such favourable conditions prevailing during any period. This behaviour had been missed in the previous occasional observations with long intervals.

Ground water lowering

An attempt to gather local experience of the problem revealed that this type of problem was recurrent in the area. Excessive settlements have also occurred in some cases where problems with high ground water occurred during the founda­

tion works. However, there was no established way of dealing with the problem.

In most cases, an attempt was only made to collect the incoming water in ditches and to pump it out. During construction of the buildings for the National Road Administration, an extra deep excavation had required the ground water to be lowered by heavy pumping in the coarse bottom layers, but this was not feasible in the current project. There was equipment for lowering ground water by vacuum

SGI Report No 54 74

Excavation halted after water started to seep in.

Sealing observed holes before backfill­

ing the excavation.

Fig. 4.2.16 First excavation at Vagverket.

Water accumulating in the pit, bottom softening up and sides caving in. An old sounding hole emitting water and soil particles from lower levels can be observed just left of the centre of the picture (arrow).

Investigations and load tets in silty soils 75

pumping in filter wells available in the area, but this would have been expensive and experience is very limited.

The Institute has previously been involved in other projects with ground water lowering in fine grained soils it has been found that problems often occur with the filters and with soil entering the mechanical pumps. It was therefore decided to postpone the load tests at Vagverket until the following year and to design a simpler and more reliable way of dealing with the ground water in the meantime. The only further activity in the test field in the autumn of 1995 was the inflation of the expander bodies in the ground anchors, which was performed at the same time as injection of the corresponding expander bodies at Vatthammar.

During the winter a new pneumatic pump system, which was intended to meet the requirements of being cheap, simple and robust, was designed and tested in the department for field investigations and measuring techniques at SGI, (Joelson and Lofroth, 1996). The system uses a type of air lift pump, which has previously been used mostly in connection with sampling of ground water, e.g. Morrison (1983).

The pump consists of a length of stiff plastic tube with two plastic "ball valves", one at each end. Two thinner flexible tubes and an electrical signal cable lead to this tube. One of the flexible tubes is connected to a control box on the ground and a supply of compressed air. The other tube is led to a ditch or drainage pipe where the water pumped out can be led further away. The stiff tube is lowered vertically into a filter well. When water reaches the lower end of the tube, it passes through the ball valve and fills the tube. When the water level inside the tube reaches a pre­

set level, one of the leads in the signal cable closes an electric circuit and the control box switches on the compressed air. The pressure increase in the tube forces the lower ball downwards into its seat to seal off the inlet and the upper ball is lifted and allows the water in the tube to be blown out through the exhaust tube. When the pump is emptied, another lead in the signal cable breaks the electric circuit, the compressed air is switched off and the pressure line is vented. The lower ball can then lift and allow new water to enter, while the upper ball falls down and prevents flushed out water from re-entering the pump, Fig. 4.2.17.

There are thus only two moving parts, the two plastic balls, in the pump. The rest of the system is all above ground. If the pumps should become clogged by incoming soil particles, they can easily be lifted up and serviced on site. No spare parts are required, only cleaning and washing. The equipment on the ground consists of a common air compressor and a control box for each pump. It can be installed in any container or workers' cabin where it is protected from the weather, the only external requirement being access to electrical power.

SGI Report No 54

76

PRESSURE LINE FOR ~ SIGNAL CABLE COMPRESSED AIR

AND VENTING

BALL VALVE

UPPER LIMIT

LOWER LIMIT

BALL VALVE

Fig. 4.2.17 Pneumatic

WATER IN pump.

Some calculations were made with a numerical programme for simulation of ground water conditions, SEEP/W by GEOSLOPE, in order to check the required capacity of the pumps to lower the ground water locally at the test field. In these calculations, a fairly high permeability was assumed in the silt above the clayey layer between 4 and 5 metres depth. The clayey layer was assumed to have the permeability measured in the oedometer tests and thus more or less act as a tight lid on top of the rest of the profile with artesian water pressures. The pumps were assumed to be installed above this low permeable layer. According to the results of the calculations, it would be possible to rapidly lower the ground water with a moderate pumping capacity.

The pumps were installed in two rows parallel to the common centreline of the plates and outside the intended excavation area and the soil volume expected to be influenced by the load tests.

Investigations and load tets in silty soils 77

The spring of 1996 was very late. Preparations for the ground water lowering started in the first week of May, when there was still frost in the ground, with the installation of 10 slotted tubes down to 3.5 metres depth. The very fine slots were supplemented with a sand filter when the pumps were installed and started two weeks later. This operation was finished just before a public holiday and a following weekend, and the equipment was thus left unattended for almost a week.

At the start of pumping, the ground water level was only about 0.4 metre below the ground surface. When the system was inspected after the weekend, the ground water level was found to be even higher and the ground was completely soaked.

The reason for this was a combination of clogging of the pumps and rainfall during the period. The pumps were cleaned and restarted, but the pumping was slow and the pumps had to be cleaned repeatedly.

The sand filters obviously did not function properly and new filter wells with geotextile filters were installed outside the old wells. The latter were left as observation holes for the location of the free ground water level. The pumps were taken up and cleaned, and were then reinstalled in the new filter wells, Fig. 4.2.18.

This operation was completed about one week later. In order to speed up the rate of ground water lowering, an electroosmotic field was also constructed. This consisted of rows of reinforcing bars pushed down 4 metres into the soil. One bar was pushed down beside each filter well and these bars were then electrically interconnected to form the outer row of cathodes. The anodes were pushed down approximately in the middle between the cathodes and about 1 m inside to form an interconnected inner row. By connecting the two rows to a source of direct current, the flow of water from the centre of the field towards the filter wells was to be accelerated.

The only readily available source of direct current is welding generators. The largest such equipment that could be connected to the available supply of alternating current was switched in. However, for safety reasons the output voltage of welding generators is limited to about 55 Volts. The resistivity in the ground was high and hence the amount of current passing through the electroosmotic field was very limited. Only a fraction of the capacity of the welding generator was thus utilised.

Before the current was switched on, the area was fenced in, Fig.4.2.19. The field was then put under daily supervision. The ground water level was also observed daily in all old filter wells and the pore pressures in the piezometers around the test

SGI Report No 54

78

Installation of a filter well in a hole predrilled with a screw auger and lined with a plastic tube.

Filter well installed and plastic liner with­

drawn. Earlier filter well remains for ground water observation.

Installation of a pump in a filter well.

Fig. 4.2.18 Installation of filter wells and pumps.

Investigations and load tets in silty soils 79

The fenced-in test field during the ground water lowering. The welding generator is placed in the container

Water streaming from the outlet pipes during the ground water lowering.

Fig. 4.2.19 Ground water lowering in progress.

SGI Report No 54

80

field were read off at the same intervals. Two open holes were also later drilled down to 2 metres depth between the locations of the test plates in order to avoid any unpleasant surprises.

After installing the new filter wells and re-starting the pumps, the ground water level started to fall at a rate of 0.1 to 0.15 metres per day. Two days later, when the electroosmosis was switched on, this rate increased by about 50 % to about 0.2 metre per day. The ground water level continued to fall for five days and then halted momentarily for about one day because of rain. It then continued falling.

Seven days after the start of pumping, the intended foundation level at 1.7 metres below the ground surface was passed and it was decided to start the excavation.

Excavation was performed in three separate square holes, one for each test plate, with 1 m longer sides than the plates. Excavation was started by using an excavator down to a depth of 1.4 metres below the ground surface and was then continued by careful manual excavation in order to obtain smooth base surfaces with as little disturbance as possible. The electroosmotic field was shut off during work with the excavator and the free ground water level was closely monitored throughout the operation. The ground water level continued to fall during the excavation period, which lasted for about a day and a half, and reached 1.9 metres below the ground surface. After excavation, the moulds for the concrete plates were placed in position after which the reinforcement and instrumentation were to be installed. It then started to rain heavily. The excavations were covered with tarpaulins and ditches were dug to lead the water to the pumping wells. This was successful insofar as the excavations remained dry and the work could continue, but the free ground water level started to rise again. By the time the instrumentation was finished and the concrete was poured, the ground water level had almost reached the excavation depth and free water was observed in deeper pockets outside the moulds.

The day after casting the plates, the 0.5 metre high moulds were raised 0.5 metre to form a wall against the soil and protect the instrumentation, whereupon the space outside the plates and the moulds was filled in with soil and the pumping operation was stopped. The free ground water level was then left to find its own natural variation.

The free ground water table was then again observed during the actual plate load tests which started about a week later. This testing period, which lasted for about two weeks, started with heavy rains and ended with warm and dry summer weather

Investigations and load tets in silty soils 81

in the latter part of June. During this period, the free ground water level rose from about 2.0 to 1.3 metres and then fell again to 1.7 metres below the ground surface, Figs. 4.2.20-21.

The combined observations indicated that successful installation of the test plates depended both on effective pumping and on luck in terms of a relatively dry period.

In an ordinary case of foundation, the groundwater level could have been controlled more effectively by pumping also closer to and below the foundation with a very moderate disturbance to the soil.

The observations of the pore pressures in the piezometers at 3 and 5 metres depth outside the test area reflect the variations in the free ground water level inside the area but with smaller amplitudes, Figs. 4.2.22 - 23.

SGI Report No 54

82

Date

-Fig. 4.2.20 Measured variation in free ground water level in the test field at Vagverket during ground water lowering and load tests.

10

Fig. 4.2.21 Precipitation in the Borlange area during ground water lowering and load tests.

Investigations and load tets in silty soils 83

3

Fig. 4.2.22 Measured variation in pore pressure at 3 metres depth just outside the test area at Vagverket during ground water lowering and load tests.

5

Fig. 4.2.23 Measured variation in pore pressure at 5 metres depth just outside the test area at Vagverket during ground water lowering and load tests.

SGI Report No 54

84