Little research work has been carried out on the behaviour and performance of light raft foundations for low-rise buildings on soft ground.This has in part led UK guidance on the design of such rafts to be conservative and, occasionally, contradictory.
This paper describes the results of an experiment on a full-scale downstand edge-beam raft foundation built at the Engineering and Physical Sciences Research Council soft clay test site at Bothkennar in Scotland. The raft was loaded around its perimeter in two stages: to design load and then to 33% overstress. Results obtained from settlement points on the raft, contact pressure cells under the raft and from extensometers and piezometers installed in the ground under the raft are described.
Measurements showed that a significant proportion of the applied load was taken by the slab.At design loading, the raft's performance was comfortably within criteria for acceptable performance.
It is concluded that design methods that allow only for the width of the downstand to resist wall loading are likely to be very conservative.
Increasing environmental and demographic pressures are concentrating many new developments in the UK into areas of land previously considered as marginal; either because of cost or because economic construction technologies were not available. Soft, natural ground, such as alluvial clays and peat deposits form one such area.
Research into construction on soft ground has historically tended to address the behaviour and performance of relatively heavily loaded structures, such as embankments, multistorey buildings or the use of deep foundations. Little research work has been carried out that specifically addresses the use of raft foundations for low-rise building on soft, natural ground.
Partly as a consequence, published UK guidance on the design of light rafts for low-rise buildings tends to be conservative, and is found in a diverse range of publications which are sometimes contradictory. For this and other reasons, mostly related to uncertainties in performance, raft foundations are not frequently used for low-rise developments on soft ground; a piled foundation solution usually being preferred. This is despite it being generally accepted that raft foundations provide good compliance with services such as drainage, and gas and water supplies, unlike deep foundations, where the settlement of ancillary services commonly exceed both the rate and magnitude of settlement of the foundations.
To help to address these difficulties, a full-scale downstand edge-beam raft foundation has been built and loaded to typical domestic service loads at the EPSRC soft clay test bed site at Bothkennar, Scotland.The primary objectives of the full-scale experiment are to:
l provide a benchmark for comparative field and numerical studies l investigate current design philosophies and design tools and issue guidance on their applicability l address concerns over performance and cost
This paper describes the planning, design and long term performance results from the foundation loading experiment.
Planning, design and constructionSite selection and ground conditions
The Bothkennar site was chosen for the experiment for a number of reasons, but principally because it is one of the most widely investigated and characterised sites in the UK. The ground conditions and properties of the soft clay at Bothkennar have been extensively reported elsewhere (by, inter alia, Nash and Lloyd,1989, Hawkins et al,1989 and Nash et al,1992).The 8th Geotechnique Symposium in Print (Institution of Civil Engineers, 1992) gives comprehensive descriptions of the site and its characterisation, sampling and testing.
These investigations found that, beneath a relatively thin desiccated crust (about 1.5m thick), the Bothkennar site is underlain by soft, normally consolidated clay, broadly consistent across the site, varying in thickness between 12m and 22m, overlying gravel. Towards the base of the crust, there is a layer of shells in silty clay that is relatively free-draining. The soft clay has an undrained shear strength (c
u) that shows signs of increasing linearly from about 18kPa just below the desiccated crust to 55kPa at depth (measured using insitu vane). Bothkennar clay has a low permeability (around 10-9
m/s), moderate sensitivity (typically 5), possibly as a result of inorganic and organic cements, and a Plasticity Index typically of 40%.Figure 1 shows a geotechnical profile from Nash et al (1992) that is consistent with the ground conditions under the raft.
Raft design Two types of raft foundation are commonly used for low-rise buildings in the UK: plane slab and stiffened edge-beam (or 'semi-rigid') rafts. Plane slab rafts are not widely used except in ground conditions that show little variation in material properties or have a very thin desiccated crust, such as Canvey Island, Essex. Semi-rigid rafts typically comprise a reinforced, thickened edge-beam running around the perimeter of the foundation and under load-bearing walls, tied together using a reinforced concrete slab typically 150mm to 200mm thick.The semi-rigid raft is more commonly used because of its suitability in a range of ground conditions, including soft ground with widely varying soil properties; because of their more widespread use, a downstand edge-beam raft was chosen for the experiment described here.
The design of the experimental raft had to strike a balance between two conflicting requirements: to realistically represent a low-rise building foundation and yet be geometrically simple enough to allow analysis, economic instrumentation and interpretation of the results. A simple geometry was selected for the experimental raft, since relatively complex shapes and forms are being investigated through case studies of real buildings.
The experimental raft is an 8.1m by 8.1m square foundation with a thickened edge-beam around its perimeter.As mentioned above, there are different approaches to the structural design of such rafts. Atkinson (1993) describes one such design procedure where the full wall loads are carried by the downstand edge-beams. An alternative and more economic approach is that described by Curtin et al (1994) which allows for spread of the load across some of the slab area.For the purposes of this experiment, the slab was designed using the approach of Curtin et al (1994) with a design edge loading of 37.5kN/m (selected from Cement & Concrete Association (1981) for typical two storey property (gable end wall)). Using C35 OPC concrete, this results in the crosssection shown in Figure 2, with A252 mesh reinforcement in the slab and three 16mm diameter high tensile steel bars top and bottom in the downstand edge-beams, tied together using shear links at 200mm centres. The surface of the raft is at about ground level; therefore the foundation partly exploits the strength of the desiccated crust.
Instrumentation was selected in order to monitor the following throughout the experiment:
l the transfer of load from the raft into the soil l changes in total vertical and horizontal stress in the soil l changes in pore water pressure l vertical movement of the raft and the surrounding ground l vertical movement of the soil profile beneath and immediately adjacent to the raft l horizontal movement of the soil profile adjacent to the raft l strain in the raft concrete
This paper concentrates on the response of the raft to vertical loading, primarily in terms of vertical displacement, but also in terms of changes in pore pressure and contact pressures beneath the raft.
Vertical movement points
Thirty-nine stainless steel domed nuts, either screwed onto a threaded bar embedded into the raft or stuck directly onto the slab using epoxy resin adhesive, were installed in a grid pattern (Figure 3) for use as levelling points.Vertical movements are monitored relative to a deep datum by precise levelling. In addition to the slab levelling points, surface settlement rods were installed around the perimeter of the raft (600mm long rods driven into 50mm diameter hand augered holes approximately 350mm deep, sleeved and backfilled with a dry sand/cement mixture).
Five pneumatic piezometers were installed beneath one quadrant of the raft at 1.5m, 3m and 5m depths (Figure 3). Any changes in the pore water pressure detected by these instruments can be compared to a remote standpipe piezometer installed at 3m depth some 25m away from the raft.
Changes in contact earth pressure
Eleven vibrating wire (VW) embedment type earth pressure cells (Tyler, 1976) were installed at various locations over one half of the raft (Figure 3) prior to the concrete blinding being poured.All changes described here were measured relative to the 'locked-in'pressures measured immediately prior to the first stage loading being applied.
Construction and loading The raft foundation was constructed between 10-20 March 1997.Once the raft had been completed the instruments were monitored at approximately fortnightly intervals until first loading on 4 June 1997, when 104.5t of concrete kentledge was loaded around the perimeter of the raft.The kentledge was placed on top of dwarf walls, built near the edge of the raft (Figure 2) to simulate the type of loading applied by a traditional domestic low-rise building. This gives an average line loading of 34.2kN/m run around the 30m of loaded perimeter.No load was applied to the interior of the raft, as such loadings are typically small (1kPa-2kPa) and, by reducing differential movements, would be beneficial in terms of the raft's overall performance. On 5 November 1997 an additional 47.6t of kentledge was added around the perimeter raising the average line loading to approximately 50kN/m run (Cement & Concrete Association (1981) suggests a value of 50kN/m run for a typical two-storey party wall).Figures 4 and 5 show the loaded raft after Stage 1 and Stage 2 loadings.
Following the stage 1 loading, which took place over about six hours, the instruments were monitored regularly over the following two days. Subsequently, the instruments were read at 7, 15, 31, 60, 90 and 150 days. Stage 2 loading was carried out 154 days after initial loading and was completed in approximately two hours.Monitoring has subsequently taken place at intervals of 1,2, 7,35,75,147,261,350 and 471 days after this increase in load.
Results and discussion
The measured movements of the slab have been split into four areas: the centre lines of the dwarf walls; the outer edge strip; the middle strip; and the centre. Plots of measured movement against time for these areas are shown in Figure 6 .
Stage 1 loading
Stage 1 loading on the raft is effectively design loading; the applied perimeter load of 34.2kN/m being very close to the design load of 37.5kN/m.
Average movements for the four areas described above, together with maximum and minimum movements of any of the points on the raft, are shown, to a log scale, in Figure 7. It is clear from the results shown in Figure 7 that, in general, about 50% of the raft movement took place within two days of loading.About one quarter of the raft movement during Stage 1 had occurred by the time the loading was complete, about five hours after the start of loading.While much of this movement will have been undrained settlement, the rapidity of settlement is also, in part, due to the relatively free draining shelly layer; piezometers at 1.5m below ground level barely registered the effects of loading.Bearing in mind that the 'total' settlement measured during the Stage 1 loading is a lower bound, since settlements had not ceased at the start of Stage 2 loading, it is clear that the 'immediate'undrained settlement is a relatively small proportion of the total settlement, perhaps of the order of 15% to 20%.
While it is clear that movement is ongoing right up until Stage 2 loading, the rate of movement has slowed significantly.At this time the performance of the raft can be judged in terms of total and differential settlement and distortion. Figure 8 shows contours of settlement of the raft, together with sections through the centre of the raft and along the edge. The raft has settled in a markedly symmetrical manner, with maximum settlements around the edges, particularly towards the corners.
Maximum and minimum measured settlements of 15mm and 6.5mm respectively have been measured, giving a maximum differential settlement of 8.5mm.This is well within acceptable levels for non-sensitive structures; for example, Eurocode 7 (CEN, 1995) suggests that total and differential settlements of more than 50mm and 20mm respectively may be acceptable provided relative rotations remain within acceptable limits. Across a diagonal, the raft has undergone a maximum relative rotation of about 1/680; this is significantly less than the suggested limit of 1/500.
While piezometers A and D measured only very small increases in pressure, probably due to the changes in groundwater level, rather than the increases in pore pressure due to foundation loading (Figure 9), piezometers B and E measured increases of 7.5kPa and piezometer C measured a peak increase immediately after loading of 5kPa. This change in pore water pressure immediately after loading represents the undrained condition. These changes are comparable to average vertical stress increases calculated assuming a simple 45degrees load spread from the underside of the raft. This suggests that the raft is effective in distributing the load.
Almost all of the VW earth pressure cells registered an increase in pressure due to the foundation loading.The exceptions were cells 9701 and 9703.Cell 9701 is located in the centre of the raft where it is possible that hogging of the foundation due to perimeter loading caused the resultant reduction in pressure.Before loading commenced it was known that cell 9703 was showing wildly fluctuating and inconsistent results.As a consequence all results from cell 9703 have been ignored. Figure 10 shows contour plots of the change in contact pressures between the raft and ground immediately after loading.The overall trend is a symmetrical pressure distribution, with a concentrated pressure zone at the corner where there is an intersection of load from two directions and the highest pressure changes occurring mid-side.
Stage 2 loading Figure 11 shows the settlement versus time relationship over the duration of the project. The increase in settlement due to the Stage 2 loading can clearly be seen by the acceleration in settlements around 154 days.As before about 50% of the raft movement due to the Stage 2 loading occurred within several days. In the longer term it is also evident that the differential settlement across the raft is slowly increasing with time, with maximum and minimum settlements of 24mm and 9mm respectively being measured. Applying the same performance criteria as before, these values are well within acceptable limits defined by Eurocode 7.
However the maximum relative rotation across a diagonal is now approximately 1/400. This is greater than the Eurocode limit of 1/500, though it is reasonable to assume that in a real situation, the stiffening effect of the superstructure would be significant, resulting in smaller relative rotations.
Both the piezometers and VW earth pressure cells responded in a similar manner to that experienced in the stage 1 loading.Figures 12 and 13 show changes in pore pressure and the contour plots of the change in contact pressures between the raft and ground respectively.
It is concluded that:
There has been significant redistribution of stresses under the raft.From the contact pressures, it is estimated that about 40% of the applied load has been carried by the downstand beams, the remainder being resisted by ground under the slab.
Under design loading, the raft's performance was comfortably within criteria for acceptable performance.At 33% overstress, the raft shows no signs of structural distress, and acceptable total and differential settlement criteria are comfortably satisfied. However, the raft's relative rotation was slightly greater than the Eurocode 7 suggested limit of 1/500.
The amount of stress redistribution and the satisfactory performance of the foundation under design load suggest that a design using only the downstand to resist the wall loading is likely to be very conservative.
The experiment described here forms part of a series of experiments at the Bothkennar site (Ground Engineering January 1999), in which the performance of different shallow foundation types for low-rise buildings will be assessed. In parallel with the work described here, a similar raft has been built on ground incorporating stone columns.The next phase of the research programme involves the construction of a plane slab raft. The work described in this paper forms part a programme of research funded by the Department of the Environment, Transport and the Regions.
References Atkinson MF (1993).Structural foundations manual for low-rise buildings.E&FN Spon, London.
Cement & Concrete Association (1981).House foundations for the builder and building designer.C&CA, Slough, England.
Curtin WG, Shaw G, Parkinson GI and Golding JM (1994).Structural foundation designers'manual.Blackwell, London.
European Committee for Standardisation (CEN) (1995) ENV 1997-1. Eurocode 7 - Geotechnical design - Part 1: General rules.BSI, London.
Hawkins AB, Larnach WJ, Lloyd I M and Nash DFT (1989). Selecting the location, and the initial investigation of the SERC soft clay test bed site.Quarterly Journal of Engineering Geology,22,281-316.
Institution of Civil Engineers (1992).8th Geotechnique Symposium in Print.Bothkennar soft clay test site: characterisation and lessons learned.Geotechnique, 42,159-378.
Jardine RJ, Lehane BM, Smith PR and Gildea PA (1995). Vertical loading experiments on rigid pad foundations at Bothkennar.Geotechnique, 45,573-597.
Nash DFT and Lloyd IM (1989). SERC soft clay test bed site, Bothkennar. Report on sample descriptions and photographs.
Report UBCE-SM-89-5.Bristol University.
Nash DFT, Powell JJM and Lloyd IM (1992). Initial investigations of the soft clay site at Bothkennar.Geotechnique, 42, 241256.
Tyler RG (1976).A vibrating wire soil pressure gauge.Tunnels and Tunnelling,8(7),73-78.
The authors would like to thank EPSRC for the use of the Bothkennar soft clay test site.The research described in this paper has been carried out for DETR under the Technology and Performance Business Plan.