Your browser is no longer supported

For the best possible experience using our website we recommend you upgrade to a newer version or another browser.

Your browser appears to have cookies disabled. For the best experience of this website, please enable cookies in your browser

We'll assume we have your consent to use cookies, for example so you won't need to log in each time you visit our site.
Learn more

Settling the score

Report on the BGS/Ground Board meeting 'Brownfield sites: foundations on fill', at the Institution of Civil Engineers on 27 October 1999, by Trevor Hardie, Geotechnics.


Main speakers at the meeting were Andrew Charles of BRE and Len Threadgold of Geotechnics.

There were also two brief presentations on recent research and a case study. The meeting was chaired by British Geotechnical Association (formerly the British Geotechnical Society) chairman Quentin Leiper.

Charles began by presenting the results of BRE research undertaken on behalf of the Department of Environment, Transport and the Regions. Risk assessment for brownfield sites is mainly concerned with hazards to human health and increasing attention is being given to environmental issues.Many brownfield sites are covered in considerable thickness of fill and the associated geotechnical problems may affect the economic development of the site.

A survey carried out by BRE in 1992 indicated that nearly 20% of low-rise construction in the UK took place on filled ground. Pressure on the restriction of development on greenfield sites is likely to increase substantially the amount of building on filled ground.

Much filled ground used for development has not been compacted to an engineering specification and has primarily arisen as a byproduct of human activity, generally associated with the disposal of waste materials. These areas vary from large opencast mines to relatively small holes backfilled with materials including colliery spoil, pulverised fuel ash, industrial/chemical waste and urban waste.

Typically, little control has been exercised during filling operations, resulting in large variations within the fill.Ground treatment will often be required before development.

The main geotechnical hazard for buildings on fill is long-term total and differential settlement.

The main settlement problem is usually associated with effects other than the loading from the building, including self-weight of the fill, creep, collapse settlement, biodegradation, chemical reactions, dynamic loading and settlement of the natural ground beneath the fill.

This is a large area of research for the BRE and Charles presented the results of research on the following three key issues:

Identification and quantification of potential for collapse compression of fill on inundation In this phenomenon, usually referred to as collapse settlement or collapse compression, a reduction in volume occurs in a partially saturated soil when the moisture content increases, without any increase in applied stress.

The mechanisms and processes at work during collapse settlement may be complex, but from some points of view the process is quite simple and can be demonstrated in oedometer tests.

However, predicting the magnitude of the collapse potential for a fill insitu is not so easy.Many factors can affect collapse potential, such as the nature of the fill material, its grading, moisture content, dry density, percentage of air voids, stress level and stress and moisture content history.

Site history can also be important. Has the fill been submerged in the past? If it has, there is a low chance of collapse settlement occurring. However, it should be assumed that collapse settlement can occur, if the previous moisture content or water level history of the fill is unknown. The increase in moisture content of the fill which triggers collapse compression can be caused either by downward infiltration of surface water or by rising groundwater level. Settlement of fill may cause drains to fracture, causing ingress of water and leading to much larger settlements.

Collapse settlements have been associated with loess soils and natural silty sands, but two major field experiments carried out by BRE in the 1970s demonstrated that collapse compression on inundation is also a major hazard at opencast mining sites. At Horsley, in Northumberland, a 70m deep opencast pit was dewatered as part of the excavation process. The water level rose 35m after backfilling of the pit, leading to surface settlements of 500mm, occurring where the water was rising. At Corby, water was introduced into backfilled opencast excavations via surface trenches, also leading to surface settlements.

Engineered fills are increasingly being placed specifically to make brownfield sites suitable for building development and eliminating collapse potential is a critical consideration. If the air voids are reduced to 5% or less, collapse compression is highly unlikely.The adequacy of the commonly adopted 95% relative compaction criterion is much less certain.

Charles's main conclusions were that:

Poorly compacted or excessively dry fill is always likely to be vulnerable to a reduction in volume on inundation.

Collapse compression on inundation of unsaturated fill is a major hazard for buildings on nonengineered fill.

Evaluation of likely differential settlement associated with variations in depth of fill There are many causes of differential settlement and changes in the depth of fill is one. It is differential rather than total settlement that damages buildings.Where the depth of fill changes dramatically, there could be potential for unacceptable long-term differential movement. These areas, where buildings should not be built, should be defined.

There are major implications for safe and economic redevelopment of many brownfield sites. If unsafe decisions are reached, serious building damage can result. However, an over-conservative approach will needlessly sterilise large areas of land and could make a site uneconomic to develop.

The general advice is to not build across the edges of filled ground so that a structure is partly founded on natural ground and partly on fill.While this is good advice, it is not sufficient in all cases. Indeed the economics of a development may be affected if construction does not take place in certain areas of a site.

One way of defining acceptable performance criteria for a building might be to look at the tilt, excluding development from areas where tilt exceeds what a building can withstand, say 0.2%. Finite element analyses can be used to define the exclusion zones in terms of unacceptable tilt, but it is questionable whether such analyses are realistic for the rather unusual 'loading'conditions.

Identification and specification of appropriate ground treatment for non-engineered fill The poor load-carrying properties of many non-engineered fills have been associated with their heterogeneity and their loose, poorly compacted condition.Ground treatment should aim to overcome these problems.An important objective is to eliminate, or at least minimise, volume change within the fill during, and especially after, building; eliminating or at least mitigating collapse potential.

Although proprietary techniques such a vibro stone columns are commonly used, earthmoving methods involving either excavation and recompaction or pre-loading can be very effective in some situations.Temporary pre-loading or surcharging puts the fill into an overconsolidation condition and is the 'natural' method of ground improvement. In simple terms, a load larger than the foundation and floor loads is applied.This is advisable but may not be sufficient where the deficiencies in the soil are not related to the building loads.

To determine the effective depth of the surcharging, an expression can be developed in terms of the vertical stress increment (. s v) produced by the surcharge at depth (z) in relation to the overburden stress (. z) at that depth, say . s v/. z = 0.25.Using elastic theory, this leads to predictions of depth of effectiveness which appear somewhat optimistic when compared with field measurements. If the depth of effectiveness is z and the height of the surcharge is H, there is some field evidence that z/H is typically in the range 1.1 to 1.3. Pre-loading can be a very effective form of ground treatment but there is a need to be realistic about effective depth of treatment. Charles' main conclusions were:

Many brownfield sites contain considerable depths of fill.

Hazards of building on fill need to be understood.

Moisture movement through the fill and collapse compression are key factors in hazard assessment.

The influence of variations in depth of fill on ground movements needs to be better understood.

The role and limitations of ground treatment need careful assessment.

Pre-loading can be a very effective form of ground treatment.

Case histories

Len Threadgold briefly detailed the results of four case histories where development on areas of filled ground up to 70m thick was carried out or is proposed. It is essential in the development of a brownfield site to consider its characteristics and whether these give acceptable or unacceptable performance. Fill characteristics are determined by carrying out:

Desk study

Intrusive and non-intrusive investigation Insitu and laboratory testing Informed monitoring Analytical monitoring The main issue for brownfield development is the amount and rate of settlement (total and differential), this being intimately related to the loading conditions.Each of the case studies presented involved placing a surcharge load to create over-consolidation within the fill so that the stress levels of the future building operated along the reload/unload section of the load-settlement curve (Figure 1).

Snatchill, Northamptonshire

The project comprised the development of opencast ironstone workings, typically up to 25m deep, that had been backfilled with arisings from excavations, predominantly reworked glacial till overlying reworked limestone. The site has been developed for housing. Ground treatment comprised applying a surcharge load of 100kN/m 2using 5m to 6m of fill, exceeding future local and general loadings. The settlement was monitored during surcharge loading and unloading using simple monitoring plates (Figure 2). Individual house locations were proved by adjacent pitting and probing and all foundation excavations were inspected and tested before construction. Typical results from one of the monitoring plates are shown in Figure 3. This showed that settlement under surcharge load occurred rapidly following application and that the movement on unloading was much less than on initial loading.Movement on reloading would be expected to be similarly low.The site has been successfully developed using this technique.

Leabrook, West Midlands

At Leabrook, development has taken place on 15m to 20m deep backfilled opencast coal pits.A variety of treatment methods were used. In some areas, spoil heaps had already surcharged the building areas and their movement had been monitored, providing valuable data. Other areas that had been only partly surcharged by spoil heaps had further surcharge placed to minimise differential settlements.The area of one building was not surcharged but was designed to accommodate settlements.At another location, vibro-compaction was used to treat 7.6m to 8.5m of fill.All of the buildings have performed adequately.

Brunswick Power Station, London

At Brunswick Power Station, development of the site involved a combination of ground treatment methods to meet tight settlement criteria.Within the area of the turbine hall 8m of fill had been placed on the old concrete base. The lower 4m had been treated using vibro-compacted stone columns and the upper 4m was dynamically compacted. The fill was then surcharged to the proposed foundation loads of 60kN/m 2.Figure 4 shows the typical response of two of the settlement plates which were placed before the surcharge load.These were monitored during filling and while surcharge was removed.A long-term load test has also been carried out on a 5m by 5m reinforced concrete slab, the resulting settlement being similar to that predicted from the calculated m vvalues from the surcharge monitoring results.

An 80ha development on backfilled opencast coal pits is proposed in Staffordshire.As part of the ground investigation, a trial surcharge embankment was constructed and monitored. During the trial, water was introduced into the made ground beneath the trial embankment by pumping water from a nearby lake into wells installed before fill placement.

The instrumentation layout of the trial is shown in Figure 5.Settlements of up to 75mm were monitored beneath the centre of the trial area on the monitoring plates and in the extensometer.The depth of influence of loading measured by the extensometer was of the order of 12m.As a comparison, the depth of influence calculated from elastic theory and that proposed by Charles (1996) were compared to the measured distribution. Elastic theory appeared to over-predict the depth of influence while there was reasonable agreement with that proposed by Charles.

Threadgold said that in each case, ground was tested to the depth that would be influenced by the proposed structures and site-specific full-scale design parameters were obtained. The ground had also been treated, minimising future total and differential settlements under building loads and generally causing rapid settlement of unsaturated fills.

Other presentations Neil Trenter presented the results of a study by consultant Halcrow of the controlled backfilling of an opencast site in the English Coal Measures.Up to 56m deep and bounded by a river, the site was backfilled with compacted mudstone, siltstone and subordinate sandstone.As part of the control procedures, some 200 insitu dry density and water content tests were made over 12 months.

Average water content was about 8% (some 2% dry of optimum, 2.5kg rammer), and the average dry density was about 2.0Mg/m 3(some 97.5% of maximum dry density, 2.5kg rammer).

After backfilling, settlement was monitored.Very little settlement was recorded during the first 400 days but, as soon as water began to rise up in the base of the backfill (shown by standpipe readings) settlement was rapid, up to about 0.8% of the total fill thickness. Settlement decreased markedly after about 600 days, when groundwater rise ceased.

Trenter pointed out that collapse settlement had been expected at the site (because of the groundwater inflow from the adjoining river) and appropriate precautions had been taken.

In other cases, what could be done to limit this phenomenon? Figure 6 shows the three factors Trenter felt were of critical importance to collapse settlement.For a given confining pressure (in this case, the depth of the backfill), the water content with respect to optimum was considered to be most critical. In practice, however, it is very difficult to raise the water content of stiff cohesive soils and argillaceous weak rocks, because of the difficulty of ensuring complete water penetration into large chunks, using bowser spraying or similar methods.Water would only penetrate the top few millimetres of the chunks and, in hot weather when water addition was most necessary, much of it evaporated in any case. Thus achieving optimum water content in such circumstances would not normally be practical.

The alternative would be to increase the compactive effort.This would have the effect of reducing the air voids and bringing the optimum water content closer to the water content achieved in the field (Figure 7). Care was needed in such cases because cohesive fills compacted to high densities at low confining pressures (such as near the surface) could heave, and so different levels of compactive effort might be required for deep fills, if the twin problems of collapse settlement and heave were to be avoided.

Hilary Skinner of the BRE briefly described some large scale testing that is being carried out in a test pit at the BRE to model the effects of a buried high wall of a backfilled opencast pit or quarry on the settlement profile.A 2m high wall with a 60degrees face was constructed from concrete blocks in the base of the 4m deep test pit.Very loosely placed mudstone fill was then put in the pit.The settlement was measured under self-weight of the fill (creep) and the water level was raised in the pit. The results showed similar shapes of surface settlement profile at each stage of testing over a wide range of maximum settlement. Finite element analyses carried out on the trial modelled the shape of the profile but positioned the settlement profile too far from the top of the wall.

Discussion William Powrie of Southampton University noted that the depth of influence data from the Poplars site could be explained by an increase in stiffness with depth. This may be due to high stiffness at small strains where the applied stress is very small compared to the overburden stress at depth.

Neil Smith of Applied Geotechnical Engineering said that in the Far East, local engineers often use water to cause inundation collapse as a form of ground treatment. Charles responded that at Corby, where this form of treatment was tried, the results were variable with large differences in the settlements.

Jan Hellings of Maunsell stated that settlements caused by water infiltration were more common and non-uniform but more difficult to predict, whereas inundation due to rising groundwater gave rise to more uniform settlement.

CIRIA is running a research project , 'Treated ground engineering properties and durability.'The 18month project being undertaken by the BRE began in April 1999 and is led by Andrew Charles.


Charles JA (1996). The depth of influence of loaded areas, Geotechnique vol 46(1), pp51-61.

Have your say

You must sign in to make a comment

Please remember that the submission of any material is governed by our Terms and Conditions and by submitting material you confirm your agreement to these Terms and Conditions. Please note comments made online may also be published in the print edition of New Civil Engineer. Links may be included in your comments but HTML is not permitted.