In its day, the mercury bath method of determining shrinkage limit was accepted as the best that could be achieved, with the technology of the time and within the prevailing attitudes to safety.
But with new technologies available and a better understanding of the health hazards of mercury that go far beyond the madness of hatters, British Geological Survey (BGS) had designed a new test that gives better, more accurate results without the use of hazardous substances.
The current test, as recommended by the British Standards Institution, is unchanged since the 1950s when it was developed by the then Road Research Laboratory (now TRL) (BSI 1990; RRL 1952; Ackroyd, 1969).
The test involves submerging wet samples in a bath of liquid mercury, then repeating this at each subsequent drying stage. The change in volume is calculated from the displaced volume of mercury using Archimedes' Principle.
The apparatus recommended in British Standard 1377 is difficult to use with fi ssured, voided, silty, weak, or highly plastic clays in an undisturbed state. Over-consolidated clays and tropical clays (Hobbs et al, 2000) usually fall into this category.
Fewer problems are experienced when testing remoulded or normally consolidated clay soils but fi ssures may still develop. Small globules of mercury can enter and remain within the specimen during drying, especially when surfaces are rough or silty.
Although tapping the specimen can dislodge large droplets, small ones barely visible to the naked eye remain within the fi ssures. Also fragments of soil may detach from the specimen and fall into the mercury bath. This results in combined volumetric and weighing errors of up to 5%.
The BS1377 subsidiary method (equivalent to the ASTM method - ASTM,1985) uses a small disc of remoulded soil. This test is even less suited to undisturbed soil specimens.
Both BS1377 methods carry a signifi cant risk of spillage of mercury in liquid form or release as a vapour, even at room temperature. Where tests are carried out, they must follow strict health and safety guidelines and use a fume cupboard (Figure 1).
However, the risks from mercury are such its use is banned in many soils laboratories and a blanket EU ban is thought to be imminent. Even at low concentrations, the metal may cause irreversible damage to nerves and brain tissue and at high concentrations can be severely disabling or fatal.
The BGS' new test method uses the fully automated Shrinkit apparatus, incorporating a Class 1 laser, 3D moving platform and a digital balance to measure 3D shrinkage of a large 100mm by 100mm cylindrical specimen (Figure 2).
Continuous measurements of sample size and weight under controlled conditions are then used to determine shrinkage limit and other parameters, without any contact with the test specimen during drying.
The size of the specimen, and the much reduced handling when compared with the TRL test, enables a more representative specimen to be tested, and also enables more fragile materials to be tested.
And although other researchers have developed similar techniques for shrinkage measurement (for example, Parcevaux, 1980; Braudeau et al, 1999) the BGS Shrinkit apparatus is the only one to have been calibrated against the original BS test.
The test has been developed to support ongoing research by the BGS to determine the real swell shrink characteristics of different soil and rock types across Britain.
Much use is made worldwide of inferred swelling and shrinkage behaviour indirectly from standard soil index test data such as plasticity, density and water content, which are widely accepted. This is partly due to the fact that direct swelling and shrinkage tests are often difficult to perform, particularly where undisturbed specimens of weak, fissured, or sensitive soils need to be tested.
Soil structure, fabric and water content contribute to test diffi culties and tend to make correlation with indirect index test data, derived solely from disturbed or remoulded soil, impossible.
Shrinkage limit is defi ned as the water content of a soil below which little or no further volume reduction takes place (Head, 1992).
Shrinkage limit has been identifi ed as one of three key characteristic water contents of a soil (the others are the plastic and liquid limits) but it is the only one to have a fundamental structural signifi cance (Sridharan and Prakesh, 1998).
The idealised characteristic shrinkage limit curve is shown in Figure 3 with volume reduction progressing steadily between liquid and shrinkage limit, and reducing markedly at water contents below the shrinkage limit. Actual experimental curves vary in shape and smoothness, but follow this overall trend.
This research forms part of the BGS' 'shrink/swell' research project investigating the geotechnical properties of major clay formations in the UK.
Detailed geotechnical, geochemical, and mineralogical analyses of samples collected nationwide are enabling BGS scientists to provide a defi nitive model of the shrink/ swell characteristics of engineering soils.
In addition to a series of forthcoming papers, the information obtained from the project provides an essential input to the BGS' national geohazards assessment.
Geological formations investigated so far include the Gault Formation, the Mercia Mudstone Group, and the Lambeth Group (Jones and Hobbs, 1998; Hobbs and Jones, 1998; Jones and Hobbs, 2005).
Shrink/swell properties of the Lias Group and London Clay Formation are currently being investigated.
The project uses several direct methods of laboratory determination of swelling and shrinkage, most of which are little used in UK engineering.
Development of this test procedure forms an important part of research, enabling samples to be tested in a safe and efficient environment.
However, it is important to verify the results with those derived using conventional techniques. The results of comparative tests carried out on various UK clay soils are shown in Figure 4 ('Tertiary' data refers to London Clay and Lambeth Group samples, and 'Glacial' to glaciolacustrine and till samples).
This research shows there is no doubt that the potentially hazardous mercury bath may be consigned to the (contaminated waste) dustbin and this new method can be used confi ently to produce consistent, accurate results, comparable to existing procedures, but in complete safety.
References Ackroyd (1969). Laboratory testing in soil engineering. Soil Mechanics. Geotech. Monograph No1.
American Society for Testing and Materials (1985). Annual book of ASTM Standards.
Section 4, Construction, Vol. 04.08, Soil & Rock: Building Stones, ASTM.
Braudeau E, Costantini JM, Bellier G and Colleuille H (1999). New device and method for soil shrinkage curve measurement and characterisation. Soil Sci Soc Am. J. Vol 63, No 3, pp525-535.
BS1377 (1990). Soils testing for engineering purposes BS 1377, Part 5. British Standards Institution.
Head K (1992). Manual of soil laboratory testing. Volume 1: Soil classifi cation and compaction tests. John Wiley & Sons.
Hobbs PRN, Northmore KJ, Jones LD and Entwisle DC (2000). Shrinkage behaviour of some tropical clays. In: Unsaturated Soils for Asia, Rahardjo, Toll, & Leong (eds. ), Balkema, Rotterdam.
Hobbs PRN and Jones LD (1998). The shrinkage and swelling behaviour of UK soils: Gault Clay (1998). By BGS Technical Report No WN/98/13.
Jones LD and PRN Hobbs PRN (1998). The shrinkage and swelling behaviour of UK soils: Mercia Mudstone (1998). BGS Technical Report No. WN/98/14.
Jones LD and Hobbs PRN (2005). The shrinkage and swelling behaviour of UK soils; the clays of the Lambeth Group. British Geological Survey Research Report RR/04/001 Parcevaux P (1980). Etude microscopique et macroscopique du gonflement de sols argileux. PhD Thesis. Universite Pierre et Marie Curie, Paris.
RRL: Road Research Laboratory (1952) Soil mechanics for road engineers. HMSO, London, 541p.