Briefing: Geotechnical engineering and building research: the early days of soil mechanics at BRS Free

The achievements of Charles Frewen Jenkin (Fig. 1) were remarkable. It was he who brought soil mechanics to building research. Son of Professor Henry Charles Fleeming Jenkin, the first Professor of Engineering at the University of Edinburgh, he too became a first Professor of Engineering Science at Oxford University, when that Chair was formed in 1908. In his inaugural lecture he argued that while Britain had been first with the Industrial Revolution, other European countries, in order to catch up, had formed colleges to teach science and engineering, where there was perhaps an excessive concentration on calculation as a means of explaining behaviour. He felt that there was a need for practical work, for laboratories with models to test and an understanding of the art of measurement. He was essentially a practical man himself. After graduating in mathematics at Cambridge in 1886, he served as an apprentice in the workshops of Mather and Platt, followed by industrial experience with the London and North Western Railway at Crewe; the Royal Gunpowder factory at Waltham Abbey; Nettlefold's iron and steel wire factory at Newport and Siemen's Electrical Engineering Works at Woolwich, where he became an acknowledged expert on electric railways.

The Government Department of Scientific and Industrial Research established various experimental stations, such as the National Physical Laboratory, and in 1921, the Building Research Station, the first in the world to be devoted to building research. This station began life in temporary buildings in East Acton, but in 1925, moved to Bucknalls Mansion, a large house in its own grounds that was built in 1855 near Watford, northwest of London.

The British Association for the Advancement of Science used to hold its annual meetings in various parts of the country, and it was invited to hold its annual meeting of 1925 in Southampton, where difficulties were being experienced with the docks. Mr Wentworth-Shields, chief engineer for the docks, described problems with the retaining walls and during the subsequent discussion it was agreed that research was needed into the determination of earth pressures so that the design of retaining walls could be improved. The British Association accepted the task of looking into this problem and established an Earth Pressure Committee, chaired by Mr Wentworth-Shields. In 1926 this Committee sought advice from Dr Stradling, the Director of Building Research, for suggestions on how the required research could be carried out and who could do it. He suggested Professor Jenkin.

Professor Jenkin accepted the challenge with enthusiasm. Starting from first principles, he had boxes made with movable ends, into which he placed dry sand. Release of the end wall in various ways, such as rotating about the bottom or top, or moving out bodily, caused the dry sand to fail in different ways. The measured forces on the wall were not repeatable, and it was concluded after a great number of experiments had been carried out, that sand does not behave in the manner assumed by Rankine and that its ‘angle of friction’ depended enormously on the closeness of packing.

Professor Jenkin soon decided that he needed better facilities and to be away from the constant demands of his students. Explaining this to Dr Stradling, he asked if he could come to the Building Research Station to continue his research. He resigned his chair in June 1929, two years before his statutory date for retirement, and his apparatus was transferred from Oxford to BRS. Jenkin continued his work with dry sand, studying the effect of density variations and of dilatancy, until he had developed a revised wedge theory, publishing his results in three papers during 1931–1932.1–3 

Professor Jenkin recognised that the real practical problems did not lie with dry sand, but with wet clay. He therefore began a programme of studying the strength and deformation properties of wet clay. He chose kaolin as his clay, which he was able to buy in the form of a dry powder. This he mixed with water to form a slurry which was then subjected to a vacuum to draw out air, and consolidated in a filter press of 9·65 in² (62·26 cm²) area, which was in effect a 3½ inch (88·9 mm) diameter oedometer that could produce a vertical stress of 103·6 lbs/in² (714·5 kN/m²). Curves of pressure versus water content were found to be reproducible. The 3½ inch diameter cakes from the filter press were ⅜ inch (9·5 mm) thick. Cylindrical samples were made by pushing a 1 inch (25 mm) diameter cutting tube through the cake, slightly coning the end with a rotating hand tool to produce an outward cone, then pressing it into another cake until the required length of sample was obtained. They had remarkably consistent reproducible water contents throughout their length and, as subsequent tests showed, were saturated. These samples he tested in laboratory equipment he designed and had made in the BRS workshops. All were advanced for their time and included a triaxial compression apparatus, an autographic hysteresis loop apparatus that applied unconfined compression and an autographic ring shear device. The triaxial cell, shown by Fig. 2, and called by Jenkin a conjugate pressure apparatus, had a brass cylinder to retain the lateral pressure, and the vertical compression force was applied through an internal spring by a threaded rod that could be screwed by turning an external handle, thereby avoiding the plunger friction problem of later machines. The movement of the head of the sample, and hence the compressive strain, was measured by a second threaded rod carrying an electrical contact at its lower end. This was screwed down manually so as to just touch the sample head, as indicated by an electric bulb controlled by the contact. Samples were made at least twice as long as their diameter. To reduce barrelling, the end pieces of the apparatus were made slightly conical, and the sample ends trimmed to a matching female cone with a rotating hand tool, so that throughout the compression, the ends of the sample spread, keeping it more nearly cylindrical. In calculating the compressive stresses, the cross sectional area was assumed to increase as the sample shortened, remaining a right cylinder of constant volume.

The unconfined hysteresis loop apparatus, shown by Fig. 3, used the torsional elasticity of a long $316$ inch (4·76 mm) diameter solid steel rod to measure load. The ends of the rod were clamped to the base and its centre could be rotated by an attached light steel tube that was initially vertical, carrying at its top a smoked glass plate. Deflection of the sample was measured by a long counterbalanced recording arm carrying a stylus that could move over the smoked plate, thereby scratching a load/deflection curve. The apparatus could be used to apply cyclic loading or simply used to make unconfined compression tests.

A whole series of repeatable unconfined compression tests made in the conjugate pressure and the autographic apparatus at water contents up to 78% produced a unique curve, shown in Fig. 4, relating unconfined compressive strength to water content. Lower water contents were obtained by allowing the samples to dry, right down to oven dry, but below the critical shrinkage point, no further strength increase was obtained: if anything there was a slight drop in strength when all the water had been driven out.

A series of tests with samples covered by a very thin rubber tube under conjugate pressures from −1 to +1 atmosphere revealed no change in strength. This was taken to imply that the conjugate pressure was transmitted hydrostatically by the water in the clay and indicated that the samples were saturated.

Some unconfined compression tests were made on a very fine sand with 81% particles of 0·05 mm size and the rest smaller. A saturated slurry was poured into a sample tube, slightly consolidated and extruded on to the platens of the unconfined compression apparatus. Rapid application of a small load stiffened the sample so that the test could be carried out. As the load increased, the sample took on a drier, matt appearance and the surface appeared to get steadily drier until failure occurred on a diagonal shear plane. This behaviour was explained at the time as due to dilation drawing water into the surface capillaries, causing an internal pressure deficiency that increased steadily as the menisci were pulled deeper into the sample by the dilatancy. This caused the normal force between adjacent particles to increase, so increasing the strength of the sample. Ultimately the point was reached when the water surface could get no deeper into the capillaries without the film breaking at points on the sample surface, allowing air to enter along an incipient shear plane, allowing of shear. Compression strength at failure of a test was 6·7 lb/in² (46·2 kN/m²) and the measured inclination of the shear plane was 27·1° to the vertical, giving Ø = 35·8°. This implies that the confining pressure was 2·4 lb/in² (15 kN/m²). With the clay, composed of much smaller particles, no air was drawn into the samples so that they were under the confining pressure from the surrounding air. The effect of this confining pressure was illustrated by a compression test in which the sample could be surrounded by water. When the water was added, the strength fell virtually to zero. Looking back, we can see that this explanation indicated that the importance of effective stress was appreciated, but at that time pore pressures were not measured. The pressure at which air can enter a saturated porous material depends on the size of the pores. A low air entry coarse filter with pores sizes 0·02–008 mm has a blow through pressure of 0·4–4·0 lb/in² (2·8–28 kN/m²) indicating agreement with the air entry value of 2·4 lb/in² found indirectly for the fine sand. High air entry filters that will withstand atmospheric pressure have pore sizes less than 0·001 mm, a size likely to be greater than that in the clay.

A point that concerned Professor Jenkin was the strength at zero normal load, and to determine this he devised the ring shear apparatus, shown in Fig. 5, that utilised the same autographic arrangement as the hysteresis loop apparatus. It had a counterbalanced top ring, so that a sample could be sheared without any normal load being applied across the shear plane. Labels on the photograph of the apparatus show the torque rod (L) and the lightweight tube (K) carrying the smoked glass plate at its top.4 

Professor Jenkin felt the cold and when at his desk he used to keep his feet and legs in a hay box. He was clearly suffering poor circulation and heart problems, which reached the stage in 1933 to make him decide that he must retire. To round off his work, he wrote Note 7505 which embodies the conclusions arrived at from the experimental work on clay and a short account of some tentative theories that had suggested themselves as the work had developed. It is said that he left the laboratory saying he was going home to die. He certainly retired, but in fact did not die until 1940, aged 74.

Amongst the staff who had joined BRS earlier, in 1927, was L. F. Cooling, a physics graduate from Birmingham University, whose photograph is given in Fig. 6. He worked in the brick and stone section, studying the problem of Whitehall coping stones, and measured the pore sizes of the various stones by using a suction plate apparatus. When Professor Jenkin retired, his two assistants, Bevan and Smith were joined by Cooling. His knowledge of pore sizes and pore suctions had considerable relevance to soil behaviour, although no attempts appear to have been made to measure pore pressures at that stage. Cooling became head of the research work and in September 1933 the unit was given the distinction of being made into a section. The new section began a programme of research on behalf of the Road Research Board which had been formed in April 1933. Previously experimental work for roads had been carried out by the Road Department of the Ministry of Transport which had an experimental station set up in 1930 at Harmonsworth, near what is now Heathrow Airport. This became the Road Research Laboratory under DSIR and Dr Stradling was made Director of both BRS and RRL.

The publications of Terzaghi, particularly those associated with his connection with the US Bureau of Public Roads and his methods for classifying soils were followed, determining the ‘lower liquid limit’ and the ‘lower plastic limit’ as well as the centrifuge moisture equivalent, i.e. the amount of water retained when the sample had been subjected to 1000 g for one hour. A lorry mounted device with a big screw jack that used the weight of the lorry to push down, was made for taking ‘undisturbed’ samples, in core tubes that were forerunners of the ubiquitous ‘U4’ sample tubes (U100 since metrication). The 2 ft (0·61 m) long tubes were split, with a cutting nose of $4316$ inches (106 mm) turned in to $4116$ inches (103 mm) at the cutting edge to give clearance to the sample as it passed into the sample tube. It was twisted half a turn to shear it off, then pulled out, split open and the sample slipped into a close fitting tin, closed by a cap and sealed with insulating tape for transport back to the laboratory. Samples had been obtained from depths of 6 ft (1·8) in waterlogged clay and firm sandy soils.

Tests were made to see the effect of remoulding. London clay at a water content of 29% showed a reduction of unconfined compression strength from 31 to 22 lb/in² (214 to 152 kN/m²). Assuming that the shear strength was half that in unconfined compression, good correlation was obtained between the shear strength of clay measured in the ring shear apparatus under no normal load and by unconfined compression tests: work described by Cooling and Smith.6 

A table was drawn up to record the field work, and each new job was given a location number. The first of these, dated 6 September 1933, was Denham by-pass road, where samples were taken of clay. Twenty-seven locations developed during the next year, the majority being road jobs. Two exceptions were the settlement and cracking of buildings at Combermere Barracks, Windsor; and a pile test pit at BRS. During the following year, 1934/5, more interesting jobs involved the failure of a retaining wall near the LMS tunnel at Kensal Green; three embankment slips on different parts of Southern Region Railways; and testing samples from British Guiana and from Paris. It was recognised that road foundations were affected by capillary pressures, changes in water content and variable loading. A start was made measuring movements of road foundation by periodic levelling from stable datums, field measurements were made of the settlements of large buildings being constructed on clay, and samples had been obtained from the sites of retaining wall failures and from embankments that had failed. The soils section was transferred to the Engineering Division of BRS. Eight embankment slips had been investigated before the failure of Chingford Reservoir, entered as Location 62 on 13 August 1937.

By 1935 the new section had expanded from the laboratory experimental work of Jenkin to field studies of the behaviour of structures and earthworks. The techniques and equipment developed by Jenkin were used to measure the shear strengths of clays in the studies of full-scale failures, and the scene was set for further studies of the behaviour of civil engineering structures. Preparations were in hand for the first international conference on soil mechanics and BRS Assistant Director had suggested that Cooling should attend.