Report on 'Learning from pile tests', held at the University of Nottingham on 27 April, by Richard Bennett, researcher, Pavement & Geotechnics Research Group, University of Nottingham.
Speaker Dr Ken Fleming of Kvrner Cementation began by reminding the audience of the First International Conference on Soil Mechanics, held at Harvard University in 1936. A great topic of debate at that time concerned dynamic pile driving formulae, which were recognised as inconsistent and exhibiting poor correlation with static load tests. Indeed soon afterwards Terzaghi (1943) concluded that despite their popularity with practising engineers, such formulae had no basis in fact.
Fleming suggested that modern dynamic test methods still exhibited serious problems. Loading times of around 20m/s meant that dynamic methods could not detect consolidation and creep, which were the very soil factors that determined the settlement of a pile. They were also unable to incorporate the effects of pore water pressures. Possibly the most significant modern advances were in the automatic construction, monitoring and testing of piles, which guaranteed consistency, he added.
Fleming described the modern configuration of static load tests (Figure 1) which have benefited to a great extent from the consistency generated by automating the tests. Instrumentation now allowed loads to be applied to a pre-set 'recipe' and maintained to within 0.2%. Automated tests offered improved safety with pre-programmed shut down routines, and remote control. He said that the habit of static load testing had evolved in the UK possibly because contractors were keen to prove the load bearing capacity of their piles. This in turn allowed a great deal to be learned about piles. But he criticised the traditional approach to static load testing, which involved the selection of erratic load holding periods, making the tests unreliable and comparisons between results very difficult. Most published data of these tests was misleading, he thought.
With tight load control and a sensible recipe of loading, Fleming said, it was possible to find out what happened to the pile by replicating the settlement curve with a pair of linear fractional functions, so called because of the way they changed in time. All settlement curves could be replicated in this way and used to project an asymptote. A smooth curve was created and this process effectively removed the effect of the erratic load holding periods. Models had been constructed from linear fractional functions (Figure 2) for base resistance, shaft friction, and elastic shortening. The hysteresis loop generated by linear fractional functions had been found to track the settlement recovery effectively. This approach had been successfully applied to over 1,500 load tests, with some inter- esting results. For example the dimension- less parameter for shaft friction, Ms, which is norm-ally assumed to fall within the range 0.0005 to 0.005, had actually been found to be near constant in the region of 0.001. Back cal- culation using a programme based on linear fractional func- tions could be used to define the load in the pile, load in shaft, load on base, and to calculate the settlement recovery.
A number of examples were given to dem-onstrate that this method could be used to identify anomalies dur- ing construct- ion. It was used to detect the problem of poor base construction (Figure 3). In this case a pile was bored by rotary drill, where excav- ated spoil fell back into the base of the bore. The effect of debris in the base was revealed from back analysis by a steep or vertical end portion of the load/settlement curve. A similar effect could be generated by heave from adjacent piles, particularly in soils overlying a rocky or sandy base layer.
Fleming noted that another common problem occurred in ground with a blocky rock structure, such as mudstone or chalk. In some cases high end bearing during driving occurs, but with low end bearing and high shaft friction during loading. Figure 4 shows the case of a driven precast pile, only 5.3m long and driven by a hydraulic hammer with a set of about 30mm per 10 blows and a 0.5m hammer drop. The pile was founded into mudstone and the overlying soils were of low strength. The net effect was a low end bearing value and a higher than anticipated shaft friction. The mudstone had a blocky structure with the blocks assessed by coring as being smaller than the pile width. The inter block material was soft clay. It was believed that in this situation the blocks were pushed downward and effectively became an extension to the pile. As they were pushed downward they developed shaft friction in the same manner as the pile (Figure 5).
A 10.7m long pile installed in beach deposits by a continuous flight auger was then used to demonstrate the effect of drilling disturbances. Load testing revealed very low shaft fraction of 180kN. In this case Dutch cone penetration tests revealed the auger had been overdigging and pulling material in from the sides. Fleming said this behaviour could be predicted from Terzaghi's theory of shaft stability. With the onset of overdigging the angle of friction of the soil went down, and the horizontal active earth pressure increased, pushing more soil onto the auger. He stressed it was important to avoid this phenomenon and that instrumentation was applied to augers partly for this reason.
Another example, revealed during a piling competition in Italy, was presented to illustrate the effect of base grouting. Six contractors were invited to install piles through a silty clay into sand. One competitor used base grouting, with significant results. In this case, the effect of base grouting was to reverse shaft friction which resulted in a stiffer pile with no increase in load. This condition was therefore directly analogous to prestressing.
Another common problem was damage to precast concrete piles through overdriving, Fleming said. Engineers who were enthus- iastic to achieve the requirements for set, often damaged piles by overdriving. This was part- icularly prone to occur in clay or ground which alternated between hard and soft driving resistances. The results of such actions could be revealed by the load test as excessive apparent elastic shortening.
Fleming then described a model for soil dilation caused by shear at the walls of the pile (Figure 6). This consisted of springs which represented the horizontal active earth pressure, and balls which represented the shear surface. When loaded rapidly, the forces in the springs were high, when loaded slowly the forces were low. He suggested this was currently the most successful model for soil dilation during piling, and representative of the behaviour of piles (Figure 7).
Fleming concluded by reminding the audience of the importance of effective piling, using the example of Mexico City. It is obvious on visiting the city which structures have been piled: they have steps going up to the entrance, non-piled buildings have ramps going down. A number of notable structures, such as the Cathedral and Palace of Fine Arts, have sunk by up to 7m. He reiterated that with careful load tests, the information obtained would be of high quality and a great deal would be learned about the pile, ground, and construction anomalies.
Fleming, WGK, 1992. A new method for single pile settlement prediction and analysis, Gotechnique, 42, No 3, pp411 - 425.
Terzaghi, K, 1943. Theoretical Soil Mechanics, Wiley, New York.