Test interpretation

Load–settlement relationship

Until a few years ago the aim of a load test was essentially the determination of the bearing capa city, to be employed in a capa city based design. Recently, attention has been switched to the settlement prediction and, accordingly, the scope of pile load tests has broadened to include the determination of the whole load–settlement relationship.

The static ver tical load test is gen erally confused with the Ideal Load Test (ILT) (Figure 7.9a). As we have seen in §7.1.2 above, in practice the load is applied to the pile by a hydraulic jack; the reaction system can be a kentledge resting on supports (Figure 7.9b) or a beam an chored to the soil by tension piles or ground an chors (Figure 7.9c). Recently the so­ called Osterberg cell (Figure 7.9d), providing a “self­ reaction”, is becoming increasingly pop ular. The setups illus trated in Figures 7.9b– 7.9d differ from the ILT because they apply to the ground a load system with zero resultant. This is not without con sequences on the load–settlement relationship and on the ultimate bearing capa city, as compared to that of ILT.

In the case of a test with kentledge Poulos (2000) notes that the stress arising in the subsoil from the weight of the kentledge tends to cause an increase of the shaft friction and end bearing of the pile. As the load on the pile is increased by jacking against the kentledge, the stress will reduce and some upward displacements tend to de velop in the soil, while the pile undergoes settlement. The pile head stiffness is thus

158 Present practice: vertical loads

overestim ated, while the pile’s ultimate capa city may be rel at ively close to that of the ideal test.

Figure 7.10 reports the results of a parametric study on the load–settlement curves of a pile subjected to an Ideal Load Test and to a test with kentledge. The curves have been obtained by non­ linear finite element elasto­ plastic ana lyses assuming the soil to be a uniform sand with different values of the friction angle φ′ (Mohr­ Coulomb model) and test piles with d = 1 m and L / d = 10, 20 and 50. The test with kentledge overestim ates the initial stiffness of the pile, the more the higher the ratio L / d.

On the contrary, at rel at ively large displacements (w = 10% d) the discrepancies decrease and eventually the value of the ultimate capa city is prac tically unaffected by the influence of the kentledge. Similar trends have been found for undrained clays. The effect of inter action between reaction piles and the test pile is again an over­ estimation of the pile head stiffness. The overestimation may be very significant for slender piles and reaction piles close to the test pile (Poulos and Davis 1980; Poulos 2000; Kitiyodom et al. 2004). Some further results are reported in Figure 7.11, which refers to a pile with d = 1 m, L / d = 20, two reaction piles ident ical to the test pile and different values of the spacing s between the test pile and the reaction piles (s / d = 4, 6, 10). Figure 7.11 is based on the same soil model adopted for the ana lyses reported in Figure 7.10; sim ilar trends have been found for undrained clays.

Figure 7.9 Various load tests setup and ideal test.

Load Fixed point Level Level Test pile pile pile Anchor c) Reaction piles ground anchors pile pile pile

a) Ideal test b) Test with kentledge Spreader beam Jack Jack Test Test Tell tales d) Osterberg cell Osterberg load cell Test Reaction support Test Jack Kentledge

In the case of a test pile jacked against ground an chors, Poulos (2000) has shown that the overestimation of the pile head stiffness is significantly less than when reac­ tion piles are used, especially if the an chors are located well below the base of the test pile.

Summing up, the test setups with kentledge or an chor piles are totally suit able for the determination of the bearing capa city. If, on the contrary, the main purpose of the test is the determination of the load–settlement beha vi our, and especially the initial stiffness, substantial corrections are needed in all cases. Without these corrections, any ana lysis based on the load test on single pile can be misleading and unconservative.

160 Present practice: vertical loads

Bearing capacity

As reported in §4.2, it is gen erally accepted that the ver tical bearing capa city of a pile is conventionally defined as the load cor res ponding to a settlement equal to 0.1d. If the load test has attained such a settlement, the bearing capa city is deter­ mined directly in an unambiguous way; for prac tical reasons, how ever, such a large

Figure 7.11 Load test with tension piles vs. Ideal Load Test. Q [kN] 0 0 20 ILT L/d = 20 s/d = 4 ILT L/d = 20 s/d = 6 ILT L/d = 20 s/d = 4 φ = 23° φ = 27° φ = 31° φ = 35° φ = 23° φ = 27° φ = 31° φ = 35° φ = 23° φ = 27° φ = 31° φ = 35° 40 60 80 100 w [mm] 0 20 40 60 80 100 w [mm] 0 20 40 60 80 100 w [mm] 2000 4000 6000 8000 Q [kN] 0 2000 4000 6000 8000 Q [kN] 0 2000 4000 6000 8000

settlement is not always attained. In this case, the simplest way of determining the bearing capa city, as defined above, is that of extrapolating the load–settlement curve to a settlement w = 0.1d.

Chin (1970) noted that in most cases the load–settlement curve is well approxi­ mated by a hyperbola:

(7.1)

where w is the settlement under the load Q and m, n are two constant to be deter­ mined by fitting the curve to the experimental results. Eq. (7.1) may be written:

(7.2)

suggesting plotting the experimental result in a graph (w / Q, w) and to interpolate a straight line among them (Figure 7.12). The intercept on the ver tical axis is equal to m, while the inclination over horizontal is equal to m. Once m and n have been determined, Eq. 7.1 may be used to calculate the bearing capacity:

If the test has been kept to a sufficiently high max imum test load, the extrapolation is very reliable.

The same technique may be employed to extrapolate a proof load test to estim ate the bearing capa city. In this case, the max imum load rarely exceeds 50% of the bearing capa city and the results from the extrapolation are much less reli able, but gen erally conservative.

Figure 7.12 The procedure suggested by Chin (1970) to interpolate a hyperbola through the experimental results of a load test.

162 Present practice: vertical loads Transfer curves

Transfer curves of the side shear (τ–z) and base resistance (p–z) can be obtained by in ter preting the results of load tests on piles instrumented to meas ure the displace­ ments or the strain along the shaft. The pro ced ure is illus trated in Figure 7.13.