Foundations of buildings ensure the safety and serviceability are upheld. A common type of
foundations used in the Netherlands is pile foundations which are typically founded on a thick sand layer in the subsurface. These pile foundations are the focus of this thesis and more specifically the methods used to design such foundations. This thesis aims to give a better understanding of the resistance of closed-ended piles in a layered soil by comparing existing pile design methods to the measured forces from the Drukpaal jacking machine. Drukpaal jacks precast concrete piles into the soil and the records collected give a valuable insight in the pile resistance in various locations and soil profiles with layered soil in the Netherlands.

The design of a pile foundations consists of the combination of two different types of methods. One will predict the base capacity of the pile foundations and the other will predict the shaft capacity. These two capacities together give the predicted bearing capacity of the pile. However, multiple base capacity and shaft capacity methods exist with varying influence zones and with that varying degrees of accuracy. In this thesis the different base and shaft capacity methods used are all based on the tip resistance as measured by a CPT. The following base capacity methods are included: Koppejan, LCPC, Filter method and the Filter method as adapted by Munta de Boorder (in this thesis referred to as the Munta de Boorder method). And for shaft capacity the NEN method and the ISO method are used.

This thesis aims to answer the following research question: What combination of capacity prediction methods for shaft and tip resistance works best to estimate the total force over depth as measured by Drukpaal?.
This question will be answered by a series of simulations of different projects completed by Drukpaal, which is a company which jacks piles into the ground and records the required force over depth.

The measurements and simulations for the various sites were compared both visually and
statistically. The Filter method with ISO gives the best visual fit.
For the statistical analysis the coefficient of variance (COV) is taken as well as the root mean squared error (RMSE) of the ratio line of the simulation (simulation divided by the measurement). Here the COV will account for the shape and the RMSE will account for the overall error. The sum of these two values is used to determine the best fit, since both of these values should be as small as possible. From this analysis it is found that Filter method in combination with ISO is overall the best combination, followed by Munta de Boorder with ISO and Filter method together with NEN.

The last aspect to be looked at is the location dependency of the combination of methods. From this it was found that the base capacity methods in combination with NEN vary a lot over the measured locations and these combinations are thus location dependent. For the base capacity, the LCPC method was also found to be location dependent in combination with ISO. So even though this combination worked really well for one of the projects (Gorinchem) it is not dependable to use because of the varying degrees of accuracy. The other three combinations; Koppejan, Filter method and Munta de Boorder together with ISO, were all found to be location independent. However, Koppejan with ISO garnered quite high values for RMSE and COV, meaning that this combination is not very accurate.

Overall, Filter method with ISO was found to give the best fit which was location independent. Munta de Boorder with ISO followed and Koppejan with ISO was found third best fit. The other combinations were found to be location dependent and thus are, based on the work in this thesis, less suitable for application in The Netherlands.

This research looked into the influence of negative skin friction on timber piles in Amsterdam. Tensile load tests were performed on new, instrumented timber piles at Overamstel in Amsterdam. The negative skin friction calculated from the test data was compared to the negative skin friction from NEN 9997-1 for a single pile. The test results came close to the values from NEN 9997-1. A reduction of the factor required for timber piles when calculating negative skin friction using NEN 9997-1 seems necessary, but further studies are needed.

Alpha factors were derived from the test data, and comparison to the alpha factors from compression load tests confirmed the influence of tapering on the timber piles.

The research also confirmed the suitability of fibre optic sensors for pile load tests on timber piles.

In the city of Amsterdam, many structures, such as houses,bridges and quay walls, are founded on wooden piles. In order to gain insightinto the safety of such structures, assessment of the foundations is required.As part of an experimental framework assessing the safety of bridges and quaywalls in the city of Amsterdam, a number of piles are instrumented with fiberoptic sensors and load tested in compression. These tests aim to providedetailed information on the behavior of timber piles subjected to loading. Thisinformation can be used to determine the geotechnical bearing capacity of suchpiles. An in-depth analysis is conducted on 8 timber piles tested within thisframework.  The conducted analysis has resulted in a variation ofoutcomes. Without including the effects of residual loads, an average baseresistance of 130 kN is observed. Upon inclusion of residual loads the averagetrue base resistance increases to a value of 188 kN. The average shaftcapacities in the bearing sand layer for scenarios excluding and includingresidual effects are 48 and 60 kN respectively.  The correlation factor α_p is derived usingthree cone resistance averaging techniques. The Koppejan method hasconsistently resulted in the highest derived α_p factors withvalues of 1.09 and 1.61 for scenarios excluding and including residual loadsrespectively.  The scenario excluding residual effects has resulted in anaverage α_sof 0.009 for the bearing sand layer. Upon inclusion of residual loads, thederived α_sfactor for the first sand layer amounts to 0.012. This increase in shear forcesis also observed in the Pleistocene peat layer. Therefore, residual loadsredistribute the capacity of the piles by increasing the shaft resistance inthe bearing sand and Pleistocene peat, while simultaneously decreasing theshear stresses in the Holocene layers above.  Conclusively, timber pile characteristics such as variationin geometry and mechanical properties have significant effects on the capacityof the piles. Additionally, these variations result in fluctuations in thecalculated load distribution along the pile. Local smoothing of thesefluctuations results in higher apparent loads, most specifically at the pilebase. Therefore, smoothing algorithms are not implemented in this analysis. Thevariation in diameter along the entire length of each pile directly affects theload distribution. Despite this influence, no trend is observed for thevariation of α_p factors with respect to pile tip diameter. Anapparent relationship between the tapering of the pile in the bearing sandlayer and the derived α_s factors suggests that tapering effectivelyincreases the shear forces along the shaft in that layer.  Furthermore, the usage of fiber optic sensors on woodenpiles has proven to be effective. The variation in local behavior of wood isclearly illustrated through the conducted analysis. As a consequence of thebiological nature of wood, the local behavior of wooden piles is best capturedby sensing technologies measuring strains at high spatial frequencies.  

Over the last couple years, a number investigations into the α factors used for the cone penetration test (CPT) based calculation methods for the base and shaft capacity of driven piles have been carried out. These investigation, express different concerns and limitation of the currently used CPT based calculation methods. Prompting the need for further investigation into the consistency of these methods.For the base capacity calculation methods, the αp factor used is accompanied by a cone resistance (qc ) averaging technique. However, some limitation have been found for the currently used qc averaging technique, which affects the consistency and accuracy of the base capacity calculations when a constant αp factor is used, namely, a dependency with penetration depth. Presentiments by Randolph (2003) & White and Bolton (2005), suggest that a single, constant αp factor can be used when an appropriate averaging technique is applied in combination with the inclusion of the residual loads present in a pile. Hence, a new CPT averaging technique was developed during this study. The goal of this new averaging technique is to find an accurate and consistent averaging technique, which can be used in combination of a single, constant αp factor. The new CPT averaging technique was calibrated using a series of high quality CPT laboratory tests with varyingsoil deposits. Comparisons were then made between the new and existing averaging techniques, as well as, base capacity calculation methods by applying the different averaging techniques to the Deltares pile load test database. Lastly, the effect of residual loads were investigated by applying the averaging techniques to 4 well documented pile load tests where distinctions between residual loads were made. The investigation carried out in this report concluded that the new CPT averaging technique, developed in this study, was more accurate than the other investigated CPT averaging techniques, when comparisons were made with the CPT laboratory tests. This was also the case when comparing the CPT averaging techniques applied to the pile load database. Although the pile load database demonstrated that the new averaging technique had the least spread in results, a dependency with embedment length in the sand bearing layer was still present. This effect was not removed when pile load tests including residual loads were considered, as predicted by Randolph (2003) & White and Bolton (2005). However, the limited number of these tests calls for further research in order to confirm any conclusions.For the shaft capacity calculation methods, the αs factor used by the Dutch norm in particular, NEN-9997- 1, was investigated. This is because the formulation of the NEN-9997-1 calculation method is believed to be too simplistic (van Tol, Stoevelaar, Bezuijen, Jansen, & Hannink, 2013). Comparisons between a variety of shaft capacity calculation methods were made by applying the calculation methods to the Deltares pile load test database and to the 4 well documented pile load tests where distinctions between residual loads were made, in order to look into the effect of residual load on the accuracy of shaft capacity calculation methods.Additionally, results from the research carried out in this report into the αsfactor used by the Dutch norm, concluded a presence of a strong correlation between the αs of sand layers and friction fatigue terms used in other shaft capacity calculation methods. This can be used in the future to improve the current shaft capacity calculation method used by the Dutch norm.