Since its introduction, Eurocode 7 has acknowledged the observational method as a valid design approach for geotechnical structures. The observational method is defined in five paragraphs. Although several case studies have concluded that the observational method is useful, the method is still sparsely used. This study compares the actual implementation of the observational method in the Veda embankment project according to the definitions given in Eurocode 7. The outcome of the observational method as a design approach, and the major deviations compared to the definitions in Eurocode 7 are discussed. It is concluded that the design of a multi-staged embankment, where the building process is governed by consolidation in the subsoil, can only partly be adapted to the observational method as defined in Eurocode 7. Furthermore, when this is the case, the possibility of combining "design by calculations" and the observational method is discussed.

One of the essential inputs in settlement prediction models is the soil modulus, which may be obtained from laboratory tests or estimated from in situ measurements. The total uncertainty in predicting the confined modulus of a sandy soil is quantified with data from side-by-side in situ testing using the standard penetration test, the static cone penetration test, the light dynamic probing and the laboratory oedometer test. To estimate transformation errors, correlations are proposed between in situ and laboratory data. The results indicate that similar magnitudes of total uncertainties are associated with the in situ methods, which are approximately twice as high as those from the direct oedometer method. The quantified uncertainties are an important input for reliability-based designs of foundations under similar soil conditions.

In light of a restricted budget, there is a need to stress on the potential savings in conducting qualitative geotechnical investigations. This paper presents a case study on how site investigation efforts can be linked to potential savings in designing a sheet pile wall in central Sweden. The uncertainty in the undrained shear strength is a measure of investigation effort and have been evaluated from multivariate information, several investigation methods. A multivariate analysis (MVA) procedure based on Bayesian statistics was used to cross-validate information obtained by different investigation methods, thus allowing the uncertainty or effort to be updated (reduced) when additional investigations are included in the analysis. The uncertainty was evaluated for two sets of investigations, one of which included additional measurements and hence less uncertainty. A reliability-based design method, FOSM, was then used to study how the additional investigations affected the evaluated uncertainty and one design constraint of the sheet pile wall, namely the depth of penetration. The results show that the depth of penetration can potentially be reduced by approximately 11% of the total wall area.