Review and validation of settlement prediction methods for organic soft soils, on the basis of three case studies from the Netherlands

Abstract

Organic soft soils pose technical challenges in construction industry due to their extraordinary compressibility and deformability, which is an issue of particular relevance in the densely populated Netherlands, of which more than half of the land surface is covered by such soft soils. Uncertainty and variability in settlement predictions for constructions on soft soils are large, however: in the order of tens of percent. Therefore this research was performed, reviewing and validating one-dimensional settlement prediction methods for organic soft soils, in particular peat, with focus on long-term settlements. Special interest for validation concerned relatively unknown and uncommon, but seemingly simple and easy to use, empirical settlement prediction methods or compressibility parameter correlations. The main goal was to designate the best performing 1D settlement prediction method with an optimum balance between accuracy, usability and time investment, taking also into account the soil parameter determination including preceding sampling and testing procedures. This goal was met in part based on thorough literature review of seven different settlement prediction methods or models: Terzaghi, Buisman, Koppejan, Bjerrum, Fokkens, De Glopper, Den Haan (a,b,c- Isotachs), and in part based on subsequent validation of the latter four of those methods. Validations were performed on the basis of soil test data and field measurements, collected in the course of three actual field research or construction projects in the Netherlands. The literature review additionally comprised a critical review and comparison of different soil compression test methods: incremental loading (IL) oedometer, constant-rate-of-strain (CRS) and K0-CRS tests. Furthermore, previous studies regarding the reliability of settlement predictions, addressing many error sources including sample disturbance, were reviewed and summarised. In one case, soil compressibility characteristics had to be determined from raw IL and CRS compression test data, for which various methods of soil parameter determination were studied and performed. This includes in particular the determination of the preconsolidation pressure, for which four different methods were applied and mutually compared: Casagrande’s method, Butterfield’s method, the work-per-unit-volume method [Becker et al., 1987] and the pore pressure method. The main conclusions from the settlement prediction methods validations are that the soil compressibility parameter correlations and the method of Fokkens are best to be disregarded for use in engineering practice because of their unreliability or their considerable practical limitations. Of the remaining settlement prediction methods, the a,b,c-Isotachs model theoretically provides the most sound and versatile description of soil behaviour upon compression as well as unloading. However based on the data available and the results obtained, the Bjerrum model yielded the most accurate results in practice, with final settlement estimations for a more than 350 years old dike within 12% of the actual settlement.