Soil Interpretation in Groningen

Abstract

In the northern part of the Netherlands, the exploitation of gas fields has been inducing small earthquakes, causing damage to existing buildings.
With the aim of preventing consequences to people and structures, the VIIA Groningen project deals with CC2 and CC3 buildings retrofit and provides reinforcement measures when necessary.
As part of the structural response assessment, after the NPR 9998 (2015), and eventual special cases from the latest NPR 9998 (2017), non-linear time history analyses (NLTH) are executed, comprising seismic ground response analysis (SRA).
The propagation of seismic waves through a 1D soil column is highly dependent on the characteristics of the materials constituting the soil deposits. Hence, it is essential to correctly interpret the soil properties, in order to achieve realistic representations of the in-situ conditions.

To interpret the soil layering at a particular site, the Cone Penetration Test (CPT) is commonly used in Groningen. It offers a quick, economical and reliable measurement of ground conditions. However, the CPT-based correlations used to estimate soil properties can constitute a source of uncertainty if not coupled with full-scale testing and laboratory measurements.
The present thesis, thus, deals with the verification and the improvement of two CPT-based correlations used for soil interpretation in Gronigen specifically. The research study focuses on the mathematical models related to the plasticity index (PI) and the undrained shear strength (Su) of soft soils present in Groningen.

A comprehensive database of factual data was compiled in order to group various test types and provide a best-estimate of soil properties for different soil types using geotechnical and stratigraphic considerations. Secondly, a statistical characterisation of data-sets was performed to obtain insight on the correlations performance in relation to the in-situ and laboratory measurements. Based on the outcomes of the statistical comparison, analytical and regression analyses were carried out with the scope of improving the correlation that was deemed to be inadequate. Additionally, a sensitivity analysis was executed to investigate the influence of three relevant soil properties on the seismic ground response from a typical soil profile from Groningen. The parameters assessed are: plasticity index (PI), undrained shear strength (Su), and shear wave velocity (Vs).

Results indicate that, among the considered CPT-based correlations, the equation for PI from Cetin and Ozan (2009) is adequate in some cases. The geotechnical units sandy Clay and Loam show good correspondence with the factual data. On the other hand, the PI predicted with such relation tends to be lower than the laboratory measurements for the remaining soil units (e.g. clean Clay, silty Clay, OC Clay). Conversely, the PI behind the models implemented in the NPR 9998 (Bommer et al., 2017a) are in closer agreement to the factual data, however, the
PI from some soil units can be further improved with the findings from the present research.

For the interpretation of Su, the SHANSEP model from Ladd and Foott (1974) is frequently used. The available factual data showed a poor correspondence with the predicted Su values. Therefore, the SHANSEP model was further studied to calibrate its parameters for different soil types. From the available triaxial consolidated undrained laboratory tests, best-estimate of SHANSEP coefficients were obtained for the main soil types (clean, sandy, and silty Clay). New Su values were validated with the in-situ and laboratory measurements. In this context, it is confirmed that the dependency of Su on the overconsolidation ratio (OCR) is crucial. Moreover, the estimation of OCR from CPT measurements (following the Mayne, 2014, procedure) is found to be partially inaccurate within the SHANSEP framework and needs to be studied in more detail.

Engineering aspects related to the topics of the research are discussed and considerations regarding the applicability of the new correlations are provided. Furthermore, the present study gives indications about the usefulness of a number of test types, suggesting direction for future soil investigations. In addition, look-uptables for PI and Su, based on the outcomes of the present research, are provided as part of the recommendations for implementation in the soil parameter interpretation for the Groningen region.