A Python-based automation for estimating undrained shear strength from field vane test

Loading...
Thumbnail Image

URL

Journal Title

Journal ISSN

Volume Title

School of Engineering | Master's thesis

Department

Major/Subject

Mcode

Language

en

Pages

0

Series

Abstract

In Finland, geotechnical designers often use Novapoint in Civil 3D to access ground investigation database. They retrieve field vane test (Siipileikkauskoe) results to manually calculate reduced undrained shear strength. This process is time-consuming, prone to error, and hard to manage when working with large datasets or multiple cross-sections. Also, the absence of visualisation of the reduced undrained shear strength on the soundings makes clay layer definition difficult. In addition, the effect of the stress state is often overlooked in designs. This thesis addresses these challenges by developing a tool named “Su from FVT”. The tool extracts data, handles missing values, and supports three different reduction methods. It also provides visualization of Su across stress states. The tool applies interpolation or empirical correlations derived from Finnish soil databases. The fineness number from the correlation model tends to be higher, which can overestimate strength if not carefully reviewed. The study adopts a design science research approach to design, develop and evaluate the tool with a case study to illustrate its performance in terms of time, precision, and reproducibility. On average, it takes 30 seconds for the tool to process a typical task, using about 60 MB of memory. The case study showed a 60% reduction in calculation time for the studied cross-section. It also emphasized the importance of observing soil strength across stress states to support designers make project-specific choices. Relying only on field vane test-based undrained shear strength for all types of designs could lead to under or over estimation of soil strength. Overall, the tool helps to quickly estimate and assess undrained shear strength from field vane test. However, its reliability depends on the quality of the correlation models and the completeness of the input data. Hence, designers must always review the outputs against local conditions.

Description

Supervisor

Solowski, Wojciech

Thesis advisor

Napari, Matias
Kudsk, Lasse

Other note

Citation