Sulphide soil is a fine grained organic soil with poor geotechnical properties. It is a predominant soil type along the Baltic coast of Sweden and Finland, among other parts of the world, under anaerobic waterlogged conditions. When sulphide soil gets in contact with air it oxidises, becoming an environmental hazard due to potential leaching of metals and acid drain. Geotechnical structures built on sulphide soil often face significant economical and environmental costs since they heavily rely on expensive soil reinforcement or stabilisation to reach the required bearing capacity. On the other hand, replacing the sulphide soil with a more suitable material involves significant transportation costs and emissions, as the excavated sulphide soil require careful disposal at dedicated landfills.

Common structures found on sulphide soil are roads and railways, which are not subjected to high loads, but they undergo a significant amount of cycles with relatively small amplitudes. Optimising roads and railways in terms of economical cost and environmental impact includes the account of the effects from acting cyclic loads. Up to now, the static mechanical behaviour of sulphide soil has been studied., but not its cyclic mechanical behaviour. Parameters for the calibration of suitable constitutive models are needed to define the cyclic mechanical properties of sulphide soil and make correct predictions of deformations and pore water pressure development. For this purpose, advanced laboratory tests on high-quality representative samples are essential often acquired through undisturbed sampling. However, undisturbed sampling of sulphide soil can be challenging, as there is a constant risk of sample disturbances.

Laboratory results on undisturbed sulphide soil samples are known to suffer from an unavoidable scatter in the results when compared to similar studies on other soils types. To mitigate the scatter, this work proposes to base a portion of the soil characterisation on reconstituted samples. The slurry deposition method has been modified to produce homogeneous and repeatable samples of sulphide soil which complement the undisturbed samples and help to produce more concise laboratory results needed for the investigation of the cyclic behaviour. The applicability of the slurry deposition method was assessed by comparing the initial index properties and stress strain-behaviour of the reconstituted samples against that of the undisturbed samples. The index properties of the tested reconstituted samples precisely and accurately matched the average results of the intact samples. Results from isotropic compression test performed in the triaxial device show matching normal compression lines and unload-reload lines for both types of samples. Drained triaxial tests performed on neither type of samples exhibited a distinct failure point under the failure criteria of critical state soil mechanics. The overall undrained triaxial behaviour of the reconstituted and undisturbed samples were found to be comparable with good agreement between the derived critical state lines, but the former presented a slightly lower effective friction angle. In conclusion, the results from this study suggest that the adopted slurry deposition method is suitable for reconstituting sulphide soil samples, producing repeatable samples with behaviour comparable to in-situ conditions.

The scope of work in this thesis is limited to monotonic loading tests of sulphide soil. The continued research is planned to include cyclic loading tests. To prepare for the cyclic laboratory data, the obtained monotonic results of the reconstituted and undisturbed samples were used to calibrate parameters for the constitutive model Clay-Hypoplasticity. Once cyclic laboratory data is available, the performed calibration will allow for predictions of the strain accumulation and pore pressure development from cyclic loads on roads and railways, using the High Cycle Accumulation model. A simplification of this model following a laminar structure has been included in this thesis, which serves as a proof of concept, until cyclic laboratory data is available.