Mineralogical assessment of rocks of hydropower tunnels subjected to swelling

Hydropower tunnels located in weak and swelling rock masses are prone to stability problems and failures. Co-operating partner Statkraft is supporting further research for the assessment of swelling minerals through mineralogical investigation techniques. This master thesis has aimed to evaluate traditional investigation techniques as well as potential alternative methods for assessing swelling minerals in the laboratory. In addition, it was aimed to evaluate the modified preparation method and whether this can be an established method for soft rocks. The tested rock samples were collected from the Moglicë hydropower plant in Albania. A swelling potential was found in all tested samples through the oedometer swelling pressure test. Low swelling was classified in most samples, and medium swelling was found in the fault gouge rock. The swelling was likely caused by the presence of smectite clay and zeolite minerals in the rock specimens. It is interpreted that the zeolite minerals caused the increased swelling in the fault gouge material. The study implied that mineralogical assessment should depend on various laboratory methods, rather than the reliance on one method. Mineral identification errors were observed in all the tested laboratory methods to a varying degree. The X-Ray Diffraction(XRD) method showed the best results related to the detection of swelling minerals with a good correlation to the measured swelling pressure [MPa]. The findings of the two alternative methods, Automated Mineralogy System (AMS) and Hyperspectral Imaging (HSI), were ambiguous. In contrast to XRD, both methods provide beneficial visualization of the sample mineralogy. Limitations were discovered in the mineralogical composition and quantification, thus making the methods unsuitable as sole assessment methods. The study uncovered the potential of the modified crushing method and recommends this to be an established preparation method for soft and weak rocks at the Department of Geoscience and Petroleum at NTNU. Compared with the standard samples, it appears that the modified samples preserved soft minerals better, and also reduced the disintegration of rock material. A large loss of material was discovered in both crushing methods. This was assumed to be due to the loss of hard silicate minerals and soft clay material. For future projects, alterations should be made to the modified method to make the crushing procedure less time-consuming, more automated, and reduce the loss of material. The rock mass quality of the samples is mainly weak and disintegrated, making it difficult to perform adequate rock mechanical tests. Three samples of varying lithological character were tested by the Uniaxial Compressive Strength test (UCS), the results were strongly influenced by weak minerals and fractures, and thus not reliable. The mean tensile strength of the rock mass was 2.35 MPa, obtained from Brazil tensile strength test. The number of samples is below the ISRM recommended amount for both tests. Continued investigation is necessary to fully assess the potential of the different mineralogic and rock mechanical laboratory methods. Additional testing using a larger number of samples is recommended as this may reduce uncertainty. The modified preparation method should be adjusted and applied to soft rocks in future projects. Both AMS and HSI have shown potential with mineralogical classification along with XRD. It would be beneficial to improve post-processing and machine learning approaches for the laboratory data as mineral quantification has revealed limitations.