Boulard, Coline

Master thesis supervised by Van Noten, Koen; Zeckra, Martin (Université libre de Bruxelles)

Abstract: Geothermal energy is becoming a major stake following global warming and the need to decarbonize our society. In Belgium, one of the biggest problems in terms of decreasing our energy demand, remains the heating and cooling of buildings. Currently, a lot of studies are made to investigate if it is possible to use open or closed geothermal systems especially in city districts or for large buildings with the purpose to cool them in summer and heat them in winter. Prior to installing geothermal systems, expensive drillings need to be made to know the depth to bedrock of interest. The objective of this master thesis is to study non-destructive geophysical techniques to predict the sediment thickness overlying the bedrock. This master thesis is framed within the GeoCamb research project in which the Royal Observatory of Belgium participates, and which aims to investigate geothermal energy in Brussels and in Walloon and Flemish Brabant. It is aimed to compare the differences in sediment thickness using the Horizontal-over-Vertical spectral ratio (H/V) method, which consists in the analysis of the ambient seismic noise recorded on three components (one vertical and two horizontal) of a seismometer. The H/V spectral ratio is calculated by dividing the spectra of horizontal components by the vertical component which gives a spectrum composed of a peak corresponding to the fundamental frequency (f0) of the ground. This frequency can be converted to sediment thickness with a simple powerlaw equation that already exists for Brussels from prior work. By removing the obtained sediment thickness from the topography, a prediction of the altitude of the bedrock is determined. In this study, 4 buildings in Brussels (BE) that are of interest for geothermal energy are investigated: The European Paul Henri Spaak building, the Chirec Delta Hospital, the Cultural Center of Auderghem, and the Solaris building. The aim was to calculate the sediment thickness using the H/V method and to compare the predicted bedrock depths with those of nearby boreholes using 38 measurements at these four locations. The existing sediment thickness to bedrock conversion equation was constructed by ambient noise measurement above boreholes from when bedrock depth was known. It would be interesting to find out if this conversion law still fits with the investigated four locations. In this study it is also investigated if it is better to record ambient noise during the day or at night or if it does not matter. Also, in order to use the H/V ambient noise technique, it is common to clean up the data prior to computing the H/V spectrum. It would be interesting to know if it is really necessary to clean them or not. After comparing the different results obtained, this study shows a good correlation between the prediction of sediment thickness in the 4 locations and the boreholes in the vicinity, suggesting that the existing conversion law works. We can also add that there is a considerable difference between the bedrock depth calculated with the HVSR technique and the data visible on Brugeotool. The extrapolation used by Brugeotool to fill in the places where there were no borehole measurements and the 3D geological model of the top of the Brabant Massif should be updated to remove this difference. The study demonstrates also that there is no obvious difference between the variability of f0 at night and during the day. That would imply for future references, that the stability/uncertainties of measurements rather rely on the location than on the time of day, when the measures are taken.On the other hand, there is less variation of f0 over time when the measurements are taken during the day. This could be due to the fact that there is more noise, and that this energy would be required to have more stable data. One can also detect that the uncertainties on the sediment thickness prediction (taken from the error on f0) are each time larger than the difference in sediment thickness prediction between day and night. This study displays as well the value of cleaning the H/V spectrum before using it. Actually, even if sometimes the measurements are good, it is preferable to clean the data to avoid big aberrations and to have the most accurate data possible. This has strong implications for researchers that want to investigate the resonance frequency with automatic procedures, without considering how to clean the data.

Keyword(s): H/V spectral ratio analysis ; Bedrock depth ; Brussels' geology