Contrôle de l'incertitude par une approche multidisciplinaire lors de la caractérisation environnementale des sites urbains contaminés

Abstract

The remediation of brownfields is a strategic challenge for sustainable urbanization at the global scale. However, several obstacles affect brownfield remediation. These barriers stem from the same source: uncertainty. Control of uncertainty in environmental characterization is a key factor to overcome these obstacles. Nevertheless, uncertainty control has been insufficiently studied and is rarely mastered or even nonexistent in practice. This doctoral research focuses on uncertainty control, through its quantification and minimization. Thus, the primary research goal was to develop and formulate an approach that could provide control on the overall uncertainty associated with brownfield characterization studies. Theresearch strategy was based on improving the characterization of site heterogeneity across three interrelated concepts: 1) the internal geometry of the site, 2) the representativeness of sampling and 3) the spatial distribution of the contamination. A real brownfield site, having a very large physical and chemical heterogeneity due to the presence of a heterogeneous backfill (urban fill), was used to carry out the research and to validate the developed approach. Internal site geometry was characterized using a multimethod geophysical survey, including magnetometry, electromagnetic induction, electricalresistivity imaging, and ground-penetrating radar. Sampling representativeness was studied by comparing grab sampling to an alternative sampling procedure developed herein based on the principles from Gy’s Sampling Theory. The distribution of contaminants was characterized according to a geostatistical procedure based on conditional simulations that optimizes the location of additional sampling stations. The multi-method geophysical survey has improved the mapping of the internal physical structure of the fill, both vertically and horizontally, while allowing a delineation of zones of distinctive heterogeneity. This improved mapping of heterogeneity across the site allowed an optimized sampling strategy that took into account each of the distinctive heterogeneity zones. The developed alternative sampling procedure has quantified the reproducibility of sampling at each of its sub-steps (from field to chemical analysis in laboratory). The quantification of this reproducibility at each sampling sub-steps ensured not only a clearer understanding of the heterogeneity at the sample scale but also the optimization of any sampling procedures to achieve a desired representativeness. Moreover, it was shown that the alternative sampling procedure was more reproducible by a mean factor of 10 compared to conventional grab sampling. The geostatistical procedure determined the spatial structure of contamination in the urban fill and quantified both the local uncertainty and the overall uncertainty specific to a given sampling plan. Moreover, the developed procedure was used to optimize the location and number of sampling stations during additional stages of characterization following a rational algorithm. It was also used to categorize the site into separate soil volumes (volume to be treated, uncertain volume and low risk volume) according to the probability of exceeding a regulatory threshold. This categorization of the site is based on two decision-making bounds which are adjusted according to the heterogeneity of the site. In short, the developed approach in this doctoral research provides a formal and rational framework for characterizing the heterogeneity of a brownfield, both at sample scale and field scale. This characterization of heterogeneity leads to the control of the overall uncertainty during environmental assessment, namely by its quantification and its reduction, and the achievement of an optimized sampling plan.

Date	4 May 2016
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Jean-Sébastien Dubé (Supervisor)