Geotechnical parameters estimation in iron ore tailings piles via bayesian models
Geotechnical parameters estimation in iron ore tailings piles via bayesian models
As a result of the several events involving dams around the world, the construction of upstream dams was banned, new regulations have been established, and revisions to waste disposal norms have been made. New technologies were developed for the tailings treatment, allowing alternatives for their disposal, such as the construction of different stacking methodologies. Several measures are taken in order to assure the structures safety, such as carrying out geotechnical tests throughout the construction and technological control, which ensure that the project technical specifications are complied. Some tests must be carried out in the laboratory and require a long execution time, reducing the information number of materials physical and geomechanical parameters. In this way, important geotechnical parameters for the complete understanding of the behaviour of the tailings are restricted to a few points of the structures, and with deterministic results. Despite to work with a distribution of samples that represent the structure in a significant way, as they may be subjected to different levels of stresses, humidity, granulometry, among other conditions, using values from specific locations to assure the safety of the whole construction can lead to erroneous estimates in unsampled locations. In this article, data from piles of different locations in MG/Brazil extracted of Geolabor, an ecosystem of applications for managing geotechnical data, were used to estimate different geotechnical parameters from the technological control at non-tested points. Multiple spatial and non-spatial regressions were implemented, and the results obtained were satisfactory, showing the importance of the methodology developed and implemented with Geolabor.