Randlab is a research consortium devoted to the development of innovative geological reservoir characterization and modelling methods. The consortium is hosted by the University of Neuchâtel in Switzerland.
One of the main focus of our work is Multiple-Point statistics (MPS). In many engineering applications, modelling time series, 2D, or 3D fields is required to evaluate uncertainty or to model physical processes (e.g. fluid flow including groundwater, oil, gaz, CO2). In this framework, MPS allows simulating complex patterns in a straightforward manner. The basic principle is to provide a training data set from which non parametric statistics can be inferred and used to control the generation of random fields which mimic the structures present in the training set.