Computational Optimization Using Spherical Geometric Modeling for Wave Propagation on Time Domain
Seismic Modeling; Computational Costs Reduction; Energy Sustainability; Data Processing
The computational costs reduction enables large data volumes projects. In seismic simulations, processing data rapidly allows for adjustment of predictions and refinement of geophysical analyses in a shorter time. Therefore, it is possible to contribute to energy sustainability, as energy consumption and financial costs related to more complex projects are reduced. In this context, this work proposes the creation of methods applied to wave propagation in the time domain capable of reducing computational costs during simulations. Initially, and considering modeling simulations, spherical geometric modeling is proposed to delimit the regions of a computational grid within which an acoustic wavefield is propagated. This method calculates a spherical region that delimits the Laplacian calculation for solving the acoustic wave equation in each time interval. This prevents calculations from being carried out in regions that are not of interest, reducing the computational cost. A three-dimensional model capable of constructing spherical coordinates was simulated during a predetermined period, which varies from the source trigger to the acquisition by the receiver. The tests were performed by varying the source position within the grid (with the receiver position fixed), using two different types of discretization, as well as three different numbers of points for the grid. Our initial results achieved a gain of 34% on data runtimes.