Benchmark comparison study for mantle thermal convection using the CitcomCU numerical code

The mathematical approach for mantle studies relies on both analytical and numerical methods, but the absence of known analytical solutions for most of the scenarios makes a wide variety of numerical codes that cannot be easily tested. For theses codes to be tested, there is a benchmark, which will provide comparison data from different researchers. This data is presented as a single result with deviation calculated with the similar data provided by the researchers. These benchmarks set standard models. This benchmark comparison study contains numerical simulations calculated using the 3D finite element parallel code CitcomCU for mantle thermal and thermochemical convection reproducing some of the benchmark cases proposed by Blankenbach et al. (1989). The mantle is treated as an inelastic, uncompressible and viscous fluid using the Boussinesq approximation with an infinite Prandtl number. This benchmark comparison allowed to verify the reliability of the numerical code for qualitative studies and observed limitations of the code were presented for a better quantitative result. The numerical solutions indicate that the linear interpolation of the thermal structure between the nodes results in larger deviations mainly in zones with relatively high gradient. The proposed method to this would be to increase the grid resolution in those zones seeking for a balance between grid resolution and consequent simulation run time.