A novel particle method, Lagrangian gradient smoothing method (L-GSM), has been proposed in our earlier work to avoid the tensile instability problem inherently existed in SPH, through replacing the SPH gradient operator with a robust GSM gradient operator. However, the nominal area of each L-GSM particle determined by the relative location of particles is always inconsistent with the real representative area of it in simulation, especially in large-deformation problems. This is why the earlier L-GSM model has to be limited to the solid-like flow simulations where the deformation is not very serious. In this work, a conservative and consistent Lagrangian gradient smoothing method (CCL-GSM) is developed for handling large-deformation problems in hydrodynamics with an arbitrarily changing free surface profile. This is achieved by deriving a conservative and consistent form for the discretized Navier–Stokes governing equations in L-GSM, which even holds in the neighbor-updating or ‘remeshing’ process. Special techniques are also devised for free surface treatment, which is important to restore the conservation and consistency manner of CCL-GSM simulation on free surface boundary. The effectiveness of the proposed CCL-GSM framework is evaluated with a number of benchmarking examples, including dam break, wall impacts of breaking dam, water discharge and water splash. It shows that the CCL-GSM model can handle the incompressible flows with complicated free surfaces effectively and easily. The results comparison with experiments and SPH solutions demonstrates that the CCL-GSM can give a desirable result for all these examples.
CITATION STYLE
Mao, Z., Liu, G. R., Dong, X., & Lin, T. (2019). A conservative and consistent Lagrangian gradient smoothing method for simulating free surface flows in hydrodynamics. Computational Particle Mechanics, 6(4), 781–801. https://doi.org/10.1007/s40571-019-00262-z
Mendeley helps you to discover research relevant for your work.