Parallel GPU-accelerated adaptive mesh refinement on two-dimensional phase-field lattice Boltzmann simulation of dendrite growth

Natural Convection on Dendrite Morphology: A High–performance Phase–field Lattice Boltzmann Study

“…37,38) To enable dendrite growth PF simulations in a much larger domain, we developed parallel computing in multiple GPUs with dynamic load balancing for the block-structured AMR scheme (parallel-GPU AMR) for purely diffusive growth 39) and growth with liquid flow and solid motion. 40,41) In this study, we performed high-performance PF-LB simulations of dendrite growth with natural convection by applying parallel-GPU AMR 41) to investigate the effect of natural convection on dendrite growth and fragmentation in 3D.…”

Geometry-structure models for liquid crystal interfaces, drops and membranes: wrinkling, shape selection and dissipative shape evolution

“…An adaptive mesh refinement (AMR) method optimizes the numerical performance without wasting computational resources. [231][232][233][234] AMR is a powerful computational framework that is commonly applied in tracking front propagation, 235,236 dendrite growth [237][238][239] and interfacial flows. [240][241][242] Another advantageous numerical approach to record the interfacial movement is the level-set method.…”

High-performance GPU computing of phase-field lattice Boltzmann simulations for dendrite growth with natural convection

“…Adaptive mesh refinement (AMR) is a powerful numerical scheme for reducing the computational cost of PF simulations [49][50][51][52][53]. Recently, we implemented AMR in parallel computing with multiple GPUs (parallel-GPU AMR) for the PF and PF-LB methods [54][55][56]. Here, block-structured AMR, dynamic load balancing, and multiple-grid methods were introduced to further reduce the computational cost of PF-LB simulations.…”