digilib@itb.ac.id +62 812 2508 8800

In this study, VPM-LSMPS solver is enhanced with adaptive resolution to improve its ability in simulating multi-scale flow phenomena. The numerical characteristics of VPM-LSMPS are first analyzed through parameter variations in elliptical vortex and Lamb-Oseen vortex simulations in 2D. The convergence rate is primarily determined by the particle size with order follows the LSMPS polynomial order, while the time step is selected to satisfy Lagrangian stability (CLCFL < 1.0) and meet the required diffusion stability range (0.05 < ? < 0.5). The adaptive resolution framework is implemented using a multi-block tree structure, with refinement controlled by feature-based and Laplacian-based criteria. The multiresolution approach achieves a 50% reduction in computational time while maintaining accuracy. To validate the model, simulations are conducted for benchmark problems involving flow over a bluff body in both 2D and 3D. The results show good agreement with expected trends; however, further accuracy improvements are needed, particularly by employing finer resolution settings. Future research should explore the solver’s performance in a wider range of complex flow scenarios to further assess its capabilities.