1: \begin{abstract}

2: This study presents an innovative direct numerical simulation approach for complex particle systems with irregular shapes and large numbers.

3: Using partially saturated methods, it accurately models arbitrary shapes, albeit at considerable computational cost when integrating a compatible contact model.

4: The introduction of a novel parallelization strategy significantly improves the performance of the contact model, enabling efficient four-way coupled simulations.

5: Through hindered settling studies, the criticality of the explicit contact model for maintaining simulation accuracy is highlighted, especially at high particle volume fractions and low Archimedes numbers.

6: The feasibility of simulating thousands of arbitrarily shaped convex particles is demonstrated with up to $1934$ surface-resolved particles.

7: The study also confirms the grid independence and linear convergence of the method.

8: It shows for the first time that cube swarms settle $13$ to $26$\% slower than swarms of volume-equivalent spheres across different Archimedes numbers ($500$ to $2000$) and particle volume fractions ($10$ to $30$\%).

9: These findings emphasize the shape dependence of particle systems and suggest avenues for exploring their nuanced dynamics.

10: \end{abstract}

11: