Design and optimize a novel hydrocyclone by combining mechanistic and data-driven models

New hydrocyclone designs can significantly enhance separation efficiency in applications such as water treatment and particle classification. Therefore, various hydrocyclone geometries with different inlet and cone configurations are explored through a validated mechanistic model, leading to a new cyclone design. The proposed design features a laminar spiral inlet and a straight-to-convex cone, achieving reductions of 44.2 % in separation sharpness and 58.4 % in water split. Its double-cone configuration reduces tangential velocities in the upper conical section, while the convex lower cone broadens the separation region, maintaining relatively high tangential velocities near the spigot. These effects reduce particle accumulation near the spigot and improve separation performance. Moreover, the laminar spiral inlet mitigates short-circuit flow near the vortex finder. To further enhance separation efficiency, the novel conical section is optimized by integrating mechanistic and data-driven models. Compared to the initial novel design, the optimized version exhibits an 18.2 % reduction in separation sharpness and a 16.2 % reduction in water split by optimizing four geometric variables characterizing the conical section. Internal flow field analysis confirms that the optimized configuration establishes favorable tangential velocities in the conical section, guiding fine and coarse particles along optimal paths and improving overall performance. • A novel design for the hydrocyclone is developed using the mechanistic model. • Mechanistic and data-driven models are combined to optimize the novel design. • The separation efficiency of the optimized novel design is improved significantly. • The laminar spiral inlet decreases the short-circuit flow near the vortex finder. • The longer and convex cone promotes the proper directional migration of particles.