Multiscale modeling and intelligent optimization of transient cuttings transportation during in RMR system

Efficient cuttings transportation is critical for the efficacy of riserless mud recovery (RMR) systems, yet transient hydrodynamics during pump switching remain poorly understood, with precision simulations often limited by idealized boundary conditions. To address this, a novel multiscale framework was established by coupling a System of Fluid Dynamics (SFD) model with a CFD-DEM modeling, where the SFD model imposes realistic, time-varying boundary conditions for a precise resolution of particle-fluid interactions. The integrated model systematically investigated the effects of key operational parameters, including valve switching time, pipe inclination, particle size, and inlet velocity, on cuttings bed thickness. To overcome the computational expense of this precision framework for real-time guidance, simulation data were used to develop a Particle Swarm Optimization-Backpropagation(PSO-BP) hybrid intelligent model that accurately predicts system performance and facilitates the rapid identification of optimal operational windows. Thus, the proposed framework provides a precision method for analyzing transient hydrodynamics, whereas the intelligent model offers direct quantitative guidance for optimizing field operations.