“…For instance, the model deals with empirical phenomenological parameters that are hard to measure [23,24] and fails to predict particle migration in curvilinear torsional flows [25][26][27][28]. Another problem is related to its numerical implementation, since the model becomes singular as the local shear rate approaches zero, as occurs near the symmetry plane of pressure-driven flows through tubes and channels [29,30]. However, because of its relative simplicity, good accuracy and low computational cost, the diffusive flux model has been extensively used to study particle migration in many different problems involving complex flows, such as flows through two-and three-dimensional bifurcations [31,32], free-surface flows [33,34], and flows of suspensions with non-spherical particles [35,36].…”