Adding heavy particles to turbulent flow between rotating cylinders (Taylor-Couette) should increase drag. More mass in the flow means more momentum to transport, more collisions, more dissipation. The expectation is monotonic: more particles, more drag.
Note (2026-11-09): I checked the abstract of the Jiang, Lu, Ma and Chong paper. It varies particle Stokes number and Froude number (settling) at a fixed Reynolds number in a two-way coupled Eulerian-Lagrangian simulation, and reports an abrupt change as the influence of particle inertia and settling grows. The essay's control variable, a critical particle loading, is not what the abstract varies, and its statement that the coexistence regime is narrow is not in the abstract. The abstract also reports drag reduction non-monotonic in Froude number for light particles and significant drag enhancement for heavy particles at large inverse Froude number.
The response is not monotonic. At moderate particle inertia and gravitational settling, particles suppress the coherent Taylor vortex structures that organize the turbulent flow. The vortices — large-scale rolls that dominate angular momentum transport — are weakened by the particles, and drag decreases. The particles are acting as vortex dampers.
Beyond a threshold particle loading, the system switches discontinuously. The Taylor vortices do not gradually weaken further — they collapse entirely. In their place, particle-induced turbulence emerges: a different kind of disorder, driven by particle wakes and collisions rather than rotational shear instability. This new turbulence enhances drag through a completely different mechanism. The drag increases again, but for new reasons.
The transition is abrupt. The system occupies one of two states — vortex-dominated turbulence with reduced drag, or particle-dominated turbulence with enhanced drag — and jumps between them at a critical particle loading. The intermediate regime where both coexist is narrow.
The structural point: particles do not simply modify turbulence — they can replace it. The drag response is non-monotonic because the mechanism of turbulence itself changes. The same particles that suppress one form of disorder (Taylor vortices) create another (particle-driven turbulence). The system has two turbulent states, and the particles select which one operates.