On the choking mechanism in supersonic ejectors: a one-dimensional analysis of Reynolds-averaged Navier–Stokes simulations

Ejectors are passive devices used in refrigeration, propulsion and process industries to compress a flow without moving parts by exploiting the energy released during the expansion of another stream. The engineering modelling of choking in these devices remains an open question, with two mechanisms – Fabri and compound choking – proposed in the literature. This work develops a quasi-one-dimensional framework based on common governing equations, enabling a direct comparison between the two using axisymmetric Reynolds–averaged Navier–Stokes (RANS) simulations as reference. The compound formulation incorporates wall and inter-stream friction and a local pressure-equalisation procedure that enables stable integration through the sonic point, together with a normal-shock reconstruction. The Fabri formulation is assessed by imposing the dividing streamline extracted from RANS, isolating the sonic condition while avoiding additional modelling assumptions. The calibrated compound model predicts on-design secondary mass flow typically within 2 % of the RANS simulations, rising to 5 % for a strongly under-expanded primary jet due to the equal-pressure constraint. The Fabri analysis attains less than 1 % error in on-design entrainment, but exhibits high sensitivity to the dividing streamline and closure, which limits predictive use beyond on-design. Overall, the results show that Fabri and compound mechanisms can coexist within the same device and operating map, each capturing distinct aspects of the physics and offering complementary modelling value. Nevertheless, compound choking emerges as the more general mechanism governing flow rate blockage, as evidenced by choked flows with a subsonic secondary stream.