How the ejector energy balance works:
The motive gas accelerates to sonic velocity in the nozzle throat (choked flow) with pressure decreasing by approximately 45% (depending of gas properties). The gas continues to accelerate to Mach >1 in the outlet section of the nozzle, with the pressure reducing to that required to draw in the entrained gas. The ejector nozzle is also known as a De Laval nozzle.
Pressure reduction of the motive gas has thus been converted isentropically into usable kinetic energy. Some energy is lost because of isentropic efficiency <100%.
The entrained gas is drawn into the mixing section and also accelerates to sonic velocity (there are cases where it does not reach sonic velocity but these are not described here). Once again there is some energy loss due to isentropic efficiency <100%.
The motive gas and entrained gas have different velocities. Momentum is preserved as they mix, but kinetic energy is lost. In general, the greater the ratio of motive gas pressure to entrained gas pressure, the greater the loss of kinetic energy.
The mixed gas is still supersonic as it leaves the mixing section. A shock wave then occurs as it enters the diffuser. Flow becomes subsonic and a step change increase in pressure occurs.
The gas decelerates in the diffuser with kinetic energy being recovered as increased pressure. As above, some energy is lost due to isentropic efficiency <100%.
Friction against the internal metal surfaces of the ejector causes additional energy losses.
Link to equations relevant to the energy balance.
The final stop in procuring an ejector is at the vendors, but modelling and calibrating the energy balance takes us a long way forward on this path. The final design is complex but initial process design does not have to be so mysterious.
An understanding of the energy balance and geometry enables an intuitive feel for control methods and constraints.