When the kinetic wave has insufficient momentum to couple with a field generator, the curl free generator may operate with low efficiency (F = P/c), or the generator might be reconfigured to weakly couple with a gravity field using different forward looking geometry. Also the possibility of adding a curl to the generator may exist when a suitable magnetic field is available for coupling, by making a change to the interference pattern in the generator.

The general case should have a kinetic stress energy wave, with no alternating components, convex geometry and appearing nearly stationary standing wave when viewed from the field generator.

Concave fields like gravity are always attracting each other. Gravity potential fields and convex kinetic fields diminish each other as shown in Lagrangian Densities, suggesting a repulsion. If true then kinetic fields would be expected to attract each other.

In this representation the efficient propulsion curl free field generator would produce a non alternating convex standing wave form when interacting with an external gravity field, and a non alternating concave standing wave form when interacting with a kinetic wave.

All of this is subject to testing.

A proof of the energy flow can be made with a Josephson Junction detector in which the superconducting component generates an electrical potential as a curl free vector wave passes through it.

Proof that it is a stress wave is demonstrated by putting a shield and barrier between generator and detector.

Not all curl free vector potentials occur in wave forms. In these cases the proof of energy flow can be made by generating the vector field with interference components of very slightly different wave lengths. Then the wave character is not detectable in one place but fully developed in some other distant place where the components are out of synchronization.

By suitable management of the generator components the entire curl free power flow can acquire a curl and be decomposed into electromagnetic waves at a distant place.

The intent of this answer is that variability should be prevented as much as possible. Power flow should be curl free and producing vectored vacuum stress of geometry not propagated in waves, for best chance of break through propulsion. Then there are several stress geometries to choose from in experiments.

In as much as the propulsion field is curl free the generator can be fully contained within a high speed vehicle. Most of the kinetic vacuum stress energy would also be contained inside the vehicle, so there is no need for external field generators unless the vehicle is found to be super conducting to an extent that the field cannot pass through it. The curl free fields pass through all other materials of construction.

Much is published about cavity thrusters, but the curl free character is not well developed, leading to low thrust efficiency and much of the power going out as heat. In all cases the momentum is conserved with components of the momentum going backward in vacuum stress.

In speaking about vacuum stress, the resulting local strain can be positive or negative, but there there is no scientific sign convention for vacuum strain. When local strain is described as large scale curvature, then a gravity field can be locally represented as strain concave to the source, tending to close an event horizon. Next applying the Lagrangians to dynamic sources the kinetic curvature should represent strain convex to the source, tending to open up local spaces, and in an extreme case opening wormholes in a larger dimensionality.

Two concave curvatures of gravity attract each other. So a convex curvature should repel a concave curvature as in the Lagrangians. By the same reasoning two convex curvatures should attract each other when vectored in the same direction.

Kinetic field energy is represented as a curl free convex vector field pointed forward, so on first estimate the thrust field generator might produce a curl free convex vector field pointed backward, reversing the sign.