Copyright 2015 Robert Clark
Blue Origin successfully landed their New Shepard rocket after reaching suborbital space:
Observing the last portion of the video showing the landing, deviations from the vertical are visible but the ability to hover allowed it sufficient time to correct.
Comparing this to the SpaceX Falcon 9 failed attempts at landing it is apparent the inability to hover for the F9 did not allow it sufficient time to make the needed corrections.
SpaceX has said they want their next test landing to be on land at the launch site. My opinion, they might succeed on the next test or two but they will always have failures without hovering ability.
Merlins in a pressure-fed mode.
Achieving hovering is not even difficult. In the blog post "Hovering capability for the reusable Falcon 9, page 2: Merlin engines in a pressure-fed mode?" I suggested giving the Merlin the ability to run in a pressure-fed mode. The question was whether this was technically feasible. I found in fact that this process of giving a turbopump powered engine a pressure-fed mode, called an idle mode, had been successfully tested during the Apollo days on the J-2 upper stage engine.
In giving the J-2 an idle mode though, it was changed from the gas generator cycle that is used by the Merlin 1D to a tap-off cycle:
However, there is an engine that uses the gas generator cycle and has an idle mode, the LE-5 upper stage engine of the Japanese space agency:
Development of the LE-X engine.
In this idle mode though the thrust is significantly less than at full thrust, only 3% in the LE-5 case. If it is a similar low percentage for the Merlin's then all 9 engines would have to be used in this idle mode to allow it to hover on landing.
The idle mode has an additional advantage since it does not use the turbopumps. It could be used to burn both residual liquid propellant and gases in the tanks. This would mean much less residual fluid would be left in the tank. This then reduces the amount of propellant that needs to be kept on reserve for the landing.
Elon Musk has also recently said in his Twitter account that the F9 first stage has single-stage-to-orbit (SSTO) capability. For an SSTO the residuals in a first stage can subtract a significant amount from the payload it can deliver to orbit. Then the ability to run in an idle mode with minimal residuals left over can significantly increase the payload for an SSTO. So this would be a further advantage of giving the Merlins an idle mode.
Hovering by use of flexible nozzle extensions.
In the blog post "Altitude compensation attachments for standard rocket engines, and applications", I discussed another method of achieving hovering capability, attaching nozzle extensions to the bottom of the engines that would allow restriction of the thrust. The flexible high temperature materials already exist in the reentry materials used in NASA's Inflatable Re-entry Vehicle Experiment (IRVE). This has the advantage that the nozzle extension would have to only be applied to the one central engine to reduce its thrust on landing.
However, the extendable nozzle attachments also have an advantage to the SSTO case. By using an extension that can be retracted at launch and fully extended at high altitude, you can get engines usable at sea level that can reach the high vacuum Isp's usually reserved for upper stage engines. In this way the 311 s vacuum Isp of the Merlin 1D can be raised to the same level of 340 s as the Merlin Vacuum. An increase in the vacuum Isp to this extent can as much as double the payload of a SSTO.
Note that both of these techniques, idle mode or flexible nozzle extensions, would mean hovering capability can actually increase the payload rather than reduce it.