Suppose somehow the base heat shield fails to separate, is that a LOC event? or just a harder but survivable landing? (I understand that the the heat shield covering parachutes failing to separate is most likely a LOC event)
It would likely be a "harder but survivable" landing; that is the case for Soyuz (and I would assume Shenzhou) in similar situations. For Soyuz/Shenzhou, a failure of the base heat shield to separate means that the landing retro-rockets are not exposed to provide the final "cushion" before touchdown, which is broadly similar to the role of the airbags for Starliner.
About the trunk or service module, suppose that fails to separate, it is obviously not protected and will be destroyed on entry, which would ultimately expose the heat shield, is that possible? or a failure to separate would mean LOC? (I understand Soyuz was designed where one part of the entry module will separate due to structural failure from reentry heating if the pyrotechnic separation fails)
(bolding mine)
Small note: Soyuz wasn't
designed to ensure its service module's attachment points would "burn off" safely during reentry after a failed separation. Rather, that was discovered the hard way in practice, on the
Soyuz 5 mission in 1969. As designed, a failure of the service module to separate was considered a guaranteed LOC event. As recounted in
this article about the Soyuz 5 incident:
When they realised what had happened—or, more accurately, what had not happened—several flight controllers buried their faces in their hands. One officer removed his cap, dropped three rubles into it, and passed it along the line; within minutes, it had filled with coins for Volynov’s young family. The cosmonaut was effectively plummeting back to Earth, nose-first, with the least-protected section of his craft exposed to the greatest thermal stress. Moreover, he was exposed to G forces in excess of nine times their normal terrestrial load. Against such overwhelming odds, it seemed that Boris Volynov’s fate was sealed.
As the article (very much worth a read in its entirety!) goes on to detail, the subsequent re-entry was extremely violent and damaged the capsule in ways that would cause further complications (the parachute lines got tangled and one of the landing retro-rockets failed to fire). The survival of the spacecraft and its pilot was very much "by accident" and to the surprise of the designers; they had no expectation beforehand that the more limited application of heat-shielding ablator on the front of the descent module would be sufficient to buy the crew enough time to wait out the burning-off of the service module connectors.
It seems that in the aftermath of the incident, the Soviet/Russian space program may have attempted to retroactively consider Soyuz's empirically-demonstrated resilience to this failure mode to be a designed failsafe. When it happened again nearly 40 years later on the
Soyuz TMA-10 mission, Roscosmos neglected to disclose the occurrence to NASA (as no US or USOS-partner astronauts happened to be on board to witness it), on which account NASA was reportedly furious when it
happened again on the next flight,
Soyuz TMA-11 (which had Peggy Whitson onboard as well as South Korea's first astronaut). That led to more detailed engineering studies jointly by NASA and Roscosmos which concluded, at length, that the service module could in fact be "counted on" with some reliability to burn off and detach before the forward heat shielding ablated to the point of an unsurvivable failure.
As for Starliner (and Dragon)...
Based on the hard-learned lessons of Soyuz, I would imagine that Boeing and SpaceX's engineers put at least some effort into evaluating how their capsules would fare under similar failure modes. However, it's not clear that Starliner and Dragon's designs would be as "favorable" under such a situation as Soyuz's happened to be.
Dragon's trunk, for instance, is attached to the main capsule via attachment hardpoints located
on the base heatshield surface, as shown in
this picture (picture sourced from
this Everyday Astronaut article). Unlike Soyuz, whose service-module attachment points clamp onto the outside of the descent module and thus protrude into the hot reentry airstream, Dragon's attachment points would be "protected" by the trunk's walls. So there's a good chance Dragon's trunk would
not "burn off" and detach like Soyuz's (not until well after the crew module's forward heat shielding had failed, at least) - making this a near-certain LOC scenario. This is likely why Crew Dragon changed the separation timeline to detach the trunk
before making the deorbit burn, leaving it in a stable orbit where it will deorbit within a few weeks (but buying time to diagnose the problem if it fails to separate) - versus Dragon 1, which detached the trunk after the burn to avoid leaving debris in orbit.
Unlike Dragon (but like every other capsule historically), Starliner doesn't have the luxury of detaching its service module before the reentry burn, because the OMACS thrusters on the SM itself are responsible for doing that burn. Based on pictures like
this one (sourced from
here), it looks like Starliner uses an attachment system similar to Dragon, i.e. with hardpoints on the base heat shield itself rather than external "claws" like Soyuz. So I suspect it would also be unlikely to burn off cleanly in a face-first reentry. The OMACS/RCS doghouses would
maybe be able to "catch some hot air" and burn during reentry since they'd probably protrude out into the plasma stream, but whether that would create enough aerodynamic force to rip the service module free from any failed disconnection points is a more complicated question - never mind what such a "tearing free" would do to the structural integrity of the crew module and the base heat shield. In any case, it's clearly not a scenario with any significant chance of survival, meaning that Boeing would've had to treat it as a single point of failure in their analysis. (In fairness, I suspect Dragon's
heat shield trunk* separation is similarly considered a single point of failure from a safety certification standpoint - as there's likely little that could be done to troubleshoot a separation failure even while the capsule remains in a stable orbit. The likeliest recourse would be to return to ISS, but there may not be delta-v for that at that point in the mission.)
*Edited 6/3 to correct a typo (Dragon has a separable trunk, not a separable heat shield, and this is indeed what I meant in the original writing). Thanks kdhilliard for bringing this to my attention via PM!