Abort options for Starship and Starship/SuperHeavy

[CorvusCorax]

...
This has implications for abort situations. A sudden loss of thrust on ascent but within the atmosphere (which might happen on P2P flights without SH, ascent from Mars, or a launch abort where SH thrust is terminated) would leave Starship in an aerodynamically unstable configuration, and close to no control authority from the flaps.
...

Is that fixable without turning it into an airplane?

and

Looks like this all ran its course. Locked. If someone actually wants to talk about the LES, then a new thread can be started.

As shown in this thread:
https://forum.nasaspaceflight.com/index.php?topic=56619.0

Starship is aerodynamically unstable when at an AoA below 20° - as such loss of thrust during ascent would likely cause the vehicle to tumble and crash or break apart.

This pulls into question many of the discussed "Starship is its own abort system" arguments.

Right now I see only two possible options.

Option 1: The propulsion system is made so reliable that loss of thrust during ascent is sufficiently unlikely that no other abort system is needed and propulsive landing is equally reliable.

Option 2: A part of the Starship payload/crew section can be jetisoned and land on its own, for example on parachutes. This would likely require a separate thrust system and parachutes or equivalent (wings, propulsive landing), as such a yet to be designed dedicated dedicated LAS.

Option 3: A significant Starship redesign that allows additional abort methods (such as gliding to a runway like SpaceShuttle)


==============
Option 1 is a no-brainer that does not need to be discussed. This is almost certainly the way to go for cargo. It will also be the way to go for landing on other planets where there is no or very thin atmosphere and as such no parachute/aerodynamic abort mode possible. This has been discussed to death in the old closed thread.
There are some who are very adamant that this is the only option Starship might ever need. This opinion, while valid, is hereby declared off topic for this thread, as it derails the technical discussion about actual abort systems. Please don't delute the discussion about possible abort options into a discussion whether starship needs abort options or not. Off topic. There's 52 pages of that right here: https://forum.nasaspaceflight.com/index.php?topic=43438.0
This includes the notion that "commercial airplanes have no abort system/parachutes either" - yes, we know that by now.
==============
Option 2
I think this is where the most productive discussion can be. What would be possible designs. Which abort scenarios would they cover and which do they not? What would be the payload penalty? What would be additional marginal cost to launch (if any)? Can we come up with designs?
==============
Option 3
I would also like to discourage discussions of approaches that are a too radical deviation from the Starship design. The borderline is a bit fuzzy, as SpaceX itself has done quite radical redesigns of Starship over the years, but the design has by now matured quite a bit.
I think a radical redesign - such as a spaceplane or similar - while possibly safer or more convenient for P2P transportation - would no longer be called "Starship". It's also not really on SpaceX roadmap - yet.


So let's focus on Option 2 please, shall we? :-)



It's clear that the current designs - especially nose designs - of Starship are for the cargo section. Some have a slit for a sat dispenser, giant payload bay doors have also been discussed. But the crew version was always different with large windows, also forward facing windows. These windows would be primarily on the "upper" side of Starship (away from the heatshield) since the thermal loading there is low enough to allow windows in the first place.

My proposal for a LAS is a "half fuselage" shaped reinforced section in the upper half of the nose (if Starship were to be placed horizontal) that is
1. large enough to include all inhabited seats during takeoff
2. does not include the "luggage bay" or other storage -- similar to how passenger airplanes have the people on top and the luggage below and sometimes behind

This section would be shaped so that it is
1. Jetisonable in all stages of flight
2. Shaped aerodynamically stable, so it does not tumble when freeflying
3. Propulsive - with abort motors that can get it far enough away from a failing starship quickly enough - there might be solids/pyrotechnics involved - this would also allow pad abort

4. Has some sort of landing capability - for example really large parachutes.  How big and how heavy would this be? We are talking a fuselage section let's say 6m wide and 4m high and 20m+ long with 100+ people inside (maybe  on 2 decks)

Open question:
- What would that weigh?
- Is parachute landing feasible for something that size and weight?
- Is a glider-style landing feasible? If so, what would be landing speed (assuming shape with body lift and some aerodynamic control surfaces, but no thrust and very small or no actual "wings") And runway requirements? Could an ocean landing work?

- What would be the mass penalty for Starship?
- What would be constraints on maximum number of seats carried with and without LAS?


- Would there be alternative approaches?
Suggested has been for example - a Dragon2 in the nose - which would probably be a good stopgap measure with existing technology but raises questions regarding complexity/heatshield compatibility and seats only 7 ppl max,

Have at it

[frith01]

My guess would be option
 4.  Sign a waiver for the first few years, use as-is. 

Once they get to manned point to point travel (2030's), there may need to be something addressed. 

Other likely options:

5.  Drogue chute to stabilize in case of loss of control.  parachutes for very un-graceful landing.
6.  alternate / emergency hot thruster controls similar to Lunar Starship, but they could be behind steel panels on tiled side, and just move panel & blow out the tiles when needed. 
7.  reinforced + crumple zones around manned cabin inside starship to extend the range of survivable scenarios on manned missions. (probably only good up to 300mph, but it sounds good)

[JayWee]

Is that fixable without turning it into an airplane?

Disclaimer: I know absolutely nothing about aerodynamics, plane design, etc...

I understand you want to focus on option 2, but I'd still like to ask about Option 3 (SS Redesign):
What I was trying to understand there is whether it is possible to have a starship-like body (a 9m steel tube with ogive ending) and "fiddle" with the aerodynamic surfaces to gain stability at lower AoA. Something like moveable/deployable canards/wings.

[M.E.T.]

I thought you were going for a small, more or less Dragon sized escape pod in the very tip of the nose, for early flights until reliability is well established. For small crews up to say a dozen people. That seems feasible.

Trying to have an escape pod for 100 passengers seems…difficult.

[meekGee]

Is that fixable without turning it into an airplane?

Disclaimer: I know absolutely nothing about aerodynamics, plane design, etc...

I understand you want to focus on option 2, but I'd still like to ask about Option 3 (SS Redesign):
What I was trying to understand there is whether it is possible to have a starship-like body (a 9m steel tube with ogive ending) and "fiddle" with the aerodynamic surfaces to gain stability at lower AoA. Something like moveable/deployable canards/wings.

...and landing gear, yes? 

[CorvusCorax]

Is that fixable without turning it into an airplane?

Disclaimer: I know absolutely nothing about aerodynamics, plane design, etc...

I understand you want to focus on option 2, but I'd still like to ask about Option 3 (SS Redesign):
What I was trying to understand there is whether it is possible to have a starship-like body (a 9m steel tube with ogive ending) and "fiddle" with the aerodynamic surfaces to gain stability at lower AoA. Something like moveable/deployable canards/wings.

The main issue is the far-back center of mass in a launch abort situation. Because of the full tanks, the vehicle would want to fly tail-first. You need quite significant control authority to overcome that, which means pretty big control surfaces (fins) at the far back end, to overcome the torque of the nose when it starts going out of alignment with the airstream.
But those would create so much passive stability, that you would have big issues entering a high angle of attack for reentry. you'd be locked in a low AoA configuration, which would reenter like a lawn dart and hit the dense atmosphere at high hypersonic velocities. Especially for regular reentries when the tanks are then empty.

I would highly suggest to try out different spaceplane configurations and reentry styles in Kerbal Space Program (with FAR mod aerodynamics) to get an intuitive feeling for how stability over large AoA ranges and center of mass play together.

There are possible configurations, but they are as impractical as they are inefficient, and they are all quite far away from anything that looks like Starship.

[volker2020]

It should be self evident, that a starship without functioning raptors and with lot's of fuel in it can not come down unharmed.

Than, apart from a major pressure lost, I don't see another error vector, that could lead to all engine control lost at once.

The only chance I see for a large pressure lost, would be a ruptured hull (one misbehaving vent would not be fast enough for this treat).

I guess it is fair to assume, that if you have a ruptured hull, the aerodynamics of your star ship is the least of your concerns.

[CorvusCorax]

I thought you were going for a small, more or less Dragon sized escape pod in the very tip of the nose, for early flights until reliability is well established. For small crews up to say a dozen people. That seems feasible.

Trying to have an escape pod for 100 passengers seems…difficult.

Tip of the nose is actually possibly a bad idea. (Edit)
It would work well for ascent aborts, but for reentry, you'd have to cover it with heat shield tiles asymetrically.
It would likely be more advantageous to have something that flies away from where the cockpit would be if Starship were a 747. As I wrote above, this is where the forward facing windows would likely be anyway, and where one would want to sit for reentry and belly-flop.

Now obviously if initial flights are with a small crew contingent, this system could initially be much smaller. But that also means you need to design it twice, first for a smallscale system, and later for a large one. And the same method/design might not work in both cases.

A small LAS capsule - whever it is located and whatever the shape - can almost certainly land on parachute, just like Dragon can. But would that scale?

What it would do, is provide a no black-zone abort ability for basically all flight phases. It could even be equipped with an ablative single use heat shield, so the crew capsule could even survive a breakup on reentry and be able to land on parachute. The first possible abort scenario would be a launch abort, where the abort capsule rockets away from a pad mishap. But it would even be available during belly flop, or if something goes wrong with the flip and/or landing burn. Even once Starship is on the ground and has an SN10 style saving mishap, it could still get the crew away.

[BT52]

You take take HUGE assumption that u cannot design reliable system at all and u need said contingency's? Would said contingency even work? Maybe... But i mine opinion in very tiny edge cases.

Ascend; Cases like loss off engines etc or slow loss of tank pressure... All that could be engineered out and become more reliable. Also abort mode could be engineered into SS ship anyway. Then u can just split SS from booster if u detect defect/anomaly early enough.

Descend: Any loss of structural stability could be loss of craft anyway. I remember Red Bull Stratos project of highest jump at that time from 40km of altitude. That jump showed how scary uncontrollable stratosphere spin can be.  In that case if SS lose its control before it hits dense troposphere and fails to adjust into designed AOA its a loss of crew anyway. No split capsule would help u. You cannot separate crew from tip of SS suceffuly and controlably.  Also design  of whole secondary control for just 1 small edge case is quite ridiculous. 

Conclusion: instead SIMPLIFY!

Adjust its focus into avionics and SS architecture. Make them reliable. Adjust total engineering goals  with taking into account maintenance intervals etc.. Made componentry robust, not just for prototype sake but for data sake. Even if it takes couple extra tonnes from vehicle. Simplify flying hardware architecture to eliminate complex failures or complex repair work.

Next in future: Provide robotic scans of airframes, heatshields etc.  Sens components integrity via many strain, pressure, temp sensors all over the place. Use Xray with drones or robots. use microphones to detect instabilities of engines, flue flow, knocking etc. There are many many engineering tools of control and ways to do maintenance. There are many ways to diagnose vehicle without human factor to miss a detail. All of that can be  easily automated and trigger many alerts to the launch control, service centers and engineering's department.

Each SS would get at each flight its "medical state sheet" so they can predict performance or even narrow some operating windows if necessary. Although last statement is iffy because in that case just ground the vehicle and use another one. Point of SS in cheap production and repair so you have huge fleet of them anyway.

Use them just as aircrafts. Ideally operation should be little more expensive then Concorde. HEHE its total mindset and cultural change.

[chopsticks]

--snip--

Did you read what the OP wrote?

Option 1 is a no-brainer that does not need to be discussed. This is almost certainly the way to go for cargo. It will also be the way to go for landing on other planets where there is no or very thin atmosphere and as such no parachute/aerodynamic abort mode possible. This has been discussed to death in the old closed thread.
There are some who are very adamant that this is the only option Starship might ever need. This opinion, while valid, is hereby declared off topic for this thread, as it derails the technical discussion about actual abort systems. Please don't delute the discussion about possible abort options into a discussion whether starship needs abort options or not. Off topic. There's 52 pages of that right here: https://forum.nasaspaceflight.com/index.php?topic=43438.0
This includes the notion that "commercial airplanes have no abort system/parachutes either" - yes, we know that by now.

Off-topic.

[rsdavis9]

I thought you were going for a small, more or less Dragon sized escape pod in the very tip of the nose, for early flights until reliability is well established. For small crews up to say a dozen people. That seems feasible.

Trying to have an escape pod for 100 passengers seems…difficult.

Tip of the nose is actually possibly a bad idea. (Edit)
It would work well for ascent aborts, but for reentry, you'd have to cover it with heat shield tiles asymetrically.
It would likely be more advantageous to have something that flies away from where the cockpit would be if Starship were a 747. As I wrote above, this is where the forward facing windows would likely be anyway, and where one would want to sit for reentry and belly-flop.

Now obviously if initial flights are with a small crew contingent, this system could initially be much smaller. But that also means you need to design it twice, first for a smallscale system, and later for a large one. And the same method/design might not work in both cases.

A small LAS capsule - whever it is located and whatever the shape - can almost certainly land on parachute, just like Dragon can. But would that scale?

What it would do, is provide a no black-zone abort ability for basically all flight phases. It could even be equipped with an ablative single use heat shield, so the crew capsule could even survive a breakup on reentry and be able to land on parachute. The first possible abort scenario would be a launch abort, where the abort capsule rockets away from a pad mishap. But it would even be available during belly flop, or if something goes wrong with the flip and/or landing burn. Even once Starship is on the ground and has an SN10 style saving mishap, it could still get the crew away.

I'll weigh in and say VERY small single person escape capsule is probably the best.
Is it possible to get some major simplifications from a small capsule?
1. Smaller single chute. Maybe a backup chute.
2. Less weight of ablative heat shield.
3. Probably solids are the smallest and most reliable for escape velocity.
4. Don't put all your eggs(people) in one basket(capsule). Maybe one out of ten in crew dies. Better than all ten dying.
5. Maybe small enough to air bag for final impact on land.
6. So for risky flight everybody goes up and down in capsule. Ascent and decent. No time to get into capsule.
7. Sounds and looks like what we see in scifi. 

So what is the smallest reentry capsule that can be made?

EDIT: and the best part of small capsule is that if you don't want it(abort capsules) you can just leave it off of the ship. Easy to use small panels that bolt over holes for escape capsules. At least it seems easier than big modifications to structure and etc.

[spacenut]

I thought the Starship itself was to escape and land if there was a major failure of the booster. 

One option is to launch a Starship without crew to orbit.  Then launch Dragon capsules to provide the crew.  A Dragon was designed to hold 6 crew.  It depends on how big a crew SpaceX wants to send to the moon or Mars.  I also thought the first crew going to Mars was to be around 12 people. 

I believe it will be a while before they launch 100 people at a time.  This would be after Starlinks are put into orbit, refueling in orbit perfected, and many returns from orbit successfully as well as many launches successfully. 

I wouldn't be surprised if Dear Moon launched on a Dragon and docked with a fully fueled Starship in orbit for the lunar flyby.

Starship could contain one Dragon equipped with a 6 person seating.  This would cut into space for crew quarters and/or crew provisions.  The Dragon could be ejected and left in LEO while crew proceeded to the moon or Mars.  It would have to have the ability to eject while on the way to orbit.  Maybe installed in the Starship at an angle.  I still think that if it is going to be a problem for NASA to launch a crew on a Starship, they can always launch on a Dragon and dock with a Starship in orbit.   

[chopsticks]

First of all, putting something on the nose is a non-starter because that is where the header tank is. (Unless you want to modify the tankage of the ship as well.)

I think of this off and on, and one thing IMO that needs to be considered equally as much as abort from the pad/ascent regime is also a landing burn abort. In other words, what sort of design could be used for 1) pad abort, 2) ascent all the way to orbit, and 3) landing.

Thinking of a pad abort, the crew could perhaps be in some sort of capsule or re-entry capable* spacecraft, either a capsule or some other passively stable aerodynamic pod or canister that would be shot out of the side of the Starship. In other words, a spacecraft (or at least an aerodynamically stable pressure vessel inside a spacecraft that would land under parachutes. However, I don't see it being a good idea to launch it out at 90 degrees to the side of the rocket like spitting out Starlink sats, but maybe at a 45 degree angle upwards or so in the case of a pad abort. I'm just struggling a bit with how such a craft would be designed - the shape of it. It seems like it would be a pretty big waste of space to have such a craft nested inside the Starship, but maybe that's just the compromise that will have to be met for Earth flights with crew from launch to landing..?

*Actually, if there would be a loss of thrust somewhere in the upper thin atmosphere to space after stage sep, Starship itself could manage the reentry and the crew could eject while lower in atmosphere.

[chopsticks]

Here's an extremely crude drawing (I'm sorry) of a capsule inside the starship that could fire some sideway pointing solid rocket motors (or hypergolics) for a brief period to clear the starship, then fire the downward pointing abort motors once clear. Similar to how some missiles are jettisoned then fired once clear of the vehicle they were in.

Obviously the shape of the capsule could probably be quite simple - the astros would only need to be in it for a few minutes during takeoff and landing.

[MichaelBlackbourn]

If we’re back to postulating spacecraft inside spacecraft as a simple way to abort, just flat pack the people and reuse the pez dispenser and rapid fire them out. Boom.

[edzieba]

Putting every passenger in a Vostok-type capsule near the dorsal vehicle skin for ascent and descent is really annoying in terms of occupied single-purpose volume, but provides viable abort capability for descent (ballistic re-entry from a nominal Starship altitude and velocity would be enormously unpleasant, but probably not fatally so) and late ascent, and possible for early ascent if a strap-on solid motor is used to reach a minimum safe distance and altitude for parachute deployment (not as easy as it sounds, would require active orienting and steering). Stick the ascent/descent 'crash couches' in them and enclose with a personal life-support system - so the capsules double as emergency refuges - and hang them from rotating mounts (or have the couch rotate internally) such that they can reorient for ascent and descent, and you can at least solve one problem with all the other problems you introduce in adding them.
A cluster of armed solid motors sitting inside the vehicle along with weakened breakaway panels (or panels with armed shaped charges) seems like adding as many or more new failure modes as you mitigate with the escape system in the first place, though.

[CorvusCorax]

I thought you were going for a small, more or less Dragon sized escape pod in the very tip of the nose, for early flights until reliability is well established. For small crews up to say a dozen people. That seems feasible.

Trying to have an escape pod for 100 passengers seems…difficult.

Tip of the nose is actually possibly a bad idea. (Edit)
It would work well for ascent aborts, but for reentry, you'd have to cover it with heat shield tiles asymetrically.
It would likely be more advantageous to have something that flies away from where the cockpit would be if Starship were a 747. As I wrote above, this is where the forward facing windows would likely be anyway, and where one would want to sit for reentry and belly-flop.

Now obviously if initial flights are with a small crew contingent, this system could initially be much smaller. But that also means you need to design it twice, first for a smallscale system, and later for a large one. And the same method/design might not work in both cases.

A small LAS capsule - whever it is located and whatever the shape - can almost certainly land on parachute, just like Dragon can. But would that scale?

What it would do, is provide a no black-zone abort ability for basically all flight phases. It could even be equipped with an ablative single use heat shield, so the crew capsule could even survive a breakup on reentry and be able to land on parachute. The first possible abort scenario would be a launch abort, where the abort capsule rockets away from a pad mishap. But it would even be available during belly flop, or if something goes wrong with the flip and/or landing burn. Even once Starship is on the ground and has an SN10 style saving mishap, it could still get the crew away.

I'll weigh in and say VERY small single person escape capsule is probably the best.
Is it possible to get some major simplifications from a small capsule?
1. Smaller single chute. Maybe a backup chute.
2. Less weight of ablative heat shield.
3. Probably solids are the smallest and most reliable for escape velocity.
4. Don't put all your eggs(people) in one basket(capsule). Maybe one out of ten in crew dies. Better than all ten dying.
5. Maybe small enough to air bag for final impact on land.
6. So for risky flight everybody goes up and down in capsule. Ascent and decent. No time to get into capsule.
7. Sounds and looks like what we see in scifi. 

So what is the smallest reentry capsule that can be made?

EDIT: and the best part of small capsule is that if you don't want it(abort capsules) you can just leave it off of the ship. Easy to use small panels that bolt over holes for escape capsules. At least it seems easier than big modifications to structure and etc.

The problem with "individual escape crew capsules" for everyone on board is how this scales with the number of capsules. A single capsule can be launched with unsteered fixed solid rockets or if needed with a very simple thrust vectoring system (to make sure the thing goes up regardless of attitude of the mothership) similar to that on ejection seats.

This works well for one or two seats (see ejection seats) it also would scale to let's say up to 4 if there is sufficient spacial distancing.

It would not scale beyond that. You can't guarantee collision free escape trajectories anymore, especially if the mothership is tumbling. Worse, there might be contacts and entanglements of parachutes between different escape pods - especially on pad aborts when everything is slow and in close proximity.

From a systems complexity you add a lot of duplication without getting true redundancy for the individual passenger. You increase the chances that at least some survive thanks to multiple systems, but that doesn't help the ones who don't.

Maintenance and system checks for a large number of pods would probably also be expensive and more time consuming than a single one.

If we’re back to postulating spacecraft inside spacecraft as a simple way to abort, just flat pack the people and reuse the pez dispenser and rapid fire them out. Boom.

I would like to point to rhis article on escape crew capsules: https://en.wikipedia.org/wiki/Escape_crew_capsule - done so on various military aircraft prototypes. There are various tradeoffs between ejection seats and ejection capsules, typically the higher (thin air) and faster (hypersonic) aircraft go, the more physics favor a capsule.


For Starship, instead of putting a capsule inside the Starship hull and having it somehow get extracted in case of emergency, a much more useful setup would look like this:

You start with a standard Starships setup for crew and cargo, which includes the whole ship including flaps, heat shield, plumbing, power and propulsion, but an empty payload section - and a large open cutout where the payload bay would go.

Then you plugin a modular "payload inset" into this cutout. Which can be a slotted starlink dispenser. Or a large empty box with a giant hinged door for "big sats"  - or 2 hinged doors, an extendable boom and a 9m wide mirror to use starship as an on-orbit telescope

Or a pressurized cockpit/crew module with windows and possibly a docking port

This crew module would have to fit in the "standard payload module bay" of starship, which should fill more or less the entire useable payload volume of the nose section. It should have electrical and data interfaces to "talk" to starship and - if needed - have access to methane and oxygen (optional)

The outer mold line of this crew module (like any payload modules) would have to fit the Starship outer moldline, which is cylindrical and narrowing towards the front. This outer hull section does not include any heatshield tiles and only needs reentry protection from back-radiation from the plasma tail.

Once you have this "crew/cockpit" module that can be inserted and switched for other type of payload modules, it would be quite straightforward to modify it to make it (or parts of it) jetisonable in flight with abort motors.

It wouldn't be exactly shaped like a capsule. The basic shape would better fit the mold-line of something shaped similarly to Sierre Nevada's DreamChaser. It would make sense to go for a "lifting body" aerodynamic shape to give it aerodynamic stability post jetison.

Doing this would allow for several potential landing options, parachute, gliding, or propulsive - depending on choices such as system complexity, weight, etc...

The "top" of this crew module would obviously be standard stainless steel, optionally with windows. The bottom could be any material, including ablative heat shield materials, in which case the whole module could become a sort of "escape shuttle".

All of this without increasing the complexity of Starship per se - on the contrary it would create more similarity between different starship variants. All the extra complexity of the LAS would be in the crew module, which needs this anyway (pressurization, life support, etc...)

[CorvusCorax]

Here's an extremely crude drawing (I'm sorry) of a capsule inside the starship that could fire some sideway pointing solid rocket motors (or hypergolics) for a brief period to clear the starship, then fire the downward pointing abort motors once clear. Similar to how some missiles are jettisoned then fired once clear of the vehicle they were in.

Obviously the shape of the capsule could probably be quite simple - the astros would only need to be in it for a few minutes during takeoff and landing.

I would suggest a payload module - in this case a "crew module" roughly shaped like this:

I think it is plausible how this shape could contain a crew compartment that
1. Shares the outer moldline with Starship  (its roof)
2. Is aerodynamically stable and controllable
3. Can rocket away from Starship and land - for example  via parachute.
4. Can jetison itself during any flight phase of Starship, including on the ground

[Oersted]

Why all these new-fangled capsules and modules? - If you'd really want to have an escape capsule inside Starship just put a Dragon in there. There's room.

For a first manned Starship landing on the Moon it would also be a good abort option during descent or ascent module after a safe landing. Land a couple of vanilla Moon Starships first and then a Moon starship with the crew inside Dragon and voilá, you have a moon base. 

[CorvusCorax]

Why all these new-fangled capsules and modules? - If you'd really want to have an escape capsule inside Starship just put a Dragon in there. There's room.

For a first manned Starship landing on the Moon it would also be a good abort option during descent or ascent module after a safe landing. Land a couple of vanilla Moon Starships first and then a Moon starship with the crew inside Dragon and voilá, you have a moon base.

As mentioned before, you would have about the same complexity as in the above to get the dragon OUT of the Starship. But also have a major hazzle of getting it IN there in the first place, servicing the systems including hypergolics... I assume just firing the superdracos inside starship is probably a bad idea, so you need a separate ejection system, which needs to succesfully jetison parts of the hull, then push dragon out of there, and  THEN fire the superdracos ... And all the while you get a shape mismatch that causes you to waste a whole lot of volume.

As someone pretty famous here likes to say, Rockets aren't Legos (yes I did that on purpose! Checking if you're awake!)

Having a starship specific module would be way easier and quicker.  You could (and would) of course re-use dragon2 systems in the design, such as tanks, computers, engines, life support, toilet, parachutes, etc...  where applicable, so it could count as a "dragon derived system" but that doesn't make it a dragon.