Will a Starship go to Mars in the 2026 launch window?

[lykos]

Will they do it or not?
Can they do it or not?
Should they do it or not?

All obsolet!

When refilling is o.k. they will do it.
The sooner the better, the more the merrier.
Even legs are not needed.
They will gather soooo much data that it will be worth anyway.
Try and fail, try again and fail again...

[Twark_Main]

Yes if the arrival vInf can be managed. Looking at the plot for this opportunity showing the sum of departure and arrival vInfs (using pykep lambert_problem) seems to indicate the window is short with a budget of < 5.6 km/s.

This plot is almost showing what you want, but not quite.

   1.  It should be departure delta-v and arrival velocity, not departure v_infinity and arrival v_infinity (summing these quantities is meaningless).

   2.  You only care about any excess arrival velocity above the maximum aerocapture velocity. If the spaceship is arriving at exactly the maximum aerocapture velocity it should count as 0 m/s.  If the spacecraft arrives at half the maximum aerocapture velocity it should also count as 0 m/s.

...

Regarding the first point I think maybe I understand what you mean. The sum of the two vInf values is certainly not a delta-v budget for a mission. I'm envisioning it with the notion that it assumes Earth and Mars magically disappear. Then the sum of the vInfs is the total delta-v required to get from where Earth was (position and velocity) to where Mars was. Then when I magically reintroduce the planets I need to handle the added delta-v to get out of or into the orbits around them, starting from their respective escape velocities. Is the thinking behind that approach confused somehow?

I think the salient point that got lost here is: the point on the pork chop plot where (Vinf_departure + Vinf_arrival) is minimized is not necessarily the same as the point where (DV_departure + DV_arrival) is minimized.

The correction should really be done for each point while plotting the pork chop, instead of using the Vinf pork chop to find the (probably wrong) 'ideal' point and then doing the correction afterward.

Is there a Python file for this? I'd be happy to whip up a quick fix with the necessary correction, just post or PM me.

[sdsds]

Yes if the arrival vInf can be managed. Looking at the plot for this opportunity showing the sum of departure and arrival vInfs (using pykep lambert_problem) seems to indicate the window is short with a budget of < 5.6 km/s.

This plot is almost showing what you want, but not quite.

   1.  It should be departure delta-v and arrival velocity, not departure v_infinity and arrival v_infinity (summing these quantities is meaningless).

   2.  You only care about any excess arrival velocity above the maximum aerocapture velocity. If the spaceship is arriving at exactly the maximum aerocapture velocity it should count as 0 m/s.  If the spacecraft arrives at half the maximum aerocapture velocity it should also count as 0 m/s.

...

Regarding the first point I think maybe I understand what you mean. The sum of the two vInf values is certainly not a delta-v budget for a mission. I'm envisioning it with the notion that it assumes Earth and Mars magically disappear. Then the sum of the vInfs is the total delta-v required to get from where Earth was (position and velocity) to where Mars was. Then when I magically reintroduce the planets I need to handle the added delta-v to get out of or into the orbits around them, starting from their respective escape velocities. Is the thinking behind that approach confused somehow?

I think the salient point that got lost here is: the point on the pork chop plot where (Vinf_departure + Vinf_arrival) is minimized is not necessarily the same as the point where (DV_departure + DV_arrival) is minimized.

The correction should really be done for each point while plotting the pork chop, instead of using the Vinf pork chop to find the (probably wrong) 'ideal' point and then doing the correction afterward.

Thanks for returning to this! I agree the approach I've been using isn't correct. And I think it may get progressively worse (as a proxy for the truth) as the transit arc shifts away from the minimum.

I'm particularly concerned because it conflates each three-axis velocity vector with a scalar speed. This is almost okay when the departure velocity vector is closely aligned with the velocity vector of the departure planet and the arrival velocity vector is closely aligned with the velocity vector of the arrival planet. That's the case for a Hohmann style transfer in a coplanar circular orbit model but that's not good enough in reality, particularly for transits of non-'optimal' duration.

Is there a Python file for this? I'd be happy to whip up a quick fix with the necessary correction, just post or PM me.

I'm using the numpy, scipy, pyplot and pykep packages. A couple of the central functions are appended below. I'll gladly share the full code base with one major caveat: it wasn't written in a style intended for sharing. You likely know what I mean.... ;-)

def _doGridPoint(plD, plA, tD, tA, gridFunc):
      rD, vD = plD.eph(epoch(tD, "mjd2000"))
      rA, vA = plA.eph(epoch(tA, "mjd2000"))
      sFlight = (tA - tD) * DAY2SEC
      l = lambert_problem(rD, rA, sFlight, MU_SUN)
      return gridFunc(l, vD, vA)

def _evalLambert(lambert, vDeparture, vArrival):
  v1 = lambert.get_v1()[0]
  vD = norm([a - b for a, b in zip(v1, vDeparture)])
  v2 = lambert.get_v2()[0]
  vA = norm([a - b for a, b in zip(v2, vArrival)])
  return vD, vA

[Twark_Main]

I'm using the numpy, scipy, pyplot and pykep packages. A couple of the central functions are appended below. I'll gladly share the full code base with one major caveat: it wasn't written in a style intended for sharing. You likely know what I mean.... ;-)

def _doGridPoint(plD, plA, tD, tA, gridFunc):
      rD, vD = plD.eph(epoch(tD, "mjd2000"))
      rA, vA = plA.eph(epoch(tA, "mjd2000"))
      sFlight = (tA - tD) * DAY2SEC
      l = lambert_problem(rD, rA, sFlight, MU_SUN)
      return gridFunc(l, vD, vA)

def _evalLambert(lambert, vDeparture, vArrival):
  v1 = lambert.get_v1()[0]
  vD = norm([a - b for a, b in zip(v1, vDeparture)])
  v2 = lambert.get_v2()[0]
  vA = norm([a - b for a, b in zip(v2, vArrival)])
  return vD, vA

I certainly do! It can't be worse than mine.   I think the board supports zip files.

It's not the neatest code ever, but something like this should do the trick:


def _vInfToCaptureDV(vInf, rBurn, muPlanet):
  vEscape = math.sqrt(2 * muPlanet / rBurn)
  return math.sqrt(vInf*vInf + vEscape*vEscape) - vEscape

def _vInfToCircularOrbitDV(vInf, rOrbit, muPlanet):
  vEscape = math.sqrt(2 * muPlanet / rOrbit)
  vCircularOrbit = math.sqrt(muPlanet / rOrbit)
  return math.sqrt(vInf*vInf + vEscape*vEscape) - vCircularOrbit


The correction should really be in its own function, but just for illustration I'll cram it into _evalLambert (I used pykep's safe_radius property for the burn radius but you probably want to sanity check those values):


def _evalLambert(lambert, vDeparture, vArrival):
  v1 = lambert.get_v1()[0]
  vD = norm([a - b for a, b in zip(v1, vDeparture)])
  v2 = lambert.get_v2()[0]
  vA = norm([a - b for a, b in zip(v2, vArrival)])
  dvD = _vInfToCircularOrbitDV(vD, earth.safe_radius, MU_EARTH)
  dvA = _vInfToCaptureDV(vA, mars.safe_radius, mars.mu_self)
  return dvD, dvA


Probably want to retitle the graph too, but code-wise (modulo a few import statements) that should be enough to generate the appropriate pork chop plot with just a bit of cut-and-paste.

No LLMs Were Harmed^TM, so all typos and boneheaded code mistakes are 100% my fault.   

[daedalus1]

What's all the above maths for?
Fact,  the only starship test stand looks like the aftermath of a nuclear explosion, less than 18 months before the launch window. And every starship is now disintegrating.
I could go on. No sustainable orbital flight yet.
The moon is the priority.
So not a chance this will be happening

[sdsds]

What's all the above maths for?

For me the interesting question here is when the 2026 departure window closes for various types of Mars missions. That depends on lots of unknowns so of course anything here would be simply a guess. The goal is to make a guess at least somewhat informed by orbital mechanics calculations.

I think the salient point that got lost here is: the point on the pork chop plot where (Vinf_departure + Vinf_arrival) is minimized is not necessarily the same as the point where (DV_departure + DV_arrival) is minimized.

I adapted your code snippets and came up with the attached plot. A zip file with the python code is also attached.

[sdsds]

Following up, I'll first emphasize these calculations are based on somewhat arbitrary values for the departure LEO altitude and the altitude of the Mars arrival burn. And the calculations could be flat out wrong. That said, there are those who project that with a full prop load in LEO a v3 Starship might have a delta-v budget on the order of 8000 m/s, in which case the departure window extends into mid-January of 2027.

[Twark_Main]

Following up, I'll first emphasize these calculations are based on somewhat arbitrary values for the departure LEO altitude and the altitude of the Mars arrival burn. And the calculations could be flat out wrong. That said, there are those who project that with a full prop load in LEO a v3 Starship might have a delta-v budget on the order of 8000 m/s, in which case the departure window extends into mid-January of 2027.

I generally use 90 nautical miles (167 km) if I want to be optimistic yet proven. This was the TMI altitude of the later Apollo J Type missions. Lowering the TMI altitude from 100 nm to 90 nm is what improved performance enough to (among other things) bring the lunar rover.

For Starship they will have extensive atmospheric drag mapping/prediction from Starlink, which helps when balancing the altitude vs reboost tradeoff.

[sdsds]

[...] somewhat arbitrary values for the departure LEO altitude and the altitude of the Mars arrival burn.

I generally use 90 nautical miles (167 km) if I want to be optimistic yet proven. This was the TMI altitude of the later Apollo J Type missions.

Your earlier comment was compelling: the shape of the contours on a plot of 'Total Vinf' might not be the same as the shape of the contours on a 'Total dV' plot. Clearly on a 'Total dV' plot the values themselves change when the departure or arrival altitudes are changed. Is the concern now that the shape or location of the contours is also affected by these parameters?

In the context of the window that opens in late 2026 there are obviously still many unknowns and the conops for final refilling is certainly a big one. A major example is the idea SpaceX has floated that the final orbit for refilling might be a high apogee ellipse. If that approach were used in 2026 then predicting the orbital speed at the time of the departure burn is going to be really tricky.

[Twark_Main]

I generally use 90 nautical miles (167 km)...

Your earlier comment was compelling: the shape of the contours on a plot of 'Total Vinf' might not be the same as the shape of the contours on a 'Total dV' plot. Clearly on a 'Total dV' plot the values themselves change when the departure or arrival altitudes are changed. Is the concern now that the shape or location of the contours is also affected by these parameters?

It certainly will effect it. In the limit, as the altitude approaches infinity it will approach the v_infinity condition, so yes the chosen altitude changes the shape.

Typical Earth radius is 6371 km, so a plausible value for r_burn would be 6538 km.

In the context of the window that opens in late 2026 there are obviously still many unknowns and the conops for final refilling is certainly a big one. A major example is the idea SpaceX has floated that the final orbit for refilling might be a high apogee ellipse. If that approach were used in 2026 then predicting the orbital speed at the time of the departure burn is going to be really tricky.

Fortunately that doesn't change the shape of the porkchop, it's really just giving the ship an additional delta-v boost, specifically the difference between the perigee speed and a circular orbit at the same altitude.

At 167 km the escape velocity is 11.044 km/s and a circular orbit is 7.809 km/s, so parking in a near-escape elliptical orbit gives your ship up to ~3.235 km/s of additional delta-v.

[sdsds]

The following analysis assumes Musk will want to attempt something during the 2026 opportunity and there will be constraints (technical and social) that preclude a landing attempt.

During the brief "sweet spot" in the opportunity the required Mars orbit insertion delta-v is not high. As a reference the ISRO Mars Orbiter Mission used ~1100 m/s of delta-v for insertion into a 423 x 80000 km Mars orbit. Parker et al. indicate ESCAPADE is planning something similar; each of its two spacecraft "target an MOI Corridor that is centered about a nominal aimpoint 450 km above the surface. A thrust arc
centered around this passage is used to capture into an elliptical orbit, aiming for a period of ~60 hours." The 2026 opportunity appears to offer similar delta-v insertions. (Attached plot is for 'capture' into a parabolic orbit.)

The test objectives would be:
- validate cryo boil-off modeling for interplanetary cruise
- verify Raptor restart after interplanetary cold soak
- attempt to deploy a prototype Marslink payload into Mars orbit

With that set of objectives there would be no need for a heatshield or landing legs. It seems plausible a Starship in that configuration could depart in 2026 and have more than sufficient delta-v  for MOI upon arrival at Mars.

[steveleach]

With that set of objectives there would be no need for a heatshield or landing legs. It seems plausible a Starship in that configuration could depart in 2026 and have more than sufficient delta-v  for MOI upon arrival at Mars.

Is that with or without refuelling?

I'd assume that with refuelling it would have sufficient delta-v even with legs and tiles.

Without refuelling, my impression was that a fully-stripped down ship on a fully-expended booster was marginal for a flyby, so nothing close to the delta-v needed for capture.

[meekGee]

If current predictions of v3 first flight at the start of 2026, then they have 6 months of engineering before they need to mount the launch campaign.

That's a very long time when they're not dealing with mishaps.

So on the optimistic side, they won't have to bother with the trade-offs.  On the realistic side, however....

[sdsds]

[...]
I'd assume that with refuelling it would have sufficient delta-v even with legs and tiles.

Without refuelling, my impression was that a fully-stripped down ship on a fully-expended booster was marginal for a flyby, so nothing close to the delta-v needed for capture.

Yes I'm imagining propellant transfer in LEO because until that works it would likely be the sole focus of their beyond-LEO efforts and thus preclude any attempt at a Mars mission. It's not clear they would have enough propellant on-orbit to fully refill an outbound Ship; or how boil-off in transit might affect the available delta-v once at Mars.

[steveleach]

[...]
I'd assume that with refuelling it would have sufficient delta-v even with legs and tiles.

Without refuelling, my impression was that a fully-stripped down ship on a fully-expended booster was marginal for a flyby, so nothing close to the delta-v needed for capture.

Yes I'm imagining propellant transfer in LEO because until that works it would likely be the sole focus of their beyond-LEO efforts and thus preclude any attempt at a Mars mission. It's not clear they would have enough propellant on-orbit to fully refill an outbound Ship; or how boil-off in transit might affect the available delta-v once at Mars.

They don't need to have such a narrow sole focus though, they can do both. A one-off "chuck something at Mars" mission wouldn't significantly delay the propellant transfer work, especially if they have 2 gigabays  and 4 launch towers by then.

And, if they have got refuelling sorted, there's not that much benefit to sending a stripped-down ship when they can send a regular one with a bit more fuel.

[Lee Jay]

Having the president-elect in his victory speech raving enthusiastically for several minutes about the Super Heavy Booster grab was exceptional. Obviously Starship now has the complete backing of the new Administration. Politically-motivated delaying won't be happening for the next 4 years. That'll take us well past 2026... Now it is just physics and the pace of development standing in the way, and SpaceX is pretty good at dealing with those.

Since then:
Starships 7, 8 and 9 have all failed.
Trump and Musk have fallen out with support threatened to be pulled from both sides (cancel contracts/dismantle Dragon), meaning "politically-motivated delaying" is more probable than ever.
Ship 36 blew up and destroyed a good bit of Massey's.

Things don't always turn out how you think they will, but even I didn't predict they would go as badly as they have.

[Robotbeat]

[...]
I'd assume that with refuelling it would have sufficient delta-v even with legs and tiles.

Without refuelling, my impression was that a fully-stripped down ship on a fully-expended booster was marginal for a flyby, so nothing close to the delta-v needed for capture.

Yes I'm imagining propellant transfer in LEO because until that works it would likely be the sole focus of their beyond-LEO efforts and thus preclude any attempt at a Mars mission. It's not clear they would have enough propellant on-orbit to fully refill an outbound Ship; or how boil-off in transit might affect the available delta-v once at Mars.

They don't need to have such a narrow sole focus though, they can do both. A one-off "chuck something at Mars" mission wouldn't significantly delay the propellant transfer work, especially if they have 2 gigabays  and 4 launch towers by then.

And, if they have got refuelling sorted, there's not that much benefit to sending a stripped-down ship when they can send a regular one with a bit more fuel.

not only would it not delay propellant transfer work, but lifting something to Mars is a good use for the propellant transfer test vehicle.

Like, let’s say you launch a Starship to orbit, packed with Starlinks, deliver those starlinks to LEO, then reuse that ship for the propellant transfer demo, then since it’s full of propellant now, send it to Mars instead of trying to recover it at Earth (SpaceX is going through starship iterations so fast, it might not make a lot of sense to recover all the early Starlink launch ships). Beats scrapping it at Earth.

[steveleach]

Having the president-elect in his victory speech raving enthusiastically for several minutes about the Super Heavy Booster grab was exceptional. Obviously Starship now has the complete backing of the new Administration. Politically-motivated delaying won't be happening for the next 4 years. That'll take us well past 2026... Now it is just physics and the pace of development standing in the way, and SpaceX is pretty good at dealing with those.

Since then:
Starships 7, 8 and 9 have all failed.
Trump and Musk have fallen out with support threatened to be pulled from both sides (cancel contracts/dismantle Dragon), meaning "politically-motivated delaying" is more probable than ever.
Ship 36 blew up and destroyed a good bit of Massey's.

Things don't always turn out how you think they will, but even I didn't predict they would go as badly as they have.

If they weren't building 2 gigafactories and 3 or 4 launch towers then I'd be worried.

[Lee Jay]

Having the president-elect in his victory speech raving enthusiastically for several minutes about the Super Heavy Booster grab was exceptional. Obviously Starship now has the complete backing of the new Administration. Politically-motivated delaying won't be happening for the next 4 years. That'll take us well past 2026... Now it is just physics and the pace of development standing in the way, and SpaceX is pretty good at dealing with those.

Since then:
Starships 7, 8 and 9 have all failed.
Trump and Musk have fallen out with support threatened to be pulled from both sides (cancel contracts/dismantle Dragon), meaning "politically-motivated delaying" is more probable than ever.
Ship 36 blew up and destroyed a good bit of Massey's.

Things don't always turn out how you think they will, but even I didn't predict they would go as badly as they have.

If they weren't building 2 gigafactories and 3 or 4 launch towers then I'd be worried.

Worried about what?  Whether or not an un-crewed no-payload Starship leaves for Mars in 2026 is irrelevant to basically everything.  So I'm not worried about that either.

[Robotbeat]

Having the president-elect in his victory speech raving enthusiastically for several minutes about the Super Heavy Booster grab was exceptional. Obviously Starship now has the complete backing of the new Administration. Politically-motivated delaying won't be happening for the next 4 years. That'll take us well past 2026... Now it is just physics and the pace of development standing in the way, and SpaceX is pretty good at dealing with those.

Since then:
Starships 7, 8 and 9 have all failed.
Trump and Musk have fallen out with support threatened to be pulled from both sides (cancel contracts/dismantle Dragon), meaning "politically-motivated delaying" is more probable than ever.
Ship 36 blew up and destroyed a good bit of Massey's.

Things don't always turn out how you think they will, but even I didn't predict they would go as badly as they have.

If they weren't building 2 gigafactories and 3 or 4 launch towers then I'd be worried.

And making money hand over fist with Starlink.

This sort of blackpilling that you see people engage in here is never helpful.