Orbital Access Methodologies Part V: Boostback TSTO

While I have the topic fresh in my mind, I decided to jump into the next part of my continuing series. Though it wasn’t a conscious choice on my part, I notice that the order I went with for this series actually follows a consistent pattern. In each part of this series, we discuss methods that move more and more of the delta-V load off of the orbital stage and onto the carrier vehicle or the first stage. In the case of Air-Launched SSTO, the carrier plane removed about 1000m/s from the ~9km/s normally required for a ground-launched SSTO, thus making an SSTO design feasible. For the Pop-up TSTO design, the first stage’s vertical trajectory removes all of the gravity and drag losses from the upper stage (a savings of ~1.6km/s). For the Glideback TSTO design, by using aerodynamic lift to turn around and glide back to the launch site, some horizontal downrange velocity was added, thus lowering the delta-V requirements even further (probably saving somewhere between 1.6-2.4km/s depending on the details). The next approach we’ll discuss follows this same trend.

Two Forms of Boostback Techniques
In a Boostback TSTO system, the first stage provides not only vertical velocity to overcome most if not all gravity and drag losses and significant downrange velocity, but it also provides enough propulsive capacity to return itself to the launch site after separation. Unlike the glideback case, the Boostback TSTO approach stages at a sufficiently high velocity that at least some of the return to launch site (RTLS) delta-V has to be provided propulsively by the stage itself. Also, unlike the glideback approach, the first stage does not have to have a high L/D ratio, and in fact boostback can be used with VTVL vehicles.

The first, and most well-known form of boostback, (the form proposed for use with the Kistler K-1 vehicle, which I’ll call Propulsive Boostback) involves a first-stage rotation maneuver after staging, followed by firing the engines long enough to both cancel out all of the downrange horizontal velocity, and provide enough net uprange horizontal velocity that the stage can land back at the launch site. In the case illustrated in the presentation I linked to in the previous part (and further detailed in this report), the optimal staging velocity was found to be about Mach 5.2 (~1800m/s), at an altitude of around 52.5km, and a staging flight path angle of about 31 degrees. For this case, I did a little analysis, and I’m estimating that between the ascent phase and RTLS boostback maneuver, the total first stage delta-V would be around 5500-5800m/s. But the good news is that the upper stage would also be down in that range (ie slightly lower than 6km/s even including landing propellant for the VTVL case). The Kistler K-1 vehicle used a similar but slightly different trajectory, where the staging was planned to take place at about Mach 4.4 (~1500m/s), and at around 42km. That would result in a slightly higher required upper stage delta-V requirement, but a lower first stage performance requirement. This figure, from Barry Hellman’s report I linked to above shows an example propulsive boostback (starting at the staging point):

While Propulsive Boostback is the most well-known form of Boostback, I realized last week that there was another approach that is also uses a form of boostback maneuver. For sake of clarity, and for lack of a better term, I’ll call this approach Lift Assisted Boostback.

I thought of this boostback approach in response to some questions to my previous post on glideback approaches. Someone had asked why you couldn’t stage at an even higher velocity. I started in on an explanation about how at velocities any higher than Mach 3.2 (using the assumptions in the prior studies), the rocket would not be able to glide back all the way to the landing site, and that therefore you’d need some sort of additional propulsion event after staging in order to get home. While people typically recommend turbojets for such missions (thus switching from glideback to “flyback” for the first stage), I suggested that it might be worth just using the rocket engines in such a situation. Upon further thought, I realized that there might be more to this suggestion than I had originally thought.

Basically, if the first stage has a sufficiently good L/D, what you can do after staging is, glide downrange a bit, and then perform a turn-around maneuver aerodynamically (once you’re back in the atmosphere enough to do so), and finally relight the engines to provide enough momentum to get you back within glide back range of your launch site. By performing the turnaround maneuver, you’re using aerodynamic lift to bend your trajectory around so that the downrange (away from the launch site) velocity is now actually turned into velocity heading back home. That way, when you light your engines for the boostback maneuver, while you may be at a lower altitude, you no longer have to null-out the downrange velocity, and your propulsion system also doesn’t have to provide all the uprange velocity in order to return to the launch site.

[Update 7/1/08: A commenter mentioned that there’s a third approach that combines some of the features of propulsive and lift assisted boostback to avoid some of the key drawbacks of both. Basically, if you have a vehicle that both has good L/D, and has a propulsion system that can handle a boostback retrofiring maneuver, you have a third option that avoids hypersonic flight and excessive TPS requirements, while also keeping the first stage Delta-V more reasonable. Basically, after staging you immediately pitch over and decelerate until you’ve slowed yourself down enough that you can reorient yourself and do a glideback trajectory from there. While it adds some extra operational complexity (two rotational maneuvers), it gets rid of the TPS issues with lift assisted boostback, and gets the required delta-V for the stage down into the 3.8-4km/s range instead of the 5.6-6km/s range required for a purely propulsive boostback technique. Food for thought.]

Benefits and Drawbacks of Propulsive Boostback
The two different boostback techniques have somewhat different advantages and drawbacks. Propulsive Boostback is the form best known, so I’ll discuss some of the pros and cons of this approach first.


  1. A common benefit of both approaches over the previously discussed methodologies is that the delta-V requirements on the upper stage are much lower. Depending on the exact staging conditions, the upper stage may need to provide as little as 5800m/s, compared with at best 6400m/s for Glideback TSTO, 7400m/s for Pop-up TSTO, and 8000m/s for Air-launched SSTO. 5800m/s equates out to a propellant mass fraction of about 0.83 for a medium-end LOX/Kerosene stage, and about 0.73 for LOX/LH2. Both of these are very realistically attainable pmf values.
  2. The delta-V requirements put the two stages at a level of technology only slightly beyond that needed for small suborbital vehicles (which tend to suffer from higher drag losses than larger suborbital vehicles, and thus need a higher total delta-V for the same apogee), making the step from suborbital to this form of orbital easier.
  3. A boostback TSTO has the option of doing occasional downrange landings (if there is a suitable landing site) in instances where you need to lift heavier payloads.
  4. With the upper stage empty an unfueled, the first stage could actually self-ferry the stack fairly long distances (several hundred miles).
  5. The boostback maneuver ends up resulting in a very low reentry velocity compared to what you would expect from the staging horizontal velocity. The reentry velocities are low enough, ~Mach 2, that TPS is almost unneeded for the first stage.


  1. The first stage ends up requiring a lot more delta-V than earlier methods, but a substantial chunk of that is used for the RTLS maneuver. At low achievable propellant mass fractions and Isp, this results in a much easier to build RLV than the other approaches. However, as the achievable mass fraction and Isp increases, at some point the extra delta-V actually makes the vehicle heavier (both in total mass as well as in just dry mass) than a pop-up or glideback stage. While admittedly higher dry mass doesn’t necessarily equate to higher costs (a 1000lb dry mass stage made of 5383 TIG-welded aluminum is going to cost a lot less than even a 500lb dry mass stage made of friction stir welded Li-Al alloy, or a 250lb stage made of Unobtanium Wishalloy-X), there may be a performance point at which the boostback design no longer has sufficient cost or performance advantages over glideback or pop-up designs to justify the more complicated maneuvers.
  2. The turnaround and boostback maneuvers (often called the RTLS maneuvers) are somewhat complicated, and involve in-air relights of engines. Admittedly for a VTVL stage, your propulsion system better be rock-solid reliable anyway, so this isn’t as big of a deal for VTVL boostback systems, but every additional complication comes at a price.
  3. Boostback trajectories have more of their safety-critical operations occurring downrange of the launch site than many other approaches. This means that more attention will have to be paid during launch license applications to making sure the trajectory is tuned to keep the risk to the uninvolved public low enough.
  4. More to the point, at some point, the Vacuum IIP (the point where the vehicle would hit if it’s propulsion systems failed at that instant and there was no atmosphere) ends up loitering over some downrange site. Making sure you can have this occur over an unpopulated area is critical for getting launch licenses.
  5. Trajectory tuning like this requires extra performance margin. With enough margin, you can probably find appropriate trajectories for most launch sites and azimuths, but the more generally useable the stage wants to be, the more margin you need. The problem is that the first stage in this case is already getting near the steep part of the delta-V vs. Mass Ratio curve. Adding extra margin becomes harder and harder very rapidly.

There are probably other benefits and drawbacks I’m not thinking of, but these are a start.

Benefits and Drawbacks of Lift Assisted Boostback
While there are several big potential advantages to the Lift-Assisted Boostback, there are also some unique differences and drawbacks. Unfortunately, since this isn’t a concept I’ve seen investigated in the literature before, and as the aerodynamic turn-around maneuver is more complicated than I know how to easily analyze (and I don’t have access to a full-up 6DOF trajectory analysis program), I will only be able to give some general thoughts. If anyone reading this actually has enough time to analyze the concept in detail, they might be able to provide some more insights.


  1. By using aerodynamic lift to do the turn-around maneuver, you will end up requiring less RTLS delta-V for a given staging velocity.
  2. While it is possible to do a propulsive boostback with an HTHL stage, all of the main burns for a lift-assisted boostback system are performed at altitudes where aerodynamic control surfaces can provide some or all of the control, thus allowing you to use engines as simple as those that would be required for glideback.
  3. This approach gives you most if not all of the reduced upper stage delta-V requirements that a propulsive boostback technique without anywhere near as much of a first stage delta-V penalty. This means that this approach may stay competitive with glideback and pop-up approaches even as the level of achievable stage performance increases.
  4. Unlike propulsive boostback, your IIP never ends up stopping and loitering over any given point, because your trajectory is being bent around aerodynamically. A rapidly-moving IIP crosses a given chunk of land faster, thus making it easier to maintain a reasonable E-sub-c for launch license purposes.
  5. The fact that this approach doesn’t really require any unique capabilities not needed for glideback (glideback may assume that you have the capability to relight the engines in case you need to do a go-around at the landing site), means that you can incrementally upgrade a glideback vehicle to be able to perform a lift-assisted boostback. For a given glideback TSTO design, as you incrementally add first-stage performance, that offloads performance requirements from the upper stage, allowing it to carry more payload over time.
  6. Most of the aerodynamic maneuvering occurs at a high enough altitude and speed that it’s possibly in the hypersonic regime. In the hypersonic regime, lifting bodies are just about as good as winged stages, which means it might be possible to have a VTVL system that has a lifting body configuration. You’d use the lift for aiding in the turn-around maneuver, and some of the glideback, but would use propulsion for takeoff and landing. Thus getting some of the benefits of a winged vehicle, while avoiding the disadvantages of a VTHL system.


  1. In order to do the turn-around maneuver, your stage is going to be going fairly fast during reentry, and in order to maximize performance, you will likely end up exposing your vehicle to pretty ugly thermal environments–much worse than propulsive boostback, glideback, or pop-up TSTO designs. Nowhere near as bad as orbit, but possibly as bad as “flyback” trajectories. This requires a real, honest-to-goodness TPS system that will need to be developed and proved out. We’re talking maneuvers going on at airspeeds faster than the SR-71, so this isn’t a trivial problem, even if the duration is relatively brief.
  2. Unlike propulsive boostback, if you staged at a similar velocity, you’d end up going much further downrange before you could get back into the atmosphere far enough to start turning around. Depending on how much of the velocity you can maintain after the turn, this may require a significant burn to get the vehicle back to the launch site. In other words, at least some of the benefit you get from not having to use propulsion to null-out the forward velocity is counterbalanced by possiblly requiring a bigger burn to get up to speed to get back to the launch site. This may mean that the optimal staging point is at a lower velocity than for propulsive boostback. Or it may just mean you have to do a hotter turn-around maneuver.
  3. Since you end up going much further downrange, it may be harder to find areas remote enough to launch out of.
  4. A failed engine relight may force an emergency landing a long way from your launch site. This may require a decent amount of contingency planning.
  5. Doing a large hypersonic turnaround maneuver may end up causing a large sonic boom, which may also complicate trajectory planning.

There may be some other benefits and other problems, but those are the major ones.

Enabling Technologies and The Path Forward
Boostback TSTO designs share similar enabling technologies to the other approaches. HTHL versions could really use composite cryo tanks to allow them to fly with “wet wings”. All of the different boostback approaches can benefit from suborbital vehicles–it may even be possible to test out a lot of the techniques necessary using suborbital vehicles. The orbital stages for these approaches need TPS work just as much as any of the others–but in the case of lift-assisted boostback, even the first stage will require advanced TPS work. Altitude compensating nozzles (or Thrust Augmented Nozzles, which also have a form of altitude compensating) help a lot, as most of the RTLS burn is done at high altitudes, and for propulsive boostback, higher thrust for the boostback maneuver ends up reducing the required delta-V back by a small but not insignificant amount.

The real way ahead for both of these projects is going to involve testing out the required maneuvers with suborbital vehicles first. There are some groups in the Air Force that are really keen on using this technique as well, and they have been pushing it quite hard lately. Even sub-suborbital vehicles (like XCORs Lynx, most of MSS’s XA-0.x demonstrators after 0.2, and most of Armadillos’ nearterm vehicles) can do some of these experiments, and it would be good if the Air Force could continue working with these firms as their vehicles become available. Admittedly, I’m somewhat biased there–being a propulsion engineer for one of the companies that could benefit from such a move. But by using a boostback maneuver with a suborbital sized vehicle, the delta-V requirements for an expendable upper stage would be low enough to allow for a decent nanosat launcher (or a vehicle that could launch TPS testing reentry vehicles, which would be a great way to get the data you need before you can start building an orbital LV.

So, does anybody have a 6DOF simulator and lots of time on their hands that wants to do some extra analysis of this lift-assisted boostback maneuver? It might make for a fun Master’s Thesis.

The following two tabs change content below.
Jonathan Goff

Jonathan Goff

President/CEO at Altius Space Machines
Jonathan Goff is a space technologist, inventor, and serial space entrepreneur who created the Selenian Boondocks blog. Jon was a co-founder of Masten Space Systems, and the founder and CEO of Altius Space Machines, a space robotics startup that he sold to Voyager Space in 2019. Jonathan is currently the Product Strategy Lead for the space station startup Gravitics. His family includes his wife, Tiffany, and five boys: Jarom (deceased), Jonathan, James, Peter, and Andrew. Jon has a BS in Manufacturing Engineering (1999) and an MS in Mechanical Engineering (2007) from Brigham Young University, and served an LDS proselytizing mission in Olongapo, Philippines from 2000-2002.
Jonathan Goff

About Jonathan Goff

Jonathan Goff is a space technologist, inventor, and serial space entrepreneur who created the Selenian Boondocks blog. Jon was a co-founder of Masten Space Systems, and the founder and CEO of Altius Space Machines, a space robotics startup that he sold to Voyager Space in 2019. Jonathan is currently the Product Strategy Lead for the space station startup Gravitics. His family includes his wife, Tiffany, and five boys: Jarom (deceased), Jonathan, James, Peter, and Andrew. Jon has a BS in Manufacturing Engineering (1999) and an MS in Mechanical Engineering (2007) from Brigham Young University, and served an LDS proselytizing mission in Olongapo, Philippines from 2000-2002.
This entry was posted in Launch Vehicles, Orbital Access Methodologies, Space Transportation, Technology. Bookmark the permalink.

21 Responses to Orbital Access Methodologies Part V: Boostback TSTO

  1. Anonymous says:

    It seems to me that one of the less expensive development paths migh be to go 3STO until more experience is in. Use the pop up stage that you already have in suborbital service for one. A disposable orbital nanosat launcher for three. And incrementally do an expendable boostback second stage. Reuse the second while expanding its’ flight envelope until you lose it. Use the data for the next build.

    After fully developing the RTLS capability, go back to the TSTO in the orriginal planning. You only need to find one gotcha in the boostback concept to validate a cheap approach. In the meantime, early revenue stream.

  2. Iain McClatchie says:


    You’ve got a funny edit link in there, right in the middle of a word, just before “…requiring a bigger burn to get up to speed…”

    Might want to take that out

  3. Formerly known as Skeptic says:

    Another way to combine propulsive boostback and glideback would be to burn to kill downrange and establish uprange velocity and achieve a glideback trajectory, thus increasing your staging velocity over the glideback option, but not getting into the high speed aerodynamics.

  4. Jon Goff says:

    Good catch–fixed. I wonder what the heck caused that.


  5. Jon Goff says:

    There are many flavors of incremental tricks you can pull along the way. Combining Pop-up with boostback for a 3STO is an idea I was also looking at at one point. The problem with doing this for a full-sized booster is that expendable upper stage is likely not going to be cheap or easy, and while you’re doing that, it’s going to be hard to be much cheaper than anyone else in the business. Though, that said, having a way to do your first rev TSTO as an experimental vehicle that you can eke a little revenue out of isn’t entirely bad either.


  6. Jon Goff says:

    Formerly known as skeptic,
    That’s an interesting option. It would require good aerodynamics and all the fancier propulsion stuff, but it looks on first pass like it would be a good way of doing things. Say you’re staging at Mach 5. You do a boostback burn to get yourself back to a little below Mach 3, and then do a turnaround maneuver and glide back? Not a bad approach. That might keep your first stage velocity requirement below 4km/s, while getting your orbiter stage down below 6km/s. I imagine that there are probably other subtle nuances on this idea that also give interestingly good results.

    Oh for a 6DOF (and someone who knew how to drive it properly)!


  7. Anonymous says:

    Great installment. Thanks!


    There are some groups in the Air Force is really keen on using this technique as well, and has been pushing it quite hard lately.


    -Adam Greenwood

  8. Jon Goff says:

    Good catch. I was wrapping that up around midnight after a long drive home from LA….


  9. Anonymous says:

    Those who can’t do, edit.

    –Adam Greenwood

  10. john hare says:

    Jon Goff said…
    There are many flavors of incremental tricks you can pull along the way. Combining Pop-up with boostback for a 3STO is an idea I was also looking at at one point. The problem with doing this for a full-sized booster is that expendable upper stage is likely not going to be cheap or easy, and while you’re doing that, it’s going to be hard to be much cheaper than anyone else in the business. Though, that said, having a way to do your first rev TSTO as an experimental vehicle that you can eke a little revenue out of isn’t entirely bad either.

    I screwed up the sig on the post you replied to.

    My meaning was that once you already have a pop-up available in your VTVL sub-orbital for some small multiple of propellant costs, it would be temporarily cheaper to use it instead of starting from scratch. The likely loss of the boostback unit would be during the recovery as the launch phase would be well known already from your VTVL experience. Loss of the boostback unit during recovery would mean after the primary payload had staged for orbit. Revenue would be earned even on the lost units.

    The expendable boostback unit should have a mass ratio of well under 3. It should be easier to do it than a full size launch from the ground boostback unit. There should be considerable margins for experimentation in getting the formula right. It also the drops the disposable upper stage into the M/R 6 ratio for dense fuels early on. You could be losing much less hardware per flight than the competition very early in the game.

    My main point is early market entry. Even if the competition halves your price after you have been making money for a few years, you still have a nest egg and experience they don’t have. Your mark 2 or so is partially financed and credible. And if you find the gotchas that experience only provides, your mark 2 or so might make it into revenue service before their mark 1 gets the kinks out.

    Your orbital propellant interest ties into boostback also. During freefall before reentry, scavage the propellants from the main tanks to auxillery tanks for the boostback. The unusable propellants and margins were a major mass problem in the paper you cited. It seems possible you could have a viable microgravity propellant transfer method ahead of anybody else. And make money developing it.

  11. Jon Goff says:

    Ah, I think I get your point now. So you’re saying that once Masten for instance has an XA-1.0 or 1.5 flying, that we could do a two stage expendable nanosat launcher on top, and then slowly start incrementally upgrading the middle stage until we can get it back? It’s not a crazy idea. When taken from the standpoint of suborbital LV derived nanosat launcher, it might make some economic sense.


  12. john hare says:


    That is what I had in mind. Sometimes I forget that my assumptions don’t necessarily match those of rational people. It really wouldn’t make sense to do 3STO all from scratch.:-)

  13. Dave Salt says:

    As you know, I’m something of an advocate of bimese/trimese concepts and lift assisted boost-back would be a natural recovery method for such systems.

    As you know, true bimese/trimese would use identical units in order to take advantage of only having to build one vehicle type, so the booster(s) would have a TPS capable of orbital re-entry and so have no problem performing the up-range turn.

    Unfortunately, the devil is in the details and, as you indicate, the contingency cases will be the things that make or break any concept.

  14. Swatch says:

    Why not use Orbiter to simulate the flybacks? If properly set up, it will handle aerodynamics quite well.

  15. Jon Goff says:

    I guess I never really got enough into Orbiter to know how to do something like that. I was never able to figure out how to get it to do much more than just flight sim sort of stuff. Can you give it a preplanned routine and get it to output the trajectory and other important data? In a way that would allow you to run some sort of optimization routine?

    I honestly don’t know.


  16. Swatch says:

    That’s one of those cases where if you can write it in C/C++, you can probably do it in Orbiter. If you leave Orbiter to handle the 6DOF Physics and you have a control program, then its fairly easy to interface the two. We’re doing that over at http://nassp.sf.net, but our control program is the old Apollo Guidance Computer Software, and we just wrote a wrapper that will pass state information between the two so the AGC thinks its in space.

    A similar thing would work here. Using Orbiter’s API, you create the ‘physical’ model of your vehicle. Then you control it using whatever controller you write. The entire time time you can get data and ‘telemetry’ through Orbiter’s API (sorry, no true thermal simulation though for re-entry) and write that to a file for later viewing (and Orbiter also has some built-in functions to do this data-recording). Alternatively you could be ambitious and write the optimization routines into the control code and just have orbiter launch, analyze the data within your add-on, calculate necessary modifications, and iterate ad nausium until you come out with a good trajectory.

    Someone skilled in Orbiter (and there are many within that community) could probably generate a realistic set of simple flight models and then a good control optimization code would just be needed.

    I guarantee it can be done, but I don’t know of anyone who’s done it in an “official” way before. I think our ProjectApollo is the closest thing to legitimate engineering style implementation known to the public eye. That being said, we’ve caught ESA and NASA using Orbiter anonymously in their animations before and called them on it, so they may use it for true simulation in house.

    …but it is good for visuals too. 🙂

    ~Swatch, Your Friendly Neighborhood Timekeeper

  17. Anonymous says:

    You may wish to contact the Aeronautics and Astronautics Department at the University of Washington (Seattle) for the 1997 OB-1 design study. The OB-1 study is on this type of reusable TSTO, boosted flyback vehicle. This study was an influence in a follow up vehicle to the K-1 that Kistler proposed a few years later.


  18. Pingback: Selenian Boondocks » Blog Archive » DARPA Experimental SpacePlane 1 (XS-1) Program

  19. Pete Zaitcev says:

    I have received a notice from Leon McKinney, who wrote that he was the original designer of what he dubbed a “tossback” TSTO. It was published in proceedings of AIAA in 1986 (AIAA-1986-2219). Conceptually it was very similar to K-1, Rossiyanka, and Falcon 9R. The paper included some calculations and diagrams that could be very useful.

  20. Pete,
    Thanks for the heads up! I’ll have to check out that paper.


  21. Pingback: How Relevant is New Shepard to Orbital Launch? | Selenian Boondocks

Leave a Reply

Your email address will not be published. Required fields are marked *