A SSTO research project.

kbd512 · 2019-07-21 09:47:37

GW,

There are plenty of RK code variants available for use with MATLAB. All that's required is a quick Google search. I'm sure Josh already has access to that, but replicating the logic for the equations involved is pretty simple.

Edit:

Why go to the added trouble of vehicle trajectory optimization before first determining whether or not some realistic nozzle design can be accurately modeled for the atmospheric pressure conditions that the nozzle will operate in?

Edit #2:

An over-expanded sea-level-to-vacuum nozzle was designed many moons ago by P&W AJR for the Space Shuttle:

Nozzle Design - by R.A. O'Leary and J. E. Beck, Spring 1992

From the article (in reference to the RS-25's over-expanded nozzle design):

A Rao design resulted in a wall angle of 7.5° at the nozzle exit. By reducing this angle, additional flow turning is produced, and then, an increase in nozzle wall pressure is created. A study was performed by Pratt & Whitney Rocketdyne engineers in which a large number of parabolic-shaped contours, with a variety of different initial wall angles (qmax) and exit wall angles (qe), were analyzed. After careful analysis of these contours, it was determined that a parabolic contour with qmax=37° and qe=5.3° would produce the desired wall pressure increase with the least amount of performance loss. The wall exit pressure was raised 24 percent (from 4.6 psia to 5.7 psia) at a cost of only 0.1 percent in nozzle efficiency. Validation of the design approach was provided by subsequent testing of the SSME which demonstrated that the engine can be throttled to below 80 percent power level at sea level without nozzle flow separation.

Edit #3:

Speaking of RK code for MATLAB:

Runge-Kutta method

Edit #4:

Fluid-Structure Interaction Simulations of Rocket Engine Side Loads

Edit #5 (some past work done on a lightweight actively cooled nozzle extension, so as not to drastically increase the weight of this "bigger and better" nozzle design):

Ceramic Materials for Reusable Liquid Fueled Rocket Engine Combustion Devices

GW Johnson · 2019-07-21 13:13:49

Kbd512:

I never said not to use a Runge-Kutta routine, I just said you no longer need to acquire one anymore. Those were invented to compensate for the slow computers of that time. That problem is now no more.

Myself, I'm too old and obsolete to know anything about modern stuff like MATLAB or any of the software available today. But the simple stuff I mentioned could be done in a spreadsheet or in some computer-programming environment.

As for predicting rocket nozzle performance vs altitude with fixed-bell designs, that is quite accurate right up to the separation point, then it is fundamentally unpredictable once separation occurs. Separation can even be asymmetric across the cross-section, so that not only does thrust fall, it can even vector on you.

I gave an empirical correlation for a separation criterion that is actually rather reliable for conventional bells, curved or conical, doesn't matter. Use that. Once the ambient atmospheric pressure at any altitude exceeds the separation criterion pressure, all bets are off. Otherwise, bets are very good.

All in all, rocket performance prediction is quite simple (ramjet is NOT!!!). Once you have an area ratio Aexit/Athroat locked in, your exit Mach "Mex" is locked in. So also is the expansion pressure ratio Pexit/Pc locked in. My separation criterion converts that to Psep/Pc. If you have a definite Pc, then you have a definite separation backpressure Psep for that design at that Pc. Change Pc, and Psep changes, too.

Unseparated, the nozzle thrust is always F = Pc Athroat CF, where CF = [Pexit/Pc (1 + gamma etanoz Mex^2) - Pamb/Pc]Aexit/Athroat. Everything about a fixed design operating at a definite Pc is utterly fixed in that CF equation, except the ambient backpressure-at-altitude Pamb. But, when Pamb = or > Psep, separation begins and all bets are off. Period.

For a throttleable engine, you vary Pc: lower Pc is reduced thrust, while higher Pc is increased thrust. But, you MUST figure a CF at each and every level of Pc. The vacuum portion of it will be a constant vs altitude locked in by the expansion ratio, so long as separation does not occur.

There is ABSOLUTELY NOTHING about any of that prediction technique that requires your design to be perfectly-expanded at sea level. You can design over-expanded at sea level (just not enough to separate), and get better overall average performance all the way up to vacuum, than a perfectly-expanded at sea level design. Its thrust at sea level will be a tad less than the sea level design, but better than sea level, once past the altitude where it is perfectly expanded. I did such designs all the time, for about 20 years, for a living. That's no different than the RS-25 overexpanded design you mentioned. All of us in the industry did stuff like that.

Point is, I used to do this stuff for a living. I was really good at it, too. Plus I was good at solid propellant ballistics, which are a whole lot more involved than liquid rocket ballistics. I learned it from my friend and colleague W. Ted Brooks (now deceased), who wrote the NASA monograph on it. The stuff Ted and I (and others) did was actually another step beyond what he wrote in that monograph. NASA's understanding of its shuttle SRB's that it bought from Thiokol was (and still is) limited to what was in that monograph. And all the nozzle thrust and thrust coefficient stuff in it applies to both liquids and solids.

For the liquids or the solids, nozzle massflow is w = Pc [CD***] Athroat gc / c*. [Edit update 7-22-19: I added discharge coefficient CD to the nozzle massflow equation. This is an empirical/test item, usually around 0.98 for a good nozzle profile design through the throat. You use it for massflow, but not for expansion ratio, as the boundary layer displacement thickness effects on Athroat and Aexit pretty much cancel-out of the ratio.] c* is calculated from chamber temperature and gas properties, and easily evaluated from static test firings as the most accurate source. For the liquids, total massflow from tank storage is nozzle massflow plus the massflow dumped overboard driving the pumps. For the solids, total massflow IS nozzle massflow. And for either, Isp is thrust/total massflow, not just nozzle massflow.

In the textbooks (and the monograph), c* and Athroat are constants. In the real world, they are not. Athroat varies a bit as the throat erodes (or slags up, or both). Delivered c* is actually a weak empirical power function of Pc (that's even true of theoretical thermochemical data): c* = K Pc^m, where m is an empirical curve fit constant from test data, usually on the order of 0.01. Another way to express that is c* = (c* at 1000 psia) (Pc / 1000 psia)^m. That is as true of liquids as it is solids.

GW

Last edited by GW Johnson (2019-07-22 07:52:50)

JoshNH4H · 2019-07-22 11:34:41

Hey GW,

Yes, that was very helpful, thank you. No doubt I will have more questions when the time comes to throw everything together in one simulation.

On the topic of solvers and Matlab there are no doubt lots and lots of options available out there. Typically I will build my own, which isn't that difficult and guarantees maximum understanding of the task at hand. Normally I start with a rough and dirty method (making no attempt to economize on computational power; In the first run I'll probably set the step size to 0.1 s for maximum accuracy) and then only work in improvements to the code if I find that the runtime is excessive on the equipment at hand. You can get substantial improvements even with simple modifications, like lowering the step size and compensating with a trapezoidal rule and introducing a variable step size. Given the things I find myself modelling in Matlab and the power of a modern PC this is generally adequate. If I were to attack something much more complicated (for example modelling the dynamic response of a rotating skyhook to various peturbations) I'd need to really think about whether my programming ability and PC are up to the task.

Edit: It's multidimensional optimizations that really kill you on computational power. Taking the timing, angle, and duration of the pitchover maneuver as an example you might try to optimize by looking at 100 different possible times, 50 different angles (presumably the typical pitchover is only a few degrees and you might go in increments of 0.25 degree for precision), and 100 different durations. This is 500,000 different scenarios, and even if your initial simulation of a single trajectory can run in 4 seconds the complete optimization will take over three weeks. The best way to speed things up is not by improving your optimization method but by bringing new information to bear. By researching the gravity turn trajectory and thinking about it analytically, hopefully I will be able to reduce the volume of the parameter space substantially. After that I might look to techniques such as a "breadth-first" multistep optimization (check out a few points then hone in on promising regions for subsequent rounds to increase precision) and googling "how to do an efficient numerical optimization"

JoshNH4H · 2019-07-24 17:02:43

kbd512, GW, Bob Clark, et al:

I've been thinking about this some more, and here are some thoughts I've had about how I'm going to simulate the trajectory, in no particular order.

1. The point about doing the pitchover basically at the time of launch makes sense to me; a rocket launch is a fundamentally horizontal affair and time spent travelling vertically is basically wasted impulse.

2. I had initially thought to start the simulation with a zero-time impulse (meaning to start the rocket off with some horizontal velocity relative to the ground at t=0 as a simplifying assumption) and fly a non-lifting trajectory from there, then go back and calculate the correct angle/duration (given a fixed total horizontal impulse, pitchover angle and duration are directly related to each other). Taken together this would turn a 3-dimensional optimization into a 1-dimensional one, and while a finite duration pitchover will result in a slightly lower orbit than a zero-duration pitchover, the difference is small and easily corrected for. This will not work, because if you start with horizontal velocity and no vertical velocity (the zero-time impulse pitchover) your flight angle will be zero.

3. Therefore I need to think more carefully about how to simulate the timing, angle, and duration of the pitchover. Suggestions are appreciated.

4. Strictly (theoretically) speaking, a pitchover maneuver is not required to reach orbit. An object sitting on the equator that is motionless relative to Earth has a velocity of 465 m/s. This corresponds to a two-body newtonian orbital radius around 2 million km (in practice this is interplanetary space, but you'll be orbiting something regardless).

5. A zero lift/zero AOA trajectory is not quite the same thing as firing in the direction of motion in inertial space, because the earth and the atmosphere both rotate, and therefore a craft that is stationary with regards to the atmosphere still has rotation and velocity in inertial space. Presumably the global effect of the difference is small for reasonable trajectories but at the beginning of the trajectory the difference in implied angle is a full 90 degrees.

kbd512 · 2019-07-24 19:32:50

Josh,

1. There's a trade-off between the time spent thrusting vertically to clear the very dense lower atmosphere and the time delay until it performs the gravity turn to begin accelerating through the much thinner upper atmosphere. It makes more sense to at least clear the first 35,000 feet or so, at which time ambient pressure is ~1/3rd of what it is at sea level. Incidentally, Max-Q for STS occurred somewhere around that altitude. More importantly, when the vehicle is heaviest it may not have sufficient structural integrity to sit on a launch rail at an angle. If that's not enough of a reason, there's also aerodynamic heating rates to consider.

2. To determine when / where to optimally perform the gravity turn (part of a mass / thrust optimized ascent trajectory), you have to integrate the result, but you're probably already aware of that.

3. Well...

Optimization of Launch Vehicle Ascent Trajectories with Path Constraints and Coast Arcs

Or

Global Ascent Trajectory Optimization of a Space Plane

Or

Optimal trajectory designs and systems engineering analyses of reusable launch vehicles

If not of that answers the questions you have, NASA has tons of stuff available on NTRS.

4. Theory aside, in actual space flight we use roll-pitch programs created to optimize ascent trajectories.

5. Orbital mechanics errata aside, let's just focus on 250km circular orbits (LEO), launching from KSC (28.5 degrees). Our total dV increment is ~9.5km/s, so that's what this notional launch vehicle needs to achieve. It obviously has to have a reasonable performance margin if something doesn't happen precisely as planned, too. I think a 10% margin is reasonable.

Enjoy your 692 pages of light reading:

Orbital Mechanics for Engineering Students by Howard D. Curtis - Embry-Riddle Aeronautical University

I think it starts discussing your conundrum on Page 34 of the section on the two-body problem.

Maybe we should break this problem up and ask GW for help when we get stuck somewhere. I can't speak for you, but I don't have a PhD in aerospace engineering.

RGClark · 2019-08-11 17:46:28

Thanks. None of those refs is exactly user-friendly. I’m looking for a brief intro to the equations needed to plug in some values. A couple of differential equations would do the trick.

Bob Clark

kbd512 · 2019-08-14 20:47:36

Bob,

Could you elaborate on what you need your differential equations to do?

You're trying to design a rocket engine that a lot of highly educated and experienced people across the world have yet to successfully demonstrate, not likely for lack of trying. I'm not sure that a brief intro would provide what you're after.

Orbital mechanics is not "user-friendly"? Our good friend O.M. is the reason we were able to visit the moon.

GW Johnson · 2019-08-15 11:45:52

When I did missions for "Scout" at LTV in the mid 1970's, fancy trajectory shaping was not an issue. "Scout" pretty well sprinted away at 4+ gees.

All I had to do was run repeat trajectory code runs at launch angles fractionally different from 88-90 degrees until I hit the exact stage 4 burnout conditions I wanted with the payload.

It was a creep-up-on-a-half-interval-search sort of thing. That process could be automated now, but was not back then.

The trajectory shaping thing becomes significant enough to worry about, if you creep away from the pad at low fractional-gee acceleration levels.

As for the compensating nozzle thing, you have to do the compressible fluid mechanics, no way around that. There are two schools of thought relative to the fixed bell baseline: a true sea level design, and a design that is overexpanded but not separated at sea level.

The true sea level design performs best-thrust at sea level, increasing in thrust and Isp with altitude, but a slower rate of increase than any other design.

The not-quite separated design has lower thrust at sea level for the same gas flow rate, peaking instead at some modest altitude in both thrust and Isp. It averages a better Isp across the range of altitudes from sea level to vacuum. That's the one you want to beat with trick compensating designs.

The free expansion designs will look pretty good from sea level to some higher altitude, but get killed by streamline divergence flying out into vacuum.

Variable-geometry bell extensions offer the best thrust and Isp all the way to vacuum, but at a higher weight and a higher complexity (which is lower reliability). It is not designing the extension device, it is designing-in reliability, that is the real problem to solve.

GW

Last edited by GW Johnson (2019-08-15 11:48:07)

Calliban · 2019-11-01 15:07:51

Fake comment above from someone that wants to use your site for free advertising. The comment should be deleted and the user banned.

Calliban · 2019-11-01 15:19:43

From my limited understanding: an SSTO is a non-optimised launch concept, because it means accelerating to orbital velocity a lot of mass that doesn't need to be accelerated to orbital velocity. Most of the dV required to reach orbit is required to achieve orbital velocity. That implies a lot of kinetic energy. It also means that the same engines must be used for takeoff at Earth surface (1bar, low propulsive efficiency) as are used to accelerate to orbital velocity in a vacuum. A better question would be, what is the optimum division of propulsive requirements between the lower and upper stage? Does the lower stage need to achieve enough dV to reach orbital altitude, so that the upper stage is built to focus on attaining orbital velocity? Or would a minimal design be better, whereby the lower stages pushes the upper stage so far as the stratosphere, where the upper stage engines get better expansion ratio; most air resistance is gone and a significant portion of gravity losses are already accounted for?

Last edited by Calliban (2019-11-01 15:21:19)

kbd512 · 2019-11-01 19:14:27

The offending spam posts from users "DonerkR" and "BertonU" have been removed.

SpaceNut,

Please remove these two accounts.

GW Johnson · 2019-11-02 09:54:32

Calliban:

SSTO requires very high Isp and very low inert mass fraction, and nets you a low payload fraction, which tends to drive up cost per ton delivered to LEO. That makes for not a practical mass delivery option, but a niche market item, where cost per ton is not what you care about.

Example: LOX-LH2 Isp ~ 460 sec corresponds to Vex ~ 4.511 km/s. LEO speed is 7.9 km/s. Factored for 5% drag and 5% gravity losses, the mass ratio-effective delta vee is about 8.69 km/s. The mass ratio required is thus exp(dV/Vex) = 6.865. That corresponds to a propellant fraction of 0.854. If the inert fraction were a one-shot min of 5%, your payload fraction is 9.56%, which on the face of it looks attractive.

But, if you try to reuse the vehicle in some way, there is no way in hell its inert fraction will ever be 5%, it'll be closer to 10 or 15%. That puts your payload fraction in the 0-4% range. Not so very attractive.

Plus, the flying thrust/weight ratio is quite different early in the trajectory versus late. You need around 1.5 at least to leave the launch pad, and values under 5 are preferred to where the trajectory has largely bent over as you go exoatmospheric. That's just typical gravity-turn dynamics. From there you would like to use thrust/weight less than 1, so that thrust/weight is under 5 at LEO arrival. Beefing up payloads to resist 10-15 gees instead of just 4-5 gees makes them very much heavier and more costly to build.

But if you are using the same engine or engines that you used at launch, your thrust/weight at LEO will be very high indeed, because of the high propellant fraction (mass ratio). Too many gees is a very rough ride.

Throttling the engines way down is the "solution", but lower chamber pressure is lower Isp, no matter what trick nozzle you might use. Sort of shoots you in the foot to try to do it that way. And in a big way, too. It's a large drop in Isp to run sharply-lower chamber pressure. To have 20% thrust is roughly 20% chamber pressure. The lower Isp rapidly eats up your payload fraction, by driving up the mass ratio requirement.

It just works a lot better, with a gentler ride and engines operating near 100% thrust for "best" Isp, if you stage at the natural breakpoint: where the trajectory has bent over just exoatmospheric. That way your first stage is the thrust/weight 1.5-to-5 item, and your second stage starts out thrust/weight less than 1, and ends up thrust/weight well under 5 at LEO arrival.

This combined set of circumstances is in fact why most modern LEO launchers are two stage. TSTO. If you go beyond LEO to geosynch or elsewhere, that's usually another second stage burn, or else a separate and smaller-thrust-yet third stage. All for very good reasons of operating engines near 100% where Isp is not crippled, and still staying under 4-5 gees max as the propellant burns off.

Ugly little facts of life, but there they are.

GW

Last edited by GW Johnson (2019-11-02 10:03:30)

SpaceNut · 2019-11-02 10:12:09

As always an education on rocket design from GW. to which one should be able to work backwards to how much fuel is needed to have at launch time.

GW Johnson · 2019-11-03 09:41:50

Hi Spacenut. Actually it's more about rocket-powered flight vehicle design than it is about just rocket design. -- GW

SpaceNut · 2020-03-24 10:06:26

repost of the thermal protection needed for a launch vehicles design.

kbd512 wrote:

Quaoar,
The resolution capabilities of current display technologies is far beyond normal human visual acuity and the ability of modern purpose-built graphics chips to process frames of data from cameras is likewise far beyond the ability of humans to detect individual image frames. I think the entire cockpit could be a gigantic wrap-around display that combines inputs from thousands of cameras to show what's outside the vehicle in every direction. Furthermore, the display could have the ability to digitally magnify a scene to show distant objects that the pilots' unaided eyes could never pick up. Add in multi-spectral capability (IR and UV) like the F-35 has and you have a system that's generations beyond the human eye. When the F-35 was developed, high-frame-rate multi-spectral HD cameras were state-of-the-art and strained the capabilities of onboard computers. Today, an iPad can do the number crunching as computing technology marches on. Special UV reflective coatings could provide unmistakable markers for landing and docking maneuvers and provide visual input to assist with manual vehicle control by using the rendering system to compute closure rates. It would still be a backup to automated systems using radio frequency waves, but useful redundancy nonetheless. A Space Shuttle pilot would be envious of the view provided. In the near future, all high speed aircraft or spacecraft will essentially be flying computers with sophisticated sensor suites that monitor everything from position relative to other nearby objects to temperature to background radiation levels.
In the future, I expect we'll need integration of structures that absorb both thermal and mechanical loadings associated with high speed flight. I don't think PICA is a good thermal protection material for hypersonic vehicles if reusability is a consideration. It's an ablative and I'm not sure it's capable of withstanding the mechanical loads associated with a wave rider.
If you haven't already read it, here's a good paper on where heat shield technology is headed:
Ceramic Matrix Composite (CMC) Thermal Protection Systems (TPS) and Hot Structures for Hypersonic Vehicles by David E. Glass of NASA Langley Research Center
That's a pretty good 20,000 foot overview of where we've been and where we're going, as it pertains to reusable high temperature insulation materials and strategies. It's over a decade old, but the only thing that's really happened between now and then is that we've done a lot more testing and refinement of the concepts and materials science involved- to the point that there's now a baseline level of competence with both the space agencies and contractors for fabrication of complex structures of the scale required for crewed spaceflight. Along with NASA and the US Air Force, the various European aerospace research agencies have expended considerable time and effort to develop sharp leading edges fabricated from very sophisticated CMC's that combine the attributes of multiple different materials for reliable and durable heat shielding of hypersonic vehicles for both defense and commercial purposes. The end of the article does a good job of listing the various significant challenges associated with progressing from fabricating a coupon of material that can withstand a blow torch to having a working structure that can withstand the operational environments that a hypersonic vehicle will see.
If it were up to me, I'd keep working on integrated thermal / mechanical CMC aircraft structures and methods to combine those structures with heat pipe systems to, to the extent practical, create nearly isothermal structures that evenly distribute the heating loads over the entirety of the skin of the vehicle while maintaining the mechanical strength required to deal with the aero loads imposed by hypersonic flight.

kbd512 wrote:

Quaoar,
Hmm...
Aluminum does okay with LH2, but stainless fares even better and steel sheet is some of the cheapest stuff you can buy and robotic welding is much faster and therefore cheaper than machining Aluminum plates with giant milling machines and then bending the "Aluminum potato chips" into propellant tank gores, anodizing them to prevent stress corrosion cracking, and then friction stir welding them (which can also be done with thicker gores of stainless). The Centaur upper stage is a non-standard thickness super-thin stainless balloon design that's custom rolled to be thinner and then planed after welding to be thinner still. All that extra work and cost only achieved a 5% decrease in structural mass fraction, but as Tory Bruno said, "it's the Ferrari of upper stages". This transport vehicle will be large enough that standard thickness sheet can be affixed onto jigs / tools and robotically welded with a friction stir welding machine.
If the vehicle had to fly under power at low speeds, then Aluminum is lighter for a given level of strength and therefore more cost effective to operate, which is also the entire reason why Airbus and Boeing switched to CFRP / GFRP- because it's even lighter for a given strength than Aluminum, but again only when temperature limitations are respected. Steel is by far the easiest material to work with and so long as its been properly designed and fabricated it'll provide years of dependable service.
I think you'll still be forced to pay the piper, in terms of structural mass, if you try to make the structure lighter by using a lighter but less heat resistant airframe, in terms of TPS mass to insulate the airframe. Skip the adhesive and go straight to mechanical fasteners. Those work well and the results are highly repeatable. Between maximum practical durability and maximum feasible payload performance, I'd choose maximum practical durability for a reusable spaceship every single time. Propellant is cheap compared to a dead crew and destroyed vehicle. SpaceX has the right idea, I just think it needs a lot of refinement and they're doing that.

elderflower · 2020-04-02 11:05:56

Sabre engine, {if it can be made to work!} will solve many of the SSTO problems.
https://www.reactionengines.co.uk/news/ … conditions

SpaceNut · 2020-04-02 14:20:13

thanks for that reminder as the option of a flyback booster for the eml launch saves a tremendous amounts of funds and lowers the payload to orbit cost....

Mars_B4_Moon · 2021-06-06 06:25:12

China has built a hypersonic Mach 30 wind tunnel https://www.scmp.com/news/china/science … sonic-race

SpaceNut · 2024-05-20 09:23:59

It seems that we can make a ship but to this point I have not seen crew size, safety of crew as protection, time of use durations, consumables for those durations which it can be used for dependent on the crew size.

kbd512 · 2024-05-20 10:38:20

Gelled Liquid Hydrogen: A White Paper - Glenn Research Center

I'm reminded that at 70% CH4 loading of H2 by weight, only results in a 40s Isp decrease- 450s to 410s, but a 253% density improvement, and mixture ratio falls from 6:1 to 4:1, which is much closer to what SpaceX is running in their Raptors (3.6:1 O/F ratio). That bumps Isp by at least 30s over LOX/LCH4 in a Raptor, while reducing the fuel tank volume by 2/3rds over pure LH2. Unfortunately, no such gelled propellant could be used in a Raptor without a major redesign. It also has to be stored at LH2 temperatures, though, which means active cooling and serious insulation mass, probably including a heavy intertank section to separate the oxidizer from the fuel.

New Mars Forums

Announcement

#26 2019-07-21 09:47:37

Re: A SSTO research project.

#27 2019-07-21 13:13:49

Re: A SSTO research project.

#28 2019-07-22 11:34:41

Re: A SSTO research project.

#29 2019-07-24 17:02:43

Re: A SSTO research project.

#30 2019-07-24 19:32:50

Re: A SSTO research project.

#31 2019-08-11 17:46:28

Re: A SSTO research project.

#32 2019-08-14 20:47:36

Re: A SSTO research project.

#33 2019-08-15 11:45:52

Re: A SSTO research project.

#34 2019-11-01 15:07:51

Re: A SSTO research project.

#35 2019-11-01 15:19:43

Re: A SSTO research project.

#36 2019-11-01 19:14:27

Re: A SSTO research project.

#37 2019-11-02 09:54:32

Re: A SSTO research project.

#38 2019-11-02 10:12:09

Re: A SSTO research project.

#39 2019-11-03 09:41:50

Re: A SSTO research project.

#40 2020-03-24 10:06:26

Re: A SSTO research project.

#41 2020-04-02 11:05:56

Re: A SSTO research project.

#42 2020-04-02 14:20:13

Re: A SSTO research project.

#43 2021-06-06 06:25:12

Re: A SSTO research project.

#44 2024-05-20 09:23:59

Re: A SSTO research project.

#45 2024-05-20 10:38:20

Re: A SSTO research project.

Board footer