Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Solar Thermal Rocket Engine

  1. Jul 1, 2013 #1
    Dear visionary inventors, megalomaniac engineers and audacious explorers,

    Chemical rocket engines aren't up to our desire to hop in the Solar system: go quickly to Mars, deviate Earth-threatening objects, and so many more missions. We need a higher ejection speed to save propellant mass, but this takes more energy than chemical reactions bring. One possibility is to tap Sunlight instead of transporting the energy.

    I suggest - as others did - to heat the propellant with Sunlight; this thread shall describe it now and here with more details. The conversion to electricity first enables even higher ejection speeds, but direct heating is energy-efficient, so for a long weak thrust, the collector area is feasible - smaller and of cheaper materials than Solar cells or even a heat sink.

    Material sublimation limits thermal designs to <3000K, so only hydrogen improves the ejection speed over a combustion. My plan is to dissociate a part of this hydrogen to increase the ejection speed a bit more.


    The heater that catches concentrated Sunlight and transfers heat to hydrogen is of tungsten alloy. At 2800K =2527°C, sublimation thins it by 45µm in 14 days, from Plansee's doc.

    30mbar in the heating chamber let 23% of injected H2 split to atomic H* for performance. Expansion to 1Pa in the nozzle leaves 15.6MJ/kg from 92.8MJ/kg in the chamber, for isp = 1267s = 12428m/s.

    Mean 2800K is already a lot, as sublimation is very sensitive to it. The nozzle can't grow much because the mean free path is already ~10mm. A lower chamber pressure dissociates more hydrogen and improves the ejection speed, but needs much more heating power, as the nozzle gets inefficient - recombining hydrogen takes a big expansion. But more pressure bring during some flight sequences more thrust traded against ejection speed; for instance, 0.8bar and 2093K from the same Sunlight concentrator push *2.1 times stronger with isp=800s.

    Marc Schaefer, aka Enthalpy

    Attached Files:

  2. jcsd
  3. Jul 1, 2013 #2
    Much Sunlight is needed, so the launcher's fairing limits the concentrators, say to D=4.4m, and most uses will have several concentrators. They can consist of metal honeycomb, similar to satellite antennas, with an improved reflecting surface. One chamber per concentrator gives redundancy and eases the orientation and the ground tests.

    D=4.4m provides near Earth 16.6kW if 80% are used, enough for 195mg/s hydrogen to push 2.4N.

    The chamber's light input is hot; with D=38mm its radiation to 2π sr would reemit already 19% of the incoming power. The optics lets the Sun's image fit through the inlet, needing a strong convergence.

    The nozzle must be oriented. A sketched example follows, where rotating mirrors add no aberration; optics engineers will make it better.

    Stacking many mirrors in a launcher's fairing constraints such designs. And with several engines operating around a craft, the control must avoid collisions, as the jet impact must be hot and corrosive.

    Marc Schaefer, aka Enthalpy

    Attached Files:

  4. Jul 1, 2013 #3
    The attached sketch suggests how the chamber could be made, I believe. It's a bit exotic in 2013, as Solar thermal rocket engines still don't exist, so more brainwork is needed between this description and an operating engine.

    It's made primarily of tungsten, probably alloyed with some rhenium, though a few parts are less hot and could be of ceramic, alloyed Ta, Nb, Ni... Spark machining shall make the many thin deep shapes, and an electron beam weld the parts. Nothing exorbitant for a spacecraft.

    The ID=40mm chamber receives a 38mm Sun's image (20kW and 18MW/m2 from a D=4.4m concentrator), with 1:1.85 light beam convergence. It absorbs light at the heater's inner wall and transfers it by conduction to hydrogen in channels, 1mm wide and 6mm deep, separated by 1mm fins. Alloyed tungsten (~80W/m/K) shall drop 63µK*m2/W and hydrogen (~0.43W/m/K near 2000K) <37µK*m2/W.

    The heater's first 10mm absorb (a~0.40) ~1.3MW/m2, dropping only 130K where hydrogen is still luke-warm. The decreasing direct illumination is supplemented at depth by reflected light. Where power density has dropped enough, helical fins in the cavity increase the absorption, a converging cone as well. Hand computations are imprecise, so lengths are not mm accurate on the sketch. The heater's diameter and length can apparently shrink further; thicker walls seem possible.

    Only the cavity's bottom is at 2800K, and heat transfer must be slower there. The walls conduct only 830W axially for 1000K/100mm, making them warm slightly more upstream; the full wall section and strength holds the heater and nozzle by the inlet. Radiation transfers heat backwards more efficiently than conduction, but due to hydrogen flow, I expect the cavity to radiate less than a blackbody.

    A ruminator surrounds the incoming light cone to absorb much of the light emitted through the cavity's inlet and transfer it to hydrogen: of 4380W from a 40mm blackbody, an infinite cone would catch 3320W, but 340W less if truncated to D=200+38mm (sketched shorter). The ruminator may reach the tertiary mirror on one side but shall not heat it. It also protects the engine from concentrated light, as long as hydrogen flows.

    The reflective (a=0.02) regenerator surrounds the heater and the ruminator and feeds the leaked heat to hydrogen. Its channels can be holes, trenches... If 120mm heater length are hot, the transferred 1470W heat hydrogen by 520K, and the ruminator by 1060K. The whole chamber holds by the colder hydrogen inlet, and the regenerator radiates negligible heat. The regenerator would better use an alloy lighter than tungsten, as it seems possible.

    The 22mm nozzle throat radiates 1320W, plus some at the upper divergent; they could be widely saved by an absorbing divergent that pre-heats hydrogen.

    Marc Schaefer, aka Enthalpy

    Attached Files:

  5. Jul 1, 2013 #4
    To bring a payload from Low-Earth- (Leo) to a Geosynchronous Orbit (Gso), a Solar thermal rocket would use its small push all the way to save time, but this takes more performance than the usual very elliptical transfer (Gto).

    From an equatorial Leo (sea launch, Alcantara...) the required Delta-V is then the speed difference between the low and the high orbit, if I didn't botch it. Starting from 7675m/s at 6366+400km to avoid drag, the transfer stage must add 4600m/s to achieve the 3075m/s at 6366+35798km. This costs 741m/s more than the Hohmann Gto.

    From other launch sites, the orbit's inclination must be cancelled. The very elliptic Gto does it at apogee where speed is small, but the continuous push lacks this possibility. Fortunately, isp=1267s absorbs this as well.

    Kourou reaches 5.2° inclination at Leo. To estimate the added cost, I keep the inclination over the first 2600m/s, and compensate it over the last 2000m/s needed for altitude; the side speed to cancel is 462m/s (begin) to 278m/s (if done at Gso), so I take mean 370m/s. The engines shall push flat for 2*90° per orbit, and for 2*90°, tilted as 1,11*370m/s * sin = 1000m/s (1,11 because the tilted push extends 45+45° from the optimum point). When tilted, only 91% of the thrust is equatorial, so these 1000m/s cost 1097m/s performance, and the overhead nears 97m/s, totalling 4697m/s to Gso. A Gto would waste 15m/s from Kourou's latitude.

    Cape Canaveral reaches 28.5° (and Tanegashima 30°), giving 3779m/s side speed at 400km and 1505m/s at height, for mean 2642m/s. Here I tilt the thrust as a cosine function over one orbit. At maximum tilt, the relative components of thrust are S and C, kept for all orbits of the transfer, which is not optimum. Only 0.5*S and nearly (1+C)/2 act as a mean, so side 2642m/s versus 4400m/s along the path leave 73.5% efficiency: the total cost nears 5986m/s and the overhead 1586m/s. This wastes even more performance than de-inclining at final altitude but saves time.

    The optimum is obviously elsewhere - someone with a liking for it shall investigate. I take 5700m/s.

    Marc Schaefer, aka Enthalpy
  6. Jul 1, 2013 #5
    A Falcon-9 shall illustrate the transfer from Leo to Gso (log in to magnify the sketch). Starting from Cape Canaveral, the launcher puts 10,0t on a 28,5° 400km orbit. The Falcon may need reinforcement for the longer fairing.

    Minus the adapter, the transfer stage starts with 9350kg. To provide 5700m/s as estimated previously, it ejects 3440kg hydrogen at isp=1267s. 20 days thrust (plus some 5 days eclipse) take 10 Solar thermal engines with 4.6m concentrators. Not that huge.

    The balloon tank of thin steel, foam and multilayer insulation weighs 143kg. Polymer straps hold it to a truss of welded AA7020 weighing 205kg that links to the launcher and the payload. A cryocooler keeps the hydrogen liquid.

    Each engine weighs 100kg, of which 50kg are the concentrator and 20kg the chamber, using nickel or niobium rather than tungsten where possible.

    300kg of sensors, datacomms, control and unaccounted items leave a payload of 4262kg in Gso. That's roughly twice the capacity of chemical stages.


    Here the Solar thermal stage is expended at each launch and continues to a park orbit. It can also come back to Leo, what ESA calls a "tugboat":
    • Reusing it saves launch mass, even though the way back needs some hydrogen;
    • It can bring a satellite down to Leo for repair;
    • A lighter launch mission can bring extra hydrogen for a following overweight mission;
    • The launcher can bring the hydrogen and the satellite separately. This puts in Gso the full launcher's Leo capacity;
    • A flexible long-range vehicle between Leo and Gso can bring resupply or remote repairs to several satellites, push aside the lost ones...

    Marc Schaefer, aka Enthalpy

    Attached Files:

  7. Jul 6, 2013 #6
    Thanks for that. It seems the only thing preventing a solar thermal propulsion stage is the lack of lightweight mirrors or lenses for square kilometer scale collectors. This would seem to be an easier task than getting light weight fission reactors or plasma drives so NASA should be investigating this as well.
    An advantage of the solar thermal engine is you can throw anything in it as the reaction mass. You could rendezvous with an asteroid and use its mass as propellant to direct it to Mars for a crewed mission. Another advantage is that it could be used to redirect possible Earth threatening asteroids.

    Bob Clark
  8. Jul 8, 2013 #7
    Hi RGClark, nice to meet you here!

    The biggest advantage I see to the Solar thermal engine is its moderate collector area. It results from:
    - The intermediate isp (specific impulse, similar to the ejection speed). More than a chemical rocket, enabling high-energy missions (I haven't detailed them here), but less than ionic engines for instance. This accepts less Sunlight for the same thrust;
    - The excellent conversion (60%) of light power to kinetic energy of the ejected gas. No intermediate electricity, no mandatory ionization nor conduction losses here.

    For sure, it needs some collection area and the thrust is small, but sufficient to take-off a 10km asteroid for instance. The Leo-to-Gso is sketched on scale here, with its ten D=4.6m concentrators. For instance a nuclear reactor for electricity would need a cold sink bigger than my engine's concentrators. And as I need only a heat exchanger with no other material property, the higher temperature improves the isp (1267s) over a thermal nuclear rocket.

    The thrust also suffices for an efficient Hohmann transfer to Mercury, and even to act only at the periapsis there to lower the apoapsis. Be there in four months!
  9. Jul 11, 2013 #8
    I have such mission designs in the pipe, sure... I had investigated them in 2010 but with a wrong isp. Now with isp=1267 by a more detailed design, they still look interesting, including a Mars scenario where both the transit and the stay are short. Preset much in Martian orbit.

    I considered deflecting an Earth-threatening object by scooping its material for ejection through the Solar engine, but it would work only with clean icy material, not stone nor metal. Evaporating from a few decameters the material in situ to create thrust is safer: Sunlight concentrators, Sunlight-powered laser. Or a direct impact of powdery material brought by the spacecraft.
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook