Jump to content

Extruded Rocket Structure


Recommended Posts

Physical and chemical properties

Elemental magnesium is a fairly strong, silvery-white, light-weight metal (two thirds the density of aluminium). It tarnishes slightly when exposed to air, although unlike the alkali metals, storage in an oxygen-free environment is unnecessary because magnesium is protected by a thin layer of oxide that is fairly impermeable and difficult to remove. Like its lower periodic table group neighbor calcium, magnesium reacts with water at room temperature, though it reacts much more slowly than calcium. When submerged in water, hydrogen bubbles will almost unnoticeably begin to form on the surface of the metal, though if powdered it will react much more rapidly. The reaction will occur faster with higher temperatures (see precautions). Magnesium's ability to react with water can be harnessed to produce energy and run a magnesium-based engine. Magnesium also reacts exothermically with most acids, such as hydrochloric acid (HCl). As with aluminium, zinc and many other metals, the reaction with hydrochloric acid produces the chloride of the metal and releases hydrogen gas.

Magnesium is a highly flammable metal, but while it is easy to ignite when powdered or shaved into thin strips, it is difficult to ignite in mass or bulk. Once ignited, it is difficult to extinguish, being able to burn in nitrogen (forming magnesium nitride), carbon dioxide (forming magnesium oxide and carbon) and water (forming magnesium oxide and hydrogen). This property was used in incendiary weapons used in the firebombing of cities in World War II, the only practical civil defense being to smother a burning flare under dry sand to exclude the atmosphere. On burning in air, magnesium produces a brilliant white light which includes strong ultraviolet. Thus magnesium powder (flash powder) was used as a source of illumination in the early days of photography. Later, magnesium ribbon was used in electrically ignited flash bulbs. Magnesium powder is used in the manufacture of fireworks and marine flares where a brilliant white light is required. Flame temperatures of magnesium and magnesium alloys can reach 3,100 °C (3,370 K; 5,610 °F),[7] although flame height above the burning metal is usually less than 300 mm (12 in).[8] Magnesium may be used as an ignition source for thermite, an otherwise difficult to ignite mixture of aluminium and iron oxide powder.

Magnesium compounds are typically white crystals. Most are soluble in water, providing the sour-tasting magnesium ion Mg2+. Small amounts of dissolved magnesium ion contribute to the tartness and taste of natural waters. Magnesium ion in large amounts is an ionic laxative, and magnesium sulfate (common name: Epsom salt) is sometimes used for this purpose. So-called "milk of magnesia" is a water suspension of one of the few insoluble magnesium compounds, magnesium hydroxide. The undissolved particles give rise to its appearance and name. Milk of magnesia is a mild base commonly used as an antacid, which has some

alloy is much more safe

Edited by ox1111
Link to comment
Share on other sites

Ox1111, why shouldn't you try by yourself before spreading the legend of burning magnesium?


I wrote "I did try". What sort of value should have any text copied from people who didn't?


For your information, iron, aluminium, titanium burn more easily than magnesium - if in powder form, just like magnesium. This has nothing to do with alloying.


I had to use a blowtorch and melt the magnesium ingot before it caught fire if fed with oxygen. Without oxygen it went out spontaneously. That's fact.


Ah, and flashes were made of zirconium.

Edited by Enthalpy
Link to comment
Share on other sites

Hey, its your rocket. I can tell you that solid mag. is not as much of a problem as powder and flames are not as much of a problem as sparks. I can tell you building costs go way up and people have to be trained and cert in working with mag. I doubt you will find any welders cert. and you will need clean air welding mask I am sure by osha. You might require semiannual blood tests as well. I have a block of mag with a flint on it in my first aid kit. You shave some with a knife and use the flint and it goes up quit nice. Another issue you may have is any hardware you use or any other material that contacts the mag. may react, esp. in rain or high humidity.

Link to comment
Share on other sites

"I can tell you building costs go way up and people have to be trained and cert in working with mag. I doubt you will find any welders cert. and you will need clean air welding mask I am sure by osha. You might require semiannual blood tests as well."


=> Up to now you can copy-paste text from the Net. By the way, this seam length would be made by a machine, just as for aluminium tanks. This should be clear to everyone who has welding experience and has already seen a full-size rocket.


"I have a block of mag with a flint on it in my first aid kit. You shave some with a knife and use the flint and it goes up quit nice."


=> And the block does not catch fire. You can do the same with a block of iron and a grinding wheel.




Beyond magnesium extrusion, an additional possiblity would be pultrusion. Some companies claim to include long carbon fabric in the profile pultruded through the matrix. This would make the walls stiffer, stronger and lighter than aluminium and magnesium - provided, of course, that the profile show a reliable behaviour. Assembling them into a tank wall or interstage section required generous overlap of the fabric.


Or is it better to weave or pultrude the hollow profiles making a core and wind the inner and outer sheets of a hollow wall? Triangular profiles can be woven and constitute the core's zigzag, possibly with a foam layer between them.


I doubt a lighter tank obtained this way makes a launcher cheaper than metal extrusion does... An extruding machine costs around 100k€ and produces easily one launcher a month; many heads in parallel can make the weld joints.


Marc Schaefer, aka Enthalpy



This is how I imagine a hollow wall of composite material if not pultruded at once but assembled, with a core of elements woven or pultruded in advance:




Here the core walls can be a sandwich with foam (the skins as well), which allow them to be much thicker than a honeycomb yet strong. I've sketched one layer of prisms, but there could be more.


Foam between the prisms is what prevents delamination... But tension in the outer skin, resulting from the winding process, can help. A syntactic foam is heavier but stronger than PU foam. Maybe the foam can be pultruded with the prism at once.


With carbon fibre for instance, the wall must be nicely stiff. At a rocket tank, the inner skin would probably hold the pressure and the other ones give stiffness. To make a single-stage-to-orbit or an upper stage, the effort could pay.


A flat panel would have skins laid of fabric or crossed monodimensional prepreg, rather than wound.


Not sure how new this is... Omega stiffeners are well known, maybe the foam is new.

Marc Schaefer, aka Enthalpy



About the same assembling method can produce tubes, with a core composed of prisms made in advance, and skins wound or woven around, possibly with foam layers:




The gained stiffness enables tubes much voider than a bare skin or even a simple sandwich allow. I've sketched six prisms, but more layers and other angles are possible (and more meaningful). Could be nice for the masts of a Solar sail, the external truss of a rocket with balloon tanks...


Marc Schaefer, aka Enthalpy

Link to comment
Share on other sites

How about just a smooth inner and outer skin with instead of a truss shape inside, an injection of high density polyurathane foam?

ONE message before, LAST paragraph: "The gained stiffness enables tubes much voider than a bare skin or even a simple sandwich allow."

Link to comment
Share on other sites

I would imagine the shapes you show to be stronger and lighter. Also very costly to build, thou this is not to say not practical. Beside cost, complex shapes also have the disadvantage of complexity itself, more to fail. This is untested, but an idea you are welcome to. If you fill the cavity with shapes, like say balls, light wieght balls of super thin plastic like the type in a ball pit for kids and then fill with epoxy or super high density urethane too form a bubble honeycomb similar to the structure of bird bones.

Link to comment
Share on other sites

Light tanks and good engines go to geosynchroneous orbit in just two stages (click the drawing to magnify).

Walls are of aluminium extrusion, with a=45° and the extrusion horizontal. Magnesium would improve.



At the one-Vinci stage, the long engine shall weigh 400kg, adapters 200kg, avionics 200kg, and the supported balloon tanks 40kg per ton of propellants. Notice the fairing covers them.

The roll and injection verniers get hydrogen and oxygen at 1.5b from the tanks, burn them at 1.0b and expand to D=0.3m to achieve 4*500N and 376s, losing no efficiency and adding just 20kg. Nearly the ones from DLR.

The empty stage carries 21.5t of propellants and weighs 1680kg. Maybe an ESC-B.


At the three-Vinci stage, the short engine shall weigh 300kg. Four or six chambers could in the future share one bigger turbopump and one set of actuators.

The walls (1595kg) have B=19mm t1=t2=1.5mm of AA6005A, the dropped interstage (727kg) as well. Hydrogen heads (195kg) are 1mm AA7022 ellipsoids with 2r=3.6m. The oxygen tank (205kg) is an ellipsoid (2R=4.8m, 2r=3.4m) of AA7022, top 1mm and bottom 1.5mm. It holds at the wall by two cones (61kg) of 1mm AA7022. A truss (45kg) of AA7022 tubes holds the engines to the cone - or maybe to the walls. Foam weighs 72kg.

Carbon and ultra-thin steel may save mass, but aluminium is cheap and sound. Everything can be welded here.

Oxygen above hydrogen might have saved wall mass as the fairing would cover it.

The supported tanks weigh 2173kg and carry 53.8t of propellants, or 40kg/t. The stage weighs 3273kg empty, or 61kg/t.


At the RD-180 stage, the oxygen walls (1752kg) have t1=t2=2mm of AA6082, others (1994kg) have B=19mm t1=t2=1.5mm of AA6005A to resist 6MN*m bending plus a factor of 1.5.

Heads (652kg total) are ellipsoids (2R=5.4m, 2r=3.6m), 2mm of AA7022 below the oxygen, 1.5mm elsewhere.

The structural walls and tanks carry 259t and weigh 4437kg or 17kg/t.

The engine (5850kg), its truss (230kg) and foam (39kg) bring the empty stage to 11.1t or 43kg/t.

The fairing (3400kg) is dropped with the first stage in the performance evaluation. Aluminium extrusion that light would have needed t1=t2=1mm, so this is probably standard composite.

If an RD-180 equivalent is designed, it could usefully throttle more deeply, have wider nozzles, burn Pmdeta or Farnesane, have a turbine and gas generator of molybdenum.

Marc Schaefer, aka Enthalpy

Link to comment
Share on other sites

  • 2 weeks later...

Von Braun used internal truss when he was young. It would still work, but there are limits:


- A strong skin is nice to fly quickly through air. To resist impacts of rain, birds, chunks of ice detaching from the cold tanks... And to avoid air flow instabilities at the skin. It's already few mm thin, saving more is difficult.

- A tank must be pushed at its base, over the whole lower head. This combines less well with a thrust concentrated in a truss.

- The truss and the skin must have the same temperature at every time. Nothing damning, but one difficulty more.


What does exist: upper stages (modern Centaur) that are fully enclosed in the payload's fairing at low altitude. They have an external truss, and the tanks are ultra-thin "balloon tanks" of best steel, enough to hold the small pressure but not transmit any thrust. The combination is light.

The older Centaur described there http://en.wikipedia.org/wiki/Centaur_(rocket_stage) stands up thanks to internal pressure; this risky design is oldfashioned now.


On a lower stage, I considered a structural oxygen tank surrounded by balloon hydrogen tanks that lift nothing beyond their contents, but frankly, I dislike such thin skins exposed to the wind. With extruded aluminium I regularly compute 30-40kg/t with hydrogen+oxygen, already excellent, and with outstanding stiffness and resistance.

Edited by Enthalpy
Link to comment
Share on other sites

How applicable are developments in carbon fiber like this in rocket body creation:




Or this article on cheaper stronger aerogels:




A strong aerogel would seem a good way to combine the required insulation and lightweight structure when samwidged between two carbon fiber skins.

Link to comment
Share on other sites

Interesting! Carbon fibres are anyway the material of choice to make strong, stiff and light.


Aluminium extrusion has the advantage of classical low-tech. Cheap investment, since extrusion on demand can even be contracted to a supplier, and welding needs a machine meanwhile banal.


Carbon fibres were proposed by MAN GmbH to replace a difficult steel process for the casings of Ariane V's solid boosters. Presently, forged rings of austenitic steel are brutally cold-rolled on a specially-made machine to make high-strength thinner cylinders whose welded ends are thicker. Carbon spinning would have been cheaper and lighter, but wasn't accepted for Ariane V. It's used on Vega now and was already common on high-pressure liquid tanks that transmitted no thrust.


Then, one has to check if the desired shape and strength can be achieved: fibres work best if spun, and aligned in the stress directions, up to the locations where forces apply. Easy for a pressure vessel, less so for a tank with moderate pressure but compressive axial load. Here stiffness demands a non-uniform thickness, or hollow materials.


This is the reason for the sandwich walls and rods described here on 29 December 2012. If the new laser process can help it, it's of course highly welcome!

Link to comment
Share on other sites

  • 6 months later...

Ariane V's upper Esc-B is rumoured at 5950kg. This shattering dry mass reduces GTO (gesosynchronous transfer orbit) capacity, hampers GSO missions, nearly precludes transfers to Jupiter and beyond. Here's my alternative.



This design has no structural tanks of extruded material, as it would complicate the excellent insulation that keeps hydrogen until the apogee burn. Polymer belts hold balloon tanks in a structural truss. A superinsulation that resists the wind would enable other designs.

The elliptic hydrogen tank is of 200µm brazed maraging steel, plus 20mm foam so hydrogen warms by 0.6K after 10mn in air, and 5 plies of multilayer insulation for 0.07K after 10h in vacuum. It weighs 180kg with polymer belts. Already 10 plies would enable a week-long Moon mission; a 200 days Mars mission better has active cooling.

Two torus for the oxygen shorten and lighten the stage. With 100µm steel, 10mm foam and 3 plies, plus the belts, they sum 123kg. Electric motors and screws can adjust some belts against thermal expansion.

The outer part of the truss, similar at each level to the Soyuz interstage with 12 nodes, shall break at 6MN*m or 4.4MN. I couldn't check by hand the truss' global bucking, but only one nodes level (near the equator of the hydrogen tank) makes a straight cylinder. Welded (screwed at some places) AA7022 tubes make it, a section example being D100m*e2mm, summing 557kg.

The inner part of the truss holds half of 24t oxygen and shall break at 2*5.5g; it stiffens also the outer part. AA7022 tubes there range from D70mm*e1mm to D80mm*e1.1mm, summing 163kg. 11kg of similar stuff hold the Vinci.

The bistage conical adapter to an 8t payload is built the same way and weighs 35kg; heavier individual payloads to Leo or Gto have a special adapter.

A shell on the truss protects the tanks from wind. Its sandwich panels have guessed 193g/m2 skins of aramide composite and 10mm foam thickened to 40mm ribs on 20% of the area. D5.4m*h6.0m weigh 121kg.

The long Vinci is estimated at 280kg for want of manufacturer data, its actuators at 10kg. Vernier and roll engines (20kg) could burn gaseous hydrogen and oxygen at 1bar, or better, have electric pumps to burn the liquids at 25bar; as multiburn apogee engines, they would then outperform slightly the Vinci. They save separate tanks. Adequate chambers and nozzles exist at DLR.

The separation belts, some auxiliary tanks and pipes account for 100kg; sensors, transmissions, control and steering for 300kg; unlisted items for 100kg. The dry stage weighs 2000kg, or 71kg per ton of propellants. With a second Vinci and dropable nozzle inserts for atmospheric operation you have a single-stage-to-orbit.

Tanks holding more than here 28.2t would be nice, since this stage puts some 9.5t in GSO and 4.8t in transfer to Jupiter. Electric engines and screws can hold the shell panels, to drop 121kg just after the fairing, gaining as much payload. Maraging tubes would make a lighter (-150kg?) truss, but only if cut by laser or water jet to the shape of small frameworks; or make the tubes of carbon composite.

Marc Schaefer, aka Enthalpy

Link to comment
Share on other sites

I hope this will be on-topic: Would the extruded panels with corrugations necessarily require an 'outer' skin? Oriented in line with the thrust axis, I would think the corrugations would provide adequate stiffness against panel flutter. For thermal insulation, half of the tank circumference is naked to the air but could easily be protected with spray-on foam. This would also tend to 'smooth' the surface of the tanks and reduce friction losses.

Link to comment
Share on other sites

A light shell meant to cover a stage truss, as suggested in the Esc-B message, can look like this, depicted here with a release mechanism:



A sandwich of foam and aramide fibre composite is cheap, light and resilient to shocks - but I have nothing against honeycomb. Supersonic flight demands stiffness. The foam is thicker only at ribs to save weight.

To throw away the panels composing the shell, geared electric motors turn the screws that hold the panels. This worked well on several projects, and I prefer it to pyro devices: testable many times with the components that will fly, safe, easy storage and export.

Throwaway panels should be bigger than individual holes in the rocket's truss.

As alternatives, the payload fairing could be lengthened to cover the upper stage as well; this often reduces the bending moments on this stage and enlights it. Or a separate stage's shell can use the fairing's material and release mechanism.

When the stage uses satellite junk like multilayer insulation, its shell can be thrown away at the same altitude as the fairing, say just after.

Marc Schaefer, aka Enthalpy

Link to comment
Share on other sites

Forces are diffuse at Ariane's Esc-B upper stage, so a light truss is easier to design with a lighter material than with a stronger one of equivalent strength-to-density ratio, hence the choice of aluminium AA7022.


Magnesium alloy improves a bit over aluminium. AZ80A in T5 temper offers:

1800kg/m3, E=45GPa, 0,2% proof stress = 275MPa (tension) and 240MPa (compression).

This gives about the same specific strength as AA7022. The AZ80A can be extruded, machined, welded; turn the tubes to the optimum thickness, leaving extra material fot the weaker weld seam (if equivalent to F temper: tension 250MPa, compression unclear). Magnesium doesn't burn, as know people who tried.


The magnesium truss shape can keep aluminium's design; just put more thickness, and adapt the diameters a little bit. Because magnesium's density fits this particular task better, the design can fully exploit the specific strength, which saves 100kg roughly over AA7022.




A truss of stronger but denser materials is difficult to design here.


Titanium alloy Ti-Al6V4 would provide 4430kg/m3, 113GPa and 1030MPa (hardened condition) to 828MPa (annealed, maybe attainable at good weld seams). This exceeds aluminium and magnesium by far, almost equalling steel, but demands very short and narrow truss elements which make a cylindric truss too thin and make it weak against buckling through elliptic deformation. As it looks, long wide truss elements must themselve be trusses - not obvious to design nor produce. Isogrid tubes as truss elements are an other option.


NiCoMoTi 18-9-5 steel's specific strength equals Ti-Al6V4, and 18-12-5 outperforms it, but it's even harder to exploit. Improving over AZ80A would be a performance. Also, it demands a heat treatment after welding, difficult at this size - or an other, heavier assembling method.




Graphite composite is an obvious choice for light and strong design, excellent here as well. Exploiting the full strength would be about as difficult as for titanium, but even though the limit won't be reached, gaining mass is easy, because the material outperforms the others so much.


Marc Schaefer, aka Enthalpy

Link to comment
Share on other sites

Would the extruded panels with corrugations necessarily require an 'outer' skin? [...]


Thanks for your interest!


You're perfectly right, the extruded aluminium panels make a skin strong enough to be exposed to the wind, and naked foam is used on launchers to insulate cold tanks. I should have made it clearer.


The throwaway shell of message #40 is only for stages sensitive to wind, and was meant for the Esc-B variant suggested in message #38. This one has naked satellite junk like multilayer insulation, very useful to store propellants until the apogee kick(s) while foam would let evaporate some propellant costly brought to orbit, but the better insulation consists of 6µm thin plastic films superimposed and demands a protection.


Having foam, then multilayer insulation (MLI), then again foam is surprising, is it? We might perhaps have the multilayer insulation directly at the tanks, then only one layer of foam, if possible thrown away, but I see no good design nor operation then. At hydrogen's 20K, only helium and hydrogen are still liquid - air gases liquefy or freeze. The MLI at 20K would require vacuum (I believe this exists for small helium tanks, but is heavy and very special stuff) or be swept with hydrogen or helium all the time that the tank contains hydrogen. This is a difficult and costly operational burden; I prefer to keep 64+41kg foam at the tanks (this foam would be uneasy to throw away through the truss anyway) for simple and safer launch operations.


With the foam directly on the tanks, once the hydrogen replenishment is stopped and the tank pressurized, one has 30 min in the atmosphere before a vapour valve must be opened. That's comfort. Once in vacuum, 8+6kg MLI give one week with no propellant evaporated. Foam alone would weigh a tonne to keep the propellants until the apogee.


At launchers where hydrogen and oxygen tanks have a common head, a polymer honeycomb is said to work well. The gas in the combs freezes as hydrogen fills the tank, and they say this material keeps its shape and stays airtight; then vacuum in the combs insulates the tank. I may be wrong to distrust airtight polymers; then this honeycomb would be an alternative to foam.

Edited by Enthalpy
Link to comment
Share on other sites

A launcher would separate the panels covering a stage above the atmosphere and during a strong acceleration. These favourable conditions permit to topple the panels to separate them. This script is cleaner, and it also tolerates better a few malfunctioning motors.


post-53915-0-58230700-1377915914.png post-53915-0-50778100-1377915923.png

A pin defines the vertical and lateral position at the panel's top center B, two geared motors and screws the radial one at the top corners A, two pins at the lower corners C the radial ones plus at only one corner the lateral one - the other hole at the stage's frame being wide.

As the motors release the upper corners, springs (not sketched) push the panel's top outwards and the upper pin disengages, while the lower pins are still engaged, toppling the panel with the help of the acceleration. The panel falls to contact at D and, because the panel is curved, its rotation raises the lower corners and disengages the lower pins. The panel separates completely.

This worked properly at Mach 1.3 through the lower atmosphere. I had B and A functions at the top center with a wide flat contact, and only one C at the lower center, further inwards. Here with big panels of limited stiffness, I prefer more A and C despite this can strain the panel; one can even design the curvature to apply the panel against the frame at B also - or have a single A at the center. A's and C's can be at 1/5 and 4/5 of the width for instance.

Shape memory actuators - much smaller than for payload separation belts - are possible replacements for the geared motors and screws. Or even, wires burnt by an electric current.

Marc Schaefer, aka Enthalpy

Link to comment
Share on other sites

Trusses of shorter elements exploit slightly better the strength of a material of given stiffness. At the Esc-B sketch of message #38, two stages of 0.65m height would fit at each torus, and three around the ellipsoid's equator, with 2*30 tubes at each stage. This fits the capabilities of magnesium AZ80A; it also exploits better the aluminium AA7022, which is then as good as magnesium and more comfortable. Titanium is still uninteresting at this scale.

Global buckling of the truss puts a limit on short (hence narrow) truss elements. I estimate it by assimilation with a uniform cylindre of same stiffness per length unit: only F=0.68*e2*E according to my experiment - simple theories are knowlingly false. I could not misuse Gerrit Wolsink's free "Framework" software to determine a truss' buckling; I plan to try with Frame3DD, and would gladly read proposals.


1600 tube ends take long to weld. Turned tubes and separate milled nodes ease it and give extra thickness to the weaker weld seam.




The tubes are turned to the proper thickness, leaving more near the seam. The inner diameter can be machined more accurate at the ends. The length is very accurate, the end faces as well.

CNC milled nodes are affordable, light, and fit accurately the tubes' directions - the sketch is simplified, there would be six non-planar directions in a simple truss. They accept tubes just turned. Maybe some material at the center, plus a hole for one big screw, can hold the part when machining it.

For Tungsten Inert Gas (TIG) welding, the groove at the sketch's left is a big help: it protects the back side without argon there, defines well the weld depth, and a shallow shoulder places accurately the tube's end.

Linear Friction Welding (LFW) gives stronger seams, at the sketch's right. Can the truss be bigger than the welding machine? How shorter, and how accurately shorter, are the parts afterwards? To be checked. With the same parts shape, laser welding could be made portable and automatic.


If the skin is discarded by molten wires as suggested in the previous message, Ti-Al6V4 offers 828MPa, 7W/m/K @RT and 1.7µohm*m - a strong candidate that can be zapped by a 3V accumulator and semiconductor switches. Though, I prefer the electric motors.

Marc Schaefer, aka Enthalpy

Link to comment
Share on other sites

Linear Friction Welding (LFW) pushes strongly against an other the parts to be welded and shakes them sidewise until heat and movement forges them together, without melting.

To weld, the shaking actuator must bring many kW over <1mm amplitude hence is quick, like 50Hz to 100Hz - more would supposedly be better. 100Hz is uncomfortable for the present hydraulic actuators; electromagnetic actuators first used are uneasy with the power. Starting instead from a rotation would permit to store energy for quicker release (this improves the weld seam) and use a higher frequency. 30kW*3s need few kg flywheel. A crankshaft can transform the rotation in the shaking motion, but to assemble the parts accurately, the amplitude must decrease to zero at the end.

I suggest to revive the Stephenson valve gear for that. Here Emoscope's illustration, gratefully pinched at Wiki:

(click to see the picture move)




The slide between the two rods is moved sidewards to adjust the amplitude of the top part, down to zero at the center.

Here at a truss, it seems better to shake only the tube's free end and hold the truss node steady, so only one eccentric is really needed - with good balancing and adequate design at the slide, like adjusting the amplitude by sliding the arm's base rather than the connecting rod. The machine could catch the tube's end with one arm shaken by the slide and the node with a steady arm, connected preferably close to the crankshaft.

Other uses may have two shaking arms of opposed phases, possibly driven by the same slide of proper design.

One part or both can also receive crossed movements, say orthogonal and in phase quadrature, to shake with a circular motion that makes a more uniform cylindrical weld seam. Two crankshafts well phased, with at a part two arms converging perpendicularly to an other, achieve it.

Other designs are possible. A crankshaft with eccentricity varying continuously along the axis, plus a de-biassing part, looks fragile to my eyes. A cylindre with eccentric hole that rotates around the crankshaft's handle and whose phase adjusts the amplitude is sturdy but not obvious to control. Easier: the rotation axis is parallel to the shake, a tilted disk transforms the movement, its inclination adjusts the amplitude.

The Stephenson valve gear is well-proven. Its parts, movements and forces are sound.
Marc Schaefer, aka Enthalpy

Edited by Enthalpy
Link to comment
Share on other sites

Hydraulic actuators fit well Linear Friction Welding after all (changed my mind) and can exceed 100Hz if needed, especially if they use rotating distribution valves aided by proportional valves in series to regulate the amplitude and the mean position. But a serious difficulty here is that tubes risk to collapse instead of transmitting the shear force to the weld location. So here's a welding appartus by melting; it's semi-automatic because there are 1800 tube ends.




It's a MIG process. Two or few shell elements are fit manually around the tubes to guide the flow of inert gas and hold the electrodes in place; the rest is automatic. The fittings around the tubes can have for instance shallow grooves as gas outlets. Instead of moving a unique electrode, I imagine many (like 60) static electrodes are easier in the limited room. Each has its electric supply, and probably its mechanical feed.

I expect all electrodes active at the same time would let the molten metal flow too much under gravity, but several can operate simultaneously to get a more symmetric seam. Standard pratices, like a few points before the continuous seam, are possible. Some tungsten electrodes can start the seam if better so.

Instead of mechanically feeding the welding wires, maybe the material can be stored and brought liquid through hollow tungsten electrodes, which is more flexible since melting the base material is decoupled from supplying the feed, but this relies on tungsten not melting - it's cooled by liquid aluminium smile.png, but unfortunately not when starting the seam. This would apply to any MIG process; it's a confluence with TIG. The feed alloy must not segregate when liquid for long.

Marc Schaefer, aka Enthalpy

Link to comment
Share on other sites

  • 2 years later...

Ariane 5's main Vulcain engine is ruinously expensive old stuff, but as of 1Q2016, it shall propel the first stage of the hopefully cheaper Ariane 6.

I described a simpler multichamber bleeding expander cycle engine and a launcher:
add side boosters to adjust the Gto capacity from 6.5t to 11t.

I also describe a Vulcain equivalent that may be simpler
and less common pumping cycles too.

Several Vinci chambers could share a set of turbopumps and actuators
where the chambers exist already and the turbopumps are just scaled up.

Though, it needed to develop a new pumped engine.

Here I propose an Ariane 6 equivalent, with two short Vinci but no Vulcain, and pressure-fed reused boosters instead of expended solids.



The central stage carries 59.5t propellants. This is a lot for a last stage, but balloon tanks in a truss shall make it light enough.

The oxygen tank (250kg) is 500µm brazed Maraging steel; a thinner top would save mass. 300µm Maraging (350kg) carry the hydrogen.

10mm foam (31kg) insulate the oxygen and 30mm (219kg) the hydrogen for 10mn before take-off. Above the foam, 5 plies (11kg) of multilayer insulation keep the oxygen cold in orbit and 20 plies (106kg) the hydrogen for 5 days.

Polymer fibre belts hold the oxygen (58kg) and hydrogen (12kg) tanks in all directions to the truss.

To break at 3.1MN*m, the truss has ID=46mm OD=49.4mm L=0.69l AA7022 tubes (0.49kg), thicker (13g) at the ends welded on 18 nodes (0.45kg) per turn. The 32 stages of the 19m truss weigh 1139kg. Graphite composite could gain 750kg payload.

A shell protects the insulation in the air. 10mm foam, with 20mm thicker ribs at 10% of the surface, weigh 600g/m2, and aramide composite adds 100g/m2 at each face. This weighs 191kg but the tiles are thrown away after the fairing.

Each short Vinci shall weigh 280kg - no data available. A 70kg aluminium truss holds them.

A toroid connects the fairing, the payload, the central stage and the side boosters directly. This reduces the bending moment on the central stage. To transmit 0.5MN to or from each side of a booster, it consists or a triangular section materialized by a space truss with 12 nodes per turn and graphite composite tubes. The outer ones have estimated ID=104mm OD=113mm. The toroid weighs about 220kg, saving 0.4t over metal.

With 300kg equipment and 400kg undetailed items, the dry central stage weighs 3730kg.


2, 3, 4 or 4-then-2 side boosters adapt the performance much. Each pushes two of the toroid's six points which can be shared by two boosters: this relieves the toroid when four boosters accelerate the two full ones to be ignited later.

The Isp and maximum thrust are 2971m/s=303s and 2.33MN in vacuum, 2569m/s=262s and 1.95MN at sea level, expanding from 36bar to 47kPa. Throttling is easy, even at sea level, so some thrust more wouldn't hurt. At the end, the thrust has dropped by a half like the pressure in the chambers and the propellant tanks to lighten the helium tank, and the dry boosters weigh only 110kg per ton of propellants.

Sailing back home for reuse is described there

Fewer boosters give less speed to the central stage which isn't filled then so it accelerates well. Alternately, the Vinci could ignite before the side boosters shut off, even as soon as the vacuum permits it.




The 3.7t well insulated central stage reaches the geosynchronous orbit directly, more efficiently and safely than hydrazine at the satellite, and more quickly than a plasma engine. That's reasonably its highest energy - Mars and our Moon are easier.

Missions to far planets would benefit from an added escape stage, smaller and lighter. Electric pumps would fit best:
but my sunheat engine ("+sol" in the perf table) starting from Leo beats every chemical stage to reach the geosynchronous orbit, our Moon and the planets.
I computed with a faster spiralling transfer to Gso, and an elliptic escape without chemical kick for the planets.

Marc Schaefer, aka Enthalpy

Link to comment
Share on other sites

Europe has plans to develop a liquid stage put over one solid P120 booster made for Ariane 6 and Vega, nice, good. But sources suggest a new methane engine for it, which makes little sense.

  • At identical expansion ratio, methane gains 9s Isp over Rg-1 "kerosene", yes. But so does the denser and easily produced cyclopropane. The storable safe Pmdeta gains 3s.
  • But a useful comparison must take the same pumping power and nozzle diameter, and then Pmdeta is as efficient as methane because its expansion is better. Cyclopropane does improve over Pmdeta, but only by 200kg despite this stage is difficult.
  • Ariane 6 plans already hydrogen and the Vinci. Commonality tells to spread it to the middle-sized launcher, not to add a fuel and engine.
  • Hydrogens puts 1.5x methane's payload in orbit. It needs only 3 existing Vinci chambers and short nozzles. A common turbopump can be just upscaled from the Vinci. A common actuator set is easy. Obviously cheaper than a new engine.
  • Two US companies want methane to reuse hopefully clean engines, but hydrogen is even cleaner, and so should amines recomposition be, both in staged or gas generator cycles.
    http://www.scienceforums.net/topic/81051-staged-combustion-rocket-engines/#entry785456 (with a safer amines mix).
    The Europeans can't overtake US companies by copying them with a decade lag.
  • Two new engines would probably replace the Vulcain at Ariane 6, just like SpaceX will supposedly use two methane engines at a Falcon 9 first stage, but a seven-chambers Vinci does it better.

Here's how the launcher can look like with the three-chambers Vinci:


The design is difficult because a P120 pushes an estimated 2500kN at the end for Ariane 62, so the full liquid composite must weigh 52.5t to limit the acceleration to 4G and it must provide 6558m/s to Leo. This bad staging needs light tanks but achieves 9t in Leo, deserving a wider fairing than Soyuz ST. Details should come.

Marc Schaefer, aka Enthalpy

Link to comment
Share on other sites

  • 2 weeks later...

Here are second stage details and two optional upper stages for higher energy missions. The launcher matches Falcon-9's performance with limited development and looks cheap. It could replace Soyuz if this weren't lese-majesty.

The P120 is designed for 3 and 2.5 stages launchers, so its thrust drops little: I take 2.5MN at shutoff from A62 needs. This imposes a too strong acceleration on a too heavy upper composite. My sailback booster is an alternative

---------- Second stage

The truss transmits 4.0G when the P120 goes empty: worse than aerodynamic moments. To break at 4.1MN, it uses AA7022 tubes, Ri=34mm Ro=35.8mm, welded as 18 nodes per 0.8m stage, and weighs 718kg of which 79kg separate with the first stage. Its node diameter is 4.2m to host the 3.6m balloons; aluminium tanks welded at the truss would improve that.

The surrounding shell weighs 163kg but is thrown away earlier.

The balloon for 34.9t oxygen is of 250µm to 500µm (bottom) thick steel and 10mm foam hold by polymer belts. It weighs 225kg.

The balloon for 9.0t hydrogen is of 280µm steel and 15mm foam that gives 500s to ignite this stage after the tower's arms open. Hold by polymer belts, it weighs 335kg.

Four 750N engines control the roll while the P120 pushes, adjust the orbit and orientation after the second stage shuts off, and deorbit it. Fed from the main tanks over 10bar pumps powered by 5kg Li-poly batteries, they total 10kg.

The three-chambers Vinci shall weigh 700kg with the frame, common actuators and upscaled turbopumps. The small D=1.25m nozzles bring Isp=4345m/s=443s and 171kN; thin niobium seems possible for the uncooled section to save much mass.

300kg electronic equipment, a 50kg payload adapter and 200kg undetailed items let the dry stage weigh 2459kg, or 60kg per ton of propellants. This puts 9.1t in Leo.


---------- Chemical upper stage

It starts in orbit or almost and pushes 10kN only, taking 3 kicks at perigee for most missions. Multilayer insulation of the tanks spares active cooling.

The truss is of welded aluminium tubes with 12 nodes per 0.8m stage. The lower part has Ri=29mm Ro=30.4mm of AA7022, these 82kg stay with the second stage. The upper part has Ri=25mm Ro=26mm of AA7020 and weighs 80kg at the third stage.

The D=2.6m sphere for <=595kg hydrogen has 90µm steel (15kg), 45mm foam (48kg) that give 500s to leave the atmosphere after the tower's arms open, and 10 plies multilayer insulation (8kg) to give 50 days vacuum operation, long enough to land 1.9t on the Moon with but bigger tanks. Polymer belts (1kg) hold it to the truss.

The Ro=2.0m Ri=1.2m torus for <=4739kg oxygen has 60µm steel (12kg), 10mm foam (13kg) and 3 plies Mli (3kg). Polymer belts (7kg) hold it. Maybe the oxygen would fit in the truss instead.

The engine has electric pumps, better for that size and easier to develop
that bring 2.1kg/s of 796:100 O2:H2 to 100bar in the chambers, using 88kWe from a 114kWe 60kg fuel cell as the Toyota Mirai has - less pressure and a lighter cell would improve if available. More thrust is also possible, to target Jupiter for instance.
The fuel cell diverts 0.8% of the flux expanded separately. Four D=1m niobium nozzles expand to 20Pa for combined Isp=4842m/s=494s, wow. The engine shall weigh 30kg plus the 60kg fuel cell.

100kg electronic equipment, a 50kg payload adapter and 50kg undetailed items let the dry stage weigh 479kg, or 89kg per ton of propellants.

---------- Sunheat upper stage

Thanks to Isp=12424m/s=1267s, it brings heavier payloads to farther destinations than the chemical upper stage but slower.

Leo to Gso takes only 40 days by six engines with the inefficient spiral transfer; a year long Hohman would transfer more. The 300km Lunar orbit gets the same payload but after over a year. A transfer towards Mars needs a bigger tank than shown, and to a Martian orbit more so - but a better option would combine a chemical engine to leave Earth and capture at Mars, letting the sunheat engines change the apoapsis as described on Jul 27, 2014 in the linked thread.

Gso, Lunar and Martian transfers consume hydrogen from time to time, and taking an adjusted fraction as a gas suffices to regulate the tank's pressure. Operations at a remote planet need a cryocooler during the trip.

The truss is to break at 1.1MN compression to transmit lateral 2G to the payload and hydrogen. With 18 nodes per 0.7m stage, it uses welded AA7020 tubes with Ri=22mm Ro=23mm and weighs 243kg.

The D=4.2m h=4.4m ellipsoid for <=2900kg hydrogen has 150µm steel (68kg), 30mm foam (84kg) that give 500s to leave the atmosphere after the tower's arms open, and 30 plies Mli (61kg) that permit 40 days wait in vacuum. Polymer belts (4kg) hold it.

Six D=4.2m engines shall weigh 156kg together.

100kg electronic equipment, a 50kg payload adapter and 50kg undetailed items let the dry stage weigh 816kg.

Marc Schaefer, aka Enthalpy

Link to comment
Share on other sites

A liquid propellants launcher stage made of wound graphite composite seems light and feasible, at least on the paper. To compare, I take the same triple-chamber Vinci stage over a P120, recently estimated with metal balloon tanks in a truss. D=3.6m, H=3.9m and H=9.2m tanks make the stage but longer than a P120, so the same machine shall wind the liquid stage, but wider if needed.


  1. The balloons are wound first, as for a solid propellant but thin.
  2. They are assembled with temporary kernels, and the cylinder's inner skin is wound.
    ° Intertwining some strands from a ballon and the skin isn't mandatory (shear 100kPa) but seducing. It may need several strand spools with independent movements, as for waving.
  3. Balsa (or foam) core is glued on the inner skin. Foam can fill slits.
  4. The cylinder's outer skin is wound. It may require a different matrix polymerized at a milder temperature.
  5. A liner is put on the balloons' inner face.
    ° Perhalopolymer is known.
    ° Nickel, cobalt and alloys can be deposited.
    ° Or deposit the liner on the balloon's kernel first, before the balloon is wound.


The oxygen balloon is 870µm thick (1350g/m2) to break at 550kPa, the hydrogen balloon 500µm (775g/m2) for 310kPa. Can the machine for solid stages wind that thin? At least, carbon fabric composite plies exist with 110g/m2.

Buckling at 4.1MN determines the cylinder. Linear buckling theories are known (...by too few people) to fail, so I take an axial force of 0.68Ee2 from my experiments confirmed by:
Nasa's SP-8007, "Buckling of thin-walled circular cylinders"
hence the sandwich with quasi-isotropic (is that optimum?) skins.

The graphite inner and outer skins are each 500µm thick, contributing E=170GPa as 250µm in each direction. The balsa core is 12mm thick. The sandwich is worth an e=6mm sheet of isotropic E=170GPa. The compressive stress on 2*250µm is 730MPa.

The balsa core insulates the oxygen enough, and the hydrogen gets additional 10mm foam over the outer skin. The quasi-isotropic inner skin would contract freely by some 3ppm/K but the core suffices to hold it: radial 8kPa in the balsa stretch the inner skin and compress the outer skin by 60MPa azimutally. Well, fibre composites aren't that simple, but here margins are big.


The insulated ballons and the H=15.4m cylinder weigh 48+140+479 = 667kg. The same functions with metal took 225+335+718 = 1278kg. Gained 611kg, wow. The dry stage weighs 1927kg with graphite, that's excellent 47kg per ton of propellants. The truss wastes diameter, wound graphite doesn't.

The small compressive stress suggests that one skin and omega stiffeners, or a graphite truss, may be lighter. Though, good joints at the nodes of a graphite truss aren't obvious, while the manufacture of a wound stage is widely automatic.

Being manufactured by the same plant as solids, wound graphite stages could have the historic merit to eventually convert Europe to liquid propellants.

Marc Schaefer, aka Enthalpy

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.