Learning from Cloud Computing (or how to build an SSTO, maybe)

[jak42]

So I am not a rocket engineer, my day job is working with computer networks. But I've been thinking about some ideas for radically redesigning how a rocket could be built kind of the way cloud computing is radically redesigning the way people are building computing facilities. I haven't run any numbers on this, it is just a bunch of (probably hare-brained) ideas right now, but bear with me, and please feel free to point out any fatal flaws (of which there are likely to be quite a few).

The basic goal is to somehow reduce the structural mass fraction below 15%. There three basic ideas here that I'd like to explore:

1) Integrating the engines and TPS into one unit.
2) Rearranging the location of the crew/cargo compartment and the tanks so the tanks are no longer load bearing, allowing balloon tanks to be used.
3) Getting rid of heavy pumps or pressurization (which requires heavy tanks) and instead using a combination of vapik (sp?) to pressurize the tanks and electroactive polymers or shape memory materials to squeeze it like a water bottle.

I'll take each in turn.

1) Integrating the engines and TPS into one unit

Suppose we can build an engine with a thrust chamber about 4" long and 1.33" radius with 250 lbf such as this one:http://www.watzlavick.com/robert/rocket/index.html
Now, following the cloud computing analogy, in order to get 1Meg lbf, we would need 4K such engines. Sounds like a lot, but since they are small, we only need effectively 50 sq. ft. of rocket engine nozzle spread out across the bottom of the craft in order to satisfy that requirement. That works out to a bottom radius of around 8 ft.

Running individual plumbing lines from the tanks to 4K tiny engines would be a nightmare and an invitation for component failures like the N1. So, instead, we rather shape the engines into a large manifold with thin-walled cavities in it that route the fuel and oxidizer to the engines. Two large input ports in the manifold connect to single fuel and oxidizer lines coming from the tanks. The manifold material would ideally be a light but tough ceramic, like silicon carbide, so the engines could be transpirationally cooled, or it could be a standard high temperature metal like Inconel, and the manifold could be constructed using laser melting: http://www.sciencedaily.com/releases/2010/05/100503111519.htm.

The advantage of this approach is the the TPS is integrated with the engines. It means the vehicle is VTVL, not winged or anything, but the  engine manifold could possibly be shaped for some lift. Landing would be powered, like Masten and Armadillo. Best fuel and oxidizer would probably be L-CH4 and LOX, to avoid coking but still have sufficient density to reduce tank size (which LH2 wouldn't).

2) Rearranging the location of the crew/cargo compartment and the tanks

The upshot of this idea is to put the crew/cargo  compartment on the bottom of the vehicle immediately above the circular engine manifold with the tanks above. This allows the engine manifold to act as structural support for the cargo. The weight  of the tanks can then be radically reduced, essentially using balloon tanks. Two fuel and oxidizer lines come down either side to  the engine manifold.

One possible issue here may be safety. The crew is sitting below a big mass of flammable stuff. But the goal here is for "aircraft-like operation", and passengers on commercial airline flights regularly sit just a few feet away from wing tanks filled with fuel. If that can't be achieved, then I think we can give up on the whole idea of an SSTO or RLV.

3) Getting rid of heavy pumps or pressurization

The final idea is to try a different way of pressurizing the fuel and oxidizer. Part of this would use vapik like I think T-space wanted to do, where the pressure of gaseous, evaporated fuel or oxidizer in the tank forces the liquid contents out. The disadvantage of vapik is that the  pressure varies as the amount of liquid drains out of the tanks, since the liquid/vapor must be in thermodynamic equilibrium. To help boost the pressure and make it more consistent, we wrap small bands of electro-active polymers or shape change materials (example: http://ndeaa.jpl.nasa.gov/nasa-nde/lommas/eap/EAP-web.htm) around the balloon tanks. These materials change shape when an electric field is applied to them, for example shortening. When they shorten,  they will apply pressure on the balloon tanks and thereby boost the vapik pressure on the tank contents.

The one point I don't quite have worked out yet is where to get the power for the shape change pressurization device. Batteries are out, too heavy.  Something like lasers from the ground which have been used  for the space elevator games might work, but the photocells might interfere  with the aerodynamic  properties of the vehicle, and, besides, shape change pressurization will be most important near the end of the flight to orbit, when the vapor pressure is lowest, and that is likely to be exactly  where laser power beaming is least efficient. Maybe the best thing would be fuel cells, perhaps ones running on CH4 and O2 which could be routed from the fuel tanks.

Obviously, there are no numbers behind these ideas: how much  power for the shape change pressurization device, what the engine manifold would weigh, etc. etc. But are there any basic flaws? Comments?

[kkattula]

Nozzle size.  Robert's engine is for sea level testing. To get optimum perfomance at altitude, you want much bigger nozzle exit to throat area ratio.

Also Robert's engine, though very pretty, is almost certainly not nearly efficient enough for an SSTO.

[butters]

The premise of cloud computing is the use of a number of let's say M computers to process another number of N workloads by way of splitting workloads among multiple computers and splitting computers among multiple workloads according to user demand and resource availability.

In my view, the proposed launch system design concept is more evocative of the classic mainframe principles of hierarchical parallelism than of the virtual clustered network principles of cloud computing.

A cloud-inspired approach to launch services would be less about the design of the launch vehicles themselves than about how they are operated in the context of a dynamic and scalable launch system, possibly involving a heterogeneous population of vehicle types.

It would seem to involve a highly modular and standards based payload architecture, on-orbit assembly, and high flight rates on commodity launch vehicles which would be at least as likely to be simple expendables that can be cranked out on a volume assembly line as opposed to more sophisticated designs.

Personally, I don't see payload customers compromising design flexibility or mass overhead in order to utilize a cloud-like launch system.  I think it's a bit more plausible to consider a large orbital developer (e.g. hotels) designing a launch cloud to support their own operations than to expect several customers to design their payloads around a third-party launch cloud.

I believe that payloads drive the design evolution of launch platforms, like workloads drive the design of computing platforms.  The desktop couldn't really happen before the word processor, and the cloud fundamentally revolves around the relational database.

Most of what we've launched so far may be described as free-flying robotic spacecraft.  We're finishing up an orbital destination with the ISS, which was in some respects the ideal application for Shuttle (although there are almost certainly better launch systems for destination assembly).  And with destinations like this emerges a new class of payloads in terms of people, propellant, and provisions.

If the principles of cloud computing are coming to launch services, they're coming by way of Bigelow and SpaceX.  The launch vehicles will look familiar, just as my home server wouldn't look entirely out of place in a datacenter.  The paradigm shift will take place in the payloads and the business models. 

[8900]

Nozzle size.  Robert's engine is for sea level testing. To get optimum perfomance at altitude, you want much bigger nozzle exit to throat area ratio.

Also Robert's engine, though very pretty, is almost certainly not nearly efficient enough for an SSTO.

linear aerospike engine is designed exactly for that purpose: to solve this issue

[kkattula]

Balloon tanks.  Not actually balloons. Simply thin walled, rigid, pressure vessels that can't support themselves against gravity without internal pressurization. Can't apply force to the outside to increase pressure inside without crushing the tank. Bladders might work, but then they can't support anything else

I am not a rocket scientist (tm) but even I can see serious flaws in your concept.

OTOH, there have been previous serious concepts using multiple identical modules. I suggest you google OTRAG.

Also, combined nozzle and TPS has been suggested before, using plug nozzles. Bono's Saturn S-IVB derived SASSTO from 1967?

(And be glad I got to you before Jim )

[meiza]

Umm, the pumps are LIGHTER than any squeezing tanks which are all just variations of pressure fed - the whole tank has to be made thick as it has to be at higher pressure than the rocket engine's main chamber. It's going to be much heavier than only slightly pressurized tanks and turbopumps. Unless you have some super unobtainium for the tanks.

[tnphysics]

Note: Pumps do NOT need to be turbopumps. There are NUMEROUS other approaches with adequate performance.

[Axel]

Nozzle size.  Robert's engine is for sea level testing. To get optimum perfomance at altitude, you want much bigger nozzle exit to throat area ratio.

Also Robert's engine, though very pretty, is almost certainly not nearly efficient enough for an SSTO.

Using thousands of micro rocket engines in a reusable vehicle is at least an idea I haven't heard discussed seriously before. Maybe that is because it is plain crazy, but to a layman as myself the flaws are not obvious.

Maybe, as a guess, cooling becomes more difficult for small engines?

Is there any study about the optimal engine size? The bigger the better?

What about efficiency? Could not find ISP measurements for the Robert-Rocket...

Also as a rumor, the OTRAG guy claimed putting a lot of rocket engines close together could create a nozzle like effect without additional nozzles. It reminds me of an effect from laser launching, where the material heated by the laser pulse expands mostly directed away from the surface. And that is what a nozzle is basically for: make the exhaust go into into backward direction as much as possible (instead of expanding to the sides). Isn't it? So could a flat surface that releases rocket exhaust exhibit a similar property?

Interestingly I read other experts too believe there is such an effect, but OTRAG is suspected to have been overly optimistic about its magnitude. So theoretically it may exist, but in practice no one has measured the effect?

For the fun of a thought experiment, let's put the thousands-of-engines idea to the extreme: imagine millions of microscopically small engines. Kind of a smart material taking in oxidizer and fuel on one side and giving off exhaust at the other side. Sounds crazy. But can we prove it is impossible?

[mlorrey]

I'm glad to see someone else mentioning OTRAG. It is a better analogy for the cloud computing concept, since each OTRAG unit is equivalent to a server. The tanks are the server power supply and the engine is the CPU. You dump "Thrust Servers" as demand for thrust drops.

Carrying big balloon tanks is thus monolithic and should be scrapped.

The mass production of OTRAG units (i.e. Thrust Servers) is akin to mass production of computer servers.

You assemble them together in different configurations depending on the size of the payload and the orbital requirements.

[kkattula]

Armdillo built a multi mini-nozzle engine a few years back. Can't recall if it worked well, but they went back to single nozzles.

http://media.armadilloaerospace.com/2004_08_22/mininozzleAssembly.jpg

[D_Dom]

As long as we are brainstorming, a great visual example inspired me from the California Science Center today. I was looking at the heat shield from the Apollo/Soyuz Command Module. It looks like honeycomb structure, seriously ablated from the heat of re-entry. What if each cell were just such a micro-rocket engine as suggested by Axel?

Using thousands of micro rocket engines in a reusable vehicle is at least an idea I haven't heard discussed seriously before.

In the spirit of brainstorming, ignore for a minute the problem of supplying fuel to all those engines. My imagination was off-scale high, I had just finished watching Hubble 3-D and was satisfying my desire of being up close to actual flight hardware. Yesterday I spent a few hours answering questions from students of all ages at a booth in the hanger of Tomorrows Aeronautical Museum. Great young minds are thinking about these ideas. Excellent inspiration for this aspiring rocket scientist. Very impressive displays of technology.

[john smith 19]

As long as we are brainstorming, a great visual example inspired me from the California Science Center today. I was looking at the heat shield from the Apollo/Soyuz Command Module. It looks like honeycomb structure, seriously ablated from the heat of re-entry. What if each cell were just such a micro-rocket engine as suggested by Axel?

Using thousands of micro rocket engines in a reusable vehicle is at least an idea I haven't heard discussed seriously before.

In the spirit of brainstorming, ignore for a minute the problem of supplying fuel to all those engines. My imagination was off-scale high, I had just finished watching Hubble 3-D and was satisfying my desire of being up close to actual flight hardware. Yesterday I spent a few hours answering questions from students of all ages at a booth in the hanger of Tomorrows Aeronautical Museum. Great young minds are thinking about these ideas. Excellent inspiration for this aspiring rocket scientist. Very impressive displays of technology.

This was basically the concept beind the MIT "Micro rocket" project. Using MEMS techniques  to build an array of thrusters. My BOTE calculations suggests a 1m^2 array could equal an SSME with a T/W ratio of 1000:1. with a 100 bar chamber pressure.

It turns out electrical engineers don't make very good rocket engineers. 

Turbo machinery scales down very badly with poor sealing and speeds of 250 000 rpm. 

Had they implemented John Whiteheads concepts from Los Alamos of linked reciprocating pumps (basically a fluidic astable multivibrator) they might have delivered a working design.

[Asteroza]

As long as we are brainstorming, a great visual example inspired me from the California Science Center today. I was looking at the heat shield from the Apollo/Soyuz Command Module. It looks like honeycomb structure, seriously ablated from the heat of re-entry. What if each cell were just such a micro-rocket engine as suggested by Axel?

Using thousands of micro rocket engines in a reusable vehicle is at least an idea I haven't heard discussed seriously before.

In the spirit of brainstorming, ignore for a minute the problem of supplying fuel to all those engines. My imagination was off-scale high, I had just finished watching Hubble 3-D and was satisfying my desire of being up close to actual flight hardware. Yesterday I spent a few hours answering questions from students of all ages at a booth in the hanger of Tomorrows Aeronautical Museum. Great young minds are thinking about these ideas. Excellent inspiration for this aspiring rocket scientist. Very impressive displays of technology.

This was basically the concept beind the MIT "Micro rocket" project. Using MEMS techniques  to build an array of thrusters. My BOTE calculations suggests a 1m^2 array could equal an SSME with a T/W ratio of 1000:1. with a 100 bar chamber pressure.

It turns out electrical engineers don't make very good rocket engineers. 

Turbo machinery scales down very badly with poor sealing and speeds of 250 000 rpm. 

Had they implemented John Whiteheads concepts from Los Alamos of linked reciprocating pumps (basically a fluidic astable multivibrator) they might have delivered a working design.

You'd think someone would have done a Ph.D of s semiconductor array of nozzles using a reciprocator pump of some kind (Whitehead's linear piston type is a good one, so might be the inverse wankel as done by Liquid Piston) by now. Is it the interdisciplinary nature of the work (fluid dynamics, semiconductor/MEMS manufacturing) that's a barrier to grad students trying without a Professor as a lead investigator?

[john smith 19]

You'd think someone would have done a Ph.D of s semiconductor array of nozzles using a reciprocator pump of some kind (Whitehead's linear piston type is a good one, so might be the inverse wankel as done by Liquid Piston) by now. Is it the interdisciplinary nature of the work (fluid dynamics, semiconductor/MEMS manufacturing) that's a barrier to grad students trying without a Professor as a lead investigator?

Partly. IIRC most of the papers were group efforts although the PhD's were individual.

The tough part of Whiteheads design from a monolithic mfg perspective is the piston, specifically cutting out of the surrounding structure but keeping it attached until the last stage. I can think of a way that might work in silicon.

TBH I think silicon MEMS is too accurate. Machining to 1 mil (25.4 micrometres) would be a very tight fit. I quite like the technology of Low Temperature Cofired Ceramics. This can also build monolithic structures with operating temperatures into the 6-800c for heat exchangers, chemical reactors and burners. It's brother High Temperature CC requires kiln temperatures to 1600c.

[EDIT one of the most useful yard sticks I've ever heard of was that with 4 though clearances a car engine leaks oil. With 2-3 thou clearances (IE 50-75 micrometres) it runs smooth and doesn't leak oil. That's always helped me get a handle on what sort of level you need to be looking at for this sort of thing ]