Feeds:
Posts

## Long Endurance UAV Design

Karoliina has some thoughts on plane design, looking at High Altitude Long Endurance (HALE) UAV:s as inspiration for high L/D craft, to ultimately cruise at fast speed with little power. I disagree somewhat, and I’m sketching out why, below.
Probably the drones, like sailplanes, want low sinkrate and high L/D is secondary, because they need to just stay aloft and not go anywhere.
Power needed is P = v*D (we assume). Since D = CD*v*v, P = v^3*CD.
Lift is L = CL*v*v so v = (L/CL)^(1/2). (Note this CL is different from the L = 0.5*rho*A*cl*v^2, so CL = 0.5*rho*A*cl. It’s more practical here.)
Power thus is P = v^3 * CD = L^3/2 * CD * CL^-3/2.
The lift equals mass, so the power needed is
1. proportional to mass^1.5,
2. proportional to the drag coefficient and
3. inversely proportional to the lift coefficient^1.5.
This means the lift coefficient CL needs to be large for the craft to be able to loiter for a long time. So long wings and somewhat cambered profiles. A little drag doesn’t hurt as much as low lift so struts are a possibility.
Instead, for a piston cruiser, the L/D needs to be maximized for a certain minimum trip fuel consumption, not per time. Basically, you want to minimize delta_E = P*delta_t = P * delta_x/v so the cost function J = v^2 * CD = L * CD / CL which minimizes at maximum L/D. The CL term is less important compared to the loiterer. As a first guess this should optimize to a less cambered airfoil and smaller or shorter wings. And no struts.

## SpaceX’s Third Attempt

The first stage worked flawlessly but the staging malfunctioned. I wasn’t watching, but went to bed at 4 am local time after waiting through some launch delays, thinking that it’d take forever anyway, and of course the launch was right after.

What’s always been a mystery to me, if SpaceX is selling their Falcon 1 rocket at just about eight million dollars, why don’t they do a lot more test flights with dummy payloads? It seems that it would accelerate development a lot, since the hundreds of workers can’t be cheap to just keep working – a delay of a few months in selling some rocket flights will surely be costly.

During the webcast run-up to the launch, they showed Elon Musk doing a tour of their factory, and everything in their production processes seemed well automated and thought out, since they have to do so many engines for Falcon 9 anyway. Just to mention the totally automated copper milling for the chamber and nozzle root liner, as well as the automatic pipe bending machine for the nozzle. 80% of the hardware of Falcon 1 is produced in house from bare metal.

On the other hand, the Merlin 1 main engine has gone through many changes, with power increases (don’t remember if the turbomachinery or injectors have changed) and the regenerative nozzle at least. That means early flight testing could not have been very representative of the design or the build or integration processes. Now they seem further along in that, with the Falcon 9 first stage recently having done a full-up hold down firing of its nine engines.

## Design vs Test

There’s something fundamental about the whole issue of designing vs testing. It’s not a totally simple picture, with the current advanced computation and simulation capabilities making the boundary fuzzy. And there has always been partial hardware non-destructive testing too, like structural test models. So, can expensive destructive test flights be seen as just an extension of finding a workable design, as well as production, integration and operation processes? In that sense they all can be pooled into one, as just means of getting to some combination of capability, cost and time goals.

Even if there are no rigid mental boundaries between development and testing, one still has to make more careful judgements before doing a very expensive destructive flight test vs running a few minute configuration simulation. Of course you have to be careful with time allocation in design too, conceptual design is one tool for that, to avoid spending huge amounts of time and thus money for elaborate dead-end designs and configurations. The previous post is about that, where NASA spent a lot of design time for launchers and components that ended up too small anyway. But actually they started from very far and little assumptions in ESAS, eliminating lots of fundamental concepts in a tree analysis, pictured below.

ESAS conceptual launch vehicle design

Conceptual design is from the top down, but real hardware testing is from the bottom up. Both are needed to work in a real capability. Armadillo Aerospace’s John Carmack has mentioned innumerable times how building working hardware always discards too ambitious and overcomplicated designs (I have to shamefully admit, I have very little experience in designing built hardware, though I am in the process of changing that). If you only do conceptual design without basing anything in real hardware, you are moving on very thin ice. The NASP program was quite a good example of that.

Armadillo Aerospace's Module in tethered hover testing

In a sense, Falcon 1 is the hardware and process test platform for the real rocket, Falcon 9. Now that SpaceX seems to have their production line ready, I hope they would just do Falcon 1 test launches in rapid succession to iron out their bugs. (Of course, first do a lot of nondestructive ground testing, like for the pyrobolts this time.) Hopefully without charging money from the payload customers.

Again, when looking back to Wernher von Braun, after V-2 he proposed in USA the development of some new rockets and humongous amounts of test launches were included in the plan. Hardware got more reliable and rockets bigger and more complicated and thus expensive, which eliminated this approach, but it is an interesting historical mindset and viewpoint.

And again, too, of course, the fundamental property of expendable rockets that every flight ends in destruction, prevents economic partial testing. You can’t do careful envelope expansion or survive flight anomalies. Reusable flight vehicles on the other hand allow a lot of flight testing.

## Conceptual Design at NASA

The Constellation program has been going on for about 3 years. Kicking off with the ESAS study of a few months, it still hasn’t settled very much about the architecture. Even the number of solid segments and liquid engines on the Ares I and V launchers are uncertain – issues which mean a lot for the infrastructure. A high launcher means VAB rebuilding. A heavy launcher means new crawler ways. Everything seems to be reassessed constantly.

Ed Kyle has documented the Ares I and Ares V history, while I discussed the future of Ares V here, the picture is from that post’s presentation. At the moment the design has moved along from the first configuration in January, having added half-segments, more core length and a sixth RS-68. No move to HTPB rubber in the boosters yet or composite wet parts EDS IIRC.

Ares V Evolution, Muirhead Jan 2008

People from inside NASA have lamented the lack of conceptual design skills there, since the design keeps changing too much because of flaws being discovered.

There’s the classic story from Apollo, when Wernher von Braun simply didn’t believe the mass numbers the spacecraft people gave him, and vastly oversized the Saturn V – and it turned out that eventually all the performance was needed.

But the leaps in capabilities were huge back then. Now rocketry is routine and there is already one example of a lunar architecture to compare to. Not many new engines need to be developed for example, and a lot of the hardware is derived from STS and other flying systems.

So how is it that an agency getting 15 billion dollars a year is failing to pin down the mass numbers any better? Over ten ton sudden shortfalls in LEO mass seem to be a lot. Of course, it is a hard problem, and it’s easy to carp from the sidelines, but still…

What will the payload landed on the moon be? What propellants are used? What is the Altair’s or Orion’s mass? And work back from there to TLI mass and ultimately to launch from Earth, all with generous margins. And it has seemed that a certain cycle has formed. First a solution on Ares I is based on some logic linking it to Shuttle hardware, infrastructure or Ares V with common elements, which should save a lot of money and time and keep the workforce etc etc. Somewhat later, rumors about a severe performance shortfall on either launcher start circulating. Then after a while NASA announces a new configuration where the commonality is disrupted. And again forward we go.

The decisions made earlier are not supported anymore because new facts (performance problems) were realized later. But these decisions can’t be revisited. (Flying Orion on an EELV is one.) ESAS is referred to as having looked at all that, discarding it. Yet when some changes happen in Constellation, ESAS is mentioned as “only a 90 day study, how much can you expect from it?”. The consistency of decision justification is lacking. Ability for honest introspection is a rare thing for persons or organizations. I am just an outsider and don’t really know what’s going in inside there, maybe all is just exaggerated, but it looks troubled to me. How much can there really be progress if nobody knows what the launchers will be like in the end anyway?