Chair Force Engineer

Saturday, April 25, 2009

Test like you fly

In the business of aerospace, the phrase "test as you fly" can never be repeated too often. All hardware should be tested on the ground in as realistic fashion as possible. Flight testing should come as close to the environments where the hardware will be flown. Traditionally, new rocket designs have been tested one stage at a time, using dummy upper stages. Why then would NASA reject this test strategy for Ares I?

Perhaps it's because Project Apollo abandoned incremental testing in favor of all-up testing. Nowhere was this more pronounced than Apollo 4, the first Saturn V flight, which was the first instance when the S-IC and S-II stages had ever flown before. Of course, NASA and its contractors had extensively tested the stages on the ground before, and S-IVB had been tested on Saturn IB launches. Project Apollo is often held up as a model for how a space program should be conducted, but it's really an exception rather than the rule. Apollo was the unique product of its circumstances, and its "crash the schedule" approach should not be viewed as standard industry practice.

The current Ares test schedule calls for Ares I-X this year, a test of a 4-segment SRB with a dummy fifth segment, dummy upper stage, and avionics that don't represent Ares flight avionics. The next step is Ares I-Y in 2012. I-Y will be the first flight of the real Ares I SRB, plus an inert upper stage that, minus the engine, resembles the real Ares I upper stage. Ares I-Y will also test the Orion escape system.

As I've said many times before, Ares I-X has little to no bearing on the Ares I flight hardware and should be terminated. Ares I-Y is a far better test because it does involve Ares flight hardware, but I'm not certain there's much to be gained from flying an engine-less second stage in place of a dummy upper stage.

Recently there's been talk of an "Ares I-X Prime" which would actually test a five-segment SRB with dummy upper stage. Now we're actually getting serious about "testing like we fly." This is what Ares I-X should have looked like all along. The problem with the "Prime" test flight, in my view, is that it's being considered as a replacement for Ares I-Y instead of Ares I-X.

At this point, Ares I-X doesn't really buy NASA anything, aside from positive PR if it works correctly. Ares I-Y isn't doing much more than Ares I-X Prime would do, aside from testing the supper stage structure under flight loads and allowing for the high-altitude abort test. A wiser and more fiscally-responsible strategy would be cancelling both Ares I-X and Ares I-Y, skipping ahead to Ares I-X Prime, and then making Orion 1 (the first in-flight ignition of the upper strage and first on-orbit test of the Orion spacecraft) the last test flight before humans fly on Ares-Orion.

Tuesday, April 21, 2009

Spiral Development: The Chair Force Engineer Plan for Closing the Gap and Enabling Human Lunar Exploration

Today's big news for Project Constellation comes in the form of Aerospace Corporation's independent study of using Heavy EELV's to launch the Orion spacecraft. The short of it: there are no problems with black zones, and the launchers can launch Orion with performance to spare. But the costs of doing so won't be trivial, and EELV+Orion won't be operational until 2014 or later. That's not much of an improvement over Ares I.

Right now NASA faces two challenges that are often opposed to each other. The first is fielding a human space launch capability in a minimal amount of time after the shuttle is retired. The second is the political consideration of retaining as many shuttle jobs as possible after the shuttle retires. Ares retains shuttle jobs, but it won't be ready for another six years or more. EELV and COTS-D might be able to shorten the post-shuttle gap, but they don't retain the shuttle workforce.

Since everybody seems to have their own ideas about how Project Constellation should run, I'd like to share mine. My ground rules are simple:
1) Get a manned spacecraft flying to ISS as soon as possible
2) Whenever possible, minimize development costs
3) Take a spiral approach to development, sacrificing the arbitraty 2020 moon landing date in favor of incremental and affordable advancements.

The first step would be halting all work on the Ares launch systems to evaluate which elements are applicable to the spiral development program that I propose, albeit on a longer time schedule than the current NASA plan. Once that's been accomplished, the Chair Force Engineer plan for manned spaceflight can begin in earnest.

1) Fully fund SpaceX's COTS-D effort
This is a no-brainer. Dragon is a simple capsule designed for one mission: deliver humans and cargo to ISS. It's the furthest system along the path that can shorten the gap.

2) Replace the current Space Shuttle system with a block I Shuttle C
Shuttle C shouldn't be hard to develop, as much of the work was completed prior to 1993. Even the leftover engines from the shuttle program can be expended on Shuttle C missions. While Shuttle C would be tasked with delivering cargo to ISS, we have to face reality: it's really there as an interim measure for retaining the shuttle workforce over the long haul while not endangering astronauts on further shuttle missions.

3) Make block upgrades to Shuttle C as the budget permits
The first order of the day is to find a replacement for the finite supply of space shuttle engines. RS-68 is a good canddiate, but it needs upgrades to even come close to SSME performance levels. The new injector plate and turbopumps from RS-68A&B are a good start, but a regeneratively-cooled nozzle would be really nice.

Shuttle C is also expandable in the SRB department. If NASA insists on paying ATK to develop longer SRB's than the current ones used byu the shuttle, they can be integrated with Shuttle C fairly easily.

4) Create a manned capsule capable of returning to earth from lunar trajectories
Perhaps Dragon could be upgraded for lunar missions. Certainly SpaceX has been discussing circumlunar Dragon missions, and I wouldn't rule out a "Block 2" variant with a beefier heat shield and enough consumables for a lunar mission. If Dragon Block 2 doesn't pan out, the Orion spacecraft could be revived using Falcon 9 Heavy or a Heavy EELV as a launcher.

5) Create an Altair lander and other elements of a lunar transit system, designed for launch on Shuttle C.
I'm agnostic on whether rendezvous in earth orbit is superior to rendezvous at an earth-moon Lagrange point. The important thing about my plan is that decisions such as EML vs. LEO are deferred until the budget exists to develop lunar-capable hardware. Certainly both would be possible using Shuttle C, in-space assembly, and on-orbit refueling. It's certain that a competent lunar mission could be staged using a capsule launched on a Heavy EELV, a lander and propulsion stage that are launched unfueled by a Shuttle C, and a load of propellant delivered by a second Shuttle C.

In closing, NASA has gotten itself into a lot of trouble by avoiding the "pay as you go" approach in favor of redoing Apollo on a shuttle-era budget. Unless the agency changes direction very soon, there will be a long gap and a brain drain in central Florida. The solution is the time-honored technique of spiral development. NASA should accelerate Dragon, fly an interim Shuttle C, and upgrade Shuttle C for sustainable operations before devloping lunar hardware in earnest. Such an approach gives policymakers enough options to ensure that the US stays in the manned spaceflight business even if the lunar goal is abandoned or replaced with more ambitious exploration targets.

Monday, April 20, 2009

Shaky Math

SpaceX is delaying its next Falcon 1 launch because of "dynamic interactions" between the launcher and its RazakSat payload. A lot of commenters are coming down hard on SpaceX or wondering how this issue could have been left to simmer until the very last minute. Having observed a spaceflight program dealing with serious launch vibration issues, it's pretty easy for me to see how this happened.

Every launch vehicle users' manual contains a vibration profile for the launcher across the range of frequencies at which the rocket is expected to vibrate. SpaceX has been publishing users' guides since at least 2005, three years before the vehicle made its first successful flight. Furthermore, the flight configuration is somewhat different from the original one in the first users guides, after the change from a Merlin 1 to a Merlin 1C engine on the first stage.

The most likely scenario is that RazakSat was designed to the old vibe specs that were published for Falcon 1 several years ago (after all, RazakSat wasn't designed, fabricated, and integrated overnight.) It wasn't until all the data came back from the successful September 2008 Falcon launch that the vibe problem was discovered with RazakSat. Perhaps it affected certain structural modes of RazakSat, or maybe the vibe profile was more intense across the spectrum. Either way, it's time to go back to the drawing board.

The vibe problem doesn't require any drastic solutions. By placing a series of Softride isolators between the launcher and the payload separation system, vibrations can be damped down to a survivable level. A coupled loads analysis is absolutely necessary to examine the full launcher-softride-payload stack and determine how the isolators can be tuned for the RazakSat mission. I don't know how much time CSA Engineering would need to solve the RazakSat issue, but it would seem like the quickest possible option for getting the next Falcon 1 successfully off the pad.