SpaceX Dragon Launch Slips A Bit (Update)
Update on SpaceX COTS 2 Launch
“After reviewing our recent progress, it was clear that we needed more time to finish hardware-in-the-loop testing and properly review and follow up on all data. While it is still possible that we could launch on May 3rd, it would be wise to add a few more days of margin in case things take longer than expected. As a result, our launch is likely to be pushed back by one week, pending coordination with NASA.”
NASA Issues Statement on SpaceX Launch Date
“We appreciate that SpaceX is taking the necessary time to help ensure the success of this historic flight. We will continue to work with SpaceX in preparing for the May 7 launch to the International Space Station.”
So NET May 7?
Wonder if this also has to do with Atlas V / AHEF-2 being set for May 3rd last week.
The reason Mr Musk gives is that the software that will control Dragon during proximity operations needs a little more work. That stuff is really turning out to be their Achilles Heel – A lot of the delays to date seem to have been caused by software issues.
It’s OK if they get delays during development and testing. Once you get software right, it stays right, so this shouldn’t lead to any problems once they’re operational.
I agree. They should make sure they get it right.
However, in the past, many thought that SpaceX could begin crewed transport in the 3 years they claimed. They argued with those with experience in these issues that would say that it was going to take SpaceX much longer than the 3 years to get crewed transport flying. Believe me, there will be significantly more issues, software and hardware, flying people than cargo.
I’m all for SpaceX and other commercial ventures, but am trying to point out it’s likely to take longer than has been claimed. Things like this will show up before a crewed flight, even for a test flight, as a launch approaches.
Yes, SpaceX’s track record suggests their crew-carrying capability will slip and come on-line later than they project. But it also suggests that eventually they will get it to work.
I thought the first window after May 3rd was the end of May?
… if Atlas doesn’t clear the range.
Other online news sites are reporting that SpaceX-Dragon will now launch AFTER the Atlas V , which has been slotted forward to May 3 now. New Falcon launch date/time to be determined, but is no earlier than May 5 now if the Atlas makes its schedule , since it takes two full days to reset the Cape for next available launch.
More testing is much cheaper than a mission failure due to some software/hardware corner case.
If there is a real flaw it is unlikely more testing and documentation will discover it since testing can only identify anticipated failure modes. Several launch vehicles have failed due to software flaws despite meeting all the testing and documentation requirements. The only tests that will really improve safety are the actual maneuvers that will be performed before the rendezvous attempt.
Another of my favorite sayings is:
“It is a systems engineer’s job to make sure this thing cannot be broken, it is a test engineer’s job to try and break it”
So far the test engineers have won that contest every time (at least initially, but it motivates the systems engineers:-).
It was during Apollo that NASA decided that traditional evolutionary testing could be simulated on paper through the magic of systems engineering, by building a fault tree and inserting the failure rate of each component. This is not unreasonable if the technology is very mature and all the failure modes and rates are well understood, and if the system is normally in operation for a long enough period for random processes to dominate failures. None of these conditions is met with a new launch vehicle or spacecraft.
Being a test engineer, when management tells me that a test will be too expensive I usually respond with “And how much will a failure review board cost?”.
I usually get my way.
I’m glad to see they’re not afraid to delay test launches when they find something they could do on the ground to avoid a problem in flight.
Testing : good. Hype and unrealistic expectations of maiden rendezvous missions : bad . Lack of flexibility in Russian launch schedule: frustrating. NASA nanny culture: no comment.
Ask the Game of Thrones author: teaching a Dragon to fly where you want it to go ain’t easy…. ( Book 5: A Dance With Dragons)
ROTFLMAO, I wonder what the next excuse is going to be, maybe we need to lower the ISS orbit so the Dragon/Falcon can reach it without running out of fuel.
What? Give me a break. A few weeks slip is no big deal. Take your time, this stuff isn’t easy.
few weeks? try few years, according to the SpaceX original schedule (2008), they should have been on Dragon 5,
Frosty, At least they have been up front about all of their launch schedule changes and have given people time to change their plans, instead of hoping for a miracle right up to the last second. So, no trips across country just to see a cancellation; no resources or facilities tied up when they’re not going to be used, etc. In a space launch schedule, a few days is a long time. In this game, especially with a high profile test, better to have a delayed success than an on-time failure. Your performance record (or lack of) is everything. When most of your observers (the public and the media) can’t talk at the detailed level, Go/No-Go is your primary metric. And of course, better safe than sorry is a mature philosophy. Steve
Up front? Where have you been, If NASA behaved like SpaceX everyone would be crucifying them for being incompetent and overrunning the budget and who knows what else, Instead SpaceX gets praised for being “careful and thoughtful”. Is anyone even thinking about the amount of NASA’s (COTS) money SpaceX is wasting by yet again postponing a launch that should have happened long time ago?
SpaceX has encourage this criticism with their hyped claims. We’ve heard crewed vehicle in 3 years, reusing vehicle 10,000 times, cheap trips to Mars. It gets attention but sets expectations unreasonably high.
I agree. Clearly these inexperienced people are going to be beat by Ares-I which, thanks to firm commitment, sound planning from the top and reliable budgeting will launch on May 2nd, ahead of the first Dragon- err, second.
“Is anyone even thinking about the amount of NASA’s (COTS) money SpaceX is wasting by yet again postponing a launch”
Que? SpaceX (and all the COTS players) get paid for meeting milestones. If SpaceX doesn’t launch, they don’t get paid for their next milestone.
tommy:
You’d rather they just kept their mouths shut until launch day? If NASA wasn’t in the loop, you’d get your wish. Once Spacex sets up their commercial launch site, silence will be a good business practice. Happy?
Impatience is a killer in this game (ie: Challenger). Those of us that want Spacex to succeed can wait til they get it right.
tinker
Tommy has been spammy. Bye bye Tommy.
“Those of us that want Spacex to succeed can wait til they get it right.”
Speak for yourself! I want launch now! Now Now Now!
Paul:
LOL! Now that’s the way to express impatience!
It’s all in the delivery :).
tinker
Lolololol
Tinker, I feel certain that no matter what was going down, even a perfect mission launched ahead of schedule, some of the posters here (and elsewhere) would have no trouble thinking of negative things to say about SpaceX, and I can’t figure out why. I thought all the world loved an underdog. My only guess is that they predicted failure when SpaceX first came on the scene and they’re too stubborn to change their stance. I tried to make the point (above) to Frosty that the delay was announced well ahead of time so that people could change their plans, and just because NASA printed a press release about it, he credited NASA for the advanced notice. I wonder if these people dedicated to seeing SpaceX fail will be walking in circles carrying signs attacking SpaceX on launch day. I’m certainly not surprised that several of the responding posts employed sarcasm. Steve
Realizing how important it is to get the price down the attacks are amazing. I have no trouble with Spacex bosting some. It gets interest like a prize fighter saying I’m going to knock him out in first round.
Just
think of it this way. If NASA was doing it, it would be considered almost a decade
ahead of schedule and grossly under budget. A week or two is not much of a
wait.
Anyone really expect commercial crew to get off the ground before ISS is splashed?
YES
Yes, absolutely.
Hey Matt – YES! now lets brighten that attitude before I get out my nunchuakus
I am not sure any changes in the actual software are even needed. As I understand it the problem is simply that NASA is not yet confident the software meets specifications and wants more documentation. If anyone has details please post.
My opinion is that the root of this “wait” will afflict all vehicles that depend on such control systems … as it has already. Including ATV/HTV, although at different stages in development and system fault analysis (a function of how/when a given group adapts in its engineering lifecycle).
In comparing how GNC is does now than in the 70’s when I did it, people back then were paranoid of computer navigation/guidance – if you don’t believe me, ask Wayne Hale about autonomous landing the Shuttle with ALT tests (and before). So from the start, we focused on error modelling, and even the slightest error was grounds for review early on – we were worried about error growth and control oscillation – control theory is very touchy stuff. We had all sorts of problems, bad algorithms, bad floating point hardware (unpredictable behavior going between hardware implementations), tiny memories, bad software libraries with “surprise holes” in trig/other math functions, more error prone navaids, too slow sampling rates, gyroscope drift/loss of inertial platform …
In recent years when I examine modern GNC / autonomous control software (there are even open source versions of these), there’s the presumption that the software is mostly right, that the quality of the position fixes from navaids/instruments is reliable, that the computers are always fast/large enough … in short a world of riches, so you don’t even have to model errors, just use a random number generator in your Kalman filter to decrease aliasing … and you’re done,
So you appear to get a “good enough” answer early on … so you go with it, thinking that’s done. Then you find in flight test all the subtleties that have thus be obscured … bubble up as you encounter them as aborts, and have to backtrack. Some attempt to deal with this by setting error margins too tight, as an attempt to forestall the problem – it makes things worse, because you have more aborts and you still don’t understand the etymology of the bug – no way to backtrack it because the source was at the beginning.
So why software isn’t 100x better even with our riches … is that debugging is now 100x worse … because we rely on our 100x better resources … to avoid the human side of proving things from the ground up … which is still proceeding at the same rate as before … but no longer appreciated … because we think that our 100x better technology … debugs itself … because it is 100x better.
And when I educate this, everyone puts fingers in their ears and they hum really loud.
I would probably agree if I could figure out exactly what you said.
The irony was I was just describing the non technical, non mathematical, non statistical, non theoretical stuff. Just to understand a component like, say, a gyro (more than three in an inertial guidance unit), takes a thick book of equations to account for its details. In one case, an out of spec bearing in a gyro accounted for a 3 year delay in a guidance program. Another case, multipath problems
Before IEEE floating point, there was a high variability in implementations – for example IBM 360’s had base 16 exponents, CDC’s had base-2, the rules for truncation roundoff were different, even different versions of the 360 (like the 90/95/195) treated calculations … subjectively in certain ways. And that was just the hardware.
If you did a Monte Carlo simulation, the random number generators couldn’t be trusted, you rolled your own and proved them congruent/gaussian, sometimes reproving them when systems software changed. It was a minefield. If you depended on something, you proved it yourself.
You’d work in a variety of languages, including SPL (Space Programming Language)/JOVIAL, HAL/S (S for Shuttle), Fortran (hundreds of dialects). Submit card decks to RJE, get back several hundred feet of fanfold paper (14″ or 18″ wide).
Typically, you’d go over side by side 2 or more of these runs visually comparing runs, marking up changes in position/ momentum/ pitch/ roll/ yaw, tap numbers into calculators to check veracity. When you found an anomaly, you’d find the local expert on a given instrument, or in the worst case concoct another flight test to get data on. Sometimes 10-20 people in a room playing mathematical “blind mans bluff”.
The root problem then was no one trusted the technology. The root problem now is too many trust the technology too much.
I agree with everything noofcsq just said and most things he will say whether or not I under him or not! Lolol
Shameless parrot and copy cat
All true. There also is a tendency by some programmers at some organizations to use algorithms and code, or code snippets, developed long ago. Some of these weren’t sufficiently tested in the first place and can contain conditional failure modes that can sneak up and bite you. But the source code remains on hard drives all over the networks, and their bugs live on, too. There is some amazingly — subtly — dirty C code (especially) lurking out there, waiting for a lazy programmer. I rather hope SpaceX is using Ada.
Also, some of this software is very complex, and nearly impossible to test in the face of every potential condition or combination of conditions. Therefore, some software simply has to mature in an operational environment.
My sympathies are with the software QA people on this one. In these situations, all one can do is as much as one can do.
I remember translating obscure Fortran/assembler code into HAL/S, then into C, then into Ada, then … back into C. This over many years, on and off.
The programmers who worked daily with it, never trusted the documentation that detailed how it works. They’d come up with own untested presumptions. Those would guide testing / enhancement. Even when you’d backtrack a problem back to the original documentation, revise/retest against the document’s specifications and test matrix … they’d still presume said presumptions.
You can enhance technology. You can’t enhance its understanding. You can’t enhance the human nature aspects of the application of technology. 100x better tech limited by the 1x humans.
Use to steal fortran 77 graphic subroutines to make buggy graphic programs lol It’s does take a computer programer! Lol
I have wondered about the software and how it is being built and tested. It’s very challenging to get it right in non-critical applications; in a critical life-or-death application, the CS people I know working these kind of problems would be diagnosed as paranoid by most psychiatrists.
An interesting read is Richard Feynman’s section “Personal observations on the reliability of the Shuttle” in the Challenger Accident Report. He talks about the attitude and approach used by the programmers in the shuttle avionics.
In the big scheme of things, a few weeks or even months don’t matter. But I find it amusing how many apologists SpaceX has, is like they can do no wrong. I am sure they would find a positive spin even if the Falcon blew up on the pad (knock on wood) … they are like the space industry version of Tim Tebow or something.
They DO Something in the RIGHT direction! With most all the horses running in the wrong direction what do you expect!
Go Space !
Debug good luck!
Rahhhh rahhhhh
Little prince/Spacex satellite Group
Satellite buses massed produced to customers special custom needs.
First generation will all have the diameter of a falcon 9 core. Customer can purchase different length trunk to meet their satellite requirements. All satellite buses are provided with solar power etc. Etc.
Falcon 9 launches are offered
Spacex is now offering their Econ satellite lift option. It carries up to 4 satellite trunks into space on one launch. The trunk stack is topped with a new dragon tug vehicle. A dragon tug is simply a dragon filled with fuel.
After launch, the the falcon second stage together with dragon tug work together to place your satellite in your orbit.
Both the second stage and dragon tug reenter earth atmosphere to be recovered or park at Spacexs Leo fuel depot beam for future missions.
At this time Spacex is still working on making their falcon 9 first stage recoverable but more Satellites per launch will be available on recoverable falcon 9 Heavies soon.
If you require satellites with greater diameter than a falcon 9 we will be offering larger (falcon faring diameters). Since we build into the faring and slice the faring up into custom customer lengths you only pay your share of the launch cost.
Pricing varies do to your percentage of Payload weight, volume needs plus your choice of satellite options etc. etc.
We also provide faring adaptors for the LV of your choice should you prefer Ferrari launches prices over our Volkswagen bus launch prices.
Joe Q future Satellite bus customer.
PS I keep reading that to go to the moon we need a service module.
How hard is it to turn a falcon 9 second stage into a apollo like service module??? Don’t you just add Draco thrusters/control thrusters to a dragon trunk?????
If one is not big enough couldn’t you use two?
Out!!!!!!!
When was this 3rd flight originally planned to launch? It’s been so long ago I forgot.
Originally it was delayed so they could perform analysis on launching 2 commercial satellites at the same time, then one was dropped, then the other.
In the mean time, it’s approaching 1 1/2 years since the last Falcon 9 launch. I’m wondering if any other Falcon 9 launches will occur this year. SpaceX still needs to demonstrate that the Falcon 9 can sustain a significant launch rate.
When looking at their Spacex ksc hanger I was wondering why they don’t have two hangers there now! I supered the contruction of a helicopter hanger about half that size and it was a pretty cheap easy project. Shouldn’t they have at least two falcons ready to fly?????
Sounds like May 7th is Firm boys and girls lol
Spacex Elon Musk
Good article lol
http://www.dailytech.com/Sp…
This is Elon Musk’s vision, not private industry, leading a revolution–no one else was willing to do this before him, despite there being launch companies already entrenched for satellites and such like the United Launch Alliance; they haven’t stepped up at all to meet the needs we have despite already being right there on the front lines. Companies by themselves are not inventive nor creative and lack any real foresight, it’s always a particular individual who put their visions in motion that moves a company and the whole industry; from Alexander Graham Bell to Henry Ford and Steve Jobs. “Private Industry” is not some mythical beast that can solve any problems, it’s the revolutionary individuals who take it by the reigns and direct industry’s force that create our future.