The Guild of Icarus: Aerospace Engineering and Aeronautical Club

[Miranda.T]

March 9, birthday Gagarin

One if the most select and brave groups ever - the founding figures of spaceflight. All of those early pioneers, both Russian and American, certainly had 'the right stuff' in bucketloads.

It should of course be noted that yesterday was another historic day - SpaceX's landing of a refurbished rocket stage from an orbital mission. Things are suddenly moving along a pace, pushed by SpaceX and Blue Origin. A commercial rivalry there should be a good thing, rapidly driving forward new developments, but hopefully with safety always foremost in mind.

Yours,
Miranda.

[J. Wilhelm]

Part 1 : Stage 1 Ascent and Stage 2 Climb to Geostationary Orbit; Stage 1 Descent and landing video was lost

SpaceX SES-10 Launch and First Stage Landing

Part 2:  Same as above but with commentary + Stage 2 Separation and Payload Geostationary Orbit Insertion plus commentary

The second video gives you a lot of commentary and explanation as to what you just saw above. It is however a SpaceX sponsored web cast and as such it puts a lot of spin and hooplah into the event. I don't think that the Apollo crew, engineers and ground crew ever had so much cheering in the background  

SES-10 Hosted Webcast

I suspect the lost landing video might lead some conspiracy theorists to start spinning some "alternate facts"  

[morozow]

I suspect the lost landing video might lead some conspiracy theorists to start spinning some "alternate facts"  

We have already started. Jealous.

[J. Wilhelm]

I suspect the lost landing video might lead some conspiracy theorists to start spinning some "alternate facts"  

We have already started. Jealous.

Well. You heard it here first! They've already started!!     

[morozow]

I suspect the lost landing video might lead some conspiracy theorists to start spinning some "alternate facts"  

We have already started. Jealous.

Well. You heard it here first! They've already started!!     

There are always people who are looking for the catch in everything. And when there is an internal explanation to seek him ...

But it was just a discussion of this event on the forum of the game EVE-online. Only individual characters. Two.

[morozow]

And again April 12 the international day of cosmonautics

Ю Гагарин Поехали! Y Gagarin Poehali!

[J. Wilhelm]

Oh sweet Jesus, Mary mother of God!

http://www.bbc.com/news/technology-39803425

So what do I read on the BBC today? Apparently NASA is holding a contest (available to US citizens only) to see if someone can re-write a particular subroutine of a fluid dynamics programme suite called FUN3D. Given my experience in high speed flow (this was my exact areas of experitise), this must either be a Navier Stokes Finite Volume  solver or if not, at least an Euler Equations solver (like the one we discussed a few pages ago vis-a-vis the HOTOL hypersonic vehicle).

No listen to this: The programme is written in FORTRAN     I think I may be one of the only few dinosaurs out there who still know how to program in FORTRAN (Fortran 77, 90 and 95 in my case). In fact all of my code in my college years was FORTRAN  

The purpose of the contest is to speed up the execution of that particular subroutine by a few thousand times if possible. The prize is $55 000 USD to be split between the first and second place winners.

*Sigh* what do I do?   Of course there is no indication or expectation that I could be better than any other contestant. Other thah the fact that I'm old and I know the language   And for all I care the code is obtuse, poorly annotated, and mind numbingly convoluted. As it usually is. I HATE rewriting code, and yet that's all I did back circa 1997.

https://fun3d.larc.nasa.gov/chapter-1.html
https://software.nasa.gov/software/LAR-18968-1

I'll see if I can request the stupid thing. It'd be fun to run it in a cluster headed by my Steampunk Linux PC  

PS Another interesting story:

Chinese farmer DIY's design, build and fly their own aircraft:

http://www.bbc.com/culture/story/20170312-the-amazing-flying-machines-of-chinese-farmers

[Prof Marvel]

Apparently NASA is holding a contest (available to US citizens only) to see if someone can re-write a particular subroutine of a fluid dynamics programme suite called FUN3D. Given my experience in high speed flow (this was my exact areas of experitise),....

Go for it J.
Was Die Hölle, where's the harm?  it might be a foot-in-the-door !

No listen to this: The programme is written in FORTRAN     I think I may be one of the only few dinosaurs out there who still know how to program in FORTRAN (Fortran 77, 90 and 95 in my case). In fact all of my code in my college years was FORTRAN  

Yah, you and me both, brother.

The purpose of the contest is to speed up the execution of that particular subroutine by a few thousand times if possible.

Well, you know the math, so why not go forit? In fact you might try 2 ways
a) the way it's spec'd, all in Fortran, and
2) by using external calls to your own new code you know executes faster, like "C" ? 

Most of  "The Kids" will either stick with Fortran, or try to rewrite in JAVA and fail miserably since JAVA is 1000 times slower  !   

yhs
prof curmudgeon

[Prof Marvel]

My Dear J -

I went over to look at the links....

this *has* to be designed using parallel processing, nicht war? If so, perhaps the paralellisation is a bottleneck?
are they using flags, semaphores, shared memory, interrupts, wait-states, signals ( unix sig )?

or is it hardware specific, hardware bound or hardware independant? that alone can kill ya as we saw in the days
of the Sun "t-series". code had to be redesigned to take advantage of the "look-ahead" crap, pipes,  &etc...

If you can get a look at "the big picture" you can then see where the bottlenecks really occur and where any fixes/changes will actually help.
That "used to be" one of my specialties.

Spoiler (click to show/hide)

That part is what I "used to do" before Big Brother Ora castrated our support dept and threw all the sr guys
into the same pool with the newbie phone support people.

It was incredibly frustrating - i felt like the  depressed robot on Hitchikers Guide: overqualified and underutilized
having to field "first level" calls from CU in India and China, asking "how do I run this command" and
demanding free consulting and training that was not covered by support contract.  The best was to
                                    "teach them how to be a Unix Sys Admin over the phone"
and then explain to them how that is specifically excluded form their support contract, but I would
(gag) be happy to connect them with someone from sales to arrange to sell them a training class...

Then they would fill out a Customer Satisfaction Survey and give you a 1 (fail) out of 10 rating , which the new automatic metrics used to rate you on your job (instead of being  rated by an actual manager).

Gawd I'm glad I'm out!

Oh Dear - I looked at their links. Their process is nothing new, same old stuff touted as "new" in the 80's and 90' and 00's, just changed labels:

Agile Software - nothing new, except for the feature where it allows the customer  to continuously interfere.
                 Try that with your plumber and see how it works out.

Extreme Programming: again not much new - except for Pair programming

   Pair programming is an agile software development technique in which two programmers work together
   at one workstation. One, the driver, writes code while the other, the observer or navigator, reviews
   each line of code as it is typed in. The two programmers switch roles frequently

Oh yesssss I love having somebody back-seat driving looking over my shoulder criticizing every choice I make and
every fat finger typo that happens - We can see it works out great with Sheldon on "The Big Bang Theory"
I fear it would end in death and dismemberment!

What actually works better is "ego-less programming and peer review" where the team meets regularly and discusses selected code
without knowing who wrote it.

OMG then I see this:
http://wiki.c2.com/?TruckNumber
a bunch of real word Sheldons arguing the semantics over the "Truck Number" - ie exactly how many *essential* people you can afford to lose
on a project and not fail - thru vacation, accident, or  "just fed up and f***ing leaving"  (probably due to this exact topic being taken seriously) .

The answer is simple-
a) except for a one-man project, no single person is ever irreplaceable - it will cost time, but that's it.
b) dont put all your people in one basket vehicle

My Dear J -
On second thought, I am (again) glad I am out of it! its just SSDD.
go for the prize, but don't get invovled in the insanity!

yhs
prof marveling that he lived thru it to retire

[J. Wilhelm]

Well I know we're not gonna use vector computations  Of course, it requires parallel computation, but I suspect it is the algorithms in the subroutines which they are targetting. I'm not a computer hardware expert. That is not my department. I do physics. Direct simulation of high speed aerothermodynamics of a viscous compressible flow.  I optimize subroutines in the sense of the mathematical methods use to perform the computations. There should be a good deal of matrix and tensor operators coded in the subroutines. An upwind finite volume solver with an unstructured grid, adapting or not is standard for processing the Euler or Navier Stokes Equations in high speed flows.

There are methods to "compress" the processing of data, in terms of the mathematical physics, apart from the coding itself and no matter which computational bottlenecks you are facing in distributed (parallel) computing, it boils down to the physics of the problem to allow you the gratest possible time versus precision compromise. It may be up to the CFD subroutines to determine how fast and precise the program is.

[Peter Brassbeard]

Short of a wholesale redesign of the algorithm, you might find substantial gains reworking the order in which cells are calculated to increase the odds the data you need is in L1 cache.  But a thousand fold improvement in speed typically requires a new algorithm.  Fortran, especially run through a modern compiler, shouldn't be that bad for speed.

[J. Wilhelm]

Short of a wholesale redesign of the algorithm, you might find substantial gains reworking the order in which cells are calculated to increase the odds the data you need is in L1 cache.  But a thousand fold improvement in speed typically requires a new algorithm.  Fortran, especially run through a modern compiler, shouldn't be that bad for speed.

That's the thing. Fortran is so simple. First I need to see the First Principles theory used. I imagine that is boilerplate fluid mechanics. Then I need to see type of CFD DNS solver they're using, the grid generator and how it's used by the solver (where I suspect a lot of savings could be generated). There will be mathematical theory that can be applied to the grids in order to optimize the solution in every cell, and the solver it self, I need to look at it to see if there is a better mathematical method - though I suspect they've looked at that many times). Then I need to take a look at the subroutines. See what they're doing, and how convoluted the overall structure of the program is, and in what order things are called. Lastly all of what Prof. Marvel mentioned, would be looked at, by me or someone more educated in the hardware. Everything will be very modular, but looking at someone else's code is always excruciating.

[Banfili]

Awaiting further progress reports with considerable anticipation. 

[J. Wilhelm]

Awaiting further progress reports with considerable anticipation. 

Don't hold your breath! I don't know when I'll start and even if I could finish! I just want to take a look at it. I'd be advantageous though to have a coding expert though. I can do the physical theory, grid generators and the DNS methods but I suspect I'm a poor programmer,even if I know Fortran.

[J. Wilhelm]

Short of a wholesale redesign of the algorithm, you might find substantial gains reworking the order in which cells are calculated to increase the odds the data you need is in L1 cache.  But a thousand fold improvement in speed typically requires a new algorithm.  Fortran, especially run through a modern compiler, shouldn't be that bad for speed.

Actually a wholesale redesign is probably what is needed    that falls even more under my field of expertise  It's always easier to start anew, but understanding what the old code did so you can actually improve everything. Basically the best bet is to write a new solver.

[Banfili]

I have every confidence!
 

[J. Wilhelm]

For those mad scientists among us, I give you a genetically engineered cyborg firefly drone equipped with a tiny solar panel.

http://www.bbc.com/news/av/technology-40508099/living-dragonfly-drones-take-flight

Draper’s Dragonfl-eye is a living dragonfly with a solar-powered backpack that can control its movements using light pulses.

By genetically engineering the “steering neurons” of the dragonfly, researchers hope to be able to use the insect to get to areas larger drones cannot reach.

[J. Wilhelm]

Alright, here we go again. Elon Musk is out to surprise us again. He just unveiled a new reusable rocket hybrid he plans to have flying by 2022. That is not the surprising part. He plans to have the "joint strike fighter" equivalent of the rocket world, by having one single type of hypersonic /rocket hybrid SSTO vehicle capable of city to city travel, low orbital payload delivery, and... interplanetary travel. No joke.

He envisions nothing short of a 1999 Space Odyssey scenario with intra orbital travel,

He could be a genius or a mad man. While his achievement in reusable rocket technology is impressive, experience shows us that "joint" multi purpose vehicles are much harder to implement, because a lot of things can go wrong. For example the F-111 Aardvark (General Dynamics? Not to mention our own present day F-35 joint strike fighter by Lockheed Martin.

http://www.bbc.com/news/science-environment-41441877

[Prof Marvel]

Much as I would like to see it, I am firstly concerened about "commuter rockets" - mainly safety & pollution. rockets are notorious for using toxic fuels and polluting far more than other engines - and its all to cater to "other rich bastards" who think they are more important than the rest of us.

Also, I have become disenchanted with The Musker in general. He seems to get a wet dream in the shower and then makes huge announcements, sometimes claimng he already has "verbal government support" , which in reality is nonexistant.

Lastly, for all the happy happy joy joy stuff he has managed to pull off, (ie some small successes with his private rockets and the Tesla car)  his Tesla Electric Car is STILL powered by Chinese AA cells, and the slaves drones employees in his Nevada Tesla Sweatshop are being worked to death....

I find it is just history repeating itself ( as in the 180's -1890's prior to workers unions )

prof grumpy....

[J. Wilhelm]

To be honest with you I haven't read much about it. I'm unclear if this a single stage or multiple stage, boosters or no boosters, and just considering the fuel requirements for an interplanetary mission versus a low Earth orbit mission gives me serious doubts about the "one size fits all" claim. My understanding is that not only do you need to reach escape velocity for Earth, but then you will need to chose a transfer orbit, a fast and expensive one like the "figure 8" used for the Apollo missions or a slow cheap one like a Hohmann transfer orbit (a simple ellipse connecting the two orbits around the celestial bodies).

Take into account that as much as Musk touts himself as "chief designer," he is in fact no engineer. Others are doing all the work. He's more like Howard Hughes was, a millionaire with big dreams and bigger demands from his staff.

[J. Wilhelm]

On the battery issue it is important to follow the recent achievements by the staff of the very wise and very old Professor Goodenough at the University of Texas at Austin. Solid State batteries with liquid crystal electrolyte layers which not only triple the energy density, but prevents dendrites forming between electrodes - thus no spontaneous fires. Sadly Tesla Motors has not made an effort to acquire this technology.

[Miranda.T]

To be honest with you I haven't read much about it. I'm unclear if this a single stage or multiple stage, boosters or no boosters, and just considering the fuel requirements for an interplanetary mission versus a low Earth orbit mission gives me serious doubts about the "one size fits all" claim. My understanding is that not only do you need to reach escape velocity for Earth, but then you will need to chose a transfer orbit, a fast and expensive one like the "figure 8" used for the Apollo missions or a slow cheap one like a Hohmann transfer orbit (a simple ellipse connecting the two orbits around the celestial bodies).

Take into account that as much as Musk touts himself as "chief designer," he is in fact no engineer. Others are doing all the work. He's more like Howard Hughes was, a millionaire with big dreams and bigger demands from his staff.

Their biggest problem will be that, rather than using a small number of large engines (the Saturn V model), they will apparently be using 30-some-odd smaller engines. The massive challenge is getting all of these to work in synchronisation - the difficultly of this is what doomed Korolev's N1. Of course, technology has moved on rather a lot since the 1960s so maybe they can tame that issue.

In terms of multi-use, from the brief information I've seen the plan for the beyond Earth-orbit missions is to put the spacecraft into orbit and then move on from there at a later point - presumably after a refueling mission or two. Anyway, as I always say, before spending two to three years on a Mars mission whoever is going there should go to the moon first and use that to test and perfect the the systems that would be needed for the Mars mission. Better to spend 10-15 years thoroughly preparing for a successful mission than rush there and risk disaster.

Yours,
Miranda.

P.S. It's the 50th anniversary this year of the first flight of the Saturn V.

[J. Wilhelm]

To be honest with you I haven't read much about it. I'm unclear if this a single stage or multiple stage, boosters or no boosters, and just considering the fuel requirements for an interplanetary mission versus a low Earth orbit mission gives me serious doubts about the "one size fits all" claim. My understanding is that not only do you need to reach escape velocity for Earth, but then you will need to chose a transfer orbit, a fast and expensive one like the "figure 8" used for the Apollo missions or a slow cheap one like a Hohmann transfer orbit (a simple ellipse connecting the two orbits around the celestial bodies).

Take into account that as much as Musk touts himself as "chief designer," he is in fact no engineer. Others are doing all the work. He's more like Howard Hughes was, a millionaire with big dreams and bigger demands from his staff.

Their biggest problem will be that, rather than using a small number of large engines (the Saturn V model), they will apparently be using 30-some-odd smaller engines. The massive challenge is getting all of these to work in synchronisation - the difficultly of this is what doomed Korolev's N1. Of course, technology has moved on rather a lot since the 1960s so maybe they can tame that issue.

In terms of multi-use, from the brief information I've seen the plan for the beyond Earth-orbit missions is to put the spacecraft into orbit and then move on from there at a later point - presumably after a refueling mission or two. Anyway, as I always say, before spending two to three years on a Mars mission whoever is going there should go to the moon first and use that to test and perfect the the systems that would be needed for the Mars mission. Better to spend 10-15 years thoroughly preparing for a successful mission than rush there and risk disaster.

Yours,
Miranda.

P.S. It's the 50th anniversary this year of the first flight of the Saturn V.

Indeed. Lots of issues to tackle. Oh well. Much better see someone trying to accomplish big dreams, than people in a moribund space agency doing nothing. Half of my job as an engineer , I naively thought as a student, was dreaming. When adult life came around me, and as someone irreverently joked "took a dump on my head" I think a lot of possibilities were lost. I don't think that any Human Resources recruiters will ever know how many potential dreams are being bypassed


50 years. That reminds me how old I am... I'm a walking museum, practically.

[J. Wilhelm]

Oh yeah. I forgot. If you check the news lately, you'll find that Bombardier is now embroiled in a trade war between Canada and the US. A really serious sum of money in the billions of dollars between American Airlines and Bombardier is on the chopping block...

[J. Wilhelm]

Uber and NASA to develop Sky Taxis by 2020? I just hope they plan on developing 4 rotor VTOL vehicles and not the mostly dangerous 2-rotor design used in the V-22...

http://www.msn.com/en-us/news/technology/flying-taxis-uber-and-nasa-just-paired-up-to-make-that-happen/ar-BBEIAWl

Uber is taking a big step towards making its vision of flying vehicles a reality, announcing on Wednesday an agreement with NASA to develop a fleet of air taxis by 2020.

While the ride-hailing service and the U.S. space agency may seem like an odd couple, the relationship is necessary in order to keep everything running smoothly when Uber ultimately takes to the skies.

"The space act that we signed with NASA is initially about collaboration around air traffic management," Jeff Holden, Uber's chief product officer, told NBC News. With so many aerial vehicles flying at a low attitude, the collaboration will be necessary to help manage the skies.

Uber also announced that Los Angeles will be the second U.S. city where it will test its uberAIR service. Dallas-Fort Worth is the first U.S. launch partner, while Dubai will be the first global city.

Holden said he expects Uber will have its first flight demonstrations in 2020 and have the service commercially available by 2023. That's well ahead of the 2028 Olympics in Los Angeles, where the service could be especially useful as an already congested city prepares to host athletes and fans from around the world.

Uber's air travel initiative was announced last October with the promise of putting an end to long commutes, letting passengers hail an aircraft ride with the push of a button. In the case of Los Angeles, Uber has 20 strategically placed locations around the city for the Uber network.

The idea is to build a network of vertical takeoff and landing aircraft that would greatly reduce commutes, while also helping to ease vehicle pollution in major cities.

In a video released Wednesday by Uber, the company showed what it will be like to order a seat on one of the aircraft, which will take off and land vertically.

"It's an inspirational way to travel, too," said Holden. "You push a button, fly over the city, and you land. No volatility."