Hacker News Reader

Hacker News

Comment by charliepark | Hacker News Reader

Full thread

charliepark

This looks really interesting. "A series of eight articles … from the March 1970 issue of Astronautics & Aeronautics."

Quickly summarizing the table of contents and some of the internal content, these seem to be the main takeaways:

• Create design principles that stress simplicity.

• Test exhaustively.

• Develop mission rules, procedures, and plans for potential issues ahead of time; iterate as new information comes in.

• Execute rigorous post-mortems.

• Prioritize flexible yet disciplined management.

It's notable that this was published before Apollo 13. It'd be interesting to see what changes Kranz et al. would make to these articles after that.

View on HN

> Test exhaustively.

> It's notable that this was published before Apollo 13

One of the causes of the Apollo 13 service module explosion was damage caused during a pre-flight test. The SM included a pipe whose only role was to drain out oxygen after pre-flight launch-pad tests. It had no in-flight role. The pipe was damaged during assembly (nobody noticed) and failed to drain the oxygen tank. The test team instead decided to drain the tank by heating the oxygen using an in-tank heater and in effect boiling it off. The heater temperature gauge didn't show (it was badly designed) that the heater had in fact got so hot that it had melted electrical insulation on a fan in the oxygen tank. All of this was pre-flight.

56 hours into the mission, the crew used the fan to stir the tank, causing sparks and the eventual explosion.

All of these apply to software development as well, but especially that first point. I'm on a project right now that has a bunch of things that can be simplified. Some duplicated state, multiple code paths for computing the same thing, etc. When problems arise it's very tricky to debug and it's almost always caused by that complexity. The temptation is to just add another layer of checks or validation on top of it or make a slight tweak. But of course the best path is to just roll up your sleeves and work to remove as many moving parts as you can. Most of our bug fixes have actually resulted in substantially fewer lines of code. Anyway, the point is to always try and keep things simple, and even better to make sure that principle is followed in the initial stages of design.

I've heard or read of this before, the first bullet, and can't help but to compare this to what has happened with the 737 Max 8 current issue.

From the PDF comes: "Apollo gains a measure of simplicity from features simple both in design and operation, complex in design but simple to operate, or simple by being passive in function. The concept of simple design and simple operation is best illustrated by the Apollo rocket-propulsion systems (fig. 2-2)."

For a machine as complex as the Saturn 5 rocket, it is truly a credit to the builders that it never suffered a major failure in any of the manned missions. Same for the LM rocket which returned the astronauts to the command module. Both had to absolutely work and both did.

As John Glenn quipped "It's a very sobering feeling to be up in space and realize one's safety factor was determined by the lowest bidder on a government contract."

Gus Grissom, Ed White and Roger Chaffee :(

I do get your point, and it's amazing it worked as well as it did. Space is tough, and things are likely to go wrong, just as they still do with civil aviation.

Silly nitpick: Apollo 1 was a Saturn 1B not a Saturn 5.

https://en.wikipedia.org/wiki/Apollo_1

It also wasn't a "mission". They were performing in a test. And, in any case, the fire wasn't a fault of the rocket; it was a fire in capsule.

This document must be post Apollo 13. Copied verbatim from the linked PDF:

> Even the Apollo 13 oxygen tank rupture, by far the most critical problem of any Apollo mission to date, was overcome by relying on preplanned emergency procedures and the resourcefulness and ingenuity of the astronauts and the ground support team.

Reply on news.ycombinator.com