The Titan Sub - An Engineer's Hot Takes

I know this is the millionth Titan Sub post but I honestly have to throw my thoughts in the mix because it's my area of expertise

I'm an engineer, and I've got experience with...

System safety engineering for exactly this kind of "move fast break things" startup. And dealing with the "personalities" you meet there lol
Safe and reliable controls for robotic systems
Life-critical roboticized vehicles

Plus my dad's a submariner and took me on sooo many submarine tours when I was young, explaining the system design principles.

And if you thought that Titan Sub was negligent? Yeah. Fuckin horrifically so, to be honest. I've seen some scary shit in my career, to the point where I've quit a job over it. I've got friends and colleagues who've done the same. And this? Worse than any of that.

Most times I've seen a big safety issue? You can tell the company was trying, and there was one thing they missed or didn't understand. Maybe two. But here? There's just so much wrong. But to name a few things:

"It's got one button. And that's it."

See a difference between that and a real submarine control room? A couple hundred extra buttons? Those are pretty important.

In well-designed submarines, there's redundancy - multiple ways to do any essential submarine-control task. I've heard at least 3 is industry standard for submarines, where the loss of an essential function will be deadly. It's even common to have multiple control rooms!

And the mechanical & electrical bits that make those 3 options possible? They're physically separated so that if something like a fire damages one? The other two will be fine.

And for any high-level task, that is broken into multiple smaller ones? There are controls to accomplish each small sub-function individually in case of emergency.

One button? One? ONE? I literally thought that was satire when I first saw it. For fucks sake.

"So we run the sub with this game controller"

You've probably heard about this one.

But using a game controller for robotic applications, even safety critical ones, isn't unusual! Though it usually takes a multi-million dollar engineering effort to make a system that uses them reliable for a safety function. Exhaustively going through every possible way it could malfunction, and finding a way to address each and every failure mode.

I'd be surprised if they did that - they look like they're violating a few best practices.

First, a wired connection is standard, because it's less likely to drop. Titan's off-the-shelf Logitech controller was using 2.4 GHz WIRELESS tech. It's not impossible to do safety critical communication wirelessly, but that involves specialized safety-rated radio equipment. Just taking a look at what I believe are Logitech's specs for their 2.4 GHz wireless stuff... Uh oh. This ain't it.

Using the same frequency band as Bluetooth and Wifi? Like other devices might be using, and creating interference in that band? I'm reminded of a railway radio failure I saw, where a manufacturer's radio frequency band overlapped with a country's cell signal band. The trains would shut down on major holidays because the airwaves were overwhelmed! Luckily they were engineered well enough to just stop in place, not crash.

That's... Also bad. For safety-critical controls, you want to send the signal right away. TDMA means "Time Division Multiple Access" - AKA "Every user takes turns (time division) so we all can use the same airwaves (multiple access)"

You don't want YOUR FUCKIN SUBMARINE CONTROLS taking turns with anything. They need a dedicated and reliable communication channel. Another safety engineering principle? Isolate your important shit from your unimportant shit. That way your unimportant shit can't fuck with your important shit.

This Whole Damn Lawsuit (pdf link)

This is a lawsuit about the company firing an employee for raising safety concerns. Really valid safety concerns, IMO.

This guy seems really qualified. I'd trust his assessments.

Acoustic monitoring for the kinds of sounds that usually only come shortly before a catastrophic failure? Great - they basically made a "You're going to die now" bell. Too little, too late. Lochridge was right - the hull needs to be tested beforehand!

And flaws in the carbon fibre hull material? A famously brittle material? That isn't generally used for pressure vessels? And they kept on using that instead of any of the industry standard materials?

Invisible or barely visible tears might not seem like a big thing, but they're scary to an engineer! They're often the warning alarm that your material isn't holding up and is losing structural integrity.

I just don't understand the choice of carbon fibre at all - For a vessel subject to high forces, you'd want a material that allows for "elastic deformation" - or the ability to "spring back" to shape instead of break. For example, metal, which is what every submarine I've ever seen is made of. "Carbon-reinforced plastic collapses "catastrophically", says Professor Brizzolara."This is because the material is not ductile like metal alloys and therefore it 'catastrophically' implodes."

Out of everything here - this terrifies me the most. I have no words.

Ok, I do have some... It's bleak when "hazardous flammable materials in a submersible" is the least worrying thing on a page.

As for the viewport window, there's this thing called a factor of safety. When you engineer something, you don't just make it strong enough, you make it several times stronger than you think it needs to be.

How many times stronger depends on the industry and application. "Pressure vessels use [a factor of safety of] 3.5 to 4.0, automobiles use 3.0, and aircraft and spacecraft use 1.2 to 3.0 depending on the application"

PRESSURE VESSELS USE 3.5 TO 4.0. That window, going 4000m deep? Should have been rated to 14,000 meters at least. They went with 1,300!?!?! That's UNDER ONE TENTH of what it should have been. And I'm not sure if it was ever replaced.

Fuck anyone who doesn't realize safety standards are written in blood.

And I mean that very literally - In the US, safety regulation almost never gets enacted until people die, and those deaths are proven in court, beyond a reasonable doubt, to be due to corporate negligence. Sometimes a lot of people have to die before that standard of proof is met, as companies will dodge culpability for as long as possible. It's one of the ugliest truths about my industry. And for a company to ignore those regulations that came at the cost of people's lives? I'd say they have blood on their hands.

Yeah, in my professional opinion, fuck these guys. They fired their safety guy for doing his job, they didn't work with the certification agency. I hope they're held responsible.

Loooong story short?

I'm genuinely having trouble finding much this company did that wasn't horrifyingly negligent. I'm ending the examples here, but honestly I could keep going.

It's fuckin terrifying they were allowed to do this in the first place. And while I kinda get the "lol rip billionaires" reaction - A company should never be allowed to put people's lives at risk like this. Because if engineering negligence can kill billionaires? It sure as hell can kill any of us with far fewer consequences and far less media attention.

The warning signs were there for years before the accident, and that's the saddest part to me. This was so easily avoidable.

I feel for the guy who tried to warn people - it's a nightmare situation, to fear something you worked with is dangerous, and see your employer (or ex employer) go ahead with it anyway.

What do I figure happened, in the end?

My top guess is a hull failure, resulting from repeated stress of compression and decompression of the carbon fibre material, leading to internal tears or fractures that compromised it's structural integrity, and resulted in a sudden and catastrophic implosion. (An implosion is the opposite of an explosion - it is when something very quickly collapses in on itself, like a can being crushed. Deep underwater, the high pressure of all the water above can cause an implosion.)

Second guess? That window, if they never changed it, could have failed. Same sudden implosion in the end. But I'm waiting for a forensic accident report to say for sure.

(part 2 - Hatches & Human Factors - They bolted the sub shut from the outside - and let the passengers pilot it?!)

legal disclaimer below the cut.