Two divers critical - Hawaii

[Superlyte27]

It can and it does. The Liberty in surface mode "is designed to maintain a pre-set oxygen volume of 23% by adding oxygen if the O2 sensors detect a drop below this amount. This addition of oxygen functions automatically and is intended as a safety feature to prevent depletion of oxygen from the loop while the unit is running unintentionally in surface mode." (from the official Divesoft report cited above)

But it can't do that if the O2 is turned off.

OMG!!! EXACTLY!!!

 

[doctormike]

I think I just read my answer in Divermike's post. Breath the loop for 5 minutes to make sure it holds the setpoint. So the alarm just doesn't sound above 1.5m depth.

Again, I'm not trained on the Liberty, but from what I have read that while there are four visual alarms for low PO2 in surface mode, but the audible/tactile alarms only happen in CCR mode.

You can switch the rebreather to CCR mode yourself on the surface (so all of the alarms will be enabled), or it will automatically switch to CCR mode at 1.5 m if you forget.

 

[Superlyte27]

This is no different than packing a scrubber. Oops, I forgot to pack a scrubber. We should have an alarm for that. How interesting. There is actually an alarm for that. Guess how many times it gets ignored? Lol

 

[fuzzybabybunny]

Just like you can’t fix the rare chance that a plane will fall out of the sky.

Continuous improvements are happening all the time to make planes better and better. You can't eliminate the chance, but you can certainly make it rarer and rarer. At the end of the day every engineer knows things will fail but it doesn't stop them from doing the hard work of making it better and better regardless.

An engineer saying "the user of my device was just stupid" is an unprofessional cop-out if they actually *DIED* on your system. Are there limits to how much good engineering can prevent death? Absolutely. But this scenario doesn't feel like one to me, despite the fact that I AGREE WITH YOU that this particular diver made a lot of mistakes that I don't ever see myself making.

Years ago a plane crashed because the airspeed indicators froze up and the pilots couldn't determine if they were going fast enough to maintain lift. A number of errors resulting from this compounded and these highly trained pilots and their passengers all died. So engineers examined the problem and built heating units to keep the airspeed indicators from freezing. Then another plane crashed due to the same reasons because the crew forgot to engage the heating system. I'd imagine an automated heating system would be in the works as well as a mandated checklist regardless for the crew as well as alternative ways of determining airspeed, like GPS-based groundspeed calculations combined with meteorological estimates of regional wind speed. Examples like this of engineers making crashes rarer and rarer in conjunction with human training embody the entire airline industry.

 

[Superlyte27]

I used to build houses...
We had a saying at the closing walkthrough
“We could build you the perfect house, but then nobody could afford it”.

Take a rebreather course.
Then you’ll understand where engineers should be spending their time. Not on things that are controllable, like turning on a computer or opening a valve, but things that aren’t controllable right now, like cell linearity. cell decay, CO2 monitoring, flood tolerant sorb, etc.

 

[Superlyte27]

It’s funny... I just read a thread the bunny started about the cost of diving a rebreather, specifically a Kiss SCR.

If you don’t like the engineering of the Liberty, you’re REALLY not going to like the Kiss.
I’m certified on 5 units, including the Kiss. Of all the units, the Liberty is the one everyone is saying is already over engineered. The Kiss would be the unit with the least amount of safety features. You can even dive it without a HUD. So, just one visual alarm, instead of 4

Anyway, I’m out of here. I gotta go to sleep or my wife is going to kill me. I’m done debating on this topic, but the offer stands.

 

[telemonster]

Continuous improvements are happening all the time to make planes better and better..

Geek here with a pretty solid understanding of CCRs. No cert (have inspiration classic + hammerheads but no place around me to get cert! So I backburnered the idea.)

1. The modern controllers aren't much bigger than a normal dive computer. So power conservation is a thing. Companies are doing amazing thing with current run times on the computers but to have it sampling all the time when it's in sleep mode would be a pretty big power drain. Maybe you could sample the ADC channels that read the o2 cells every minute or so and try to catch someone pre-breathing a unit with controllers off. But it would crush batteries. Looking at a MEMS device for movement wouldn't be a huge help since boats rock around, but maybe super sharp angles could be detected then you watch o2 sensor channels in a sleep mode. Power drain would be a problem here. Batteries are already a pretty huge swing based on water temps and battery chemistry and types. Then people would just breathe down the loop to power on the unit. More bad habits

2. o2 valve was off, this unfortunately has happened a lot. Complacency. Usual checks would come into play here. Most ideas I can come up with here have worse failure modes. But yea, actively checking the tank valve electronically would be one way but now you've limited yourself in terms of tank swaps. Unit would look like the back to the future delorean when it comes to all the cables everywhere, or like a Revo (ducking.)

Pre-dive checklist should catch the failures.

 

[Superlyte27]

LOL, love the revo comment. You’re my new favorite poster.
As many know, I dove the rEvo for 4 years before I learned how much is sucked compared to other CCR’s. But that’s another story.

 

[doctormike]

Geek here with a pretty solid understanding of CCRs. No cert (have inspiration classic + hammerheads but no place around me to get cert! So I backburnered the idea.)

1. The modern controllers aren't much bigger than a normal dive computer. So power conservation is a thing. Companies are doing amazing thing with current run times on the computers but to have it sampling all the time when it's in sleep mode would be a pretty big power drain. Maybe you could sample the ADC channels that read the o2 cells every minute or so and try to catch someone pre-breathing a unit with controllers off. But it would crush batteries. Looking at a MEMS device for movement wouldn't be a huge help since boats rock around, but maybe super sharp angles could be detected then you watch o2 sensor channels in a sleep mode. Power drain would be a problem here. Batteries are already a pretty huge swing based on water temps and battery chemistry and types. Then people would just breathe down the loop to power on the unit. More bad habits

2. o2 valve was off, this unfortunately has happened a lot. Complacency. Usual checks would come into play here. Most ideas I can come up with here have worse failure modes. But yea, actively checking the tank valve electronically would be one way but now you've limited yourself in terms of tank swaps. Unit would look like the back to the future delorean when it comes to all the cables everywhere, or like a Revo (ducking.)

Pre-dive checklist should catch the failures.

1. Not sure what you mean by sleep mode, but the discussion so far about this case was about how the unit was in surface mode. In surface mode it continually reads and displays PO2, 4 places. Do you mean someone doing a prebreathe without the unit being powered up at all...?

2. Stick one of those AI transmitters that are so popular these days on your O2 first stage. Now you have data to tell if the valve is on with no wires, no problem swapping tanks.

Or, you could just actually do the preflight checklist. Tell yourself “if I’m gonna dive this thing, I’m committing to that checklist. Before every dive.”

 

[KenGordon]

I'm guessing you're not a trained programmer? Or an engineer? Trust me the pain in this conversation is mutual.

We both have the same goal in that we don't want to see anyone die. Only difference is one of us chooses to just blame people instead of entertaining the thought that maybe, just maybe, we can engineer a better system. Maybe we can prevent scenarios like this from happening through improved product engineering while at the same time NOT have your precious oxygen sensors burn up because your unit's a little wet in the trunk of your car. Maybe these two goals aren't mutually exclusive.

I am a programmer with decades of dealing with building complicated systems for users who expect the speed of light to change, time travel and other ‘but surely...’ requirements to be met.

There is a history of rebreathers trying to be clever. The Sentinal tried very hard and failed. The Poseidon tries still, AP have a bunch of features like the CO2 alarm which are more or less successful. If the Liberty fails it will be because it is too complicated and at the nanny end of the spectrum rather than because it is overly simple without obvious features.

The problem here was failure to have O2 and failure to monitor ppo2. How can we engineer that out? We can easily tell if o2 is flowing (they don’t but it would be easy enough if it helped). So now the question is ‘should o2 be flowing?’ When would that be? Normal operation. When is that? Wet? How about the prebreath? Back on the boat? Deep? How about bailed out? Moving about? Back of car, plane hold, walking between the car and the boat? Maybe I take all those inputs and feed them into some neural net which we train by taking data from all the users. Now when it alarms we have really no idea why.

Next we have the human reaction to alarms.

1 - too many alarms, never seen a real one yet, so ignore it/work round it.
2 - no false alarms so rely on the alarm spotting the problem so I don’t need to worry.

Perhaps the user in this case fell into 2.

There is occasional fierce debate about the spectrum of electronic aids in CCR. From a fully manual unit where the user understands that either they add o2 or die, through the JJ where you need to check the HUD and handsets to notice ppO2 FAILURES but which maintain setpoint, finally to machines which scream when they detect an issue.

I have a machine which relies on red lights to tell me the ppo2 is wrong. Some days I think an audible alarm would be good, but if I had such an alarm would I check for myself so often? So now I am being a bit lazy and the electronics fail completely?

Building a fool proof rebreather is not just about code, controls and monitoring. It is about the whole system including the user. The training and user attitudes are key. Planes may be safer than ever before but we still don’t let just anybody fly an airliner with 300 people on board.

Recreational divers have been using them for 20 or 30 years now and the technology has improved enormously. However, even small incremental ‘improvements’ by the biggest manufacturer have a hard time. Look at user reactions to the AP CO2 monitor.

There is also an argument made that if you can’t save everyone whose gas is off or who assembles the loop backwards then rebreathers are just too dangerous and should not exist.

Hopefully nobody buys a rebreather thinking ‘I might leave the gas off or assemble this backwards’. People should have a good think about whether that is a risk for them personally and we ought not to cast aspirations on those that do think that, or find out the hard way.