Question Perdix Shearwater AI

[grassybreakfast]

the length of the lockout is rather arbitrary and punitive

I think the idea is that they don't have enough data to be confident in their modelling of residual nitrogen in that case, so they do the safe thing and lock up for 24 hours assuming you'll be either close to 0 or be in a chamber somewhere by then.

Has the Buhlmann model actually been sufficiently tested in this regime such that confident predictions can be made? You can always program a computer to give you a number, but as a software engineer, we are taught that wrong data is even worse than no data.

 

[Belzelbub]

That sounds like a huge knowledge gap in training. A diver who doesn't even know that probably shouldn't be diving.

It’s not quite as simple as that. Many computers on the market have atrocious UIs, and the manuals are often not much help. So, the diver may not realize they have high conservatism on, and are surprised by a deco message on their computer, which they’ve never seen before, and which may not be as clear as it should be.

 

[Belzelbub]

Has the Buhlmann model actually been sufficiently tested in this regime such that confident predictions can be made? You can always program a computer to give you a number, but as a software engineer, we are taught that wrong data is even worse than no data.

To track a repetitive dive? Sure. I think so. The model has been out there for a while and is well known.

I’m mainly talking about deco stops that may not really be deco. The table posted earlier shows some clear differences among the models on the first dive. They disagree much more on subsequent dives. A highly conservative model, or setting, can result in deco where another computer treats it as NDL. 24 hour lockout reasonable in that case?

 

[grassybreakfast]

To track a repetitive dive? Sure. I think so. The model has been out there for a while and is well known.

To track residual nitrogen during the surface interval after surfacing with presumably super-saturated tissues and/or bubbles.

From Wikipedia:

The model assumes perfusion limited gas exchange and multiple parallel tissue compartments and uses an inverse exponential model for in-gassing and out-gassing, both of which are assumed to occur in the dissolved phase (without bubble formation).

Which, if my understanding is correct, means the model essentially assumes no bubble formation, which is probably not a fair assumption after deco violation.

I’m mainly talking about deco stops that may not really be deco. The table posted earlier shows some clear differences among the models on the first dive. They disagree much more on subsequent dives. A highly conservative model, or setting, can result in deco where another computer treats it as NDL. 24 hour lockout reasonable in that case?

Of course, the models aren't perfect, and we don't go from no deco to DCI from 0 min NDL to -1 min NDL. But still, that's the assumption of the model with the given setting, and getting the model to continue making predictions in conditions that it has predicted is outside of the model design assumptions would make me uncomfortable as a software engineer, and I assume that's why computers that lock up do that.

 

[tursiops]

the length of the lockout is rather arbitrary and punitive.

Actually, the length is based on data. The statistic is that (If I remember correctly) 80% of all DCS cases show symptoms within 6h, and 99% within 24h. So waiting at least 6h is to make somewhat sure you don't have a problem, and waiting 24h make darn sure. Also, you offgas nearly completely in 24h (with some deco models, 48h with others) so waiting that long to dive again is the only way to ensure you are "clean" before you start on-gassing again.

 

[Belzelbub]

To track residual nitrogen during the surface interval after surfacing with presumably super-saturated tissues and/or bubbles.

The assumption that tissues are super-saturated may be incorrect. Each model has a line that, if crossed would mean that deco is required. And on the other side of that line, some of the tissues would be super-saturated. Conservatism settings put a buffer between that line as it's generally not advised to always be on the edge. So, a deco violation could be well on the safe side of things depending on how conservative the computer is set.

For example. In the following dive, what should the punishment be if I were to have completely skipped that first deco stop. Also, looking at it, I completely missed that last deco stop. Per Subsurface, I needed 5 minutes at 10 ft. Should I have been locked out for 24 hours.

How about on this dive? All good, right?

I'm sure you've realized, but they are the same dive. I just cranked the conservatism way up in the first image. That's my point. If I opt for higher conservatism, should I be locked out on a dive that would be well within NDL? NDL on this dive never dropped below 6 minutes.

 

[Belzelbub]

Actually, the length is based on data. The statistic is that (If I remember correctly) 80% of all DCS cases show symptoms within 6h, and 99% within 24h. So waiting at least 6h is to make somewhat sure you don't have a problem, and waiting 24h make darn sure. Also, you offgas nearly completely in 24h (with some deco models, 48h with others) so waiting that long to dive again is the only way to ensure you are "clean" before you start on-gassing again.

What I was trying to say was that the lockout doesn't take the severity of the violation into account. It could be a missed deco stop with conservatism off, or a shortened deco stop with high conservatism. That may not warrant the full 24 hours. Maybe skip the next dive, that would probably make sense.

 

[grassybreakfast]

I'm sure you've realized, but they are the same dive. I just cranked the conservatism way up in the first image. That's my point. If I opt for higher conservatism, should I be locked out on a dive that would be well within NDL? NDL on this dive never dropped below 6 minutes.

Sure, if the model thinks none of your tissues are super-saturated (you are within the buffer region), maybe a lockout is not warranted. However, keep in mind that the buffer is not just for human error. It must also account for model inaccuracy, and inaccuracy due to circumstances. Obviously model accuracy doesn't improve when you change to a lower conservatism setting. It just allows you to take more risk.

But once you go into super-saturation, you basically don't have a model anymore, and I think the right thing to do is 24h lockout for more or less complete degassing.

Now we have computers that may be a bit too conservative (lockout in the buffer region), and computers that I think are a bit irresponsible (not locking up when you surface in super-saturation). If you surface in super-saturation on a computer that doesn't lock up, it's also in gauge mode. You just don't know that, because it's still giving you numbers that may not actually mean anything.

 

[rhwestfall]

Maybe skip the next dive, that would probably make sense.

No way the attorneys would accept something as flexible....

 

[Heat Miser]

I have this model - far too advanced for me, but I can read the display clearly underwater - HA! It's embarrassing to pull out this high tech computer when I am a simple AOW recreational diver. It's familiar and I can easily use the basic functions I need, which is why I hesitate to replace it. Looking to possibly sell it and get a simpler more equally matched air integrated computer. Thoughts?

The pro at the shop, lent me his to test before I bought one. He said he had never changed any of the settings. Never dived it in anything but Recreational mode on Air. But still raved about it. So clearly it was advanced enough for him straight out of the box.

I have fiddled with mine alot more. I use it for CCR. Nitrox OC, Sidemount. When on OC I like having the SAC rate calculated because I can never feel wether I'm breathing hard or slow underwater. Im also toying with attaching two AI transmitters to my rEVo and using it instead of the dream/spg.

I think what is great about it is that its so scaleable for all types of diving. Why sell?

I will say I like my Garmin Descent too, though