Diving with gradient factors for a new recreational diver

[crofrog]

Feel free to dive whatever GF you prefer.

Sure, but there are other risks than just DCS, like the boat leaving you if you decide to run 100/40 or something

 

[L13]

Yes, but if the expected incidence "is usually assumed to be one hit in a few thousand dives"(*) at GF100, and lowering it to GF70 results in one hit in a few thousand and one dive, then why bother. It maybe would matter if it makes it one hit in a few tens of thousands of dives, -- depending on one's personal take on risks, -- but we have no data to show it's one or the other, or anything in between.

*) The Theoretical Diver – Theorizing about scuba diving

Let me see if I understand your logical process:

• If we don't know the value of X
• Then X could be a value I made up
• Therefore it is illogical to behave as if X is a value consistent with what we do know and not my made up number
• Therefore we should act as if the value of X doesn't mater

 

[L13]

I get 32 minutes of total deco for 50/75.


Is the difference between a gf high of 85 and 75 significant? What is the probability of being bent at 85 vs 75?

Actually, you are right. I was trying to get your result of 27 min for the 20/85 using SubSurface and fiddled the numbers till it matched (which it didnt initially) then used those number with 50/75.

I get 31 min deco for 20/85, and 31 min deco for 50/75. But we are probably using different values for ascent and descent rate, and maybe other parameters.

To echo you question: Is the difference between a GF low of 20 and a GF low of 50 significant? What is the probability of being bent at 50 vs 20?

The answer to both questions, according to the most recent significant studies by NEDU, is that lowering the GF High reduces DCS more than raising GF low (at least till GF low is ~>50). But exact details require more research.

If you don't reduce GF high when you raise GF low, you likely get the worst of both effects.

 

[dmaziuk]

Let me see if I understand your logical process:

• If we don't know the value of X
• 
  
• Therefore it is illogical to behave as if X is a value consistent with what we do know and not my made up number

Close but no cigar. Better luck next time.

We don't know the value of X at Y=100, nor how X scales as we change Y. Therefore we assume reducing Y by N points results in a meaningful reduction of X.

 

[crofrog]

To echo you question: Is the difference between a GF low of 20 and a GF low of 50 significant? What is the probability of being bent at 50 vs 20?

I don't think that data exists, which is why I'm not saying it is "significant."

Just to be clear, I'm not arguing for or against "deep stops" or whatever gradient factors you want to use. I was pointing out that on even dives in the "tech 1" range, I don't think there is much of an actual difference, and in the recreational range even less so.

The answer to both questions, according to the most recent significant studies by NEDU, is that lowering the GF High reduces DCS more than raising GF low (at least till GF low is ~>50). But exact details require more research.

That isn't what the NEDU study said, though. It used VVal-18 and BVM(3) to build the schedule, and "gradient factors" were not a part of the study. You can use GF to modify Bulhman to make it close to what the study did, but a Bulham schedule is never going to match the schedule used in the study. Unlike Bulhman models in the study, they did not take into account the extra on-gassing that occurs with the deeper stops because they were looking at the "efficiency" of the decompression algorithms.

If you put the NEDU schedules in subsurface you'll see that the "deep stop" profile had surfacing GF of around 76% and the shallow stop profile had a surfacing GF of around 47%.

NEDU Shallow stop:

NEDU Deep stop:

Just for comparison, a 20/85 schedule has you getting out of the water 22 minutes faster than the NEDU study profiles

and a 50/75 schedule has you getting out 2 minutes faster...

So, is 50/75 safer than 20/85? Maybe you could even say "probably," but there isn't research that directly supports that, and is any of this relevant to recreational diving?

In that same vein: If we ignore the efficiency of the deco and say all we care about is surfacing GF, is a dive with the GFs of 20/75 and a total decompression time of 210 minutes more or less safe than a 75/75 dive with a total decompression time of 175 minutes?

Another interesting tidbit if you move the GF low around so that the first stop depth is similar to where the NEDU study was, and move the GF high you can get the same ending GF. It is right around ~48/77, although the profile remains well below that ceiling until the 10ft stop.

And just for others, the NEDU profiles bear little to no resemblance to a typical "technical diving profile. Here is that same 170ft for 30-minute dive done with 18/45 and Nx50 for accelerated decompression.

 

[dmaziuk]

... You can use GF to modify Bulhman to make it close to what the study did,

You can not because

a Bulham schedule is never going to match the schedule used in the study. Unlike Bulhman models in the study, they did not take into account the extra on-gassing that occurs with the deeper stops because they were looking at the "efficiency" of the decompression algorithms.

You can't make a meaningful comparison without some common base -- for staged decompression dives that is usually the TTS.

And as you say: you can't approximate their profiles with ZH-L without going over 100 on GF High. And if you believe Ross, you can't do it with VPM-B without pushing some of its parameters out of bounds either.

 

[L13]

Close but no cigar. Better luck next time.

We don't know the value of X at Y=100, nor how X scales as we change Y. Therefore we assume reducing Y by N points results in a meaningful reduction of X.

I knew I was close! (And who needs cigars, they are bad for your RMV)

We have a rough approximation of X at y =100, we know that X=0 at Y=0. We have lots of evidence and theory that X increases monotonically as Y increases. We have lots of evidence and theory that the rate of increase of X with Y increases faster as Y increases.

Given all that, any rational person can conclude a lot more about the relationship between X and Y than:

"Any made up number could be right, so why bother."​

 

[L13]

That isn't what the NEDU study said, though. It used VVal-18 and BVM(3) to build the schedule, and "gradient factors" were not a part of the study ...

I liked most of your post, but this line of thought bothersome. Just because a study does not test a particular algorithm does not mean that it does not provide meaningful data for evaluating it. It is not in any way conclusive evidence, but it does give rational people enough information to draw broad conclusions. It provides evidence validating certain theories and invalidating others. It suggests which parameters have greater impact and which have less. Etc.

If the underlying theory of Buhlmann is even close to correct:

• Gradient Factor is the equivalent of "Normalized Tissue Supersaturation".
• Higher tissue supersaturation is more dangerous than lower tissue supersaturation.
• More time spent at high tissue supersaturation is more dangerous that less time.

Deep stops are based on the theory that micro bubbles resulting from high-ish supersaturation at depth, which will expand as we surface, are more dangerous than bubbles formed at or near the surface from higher supersaturation. The NEDU data provides evidence that higher supersaturation is a greater risk. While the NEDU research did not test GF's directly, we can logically conclude:

• GF's that try to reduce deep bubble formation are less likely to prevent DCS than GF's that try to reduce peak supersaturation.
• For any given amount of time (and gas) spent on deco, time spent at deep stops will result in higher peak supersaturation and higher risk of DCS than spending that time at intermediate and shallow stops.

 

[dmaziuk]

I knew I was close! (And who needs cigars, they are bad for your RMV)

We have a rough approximation of X at y =100, we know that X=0 at Y=0. We have lots of evidence and theory that X increases monotonically as Y increases.

Everything else in this theory operates on log curves, so it's just as likely that this one's a log curve too. That would mean first halving of the M-value results in, what, 4% reduction of risk? I'm sure it's worth it, better safer than safe, right?

 

[L13]

Everything else in this theory operates on log curves, so it's just as likely that this one's a log curve too. That would mean first halving of the M-value results in, what, 4% reduction of risk? I'm sure it's worth it, better safer than safe, right?

Are you really that ignorant? Or just pretending?