Article on deep stops

[JonG1]

Also typically what dil do you use for what depths please?

 

[rjack321]

Thanks rjack thought i'd read about it somewhere but couldn't recall where.

Out of interest on CCR do you elevate ppo2 on deco above your core setpoint or stay at 6m instead of 3m for last stop or stay on loop with high o2 on surfacing?

Depends on the dive. For cave diving I usually dive ppO2 1.2 for 3 to 4hrs. I have a few 5+ hr cave dives at 1.1 Usually if I am nudging into 90+% CNS I try not to go crazy on deco ppO2. So I might bump it to 1.4 and let it decay to 1.0 then repeat (I have to do something on long decos to hold my attention anyway)

For wreck dives the BTs are short so I run 1.3 mostly and bump to 1.4 for deco but it rarely stays that high, usually decaying to 1.2-1.3. I do run live deco but my backup computer is fixed at 1.1, 1.2 or 1.3 depending on the dive. My live computer and backup are usually within 5 mins of each other either way so I suspect its just mental goofing off lol

Dil is a mixed bag. When I get commercial fills or in cave country where dil = BO I run 32% for most of my <100ft dives. A few times I have used 18/45 or 15/55 as dil/bo. Most of my wreck diving I use what I have scavenged at home or get topped off. Usually much higher helium than I would dive on OC. 16/50 is common as my standard "wreck dil" down to about 210ft. If it's not too deep but below 100ft I sometimes use 21/35ish too. I've done a few 10/70 dives below about 240ft - I wasn't sure what we'd be diving so I filled for all possible contingencies.

I've only done one BO ascent from trimix depths so essentially no data for that.

 

[JonG1]

Thanks rjack, coming back into diving after a long lay off, I have no benchmark to reference some decisions against, such as GF's because they simply didn't exist previously and having bubbled before with neuros (albeit with a PFO), I am cautious of straying too close to the line so it's interesting to reference other people's experiences.

 

[rjack321]

Thanks rjack, coming back into diving after a long lay off, I have no benchmark to reference some decisions against, such as GF's because they simply didn't exist previously and having bubbled before with neuros (albeit with a PFO), I am cautious of straying too close to the line so it's interesting to reference other people's experiences.

I have a bit over 200hrs on CCR now, both manual (kiss) and electronic (Meg) - about 55 hrs/yr mostly in 4-8C water with 45-75mins of deco.

feel way better than I did on OC, but I also use suit heat generously and find that the warmer I keep myself the better I feel ache-wise and energy-wise. I dont have any hard data for me, but I know I was very cold on my one chamber ride dive since I had partially flooded suit, it was January or February, and I recall snow on the ground. It was a ~150ft dive on 21/35 with 50% for deco and back in 2006ish I was still following some (now known to be nonsense) thoughts about oxygen windows. I dont have a profile but I wouldn't be surprised if it was 5/105 or something like that.

 

[leadduck]

Chambers are not the ocean, blood distribution among many other factors are vastly different when actually submerged.

Buhlmann's experimental dives in lakes and ocean were done to 350m (1150ft). He worked also with Hannes Keller, for his record dives to 230m (750ft) in 1957 in the Lago Maggiore, and to 305m (1000ft) in 1962 near Santa Catalina island.

Buhlmann's work was funded by Shell Oil. They were not interested in recreational diving but only in very deep mixed gas saturation diving.

He was surprised when he learned that recreational divers use his model. Nowadays the model is used in many recreational dive computers and ironically people start to believe that this is a recreational diving model and doubt whether it's suitable for deep diving ...

See also Milestones of the deep diving resarch laboratory Zurich.

 

[CandiveOz]

When I finished reading the David Doolette’s article, I asked myself; does this make sense? And after reading the post by @atdotde – the theoretical diver, I really wanted to gain a better insight.

In summary, David Doolette calculates the average of all the “b” parameters which is the inverse the M-line slope which works out to be roughly 83% and then applies it to the GFhi to determine what the GFlo would be. For him, his preferred GFhi is 85%, therefore 85*0.83 = 70% is his GFlo.

His rational to apply the average “b” of 0.83 to GFhi is that it roughly counteracts the slope of the Buhlmann “b” parameters and brings it more in line with the US Navy decompression algorithm whose compartment M-line slopes are equal to or close to 1.

Now this is a fundamental difference between Buhlmann and the US Navy. For Buhlmann the slope of the M-line in each compartment is greater than 1. Even when Albert Buhlmann himself added conservatism to his models from A, B to C he changed the “a” intercept but didn’t touch the slopes. Gradient factors, by their implementation, change both the “a” intercept and slope.

My first reaction was, if Buhlmann was wrong about his slopes and there is such a concern with deep stops, why mess around with gradient factors tweaking Buhlmann, when you could just make the US Navy algorithm open source and we can then all benefit and throw out Buhlmann and gradient factors and use a more appropriate model – if that is the case. But I guess that is not going to happen any time soon even if David Doolette has some pull at NEDU.

The problem with gradient factors is Eric C. Baker developed them specifically to generate deeper stops. Therefore, if your objective is to generate shallow stops; don’t use gradient factors. But we can only use the tools that are provided and it appears all dive computer manufactures are going down the GF rabbit hole and the only “dials” provided to change the model are gradient factors. Hopefully we will come up with a better solution in the future.

And hence, to be fair to David Doolette, he is trying to apply some rational logic to keep us safe during decompression dives. But I come back to my original question; does it make sense?

Using the data from Eric C Baker’s paper Understanding M-values. I averaged the M-values, “a” intercepts, slopes and the inverse of the slopes or “b” parameters.

Basically, I took 16 compartments and averaged them into one compartment.

Interestingly if you average the slopes and then take the inverse, you get .8342; or roughly 83%. But if you take the inverse of each slope and then average, the result is roughly 85%. Obviously rounding is a factor.

Anyway; if you take the average slope of 1.1988 and multiply it by 0.8342 you get 1. Hey is this the Navy slope of one Doolette is referring to?? Well, it’s the reciprocal of the average slope so it has to equal 1.

But my understanding, gradient factors are not applied that way. The “a” intercept is simply multiplied by the GF. In this case = 0.5081*GF. But for the slope, it is the M-value gradient which is multiplied by the GF (the difference between the slope and 1). In this case = (1.1988-1)*GF+1.

So for 70/85 versus the original average we get the following.

Notice that both the 85% and 70% gradient factors do flatten the original Buhlmann slope towards 1. But why stop there. Let’s apply 10% to GFlo.

So now the slope is really close to one, but 10/85 generates deep stops – something we want to avoid. What am I missing? What is it that I don’t quite understand about Doolette’s logic?

Is the problem in trying to tweak the Buhlmann model towards the US Navy model, is that gradient factors decrease both “a” intercept and slope when in actual fact the “a” intercept should increase to “offset” the drop in slope? Or at least keep “a” constant? I don’t know.

Suffice to say, in this paper, Understanding M-values, Eric C. Baker provides his personal GF limit of 85% M-value and 50-60% M-value Gradient for typical trimix dives. This seems to be in the same ball park as Doolette. Niether are advocating 20/90. And maybe, in trying to replicate the US Navy algorithm using gradient factors, David Doolette found the closest approximation is 70/85. It’s not perfect but maybe it’s the best of all combinations and limitations available using gradient factors.

 

[rjack321]

Buhlmann's experimental dives in lakes and ocean were done to 350m (1150ft). He worked also with Hannes Keller, for his record dives to 230m (750ft) in 1957 in the Lago Maggiore, and to 305m (1000ft) in 1962 near Santa Catalina island.

Buhlmann's work was funded by Shell Oil. They were not interested in recreational diving but only in very deep mixed gas saturation diving.

He was surprised when he learned that recreational divers use his model. Nowadays the model is used in many recreational dive computers and ironically people start to believe that this is a recreational diving model and doubt whether it's suitable for deep diving ...

See also Milestones of the deep diving resarch laboratory Zurich.

You pretty much hit my main point which was that it was originally developed for saturation diving (and chamber deco). Don's dive was a 600ft 1min CCR bounce dive with GFs included as well. All of which I suspect Buhlmann would agree wasn't part of the original concept or testing and "outside" of his model.

 

[rjack321]

When I finished reading the David Doolette’s article, I asked myself; does this make sense? And after reading the post by @atdotde – the theoretical diver, I really wanted to gain a better insight.

In summary, David Doolette calculates the average of all the “b” parameters which is the inverse the M-line slope which works out to be roughly 83% and then applies it to the GFhi to determine what the GFlo would be. For him, his preferred GFhi is 85%, therefore 85*0.83 = 70% is his GFlo.

His rational to apply the average “b” of 0.83 to GFhi is that it roughly counteracts the slope of the Buhlmann “b” parameters and brings it more in line with the US Navy decompression algorithm whose compartment M-line slopes are equal to or close to 1.

Thanks for doing this work. It's fascinating how convergent (yet tantalizingly subtly different) different means to the same end happen to end up.