Is GFlow applied to all 16 Compartments?

[CandiveOz]

I watched @Dr Simon Mitchell presentation Decompression Controversies on Youtube for the umpteenth time.
Around minute 33, he starts to discuss the problem with bubble model deep stops using gradient factors 20/90 as an example. He graphically demonstrates that while the fast compartment is off gassing at 20% of the supersaturation (Buhlmann) limit, slower tissues were still on-gassing and this was the crux of the problem. The penny dropped and I thought the simple solution would be not to apply the GFlow to the fast compartments (say the first 4) and apply the GFlow to the remaining 12. Using Mitchell's example; Compartments 1-4 would be set at 90/90 and 5-16 would be set at 20/90. The deco stops would be a bit shallower but you reduce this on-gassing problem to an extent because the intermediate, rather than the fast compartments would dictate the first stop. Of course I don’t think I am the first to come up with this idea; hence my question; is it common or generally assumed to apply GFlow to all 16 compartments in Buhlmann’s model or are the fast compartments excluded? Have any system developers programmed a Gradient Factor decompression model that has implemented the above idea or some variant (eg c1=90/90, c2=70/90, c3=50/90, c4=30/90, c5=20/90….)? I believe my logic is sound unless someone can explain a fundamental problem with the concept.

 

[KenGordon]

I watched @Dr Simon Mitchell presentation Decompression Controversies on Youtube for the umpteenth time.
Around minute 33, he starts to discuss the problem with bubble model deep stops using gradient factors 20/90 as an example. He graphically demonstrates that while the fast compartment is off gassing at 20% of the supersaturation (Buhlmann) limit, slower tissues were still on-gassing and this was the crux of the problem. The penny dropped and I thought the simple solution would be not to apply the GFlow to the fast compartments (say the first 4) and apply the GFlow to the remaining 12. Using Mitchell's example; Compartments 1-4 would be set at 90/90 and 5-16 would be set at 20/90. The deco stops would be a bit shallower but you reduce this on-gassing problem to an extent because the intermediate, rather than the fast compartments would dictate the first stop. Of course I don’t think I am the first to come up with this idea; hence my question; is it common or generally assumed to apply GFlow to all 16 compartments in Buhlmann’s model or are the fast compartments excluded? Have any system developers programmed a Gradient Factor decompression model that has implemented the above idea or some variant (eg c1=90/90, c2=70/90, c3=50/90, c4=30/90, c5=20/90….)? I believe my logic is sound unless someone can explain a fundamental problem with the concept.

It applies to all the compartments.
The idea was to protect fast tissues, so it makes sense to apply gf lo to the fast tissues.
The fast tissues have much higher limits than the slow tissues already so you are emphasising an existing feature. Originally (A set) the limits depended on the tissue half time, the longer the half time the lower the limit.

What you are saying is “What did Albert know about m-values? Maybe we should have lower slow tissue m-values, or higher fast tissue m-values.”

That might sound like I am taking the piss, but you might be generally right. One of m6 objections to GF is messing with the m values in a weird way rather than just starting with more appropriate constants, ie make the faster tissues a bit more conservative regardless. I think that would have been a harder sell though.

 

[CandiveOz]

It applies to all the compartments.
The idea was to protect fast tissues, so it makes sense to apply gf lo to the fast tissues.

Thanks for the confirmation. I agree that prior to the NEDU study, the objective was to protect the fast tissues from bubbling but I was really referring to the period after the study, when it was determined that by protecting the fast compartments was at a cost to the intermediate and slow compartments. I suspect that the general response was a reduction of spread between the GFhigh and GFlow. @Dr Simon Mitchell admits at the end of his presentation that his GF’s are now 50/70-75; a spread differential of 20 to 25 points. He seem to imply that he used wider spreads before. Certainly his example of 20/90 resulted in a spread of 70 points!

Thinking about it further, I suspect my suggestion of not applying GFlow to the fast compartments can be replicated with a combination of GFlow/GFhigh across all compartments in terms of resulting decompression profiles. Maybe 50/90 across all compartments would result in the same decompression profile of c1-4; 90/90 and c5-16; 90/20. I don’t know but it would be an interesting academic exercise.

What you are saying is “What did Albert know about m-values? Maybe we should have lower slow tissue m-values, or higher fast tissue m-values.”

I agree with the intent of your comment. And this is the risk of allowing divers to manipulate the “Buhlmann limit line” or “M value limit line” via gradient factors. What you described above is what I call a “twist” which I believe the NEDU study discredited. Using GFs you can have a twist, shift or shift & tilt, defined as follows;

1. Twist – 45/110; where GFhigh is greater than 100
2. Shift – 85/85; basically a parallel shift down
3. Shift and Tilt; 35/75 A spread between GFlow and GFhigh below 100

A twist is based on the false rational that by extending deep stops, a diver can shorten shallow stops; not a good idea and a complete rewrite of Buhlmann’s limits.

A shift is basically applying a conservative factor. It’s just a parallel shift with no change to the slope of the limit line. Everyone would agree that this is safe and within the intent of the original model.

A shift and tilt initially sets a conservative factor with GFhigh but then based on a false premise, rationalises a spread between GFhigh and GFlow as even “more safe” because it reduces bubble formation. A shift and tilt is probably just as bad a twist since the tilt deviates from the intent of the original model. Remember, GFlow was only created to generate deeper stops which the NEDU study refutes.

In my opinion, if scientists believe Buhlmann’s limits are wrong, then come up more appropriate limit lines (with proof) then allow divers to apply conservatism with parallel shifts.

One of m6 objections to GF is messing with the m values in a weird way rather than just starting with more appropriate constants, ie make the faster tissues a bit more conservative regardless. I think that would have been a harder sell though.

I agree, it may have been a harder sell prior to the NEDU study, but not so much today. That is why divers have now reduced the spread between GFhigh and GFlow. It takes time for beliefs and opinions to change but I suspect the GF spread will narrow even further. It’s nice to have the flexibility to change the Buhlmann model via gradient factors as long as we understand the risks involved...but I’m not sure we really do.

 

[KenGordon]

GF is the current bandwagon. I think that is unfortunate because it ignore everything except disolved gas. It probably does not matter but while people are complaining about computers which do something more sophisticated it leads to forgetting the reasons behind those addition warning and penalties.

There is nearly endless discussion here about m values, GF implementation details and the slope of the GF line. There is almost nothing about the consequences of re-descending.

 

[dmaziuk]

In my opinion, if scientists believe Buhlmann’s limits are wrong, then come up more appropriate limit lines (with proof) then allow divers to apply conservatism with parallel shifts.

You should read Eric Baker's "m values" paper. According to him it's not the scientists, it's the divers who want to be "even more safe". He also lists alternative approaches to increasing conservatism. Gradient factors are in the "confusion about deep stops". It has a clear problem statement, for which gradient factors are the proposed solution. They were not a solution to general "wrong Buhlmann limits", they were specifically for making it generate deeper stops.

It’s nice to have the flexibility to change the Buhlmann model via gradient factors as long as we understand the risks involved...but I’m not sure we really do.

Personally ich sprechen sie nicht so I wouldn't even understand the "L16" models if I read about them: AFAIK they're in "die tauchmedizin" and it was never translated. It's only available in its original Deutch or in re-telling by Eric Baker, Herr Heinrich, and one or two others. Forget the risks, I haven't even seen a cite for the "C" version of ZH-L16. (It may be in Powell's book that's still on my todo list.)