On a NDL dive, which computers' NDLs are not affected by GFLo?

[KenGordon]

You'll still have 16 compartments and their half-times. (And the choice of 4 or 5 minutes for the first one.) Where do you draw the line, though: by adding varying degree of conservatism w/ the sliding slope of factors, don't you change the overall off-gassing dynamics enough to not be "Buhlmann" already?

One of the most surprising, for me at least, things about ZHL-16A is that the m values are mathematically derived from the half times. So you could make up a ‘Bühlmann’ algorithm with 7, 17 or 42 compartments.

ZHL-16B changed those values slightly in the middle. Presumably that was a result of the ancient, now of course completely discredited practice, known as science and so supportable.

While GF is a bit of a hack it is possible to reason about how the resulting profiles will compare to raw Bühlmann. Once you start messing with the individual compartments you will get something else entirely, with which is the leading compartment at any point on a dive changing even more.

The idea of deep stops is to protect faster tissues. Another way to have done that would have been to reduce the m value limits of the fast tissues while leaving the rest alone. However that would have been quite hard to justify without any actual testing. If that was done I really don’t think you could call it a ZHL-16x derivative.

 

[huwporter]

One of the most surprising, for me at least, things about ZHL-16A is that the m values are mathematically derived from the half times. So you could make up a ‘Bühlmann’ algorithm with 7, 17 or 42 compartments.

ZHL-16B changed those values slightly in the middle. Presumably that was a result of the ancient, now of course completely discredited practice, known as science and so supportable.

And many(/most?) "Buhlmann" dive computers use ZHL-16C, which further modifies some A and B values, to compensate for the reduction in conservatism caused by calculating a 'live' profile moment by moment during a dive instead of using a rounded-deeper square profile, as when using tables.

I can't immediately find out who created the 'C' version, whether it was prof Buhlmann himself or not, and whether it was the result of testing or not. (I agree that Prof. Buhlmann does appear to have created the 'B' version himself, based on actual experiment.)

While GF is a bit of a hack it is possible to reason about how the resulting profiles will compare to raw Bühlmann. Once you start messing with the individual compartments you will get something else entirely, with which is the leading compartment at any point on a dive changing even more.

This really isn't the case. You still have the exact Buhlmann 16 compartment (32 if you are using Helium) algorithm tracking gas uptake and washout, calculating M-values and ceiling. The result of diving a profile generated using any kind of GF-limiting approach, whether it is by the conventional linear interpolation through the stops or per-compartment or anything else is exactly the same effective situation as a diver who has a raw-Buhlmann computer strapped on but stays a bit deeper than the ceiling throughout the ascent. A diver who does this hasn't suddenly invalidated the calculations her computer is running...

You can argue til the cows come home about the benefits/necessity of competing strategies for staying a little away from the M-value through a profile, but at the root all this rests on the 'ol Prof's hard science on people. (Unlike, say, if you use another algorithm based on bubbles in gel...)

 

[KenGordon]

This really isn't the case. You still have the exact Buhlmann 16 compartment (32 if you are using Helium) algorithm tracking gas uptake and washout, calculating M-values and ceiling. The result of diving a profile generated using any kind of GF-limiting approach, whether it is by the conventional linear interpolation through the stops or per-compartment or anything else is exactly the same effective situation as a diver who has a raw-Buhlmann computer strapped on but stays a bit deeper than the ceiling throughout the ascent. A diver who does this hasn't suddenly invalidated the calculations her computer is running...

You can argue til the cows come home about the benefits/necessity of competing strategies for staying a little away from the M-value through a profile, but at the root all this rests on the 'ol Prof's hard science on people. (Unlike, say, if you use another algorithm based on bubbles in gel...)

Imagine what would happen if you reduced or increased one of the medium compartment’s limits a bit while leaving the rest alone. Suddenly a certain subset of the dives will get quite different stops, or for repetitive dives different NDLs.

Imagine what would happen if you reduced the fast tissue limits a lot, but not the slower ones. Now you will be doing stops early, on gassing the slow tissues but NOT doing as much extra time shallow as with GF even. Sound familiar? And quite different profiles to those that were actually tested.

 

[huwporter]

Imagine what would happen if you reduced or increased one of the medium compartment’s limits a bit while leaving the rest alone. Suddenly a certain subset of the dives will get quite different stops, or for repetitive dives different NDLs.

Imagine what would happen if you reduced the fast tissue limits a lot, but not the slower ones. Now you will be doing stops early, on gassing the slow tissues but NOT doing as much extra time shallow as with GF even.

But the model is still sitting there, doing its work, in exactly the same way as it does for two different divers doing two different GF profiles, or for that diver who ascends a bit more slowly than their computer allows.

Sound familiar? And quite different profiles to those that were actually tested.

Sure. And so is everyone who uses GFs, or who anyone doesn't follow their computer exactly. Anyone who doesn't do exactly the 10m/min ascent rate that Buhlmann used for testing has already deviated from the actual test conditions. If anything, conventional GFs are way further from Prof. Buhlmann's tested profiles than any approach that aims to slow the initial ascent a bit but still get shallower faster than conventional GFs allow.

Edit for one final note - if by "Sound familiar?" you are implying that your concern is that ANY of these approaches suddenly turn Buhlmann into VPM or anything even close to it, then you are entirely mistaken. Compare any VPM profile against the posted-for-interest examples in my earlier post and check for yourself. The underlying Buhlmann model continues to add time shallow for any additional time spent deep to satisfy your GF-Hi just as always.

I think I'll check out here since I'm not sure that continuing would be constructive, and I'm not trying to persuade you or anyone else to do anything.

 

[jale]

ok redone check for AV1
air, 25M, 20min, 90/90, 24min total
air, 25M, 20min, 10/90, 24min total listing stops from 15M but with 0min stop.
Same.
For
air, 30M, 15min, 90/90, 21min total with 1min at 3M
air, 30M, 15min, 10/90, 20min total listing stops from 18M but with 0min stop.
1min less with 10/90.
And for
32%, 25M, 20min, 90/90, 22min
32%, 25M, 20min, 10/90, 24min total listing stops from 12M but with 0min stop.
2min more with 10/90
32%, 30M, 15min, 90/90, 18min
32%, 30M, 15min, 10/90, 20min total listing stops from 12M but with 0min stop.
2min more with 10/90
So, looks like it depends... or maybe the delta is too small for being pertinent...

 

[dmaziuk]

I can't immediately find out who created the 'C' version, whether it was prof Buhlmann himself or not, and whether it was the result of testing or not. (I agree that Prof. Buhlmann does appear to have created the 'B' version himself, based on actual experiment.)

I'm waiting/hoping for Powell's book on interlibrary loan request. Maybe he knows where this alleged "C" came from...

On a related note, the numbers in subsurface do not match the "B", are said to be "based on ZHL", and come with Heinrichs-Weikamp copyright. So presumably OSTC runs (ran at one point?) ZH-L16-b-double-prime or some such. (It doesn't call it "C" either.)

Imagine what would happen if you reduced the fast tissue limits a lot, but not the slower ones. Now you will be doing stops early, on gassing the slow tissues but NOT doing as much extra time shallow as with GF even. Sound familiar? And quite different profiles to those that were actually tested.

Well, you could do it by diving 10/130 (or something) and it'll still be "ZHL+GF" -- as we know the math will work just fine.

 

[atdotde]

I'm waiting/hoping for Powell's book on interlibrary loan request. Maybe he knows where this alleged "C" came from...

On a related note, the numbers in subsurface do not match the "B", are said to be "based on ZHL", and come with Heinrichs-Weikamp copyright. So presumably OSTC runs (ran at one point?) ZH-L16-b-double-prime or some such. (It doesn't call it "C" either.)

Subsurface uses C (which is already in Bühlmann's book) except it leaves out the 4 minute compartment, see http://www.nigelhewitt.co.uk/stuff/aab.jpg and compare it to Subsurface-divelog/subsurface

That said, these variations are really tiny and there are other (hidden as not exactly part of the specification) places where algorithms differ. I blogged about these here Why is Bühlmann not like Bühlmann – The Theoretical Diver The upshot is that the variation between these "Bühlmann" models are tiny compared to variations between different physiologies that one tries to model.

 

[dmaziuk]

Subsurface uses C (which is already in Bühlmann's book) except it leaves out the 4 minute compartment, see http://www.nigelhewitt.co.uk/stuff/aab.jpg

Thank you. I expect those of us who don't speak German and wouldn't know what to do with Tauchmedizin, have only seen the numbers in Baker's Fortran until now.

PS it doesn't "leave out" the 4-minute TC, you have to choose the 4 or 5-minute one for your implementation (again according to Baker). Of course in computer-minutes that's 20% (or 25, dep. on which way you count 'em) difference in the key value in your TC that's first to react to ambient pressure changes...

 

[scubadada]

Well.... I finally did a search and found where I checked this info a couple of years ago.

Shearwater Perdix AI oddities: GF settings and dive planning

The Seabear H3 did have it where the GF Lo affects the NDL.

You may also notice my post immediately following the one I linked, with an email from Shearwater that says "Note that this is consistent with saying that only GF_High controls the NDL time".

Read Stuart's post and the reply from Tyler Coen. I have MultiDeco and have had trouble understanding the dive plans at or near what would be the NDL. MultiDeco is not designed to give NDLs, but to plan deco dives. When you adjust the stop time from 1 minute to 1 second, the phenomena that Tyler writes about is clearly illustrated.

Here is a very extreme example of a 31 minute dive to 100 feet on EAN32. The first two tables show 100/100 at 1 min stops and then 1 second stops:

Then, here is the same dive at 5/100 at 1 min stops and then 1 second stops:

By the principles laid out by Tyler Coen, the 1 second stop plans show that this is an NDL dive and that GF low (even this most extreme example) does not affect the NDL time (though I'm not sure that there are not 5 seconds of deco at 10 feet at 5/100 )

 

[scubadada]

I'm waiting/hoping for Powell's book on interlibrary loan request. Maybe he knows where this alleged "C" came from...

I think you'll enjoy Deco for Divers. ZH-L12 was determined empirically with decompression trials. The M-values for ZH-L16A were derived mathematically. A was found overly aggressive in real world use. Modified set B was made more conservative in compartments 6-8, and 13 and was suggested for use in tables. Modified set C was made more conservative in compartments 5-12, and 14 and was suggested for use in dive computers. The table in the post by @atdotde lists the changes that were made in the Buhlmann a coefficients. A similar table in Powell's book on p222 shows how these changes decreased the surfacing M-values (Mo) in these compartments. The slope of the M-value line does not change (deltaM).

Unfortunately, Powell does not say exactly how these changes were determined or by whom. I assume they were in place by the 1990/1993 editions of Diving Medicine, perhaps the answers are there.