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

[stuartv]

There is no GF in your any compartment. There is current gas loading Pinsp at or below ambient pressure Pamb. As you ascend your ambient pressure goes down, causing the difference between Pamb and Pinsp to go up. Buhlmann's formula defines the maximum tolerable difference as "safe ascent ceiling".

With GFs, when you are on a no-stop dive, that ceiling is calculated by Buhlmann's formula times GF high. When there is a ceiling, it is calculated by Buhlmann's formula times GF low (at first deco stop). That's the "jump". When you hit 0 NDL, that's when the ceiling appears.

Not correct.

Every compartment has a GF. The GF of a compartment is the compartment's tissue pressure divided by the Pinsp (the inspired pressure of the specific inert gas in question), expressed as a percentage of the M-value for that compartment. That is not exactly right, but it gives the idea. When Ptissue and Pinsp are equal, there is no pressure gradient - the compartment is saturated - and the GF is 0. If you ascend until the ratio of Ptissue to Pinsp is half the M-value, then your GF is 50 (in that compartment). Again, not exactly, mathematically accurate, but conveys the idea.

The GF doesn't jump. The ceiling jumps - from 0 to 10', then to 20', and etc..

Also, it is common to have a ceiling while NDL is still greater than 0. At least with some computers. That is because your ceiling will go away before you reach it, presuming the expected ascent rate. Mandatory stops only appear when the ascent line would cross the line of the ceiling. It's easy to see in Subsurface (which is free, by the way) if you turn on the option in the planner to show the ceiling and then plan out some dives. Plan out 120' for 10 minutes, in Rec mode, on air, with GF50/80. You'll see there is a ceiling that appears after about 8 minutes or so. But, you don't quite run into it on the ascent. The ceiling maxes out around 11 feet, about 1 minute into your ascent. At that point, your leading compartment is off-gassing and the ceiling starts to reduce. Still no mandatory stop, though.

But, if you change your bottom time to 11 minutes, then you do intersect the ceiling before you get to the surface and so you get a stop (or, in this case, being in Rec mode, a warning that you are violating your ceiling).

 

[KenGordon]

I admit, in reality the influence of GFlow on NDL is at best marginal. But I thought it was worthwhile thinking about this and thus wrote a blog post about this: NDL and Gradient Factors – The Theoretical Diver (featuring my confusion).

^
Man who has written the code the behaviour of which the rest of the people are using as evidence to contradict him...

 

[dmaziuk]

Not correct.

Every compartment has a GF. The GF of a compartment is the compartment's tissue pressure divided by the Pinsp (the inspired pressure of the specific inert gas in question), expressed as a percentage of the M-value for that compartment. That is not exactly right, but it gives the idea.

Sorry Suart. Every tissue compartment has its own M-value. "GF" is the conservatism factor that modifies the M-value, it is the same for every compartment. (But the resulting modified M-value is not because they were different to begin with.) Every deco stop has its own GF expressed as the proportion between "high" and "low" and stop depth.

Robert's write-up has the problems with using this on near-NDL dives laid out quite nicely. Including the part where the whole thing was invented to add deeper stops to ZHL-computed decompression profiles to begin with -- so it's not exactly surprising that it gets progressively more wobbly as the stops get shallower and disappear.

 

[KenGordon]

Sorry Suart. Every tissue compartment has its own M-value. "GF" is the conservatism factor that modifies the M-value, it is the same for every compartment. (But the resulting modified M-value is not because they were different to begin with.) Every deco stop has its own GF expressed as the proportion between "high" and "low" and stop depth.

Robert's write-up has the problems with using this on near-NDL dives laid out quite nicely. Including the part where the whole thing was invented to add deeper stops to ZHL-computed decompression profiles to begin with -- so it's not exactly surprising that it gets progressively more wobbly as the stops get shallower and disappear.

I think Stuart is trying to say that each compartment has a current tissue ppInert which depends on the previous profile. That then lies on a line through ambient and the maximum tolerated pInert at GF100. The position up that line is where the limit would lie for some value of GF x, The highest of those x values for all the compartments is what GF99 is on a Shearwater. It is a reflection of the state of the compartment rather than an actual limit. You can imagine a hole set of GF99 numbers, one per compartment, forming a curve a bit like the Suunto DM5 software shows or the tissue graph on a Shearwater (although I have no idea if that one is absolute or relative to depth).

 

[stuartv]

Sorry Suart. Every tissue compartment has its own M-value. "GF" is the conservatism factor that modifies the M-value, it is the same for every compartment. (But the resulting modified M-value is not because they were different to begin with.) Every deco stop has its own GF expressed as the proportion between "high" and "low" and stop depth.

Robert's write-up has the problems with using this on near-NDL dives laid out quite nicely. Including the part where the whole thing was invented to add deeper stops to ZHL-computed decompression profiles to begin with -- so it's not exactly surprising that it gets progressively more wobbly as the stops get shallower and disappear.

No, the GF is the percentage of the M-value.

If you have GF50/80 on your computer and stops at 30 and up, then the GF at 30' is 50, at 20' it's 60, at 10' it's 70, and at the surface it's 80.

At 30', you will be at (approximately) 50% of the M-value in your leading compartment. And so forth.

Every tissue compartment can be described in terms of its percentage of the M-value, at any time. Therefore, it can be said that every compartment has a GF.

I didn't write the Subsurface implementation. But, I did download Erik Baker's original Fortran implementation and go through it line by line to understand it (having written Fortran for a living at one time in the distant past).

 

[dmaziuk]

I didn't write the Subsurface implementation. But, I did download Erik Baker's original Fortran implementation and go through it line by line to understand it (having written Fortran for a living at one time in the distant past).

I didn't: I haven't touched fortran since Scientific Computing 201 (which I failed by not showing up for final exam -- for some good reason or other -- and had to repeat it the next semester when they switched to teaching it in C. What a relief.

Coming from the "confusion about deep stops" paper and the formulae in the footnotes, it's fairly clear that each tissue compartment has its own tolerable "overpressure gradient" and Eric's "factors" modify that. Specifically to reduce overpressure in the faster compartments early in the ascent because that's what deep stops do. To me that looks like the forest and your interpretation: like the trees obscuring it.

 

[stuartv]

it's fairly clear that each tissue compartment has its own tolerable "overpressure gradient" and Eric's "factors" modify that.

Right. The tolerable overpressure gradient for each compartment is the M-value for that compartment.

The factor is a percentage applied to that gradient.

Every compartment has a current over-pressurization gradient, at all times, though it could be zero (saturated) or negative (on-gassing). And when you divide the current gradient by the tolerable gradient, that tells you the percentage of the way that that compartment is to the M-value. I.e. that is the compartment's current Gradient Factor.

As someone else pointed out, the compartment with the highest percentage of its M-value is the leading compartment and that compartment's GF is shown as the GF99 value by Shearwater computers (I believe).

it's fairly clear that each tissue compartment has its own tolerable "overpressure gradient" and Eric's "factors" modify that. Specifically to reduce overpressure in the faster compartments early in the ascent because that's what deep stops do.

I don't see that Gradient Factors have a "purpose" of anything related to fast versus slow compartments.

GFs are simply a tool for shaping an ascent profile to something other than continuously riding the M-value all the way to the surface (as you would with GF100/100). Some people use that tool to shape their ascent so that they do have deeper stops (for example, by using GF30/85) - apparently because they are concerned about not over-pressurizing faster compartments. Others use the same tools to achieve a different result.

I talked to 3 different very experienced tech instructors recently. One dives GF85/85. Another, GF95/95. The third, "pure Buhlmann", which I took to mean GF100/100. The first two, at least, use GFs to get shallow as quickly as possible without getting too close to the actual M-values. The third, well, I guess he doesn't really use GFs, so I maybe should have left him out of this.

Anyway, leaving the trees. Back to a big, wide open field. Yes, as I said earlier, every compartment has a GF, at all times.

 

[Pavao]

The list of computers running Buhlmann with GFs continues to expand:

Divecomputer.eu/Deep 6
Dive Soft
Garmin
Heinrichs Weikamp (not available in US)
Mares
Ratio/Seac
Shearwater
Suunto
Technical Dive Computers

Add Scubapro to the list

 

[stuartv]

Add Scubapro to the list

What ScubaPro computer supports Gradient Factors?

 

[Pavao]

What ScubaPro computer supports Gradient Factors?

The HUD, might be out this week