Will http://www.ncbi.nlm.nih.gov/pubmed/25525213 change deco procedures?

[victorzamora]

What this profile does with that deep ascent is basically turn the dive into a multi level dive with an hour of extra bottom time as compared to the other profile.

Here's something I think goes back to prejudices from a certain paradigm, but someone please correct me if I'm wrong:

Isn't the entire deco curve of a dive making a dive a multi-level dive? I mean, where do you draw the line in "multi-level" vs "deco." The "deep stop" and "bubble model" crew have been pushing deep stops as methods of controlling bubble growth. The "shallow stop" guys have been pushing shallow stops as a way of preventing oversaturating slow tissues. At what point do you decide to stop making the "deep stops" shallower, and start calling it a "shallow stop" profile? To me, it's more about "why" you're doing it than what actual depths you're doing it at that I'd be interested in and it makes the line in the sand easier to draw.

Ro: Where would you draw the line between "multi-level" and "deep stops" and why?

 

[ajduplessis]

ring…ring…ring

TOM Safe diving. This is Tom.

Bryan Hi Tom. This is Bryan. You taught my tech course last year.

TOM Hello Bryan. How's the diving?

Bryan Well, recently I've been diving to about 200ft surveying a wreck. But I've been having fairly consistent joint pain.

TOM Welcome to the tech community Bryan! You're a real tech diver now.

Bryan Ha! I guess so.

TOM But seriously. What kind of deco are you doing?

Bryan Initially VPM-B+3, but I upped that to VPM-B+4 for the last 15 or so dives,. But I'm still getting some joint pain more frequently than I'd like. I'm actually more concerned now. Do you think I should add some shallow time? Or find another model?

TOM Well, let's think about that. You tried VPM-B+3 and you have some signs that wasn't best. Then you went to VPM-B+4 and even though you added time you're still getting problems. You know Bryan, I think you might want to try VPM-B+1.

Bryan Uhhhhhhh … you mean VPM-B+5, right?

TOM No, you heard me correctly. I mean VPM-B+1. You tried longer deco and that didn't work. I think you might be doing too much deco.

Bryan Really!?? I always thought you'd need more deco if you were getting bent.

TOM Everyone seems to think that! Especially researchers…and Navy guys. But they don't do the kind of dives we do.

Bryan I can see why! I'm getting bent!

TOM (sarcastically)ha ha. Very funny Bryan. But that won't fix your issues

Bryan Ok. Sorry.

TOM Look Bryan. I'll let you in on a little secret --- there IS such a thing as too much deco!

Bryan Wow. I never knew that. Actually I've never even heard of that. Are you sure?

TOM Absolutely! Look, you had problems so you tried MORE deco. But that didn't fix it, did it?

Bryan No.

TOM Well. It's time to think outside the box. Give VPM-B+1 a try. And make sure you control your ascent rate.

Bryan Ok. I guess it's worth a try. I'll give it a shot. Thanks for taking the time Tom.

TOM No problem Bryan. Let me know how it goes. Safe diving!

--click--

TOM Heh … newbies!


… two weeks later … ring ring ….

TOM Safe diving. This is Tom.

Bryan Hi Tom. This is Bryan. Hey, Tom, I just got out of the chamber. I need some more advice ….

and your fix is to get shallow quickly and push the soft tissues instead of the joints?

 

[Redshift]

Actually, I see this as two assumptions. I'm not sure either of them is optimal.

On the other hand, I think the NEDU study should be seen as based on assumptions as well because it looks like the BVM3 model was forced by the research team into calculating the dive in a way it wasn't designed for (forcing it to calculate an ascent that gives exactly 174 minutes of ascent time from a dive that wouldn't ordinarily require it) and it gave a weird result that wasn't questioned again after that....

You imply that the deep profile was produced by some kind of fiddling. I think I noticed a similar assertion earlier. Reading David D's description quoted above from the RBW posts:

"you can specify a total decompression stop time, and an exhaustive combinations of stop depths and times (that add to the total) are tested to find the schedule that gives the minimum estimated risk."

says to me that the process was to calculate risk values for all possible profiles for that model and duration of deco and pick the lowest risk one. It searches the solution space for an optimal result.

But what if by letting the model free, it would generate a deco profile with a DCS probability of 0.1% (making up numbers) and when forced to use 174 minutes, the best of all the possible profiles would be 1%? It was forced to generate something worse, even if it was the best of the bad ones.

Indeed, I have pointed out many times that the deep stops in the NEDU deep stop profile did (as you say) protect the fast tissues from supersaturation. You seem to be assuming that this, of itself, makes it obvious that outcomes must be better if deep stops are performed. But the whole point of the NEDU study is that despite this protection of fast tissues early in the ascent, the deep stop ascent was still associated with worse outcomes.

I think people are trying to find a way to get the best of both worlds, the protection of fast tissues and not compromising the slow tissues.
I don't know the BVM3, but also found it weird that it chose to add time to deeper stops creating that profile that is not making people happy. How does it calculate the DCS probability? Could it be giving too much weight to the fast tissues and not enough to the slow ones?

You say this is because the deep stops were too long in the NEDU study. OK, lets shorten them, but keep the total length of decompression the same and compare tissue supersaturations with a "shallow stops" decompression. If the decompressions are the same length and the deep stops profile distributes more of that decompression time deeper then it is a physical inevitability that the deep stops profile will result in more slow tissue supersaturation later in the ascent. Maybe with shorter deep stops the differences in outcome between the two dives might be less obvious and more difficult to detect in a study of practical size, but why would you expect the deep stops profile to now be safer?

You seem to be putting a lot of faith in the notion that the short deep stops will control bubble formation early without causing excessive disadvantage in terms of slow tissue supersaturation later, but one of the reasons I am comfortable interpreting the NEDU study the way I do is that other studies that have tested dives with short "tech dive style" deep stops do not support your assumption. Indeed the opposite seems to be true. Ironically, bubble models do not appear to control bubble formation very well. That is why Ross spent a lot of time on the RBW thread and elsewhere trying to down play the relevance of venous gas emboli counts in decompression research.

As I implied in my earlier post to you, I can totally see that there is a "middle ground" or "sweet spot" or some other way of putting it. I just don't think it involves stops as deep as bubble models typically impose them.

Although I do not completely agree with the time imposition because it could be forcing the deco profile to become worse, I think it could be done with shorter deep stops controlling fast tissue supersaturation and leaving more time for the shallow stops to deal with the slow tissues and hopefully not lead to an increased bubble formation.

Which other "tech dive style" deep stops test dives were conducted? The VPM-B +4 comparison with GF 40/70? Or were there others?
I think what this shows is what you say further down, that bubble models do not generate the best deep stops. But GF low of 40 is creating deeper stops already. So the generalization that started to be used of "deep stops being bad" may not be correct.

Even when fast tissues get supersaturated, they don't stay supersaturated for long (because they are fast). The window of opportunity for bubble formation is therefore short and these tissues may be quite resistant to bubble formation (certainly relatively so). In the evolving pathophysiological paradigm for DCS in which circulating venous gas emboli are increasingly viewed as key players (especially when they get into the arterial circulation) it is the slower tissues where these emboli almost certainly arise that are probably the most important.

Finally, if you are asking me whether there is benefit in a middle ground in which some short deep stops are conducted early, and compensated later in the ascent by increased shallow stops to offset increased gas uptake by slow tissues I would have to say that I don't have a definitive answer to that. I doubt it (for the reasons articulated in the paragraph beginning "You say..above), but it has never been tested. But let's not lose perspective on the debate, which has been about deep stops as imposed by bubble models, and the promise of bubble models (at least initially) was that they could produce more efficient (and therefore shorter) decompression because of the deep stops, and that you could even shorten your shallow stops as a result. I really do think that notion is discredited.

Simon M

Being fast tissues more resistant to bubble formation and also, I believe, less in terms of body volume (compartments are theoretical and although some tissues are assigned as faster and others as slower, can it be quantified how much of a person is composed of each type of tissue?) it makes sense that protecting them is not that crucial and that controlling the few bubbles they may produce on ascent is also not that important.

And here you phrase this study in a more appropriate way "deep stops as imposed by bubble models", or better, as imposed by one model, the BVM3.
Not in any way trying to defend bubble models, I don't use them and think they have too many unknowns, but trying to look at what has really been done.

 

[rossh]

The GF model is just naturally more consistent with the NEDU results. Other models that I've haven't looked into may reflect the NEDU study just as well.

Really? ZHL+GF is no closer to the nedu test than VPM-B is. Neither VPM-B or ZHL + GF were tested or represented by the nedu test:

data for the above

Clearly ZHL + GF was not tested either. The GF "model" (as you imagine it) is no better matched to the nedu test than VPM-B is.

 

[rossh]

Hello again,

It seems you may have missed something in David's explanation. VVAL18 is in use in the US Navy who, I believe, have real world divers.

Simon M

No... that is a deception again Simon, and you know it !

The test model profile, is not published in the USN dive manual.

The naval command replaced it with the modified VVAL-18 M tables, which reduced it to about half the deco time involved. The model in use in the USN diving manual is designated M for modified. The test profile was almost double the USN dive manual profile. (179 vs 93 mins).



From David Doolette RBW post #236:

VVal-18M (M for modified) parameter set for the Thalmann Algorithm did not arise as a result of the NEDU deep stops study, it had been kicking around for several years beforehand. NEDU, a field activity of Naval Sea Systems Command, is a research, development, testing and evaluation facility, we make recommendations to our bosses at the Washington Navy Yard, the Supervisor of Salvage and Diving and the Supervisor of Diving, but it is them who approve procedures – and they own the U.S. Navy Diving Manual. VVal-18M was the result of a desktop exercise by Dr Ed Flynn (CAPT, USN, retired), working in the office of the Supervisor of Diving and Salvage. By adjusting some parameters of VVal-18 (developed by Ed Thalmann) he came up with a parameter set that produced similar oxygen decompression times to VVal-18, but much abbreviated air decompression times.

So, the nedu test did NOT compare real world every day USN profiles, or tech diving models. Neither test model made it into the USN dive manual.













---------- Post added December 26th, 2015 at 05:28 AM ----------

Ross,

It is a matter of clear, unassailable, public record (on this thread) that YOU raised the allegation that there were "paid trolls" operating on the RBW thread. I have asked you to provide evidence to support this allegation. Your various contorted responses, including this latest one...

...is simply bizarre. There is no other word for it. At least it puts some perspective on things for people looking on.

Simon M

Let me remind you again: You made up the allegations, all on your own - no one else said those things - no one else wrote those things - only you. Your vivid imagination played tricks on you. And now you want to blame other people and expect them to take responsibility for your fantasies and fabrications.


What you're doing here is called "proof by repeated assertion". It's a form of fallacy.


Its seems you exhibit the same problems when you discuss the nedu test and its implications.

 

[Diver0001]

Finally, if you are asking me whether there is benefit in a middle ground in which some short deep stops are conducted early, and compensated later in the ascent by increased shallow stops to offset increased gas uptake by slow tissues I would have to say that I don't have a definitive answer to that. I doubt it (for the reasons articulated in the paragraph beginning "You say..above), but it has never been tested. But let's not lose perspective on the debate, which has been about deep stops as imposed by bubble models, and the promise of bubble models (at least initially) was that they could produce more efficient (and therefore shorter) decompression because of the deep stops, and that you could even shorten your shallow stops as a result. I really do think that notion is discredited.

Simon M

Simon, thanks again for sticking with me on this. I guess when we peal away all the layers, what's at the core of this to me is the definition of a deep stop.

The study seems to be wanting to address the utility of deep stops but when you look at it from the paradigm I'm used to -- namely the paradigm that has been firmly stamped into my DNA as a diver that you need to curve the ascent -- the profile they used doesn't have anything that I can recognise as a deep stop. It has what I would recognise as a number of multi level stages but no deep stops in the sense that we're used to seeing.

That's where I can't get the gears that I'm used to working with to mesh with the gears they're using... so I just can't put the profile in a context I understand.

I think, given the large body of deco theory we already have that says you need to curve the ascent, that it's valid to question whether or not a stop can be so long that it is no longer a stop but a new stage. I think this is what Tom was trying to say too when he said you can do "too much deco".

The point can be illustrated by simply calculating the average depths for those two profiles. The average depth of the Haldane profile works out to about 39 feet. The average depth for the BVM3 profile works out to about 50 feet. If you use something like the PADI RDP to look at the NDL's for dives to 30, 40 and 50 feet then you can see that there is a dramatic drop in the NDL over this short distance. On a 200 minute dive, the average depth of the BVM3 profile is so deep that it seems logical to me that it would result in higher levels of DCS and perhaps even a little surprising that the incidence of DCS wasn't higher than it was.

Another way of looking at this to see what mainstream software says about it. On nominal, Vplanner suggests that the BVM3 divers skipped about 25 minutes of deco for this profile while the study's Haldane divers skipped none. Therefore, what's bothering me, and what has BEEN bothering me from the beginning is that I *think* (and again I'm not a scientist) ... but I *THINK* the study has been confounded by the average depth of the BVM3 profile being too deep.... I *THINK* what may *actually* be happening is related to the average depth and not the deep stops. In other words, I really have the impression that it's possible that the researchers saw deep stops as being the culprit because they *expected* it and as a result failed to look for other explanations (like average depth).

I'm just saying what I'm seeing. I'm not arguing that I must be right about this, I'm observing that there is an alternative explanation for the increased incidence of DCS in this study. And again, I'm just another diver so take all this with a grain of salt. I'm uncomfortable as it is questioning the big guns like this and taking up so much air-time in this thread, but I think it's important to view it from all angles.... don't you?

Having said all that, I firmly side with Simon on his analysis that deep stops do NOT justify shortening shallow stops. If anything else (using the avg depth idea) deep stops would argue in favour of extending shallow stops. Ever since Mark Ellyatt just about died from using Weinke's algorithm on a 250 metre dive in 2003, technical divers have had some (more than) nagging doubts about shortening shallow stops because of having done deep stops and I think Simon is absolutely on the mark when he says that this idea has been debunked. More research will strengthen this but I'm happy to take that result at face value right now.

@divetech99, I hope this sort of answers the question you asked me. I guess if It were me that I might have picked profiles that leave the average depth of both dives within the generally accepted NDL's for the run time so that there could be no question of whether or not the average depth was confounding the results. Not being a scientist I guess I'll have to defer to smarter people than me to design an experiment that makes sense, though.

R..

---------- Post added December 26th, 2015 at 12:47 PM ----------

But what if by letting the model free, it would generate a deco profile with a DCS probability of 0.1% (making up numbers) and when forced to use 174 minutes, the best of all the possible profiles would be 1%? It was forced to generate something worse, even if it was the best of the bad ones.

This is exactly what's been nagging at me but I was unable to describe it nearly as well as you did. Thanks.

 

[David Doolette]

If let run freely for that depth and bottom time, what is the deco given by BVM3? Can't it be that forcing a total deco time, if longer than what the model would choose, be producing worse outcomes? And although I understand the attempt to not have two things varying at the same time (deco profile and total time), I am not convinced that 1) they are two variables, because a deco profile has both stop depth and time included. It may not be straightforward to force one 2) that it's not possible to compare outcomes when total deco times are different. Certainly if, the deep stop model would produce more DCS one could not be sure if it was due to the deep stops or the missing deco time, but what if it wasn't worse? Interesting case to study.

Connected to 1), how does the BVM3 model decide to allocate the forced deco time? Is there a weight for different compartments? Did it protect fast compartments at the expense of slow compartments, leading to increased DCS? Because one thing is controlling bubble formation, another is actually being increasing the bottom time. We all know for example that we should leave the bottom fast, at the recommended ascent rate and that slower is not better.

It also makes sense that it's not just the supersaturation that is important, but the period as well (I think this might have been looked at during the VPM-B+4 - GF 40/74 comparison, but I can't see the plots). A short higher supersaturation may generate less bubbles than a longer lower supersaturation. Are the models trying to minimise the peak value or the integral over time?

There is no way that probabilistic algorithms can "run free" in the way you suggest. The algorithm works in one of two ways, to find the lowest model estimated probability of DCS (PDCS) for a specified decompression time, or the shortest decompression time for a specified PDCS. If you take away those constraints it is no longer an algorithm, and just a model. The equivalent concept would be to let ZH-L16 "run free" without the "a" and "b" parameters that define the safe ascent depth. In the BVM(3) probabilistic model, PDCS is a function of the sum of the time integral of bubble volume in each of three compartments (fast, medium, slow), so all are always important, it does not spare the fast compartments at the expense of the slower ones. Indeed, because the slow compartments is slow, a bubble can last a long time and contribute substantially to the PDCS, so the algorithm will juggle this with keeping bubbles small in the faster compartments. Because it is a bubble model, it will tend to skew time towards deeper stops to limit the bubble growth in all these compartments. The relatively similar length of the first few stops is primarily limiting bubble growth in the middle compartment. The calculated bubble volumes for the two schedules are illustrated (if I remember correctly) in the 2008 DAN technical diving conference proceedings which I believe are free to download. It is worth noting, that the BVM(3) estimate of PDCS for its own deep stops schedule was quite accurate, it failed to predict the much lower risk of the shallow stops schedule - but that is the essential dicotomy between bubble and gas content models/algorithm that the NEDU study was testing.

David Doolette

---------- Post added December 26th, 2015 at 12:20 PM ----------

I guess I must be missing something but what's interesting to me is that on the bubble model they tested the last three stops are in the same ball park as what the Navy tables would have said you needed using Buhlmann except they kept those divers under higher pressures than Buhlmann would have accepted for a FULL HOUR before doing those stops. Therefore, it's not surprising that this profile would give higher rates of DCS because the exposure is skewed to be very deep and very long at those depths, essentially extending the bottom time significantly before doing the required shallow stops. Likewise, the "shallow" profile they tested involved 80 minutes MORE deco than the Navy tables would prescribe in the shallow zone including a last stop which is equivalent in duration to the TOTAL deco time the Navy tables prescribe.

I am very reluctantly addressing the "deep stops" question, because all these points have been debated extensively in the RBW and Scubaboard threads, but I think you are actually missing something. What you are saying here, and in subsequent posts, is that the deep stop time in the NEDU A2 (deep stops) schedule needed to be compensated for by additional shallow stop time - that is the essence of the “shallow stops” paradigm. Under this shallow stops paradigm, you can do as deep stops as you want, as long as they are part of a longer decompression - this can be extended all the way to doing for first stop essentially at the bottom and doing a saturation decompression. However, the “deep stops” paradigm is that time spent at “deep stops” is more than compensated for by reduced time at shallow stops. The NEDU test was a comparison of these two paradigms – hence the title, “Redistribution of decompression stop time…”.

If you accept the notion that the NEDU deep stops (A2) schedule resulted in more DCS than the shallow stops (A1) schedule because of continued gas uptake into slow compartments and consequently greater supersaturation in those compartments later in decompression – and this is the most plausible explanation – then any schedule showing this pattern is likely to be similarly disadvantaged. In an analysis of the time-integral of supersaturation (to assess both the magnitude and the duration, because both are important to bubble formation and growth) in fast and slow compartments for half a million alternative schedules with different distributions of 174 minutes of stops, we found that every schedule with any amount of time allocated to deeper stops than the A1 schedule resulted in the same disadvantageous pattern of increased and longer lasting supersaturation. The hypothesis that one of these alternative deep stops schedules might have been better than the A1 shallow stops schedule is to ask the question whether that alternative schedule could be better to despite this disadvantageous pattern of supersaturation. Maybe, but on the basis of currently available evidence, I do not see that as a plausible hypothesis.

David Doolette

---------- Post added December 26th, 2015 at 01:00 PM ----------

No... that is a deception again Simon, and you know it !

The test model profile, is not published in the USN dive manual.

The naval command replaced it with the modified VVAL-18 M tables, which reduced it to about half the deco time involved. The model in use in the USN diving manual is designated M for modified. The test profile was almost double the USN dive manual profile. (179 vs 93 mins).



From David Doolette RBW post #236:

Because it is relevant to other questions raised in this thread, it is worth posting the whole context of the quote:

VVal-18M (M for modified) parameter set for the Thalmann Algorithm did not arise as a result of the NEDU deep stops study, it had been kicking around for several years beforehand. NEDU, a field activity of Naval Sea Systems Command, is a research, development, testing and evaluation facility, we make recommendations to our bosses at the Washington Navy Yard, the Supervisor of Salvage and Diving and the Supervisor of Diving, but it is them who approve procedures – and they own the U.S. Navy Diving Manual. VVal-18M was the result of a desktop exercise by Dr Ed Flynn (CAPT, USN, retired), working in the office of the Supervisor of Diving and Salvage. By adjusting some parameters of VVal-18 (developed by Ed Thalmann) he came up with a parameter set that produced similar oxygen decompression times to VVal-18, but much abbreviated air decompression times. This was incorporated into Revision 6 of the U.S. Navy Diving Manual, because a principal reason for the revision was to introduce in-water oxygen decompression for routine use. The vision was that oxygen decompression should be used for any schedule requiring more than 15 minutes decompression, and in-water oxygen would provide a bridge between air decompression and surface decompression with oxygen. In Revision 6, oxygen decompression is recommended for air schedules requiring more than 15 minutes decompression, and as the decompressions get longer, oxygen decompression is required. The corresponding air decompression schedules are for emergency use only. The risk of decompression sickness is not the only Operational Risk Management issue, so a table set with relatively short, and consequently relatively high risk air decompression schedules was promulgated. As I noted in my first post #145, the 170 fsw / 30 minute schedule is one of these “for emergency use only” schedules.

Not all U.S. Navy Diving is in the Dive Manual, and I can assure you VVal-18 is still in use to schedule air decompression – although now days in quite limited use. The VVal-18 schedule posted above is the schedule with a 20 fsw last stop – promulgated in the report from which it is replicated because a 20 fsw last stop was requested by the office of Supervisor of Salvage and Diving. The VVal-18 schedule with a 10 fsw last stop is the one we tested in the NEDU trial. Astute readers will notice that the 10 fsw last-stop schedule is longer than the 20 fsw last-stop schedule, and this is because the Thalmann Algorithm uses potentially asymmetric, Exponential-Linear, gas uptake and washout (something that was asked about in a recent post). In this algorithm, gas uptake is always exponential but gas washout switches from exponential to slower, linear washout (representing tissue gas washout in the presence of profuse bubbles) if the compartment is supersaturated above a threshold (zero in VVal-18). In this particular schedule, this slower, linear washout manifests as a longer schedule if the last stop is at 10 fsw instead of 20 fsw.

David Doolette

 

[Storker]

the Thalmann Algorithm uses potentially asymmetric, Exponential-Linear, gas uptake and washout (something that was asked about in a recent post). In this algorithm, gas uptake is always exponential but gas washout switches from exponential to slower, linear washout (representing tissue gas washout in the presence of profuse bubbles) if the compartment is supersaturated above a threshold (zero in VVal-18).

So that approach is already implemented in some algorithms. Thanks for that info.

 

[Dr Simon Mitchell]

No... that is a deception again Simon, and you know it !

Ross, in the light of David's post (after all, he does work for them), would you like to revise this confident assertion?

You made up the allegations, all on your own - no one else said those things - no one else wrote those things - only you. Your vivid imagination played tricks on you. And now you want to blame other people and expect them to take responsibility for your fantasies and fabrications.

Ross, it is a matter of clear, unassailable public record (on this thread) that you asserted that there were "paid trolls" on the RBW thread:

http://www.scubaboard.com/forums/te...-deco-procedures-post7573029.html#post7573029

You also said it here:

More thoughts on helium perfusion and solubility - Page 19

Nobody had ever claimed that before you did. It is a serious allegation. Either withdraw the comment, or provide proof of your claim. Until you do either I will keep raising the matter in all discussions with you, and frankly, I don't care if you continue to undermine your credibility by behaving in this bizarre way. It is your choice.

Simon M

 

[UWSojourner]

Really? ZHL+GF is no closer to the nedu test than VPM-B is. Neither VPM-B or ZHL + GF were tested or represented by the nedu test:

Clearly ZHL + GF was not tested either. The GF "model" (as you imagine it) is no better matched to the nedu test than VPM-B is.

My statement that you objected to was that the GF method more naturally reflects the NEDU study. As evidence for your objection you posted your chart, but without VPM-B??? That makes it difficult to render an opinion as to which better reflects the NEDU study.

So I added VPM to your chart as well as GF 53/53 (see below).


One thing to notice is both A2 and VPM-B have sagging lines (concave upward) on the GF slope. Of course GF, by definition, has a straight line. If you chart VPM-B for other dives with substantial deco this pattern (VPM keeping the GF low and continually accelerating the GF as you move toward the surface) is duplicated. In any case, after A2's initial thrust upward (due to a 1st stop of 70ft), VPM-B and A2 do indeed seem to track better than GF53/53 or GF 40/X.

But "GF by depth" is not the chart I used to form my statement. See the chart below (peak GF by tissue compartment).


It is clearly true that for compartments 4 and above A2 and VPM track pretty well, and GF and A1 track pretty well for compartments 6 and above. The biggest differences are in the fast compartments 1-3. But in those compartments VPM pushes even further away from the safest profile (A1). Although as we've shown on the deep stop threads the total time integral of supersaturation for compartments 1-3 was almost identical for A2 and VPM.

Finally, the chart below shows the time integral of supersaturation (by compartment). The chart illustrates very well the tradeoff made for "protecting the fast tissues".


Consider chart 2 (GF by compartment). A1 (found to be much safer than A2) looks rather alarming in its dramatic spike in compartments 1-3. But because those spikes were of such short duration, the time integral of supersaturation (aka the "index of decompression stress") does not get out of control as clearly shown in the chart above.

However, the cost that A2 and VPM-B paid for "controlling the fast compartments" is the rather large ballooning upward of the time integral of supersaturation in compartments 8-12 upon surfacing. That pattern repeats itself for other profiles tech divers are used to seeing.

So, yes, it does seem apparent to me that the GF model more naturally reflects the NEDU study. Obviously more naturally than VPM-B.

For even more comparisons showing the same idea, see this post and track down the links.