Buhlmann, Computers, and Helium

[chrisch]

My understanding too that stop deep and come up slow with helium.

That's why some people don't like the recreational mix idea. I think it is a step forward.

Chris

[Mdemon]

Interesting stuff Andy.

One question. We hear about Mr B. et al, but who are the guys doing research on the subject now? Is it all commercial and therefore secret? Or has it "stopped"?

Curious to know who's looking into Helium effects scientifically... Can't believe no one is...

[chrisch]

Originally Posted by Mdemon

One question. We hear about Mr B. et al, but who are the guys doing research on the subject now? Is it all commercial and therefore secret? Or has it "stopped"?...

I think there is quite a bit of commercial interest at stake here. The popularity of the Buhlmann stuff is no doubt the fact it is public domain, combined with being well tested.

There is lots of research going on, for example Bruce Weinke is still researching the RGBM model and has licenced the idea to several computer companies, notably Suunto and most recently Mares.

To be honest I think some of the best stuff is coming from the WKPP/DIR area and sadly this is lost in the background noise that this discipline generates. It is certainly very interesting.

I wish I had the money and time to read more about the whole thing. Meantime I look forward to more debate here. Great stuff.

Chris

[MattS]

Excellent article.

Just a couple points. Very much as I understand it and I make no claims as to the accuracy.

DAN's project dive explore has been running a few years now. PDE is a practical research project looking at real world and controlled profiles in order to better understand 'sport' decompression practises. The interim results concerning deep stops are really quite interesting. DAN are due to start mix experiments very soon.

Deep stops.
Buhlmann's model is a dissolved gas model, it only simulates the liquid phase. Deep stops are thought to control gas phase which occurs during the ascent. There have been plenty of theories about how a deep stop works, including that it is no more than a slowing of the overall ascent rate. The PDE results indicate that 'a very slow ascent' may be a little simplistic. DANs results show that ascending at 10m/min and stopping for a couple minutes at intermediate depths produced fewer bubbles during monitoring than making the same ascent at an ultra-slow rate.

One possible hypothesis is that bubbles continue to form and grow all the time that pressure is reducing. As the number of bubbles present increases so does the likely hood of some of those bubbles joining together (coalesce) to form a bigger, symptomatic bubble. This would cause a bend which is 'unearnt' within the limits of the traditional dissolved model. A pause in the ascent (deep stop) allows some bubbles to resolve in the lungs hence lowering the 'density' of bubbles available for coalescence.

The argument against ultra-slow ascents is that they may not provide the pressure differential to kick start off-gassing at the rate the underlying dissolved gas model assumes. Some believe that off-gassing is similar to blowing up a balloon in so far as you need more effort to start inflation than you do to maintain the inflation (that is as far as the analogy goes).
The underlying dissolved model was trialled using a relatively shallow first stop, which would produce a significant differential between compartment gas pressure and ambient. Ascending vveeerrrryyy slowy it takes longer to produce a significant gradient and you may not off-gas fast enough.

VR3
As I understand it the VR3 treats bubble control as a separate mechanism on top of the underlying Buhlmann algorithm. It continues to on-gas compartments below ambient whilst on the deep stop. Compartments over ambient will off-gas but I don't think it gives any additional credit for the fact you stopped deep. The VR3 deep stop algorithm is very similar to Pyle roughly halving depths between stops.

Buhlmann and Helium
I too was under the impression that Buhlmann got his halftimes from the military and commercial worlds. The problem here is that the military and commercial worlds mainly use Heliox and it is a bit of a leap of faith to assume Heliox decompression works the same as Anoxic trimix, Normoxic trimix or Superoxic trimix.

In the Buhlmann model the He off-gas calculation is an inverse of the on-gas calculation. On/off gassing uses symmetrical curves the same as N2. However the lower half-time of He will lead to more of the faster compartments saturating and higher levels of inert in slower compartments (compared to N2). So the decompression profile starts deeper and lasts longer. The initial short stops reduce levels in the faster compartments while later longer stops reduce levels in the slower compartments.

If you take a Normoxic profile and compare the schedule to the same profile on air the last stop can be significantly longer. This is caused to an extent by He permeating into slower compartments which N2 would not have time to get to.

With a high He mix used at high ambient pressure (in terms of dissolved gas models) the initial N2 in the compartments will be replaced by He. After a bit of time at depth N2 levels reduce below what they were at the start of the dive. So the idea of getting off He early is to trade the He which you have too much of, for N2 which you can afford a little more of.

[adam]

Hi guys,

One of the key mis-conceptions about about He on gassing and AAB's model is this 2.65 figure.

His, and most dissolved gas models, assume that He is on gassed and off gassed at a rate 2.65 times faster than N2. What the 2.65 figure actually is though is the bunsen co-efficient for lipid solubility of He over N2. Hence He is actually 2.65 times less soluble than N2.

The rate of on and off gassing is, of course, much faster in He than N2, as it is a smaller, lighter and more mobile molecule. This rate though is not linked directly only to it's solubility, but also to its molecular size (monatomic etc). I've read Buhlman, and he simply ignores (or didn't complete the research into) this issue.

OK-so we have a gas that is far more mobile across cell membranes than N2, and is also much less soluble than N2.

If you look at it in these terms it gives two effects:
1. He is much more deco "friendly" than N2.
2. As it will come out of solution so rapidly, it is critical to limit the pressure change over time curve-i.e. do "extended" deep stops.

The difference is that these deep stops will actually remove gas from tissue, not simply control bubble size. The aim is also to eliminate extravascular bubble formation during the deep portion of the ascent.

Although Buhlmann models have been modified to incorporate rate of ascent controlling deep stops (GF, Pyle etc.), it cannot actually allow for actual off-gassing on deep stops, or the rate at which He off-gasses. Thats why when you add more He, or add time at deep stops, it adds deco time and/or lowers ceilings.

Thats how I see it anyway!!

Adam

[Mark Chase]

VR3
As I understand it the VR3 treats bubble control as a separate mechanism on top of the underlying Buhlmann algorithm. It continues to on-gas compartments below ambient whilst on the deep stop. Compartments over ambient will off-gas but I don't think it gives any additional credit for the fact you stopped deep. The VR3 deep stop algorithm is very similar to Pyle roughly halving depths between stops.

In practice I can assure you that extending the deep stops on a VR3 significantly reduces the 6m stop. It can also remove intermediate stops. The computer has a look ahead stop depth and time list. So, when leaving the bottom you can see your commitment split in relation to the TTS figure. Slowing the ascent and extending the deep stops will alter the profile. Some of the deep stops will clear before you actually reach the stop depth if the ascent is slow enough.

ATB

Mark Chase

[MattS]

In practice I can assure you that extending the deep stops on a VR3 significantly reduces the 6m stop.

It is most likely just a feature of the underlying Buhlmann model. The stop forecast works by looking at the current compartment load, assuming things about the ascent profile and playing it forward. The further you vary from the programmers assumptions the less accurate the ascent prediction is.

Extending the deep stop is one way to vary from the assumption. If the compartment which is forecast to control the length of the last stop is over ambient on the deep stop, it will off gas and the last stop will be shorter when you get to it. The same thing happens if you are ascending slower than 10m/min while the controlling compartment is over ambient. You can see a similar effect by setting the last stop to 6m and following the dynamic ceiling shallower. The initial forecast assumed you would do the entire last stop at 6m but by following the ceiling you can shave several minutes off becuase the controlling compartment is off-gassing faster while you are above 6m.

Some of the deep stops will clear before you actually reach the stop depth if the ascent is slow enough.

The stop depth is controlled by the M-Value of whichever compartment is leading the decompression. At some point during the ascent the controlling compartment starts off-gassing and the deco ceiling will start to ascend. If you are ascending slowly enough the ceiling can ascend above the indicated stop level so the stop is no longer required and the deep stops recalculated accordingly.

The mechanism used to establish proximity to the deep stop might also cause a deep stop to dissapear. There must be a zone which the computer uses to establish that you are on the deep stop and the time can be counted down. If you were in this zone for 2 minutes the stop would be cleared. I think the VR3 uses a pressure differential from the indicated deep stop so the width of the zone would be relative to the depth - a deeper stop having a wider zone. I will try to remember to ask Nick how he does it next time I see him.

Giving credit for a deep stop would involve driving the underlying dissolved model past the accepted 'safe' limits - for instance setting Baker's GF HI above 100%, ignoring the time on deep stops or increasing FO2 of the final stop. I am pretty sure that the VR3 does none of these things. It sticks pretty rigidly to Buhlmann rules but adding deep stops based on the pressure difference between the max depth and the deco ceiling. The outcome is a deco profile slightly longer than straight Buhlmann.

[Andy Phillips]

MattS

I am forced to give credit where credit is due, that is a nice, clear and consise explanation of how the VR3 applies deep stops to the Buhlmann rules.

Andrew