Meaning of GF settings (was: deco calculations)

Robert C. Helling helling at
Sat Jan 5 09:15:18 PST 2013

On Jan 5, 2013, at 12:17 PM, Jan Schubert wrote:

> Please note that GF is related to Bühlmann and Suunto is not using
> Bühlmann at all but RGBM (Reduced Gradient Bubble Model). AFAIK RGBM is
> not publicaly documented but might be based or similar to VPM, both
> bubble based models. In difference to Bühlmann these models try to avoid
> the formation of bubbles in tissues right from the the beginning while
> Bühlmann assumes there would be bubbles anyway and tries to keep them
> small (smaller than a critical size). In real life this results in
> deeper deco stops at VPM and and shallower but much longer ones for
> Bühlmann. Nowadays most dive Computers use Bühlmann meaning this is more
> tested (funny side note: Definitions for the maximum tissue saturations
> in the Bühlmann model are based on tests done by Haldane using goats in
> pressure chambers).
> Anyway, back to the Meaning of GF: As said, Bühlmann defines maximum
> overpressure values for all 16 tissue compartments (there are also
> easified models using "just" 8 compartments). Bühlmann itself does not
> know of GFhigh and low. If you set GFlow=GFhigh=100% then you use plain
> Bühlmann (mostly as most dive computers and calculations ad some more
> safety beside this). To add more safety to deco calculations and to
> adapt to individual preferences (like fitness, illness, coldness, ...)
> the GF limitations has been introduced meaning we just use 75% of the
> maximum overpressure defined by Bühlmann assuming you use a GFhigh=75%.
> Sidenote: The GFhigh defines the max. overpressure when surfacing, NOT
> when being/reaching/leaving the last deco stop (this is a point many
> people get wrong). To make it more complicated^Wadjustable there is an
> additional GFlow setting which defines a (usually) lower/reduced maximum
> overpressure for the beginning of the decompression phase. This also
> tries to adapt to a (disputed) deep stop based theory and heads towards
> the bubble based models in avoiding the formation of bubbles. In other
> words: GFlow and GFHigh are "just" a linear reduction of the max.
> overpressure values for all the 16 tissue compartments. The higher the
> GF settings are the shorter and shallower the deco stops but the more
> risk in getting bend. Note: GFhigh and also GFlow can be defined with
> values more than 100% if one like to use such settings.
> And to make it even more complicated/easified there are also these 3m
> based stops introduced (10ft for imperial based people), meaning there
> is no floating deco ceiling ("smooth" using definitions by subsurface)
> but fixed stops based on depth in multiples of 3m.

To add to this from a physicists perspective: There is a very good physics/physiological motivation for diffusion based models like Bühlmann as diffusion being most natural (speed of equilibration proportional to difference). So there is no question about the tissue loading equations. There is far less evidence for the resulting ceiling and indeed Haldane and Bühlmann have done a lot with respect to real life testing (goats but also humans). The problem is, as you say, that the tissue loading is likely not the complete picture as it ignores bubble dynamics. VPM (and if RGBM does what it claims) try to add this. The downside is, that these modifications add a significant number of more parameters and their numerical values have not been empirically determined as extensively as the ones in Bühlmann etc. They have been tuned to yield deco profiles that are considered good (with deeper stops and not too much time to the surface). But they have been tuned to these profiles afaik, not with real testing (i.e. scanning parameters with lots of test subjects and measuring how often goats/people get bent, this is what Bühlmann did). 

As you write, the gradient factors have no physiological motivation whatsoever, but it was found that with those you can produce similar profiles to the ones of the bubble models were created to yield. So here, you just acknowledge that you don't have a good microscopic physical model (as the bubble models without well motivated parameters) and that all you really want are good profiles. So maybe one could call GF's less physiological but more honest.  Even further in that direction goes 'deco on the fly' and the like where you simply do a linear interpolation of assumed to be good profiles without any attempt to model anything in the body.

So, in particular with GFlow, you force comparatively deep stops that are supposed to eliminate bubble before they grow. But it is really a hack. GFhigh just adds some conservativism in the later part of the deco.


Robert C. Helling     Elite Master Course Theoretical and Mathematical Physics  
                      Scientific Coordinator                                   
                      Ludwig Maximilians Universitaet Muenchen, Dept. Physik    
print "Just another   Phone: +49 89 2180-4523  Theresienstr. 39, rm. B339       
    stupid .sig\n"; 

More information about the subsurface mailing list