Meaning of GF settings

[Robert C. Helling]

On Jan 8, 2013, at 6:45 PM, Jan Schubert wrote:

> - It (still) seems to use the ceiling-at-this-very-specific-moment
> approach, not considering a ascent which explains the ceiling deeper and
> earlier in the descent phase of the dive. Just a note, it's more
> important that we get this right for the dive planning part of
> subsurface. And if we (would) have it there, it should not be that
> tricky to apply this to this part as well!?

This would complicate the algebra even more and at least that would not match my understanding of the word "ceiling". 

For the ceiling during the descent part of the dive: For the gradient factors to work one has to use the max depth of the dive (as that is where GF_low applies). The code does not use the global max depth but the max depth up to that point (which is of course the same during ascend where it matters) or a pre-defined depth (20 or 30m) if that is deeper. I don't know what your dive computer does, do you set a target depth manually or can it look into the future? What I want to say is that the gradient factor calculation is not well defined in the early part of the dive. 

The ceiling really matters when you get close to it which also means the difference between ceiling now and ceiling once you get there goes to 0.

> - In the ascent phase the deco ceiling calculated by subsurface seems
> always behind (deeper and longer ceiling!) than the one given by the
> dive computer (look for the much longer shallower stops in the
> screenshot for example), this is something we should have a look for. My
> guess is, that there could (!) be a difference how subsurface and dive
> computers interpolate between GFlow and GFhigh. It looks subsurface is
> more conservative (assuming lower GF on the way up and much longer deco)
> as subsurface.

I implemented my interpretation of Baker's paper (or actually its last figure). That said I guess all dive computers add some sort of fudge factors for their deco curves to come out nicer and of course those are not publicly documented. Even Buehlmann did that in his book: He gives the algorithm and constants but gives tables as well but those tables differ from what the plain vanilla algorithm computes.

> - In case the calculated deco is not ended when the original profile
> "surfaces" it might be helpful to extend the time for the profile shown
> to get the calculated deco ceiling displayed completely (this would
> allow better playing with existing dives and such parameters as GF).

Good idea. Which depth would you use for the remaining off-gasing, surface or ceiling (in which case it becomes more akin to dive planning).

BTW, that reminds me of a bug the first Aladin nitrox computers had: They assumed for repeated dives that in your surface interval you were still breathing the (last) nitrox mix which is of course rarely true (and is an error in the dangerous direction). Do we do this correctly when integrating tissue loadings of more than one dive?

> - seeing the smooth profile I'm quite amazed that the non-linearities
> vanished nearly completely, I'd not have expect this before.

Could you send a screen shot of your dive with smooth ceiling as well? My guess would be that the differences between the dc ceiling and deco.c get smaller (since we always round up to full multiples of 3m).

Good night
Robert

--                                                                              
.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oO 
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";   http://www.atdotde.de