Artifacts with VPM-B ceiling on level changes

[Rick Walsh]

On 30 Aug 2015 2:32 pm, "Gaetan Bisson" <bisson at archlinux.org> wrote:
>
> [2015-08-30 07:20:18 +1000] Rick Walsh:
> > Good morning,
> >
> > On 30 Aug 2015 7:12 am, "Robert C. Helling" <helling at atdotde.de> wrote:
> > >
> > > Gaetan,
> > >
> > >> On 28 Aug 2015, at 05:31, Gaetan Bisson <bisson at archlinux.org> wrote:
> > >>
> > >> I've noticed what looks to me like artifacts regarding the ceiling
VPM
> > >> calculates for multilevel dives. See the attached screenshot.
> > >>
> > >> Basically, during manually entered ascent segments, the calculated
> > >> ceiling becomes deeper as the diver goes shallower...
> > >>
> > >> It might just be a well-known VPM "feature" but I still thought I
would
> > >> ask here if anyone knows what is going on.
> > >
> > >
> > > I figured out the origin of this „feature“: The ceilings we plot are
> > based on the allowed gradient (tissue pressure minus ambient pressure)
of
> > the tissue. This gradient is depth dependent (according to the „Boyle
> > compensation which says that (gradient + ambient pressure)/gradient^3
must
> > be constant during the ascent. When we plot, we use the gradient for the
> > current depth but that shrinks as soon as we ascend and so the ceiling
> > comes down upon the ascent. We should better plot a ceiling defined as
the
> > depth where the Boyle compensated gradient is the actual gradient and
then
> > the ceiling would no longer be depth dependent. But this requires some
more
> > work, which I have to postpone to tomorrow.
> >
> > I no longer see the feature. I think it was the sixth patch in the
series
> > that got rid of it.
>
> Thanks so much to both of you.
>
> Rick, I still seem to see the artifacts with latest master ( d22a135);
> see attached. Should that not include the commit you refer to?
>
Yes, you are correct, they are still there. In the profiles I was looking
at, they were so small I didn't notice.

> Of course, in light of Robert's explanation, it might not be worth
> getting rid of these: we are potting the model data faithfully and it's
> just how it works. We do not really have to work out something more
> convoluted to plot just so as to make VPM-B look nicer.
>
I think the value of plotting the deco ceiling really lies in plotting it
on real dives. We're not able to do that yet, but getting it right in the
planner could be an intermediate step.

Rick
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.subsurface-divelog.org/pipermail/subsurface/attachments/20150830/59722ca0/attachment.html>