V-planner or Shearwater

[DanielRJones]

Thanks @KenGordon

This is very interesting. So....I reran the 4 profiles from post #21, that used stop times of 1 min, by using stop times of 1 second. The results are significantly different:
View attachment 480387View attachment 480388
View attachment 480389View attachment 480390

16 minute dive to 100 feet on air:
View attachment 480393
So, there is still a very short amount of stop time when the GF lo is decreased, but it appears considerably less when 1 second stops times are used rather than 1 minute stop times. How does the dive computer deal with this? What does the computer use for stop times? My only Buhlmann computer, a Nitek Q, displays stop times in whole minutes. Are these actually whole minutes or are they fractions of a minute like planning with 1 second stops? In plan mode, the Nitek Q gives the same NDL regardless of the GF lo setting, only dependent on the mix and depth and dictated by the GF hi. GF lo does not kick in unless you exceed deco and cannot directly surface under the GF hi

Them 21seconds also include the traveling time if you look at the dive if you have done for 30ft/min ascent then 10ft is 20 seconds.
As long as you ascend at 30ft/min you are going to have 3 seconds of deco.
I am sure if the software could go to less than 1second you would end up with .3 of a second.

 

[boulderjohn]

Ross Hemingway's Response

Make of it what ever you like but it is clear to me that MultiDeco has an inherent deep stop bias, which is not surprising given Ross's stance on deep stops.

Ross Hemingway cannot post on ScubaBoard any more. He reached out to me via private email with this response and gave me permission to post it here:

Thanks for setting the record straight on MultiDeco GF. The gold standard of GF is the DecoPlanner, as that's where Eric Baker verified and helped to get it included. Our programs are in a near perfect alignment.

These guys trying to use a dive computer as a reference, is doing it backwards. The DC is simplified and has no spare code space or memory to include a full "forward planner", like the desktop computer does. The Dive Computer gets it about right and that all it really needs to do. The DC has fixed ascent rates, fixed rounding, and above all else, the DC is designed to solve a different problem to the desktop.

The likely difference is in the settings of the computer vs the desktop program. Ascent rate is critical, which is fixed in the DC and very adjustable in the desktop. There is the tissue saved state in each one to consider, which you can't see or adjust. Further the desktop rounds the plan and first stop up to the next full minute boundary, which gets the diver back into sync with his timer, while the DC doesn't need to do this. These are just a few of the causes of differences possible.​

 

[elmo]

Be aware (which I don't think many divers are) that running the Buhlmann GF algorithm in Multideco Ross H has incorporated an inherent deep stop bias. By this I mean a dive plan using say a GF of 50/70 on the Shearwater will produce a dive plan more like 40/70 in Multideco, so to match the Shearwater GF50/70 plan in Multideco you will need to plan with a GF 60/70, it does not make a great deal of difference just that the first stop of a minute might be 3 or 6m deeper.

This looks to be a result of how the ascent schedule is calculated, rather than gf-Lo.
Eric Baker's Deco Planner (code is public), and it appears MultiDeco (not public) and GUE DecoPlanner (not public), determines the ceiling at the start of ascent then takes the first deco stop at that level.
Subsurface, and I believe Shearwater computers, calculate the ceiling similarly but add in an extra step. Before deciding that there should be a deco stop at that level they check if the ceiling clears during ascent to it.
Neither approach is necessarily right or wrong but I like to understand how the ascent schedule is calculated.

 

[Tassi Devil Diver]

These are just a few of the causes of differences possible.

That all sounds very reasonable from Ross, pity he can no longer post on SCUBA Board, I wonder why that is? (Maybe the Chairman can offer an explanation).

Or..... just maybe...... Ross H has incorporated an inherent deep stop bias into MultiDeco as my instructor pointed out and which I have observed, now he is not going to fess up to that.

Unfortunately, MultiDeco remains one of the better desktop planning programs so I will continue to use it (please let me know if there is a better alternative), but now I am aware of the deep stop bias I will just factor it in when I use the program.

 

[boulderjohn]

Or..... just maybe...... Ross H has incorporated an inherent deep stop bias into MultiDeco as my instructor pointed out and which I have observed, now he is not going to fess up to that.

I made the comparison and posted the results earlier in the thread. The two were as identical as can be expected. Anyone can make this comparison. If it can be proved that he tampered with the algorithm that way, he would be ruined.

Why don't you publish details instead of saying you have secret information that will assassinate his character with no visible proof whatsoever?

 

[lermontov]

That all sounds very reasonable from Ross, pity he can no longer post on SCUBA Board, I wonder why that is? (Maybe the Chairman can offer an explanation).

Or..... just maybe...... Ross H has incorporated an inherent deep stop bias into MultiDeco as my instructor pointed out and which I have observed, now he is not going to fess up to that.

Unfortunately, MultiDeco remains one of the better desktop planning programs so I will continue to use it (please let me know if there is a better alternative), but now I am aware of the deep stop bias I will just factor it in when I use the program.

what are you going to use to factor that ? unless your doing long deep dives I cant see that it makes huge difference? if you plan your dive on multideco as your base line and then defer to the DC for real time tissue loading arent you going to use the shear water algorithm by default anyway?or do you stick to your slate plan ?

 

[Tassi Devil Diver]

I don't have any secret information, my instructor informed of this phenomenon, so I investigated it for myself by comparing a number of profiles on MultiDeco and on my Shearwater DC dive planning function and I found that to match the two profiles I had to increase the GF low in MultiDeco by 10-15pts. Anyone can do it and form their own conclusion.

I agree in the grand scheme of things it makes bugger all difference but just interesting to know.

 

[TrimixToo]

I don't have any secret information, my instructor informed of this phenomenon, so I investigated it for myself by comparing a number of profiles on MultiDeco and on my Shearwater DC dive planning function and I found that to match the two profiles I had to increase the GF low in MultiDeco by 10-15pts. Anyone can do it and form their own conclusion.

I agree in the grand scheme of things it makes bugger all difference but just interesting to know.

I'd be very interested about the specific profiles you found differences for between the two.

 

[boulderjohn]

Here are the details from the dive comparison I mentioned above. The dive is to 270 feet for 17 minutes using 14/55 with 50% and 100% for deco. The dive is at 5,400 feet. I left out the obvious gas switches, and I did not include the use of a travel gas. I had been diddlying around during the day and had to reenter values. The numbers did not come out exactly as they did earlier in the day with the Shearwater, but they are pretty close.

Here is how it comes out on the Shearwater:

Details are
depth -- time at depth -- (Run time)

270ft -- 17--(17)
120ft -- ASC -- (22)
120ft -- 1 -- (23)
110ft -- 1 -- (24)
100ft -- 3 -- (27)
90ft -- 2 -- (29)
80ft -- 2 -- (31)
70ft -- 2 -- (33)
60ft -- 2 -- (35)
50ft -- 4 -- (39)
40ft -- 5 -- (44)
30ft -- 8 -- (52)
20ft -- 11 --(63)
10ft -- 22 -- (85)​

Here is the Multideco plan. (I took out extraneous information for clarity)

MultiDeco 4.14 by Ross Hemingway,
ZHL code by Erik C. Baker.
Decompression model: ZHL16-C + GF
DIVE PLAN
Surface interval = 5 day 0 hr 0 min.
Elevation = 5,400ft (c)
Conservatism = GF 50/80
Dec to -- 270ft -- (4)
Level to 270ft -- 12:30 -- 17
130ft -- 0:20 -- (22)
120ft -- 1:00 -- (23)
110ft -- 1:00 -- (24)
100ft -- 2:00 -- (26)
90ft -- 2:00 -- (28)
80ft -- 2:00 -- (30)
70ft -- 2:00 -- (32)
60ft -- 2:00 -- (34)
50ft -- 4:00 -- (38)
40ft -- 6:00 -- (44)
30ft -- 8:00 -- (52)
20ft -- 11:00 -- (63)
10ft -- 22:00 -- (85)​

The plans are never more than a minute apart, which I would attribute to rounding. They end with the exact same run time.

 

[atdotde]

I think what some people do not realise that there is a lot of slack in between any model and what is going on in your body in reality. And that is not intrinsically bad as long as you are on the safe side of things. Then it only means that you get out of the water maybe a few minutes earlier or later. So for example the depth of the deepest stop: It does not make much of a difference if that stop is only one or two minutes. This is nearly the same as 0 minutes (assuming sufficiently slow ascent rate). So I would not lose a lot of sleep over one implementation stops at 27m for a minute and the other has the first stop at 24m. One or two minutes difference in a given stop time are not significant (only if they occur at a large part of the stops).

You can see this for example by playing around with your ascent rate. Given that you only have somewhat limited control of that (6m/min or 5m/min don't feel that different in the water) this has an influence on individual stop times.

Even when you nominally use the same model, there is still some slack in the implementation details that are not really defined by the model. So different implementations can come up with slightly different deco plans (I spelled this out in the example of Buhlmann in my blog: Why is Bühlmann not like Bühlmann – The Theoretical Diver ). So claiming some implementation gets (slightly) different results amounts to character assassination is not something I would do. In most of the cases, I would not even worry about the differences.

Regarding Subsurface and the depth at which GF-low applies: That program is not only a planner but mainly a dive log. So it is imperative that the deco algorithm also applies to real dives (logged by dive computers) and not just idealised planned dives. And in particular real dives lack a clear distinction between bottom time/ascent/deco. So it makes not sense for a real dive to say "this value applies at the beginning of deco" as that moment in time is ill defined (you might hold your stops one foot below the target depth then technically you never reach the ceiling and thus never deco). There fore, in Subsurface, we decided to do a running calculation of the ceiling and use the deepest of those values as the depth for GF-low. This might be slightly deeper than planners that have a clear idea of when deco starts (as that ceiling might have gone up by the time the diver reaches that depth) but in the end the differences are quite small, see above.