Simulator in OC mode

[Bardass]

Hello, i have a question about the simulator.
I noticed that it calculates some different deco plans for the same dive.

Here are comparisons of deco plans calculated by the OSTC according to several scenarios.

the file is located here:
http://lbardass.free.fr/images/plongee/Deco% 20OSTC.htm

two tabs are used.

* Variation OSTC: provides various simulations by OSTC using various actions (back to previous menu, return home, reset ...)

* Déco algo : compares deco plans by OSTC with plans generated by different programs : MV-MAP, GAP, HLP, SUUNTO.

I only used the OC algorithms.
The parameters used are provided in the header tabs.
Versions of the firmware used were 1.70, 1.74 and 1.75.
The three versions provided the same results.

* Algorithm with the ZH-L16 (GF11 and 12 default), the total length of stops is the same : 7 minutes.
Only duration varies for the stop at 6 and 5m.

* Algorithm with the L16-GF, the differences are more important.
When no parameter is changed between each simulation, the stop duration is between 10 and 13 minutes with stop at 12m canceled according to the simulations

* If i compare OSTC L16-GF (10 to 13 minutes), the results are almost the same with MV-PLAN (11 min) and GAP Buhlman mode (B or C at 10 min) : the deco plans are very close.


I still do not understand why and how the simulator by OSTC calculates some differents plans Deco for the same dive. This goes against the principle of an algorithm : for a given set of parameters, it should provide identical results.
Maybe the simulator uses other parameters such as pressure, temperature ...

I don't think that successive and consecutive simulations can explain différences : times total tier vary downward or upward so that the simulations are strung one after another.
Returns to the menu, reset deco : those actions had no effect on the stability of deco plans.

Any ideas ?

[dmainou]

Ok,

I've been there and done that and I couldn't receive a straight answer either.

http://www.heinrichsweikamp.com/read.php?2,3603

http://www.heinrichsweikamp.com/read.php?2,3504

After many trils I was able to obtain "semi" consistent results on the OSTC. To do it you need to exit all menus between simulations. I did 20 runs and averaged them to get the average profile. (The crucial bit is exiting all menus before running the simulator again)

Then went to Gue’s decoplanner and set as many parameters as possible in line with the OSTC including one called "depth safety factor" which uses the depth multiplied by 1.03 + 1 meter (close enough to the +1 meter on the OSTC default setting)

The average OSTC outcome is basically in line with GUE’s decoplanner. On the OSTC the last stop is two minutes longer and one of the deeper stops is, on some runs, a minute longer but no big deviations.

I hope this helps.

Having said that, hopefully Matthias or one of the other developers will be able to shed some light this time round.

Thanks,

D

[Bardass]

the good link

http://lbardass.free.fr/images/plongee/Deco%20OSTC.htm

[Bardass]

thanks Dmainou
when you say : "After many trils I was able to obtain "semi" consistent results on the OSTC."

i agree with you, but i don't understand why the results are "semi" consistant.

when you say : "To do it you need to exit all menus between simulations. I did 20 runs and averaged them to get the average profile. (The crucial bit is exiting all menus before running the simulator again)"

i can't agree because in my document, tab "Variation OSTC",
if you compare actions 3, 7, 10, 16 (go back to home menu), results are not the same : 11 mn or 13 mn


if you compare any same actions :
* "go back home menu", actions 3, 7, 10,16 : total stops = 11 or 13 mn
* "reset decodata", actions 4 and 11, total stops = 10 or 11 mn
* "exit simulator", enter simulator, actions 5, 8, 14, 17, total stops = 10, 11, 12 mn
* "recalculte without exit", actions 6, 8, 12, 13, 15, total stops = 10, 11, 12, 13 mn

when you say : "The average OSTC outcome is basically in line with GUE’s decoplanner"

i agree with you.
i show it in my document, tab "déco algo"



OSTC is a marvellous computer, i just want to understand

[dmainou]

Bardass Wrote:
-------------------------------------------------------
> thanks Dmainou
> when you say : "After many trils I was able to
> obtain "semi" consistent results on the OSTC."
>
> i agree with you, but i don't understand why the
> results are "semi" consistant.
>
> when you say : "To do it you need to exit all
> menus between simulations. I did 20 runs and
> averaged them to get the average profile. (The
> crucial bit is exiting all menus before running
> the simulator again)"
>
> i can't agree because in my document, tab
> "Variation OSTC",
> if you compare actions 3, 7, 10, 16 (go back to
> home menu), results are not the same : 11 mn or 13
> mn
>
>
> if you compare any same actions :
> * "go back home menu", actions 3, 7, 10,16 : total
> stops = 11 or 13 mn
> * "reset decodata", actions 4 and 11, total stops
> = 10 or 11 mn
> * "exit simulator", enter simulator, actions 5, 8,
> 14, 17, total stops = 10, 11, 12 mn
> * "recalculte without exit", actions 6, 8, 12, 13,
> 15, total stops = 10, 11, 12, 13 mn
>
> when you say : "The average OSTC outcome is
> basically in line with GUE’s decoplanner"
>
> i agree with you.
> i show it in my document, tab "déco algo"
>
>
>
> OSTC is a marvellous computer, i just want to
> understand


What I meant for "semi" is that on average each stop matches each stop generated by the desktop software. i.e. after 20 runs most of them will match the result of the desktop software, some stops will have a minute more whilst others will have a minute less. So, on average, the stop is the same.

Also on those stops where there is a one minute less deco there is a one minute more deco at the previous or next stop so it is not like it actually got lost.

I would like to fully understand as well but for that we will need to understand the code... (or have Matthias to explain the differences to us)

[dmainou]

Oh BTW,

Have you modified the following parameters on the desktop software?

• Make sure you are using ZHL16+GF rather than RGBM, VPM or any other algorithm.
• Pressure to match the OSTC reading on the main screen
• Immediate Descent (this seems to do a lot of difference on simulations)
• Ascent rate (as per the relevant CF e.g. 7m/min)
• match the High and Low GF's (I know this one is obvious)
• Last stop as per CF on OSTC
• Depth safety factor ON or offset on OSTC = zero

If the OSTC is still off by a couple of minutes you'll notice that it is likely to be on the 3 min stop. So a couple of min more at 3 meters is not something I'll get worried about.

Best regards,

D