OSTC4 not calculating Deco correctly. Possible Bug?

[Divert]

Hello, in the process of knowing my OSTC4 better I decided to plan a dive, first in Subsurface and then in OSTC4 to compare.
The dive:
50m depth with 60 minutes bottom time.
Algorithm VPM-B +3
Gas 1 Air
Gas2 NX50

Subsurface: 10 different Stops!

The OSTC Plan with same parameters:
Gases:

Settings1:

Settings 2:

[Divert]

Continuation....

OSTC Plan: ONLY 3 Stops!

I started the simulation to understand what would happen and some strange things started to happen: for example

No stop at 15m. Why???

On another run 40m with bottom time of 40 minutes The OSTC4 mandated a deco stop shallower than the calculated Ceiling!!!!

As I startet to get nervous, I switched to Bühlmann ZH-L16C and ran the 40m 40' simulation. where a full deco stop disappeared wile still on the bottom time!!! I took a video for you to watch the behavior. https://nube.buceoluegoexisto.com/s/GsbsiHG2imd4M2G

Please advice.

Humberto

[swissdiving]

I very much doubt therecis a bug. Diving the ostc4 for years now not having a problem with deco....

[Divert]

I very much doubt therecis a bug. Diving the ostc4 for years now not having a problem with deco....

This is so nice to hear! Do you dive also with VPM?

I am so willing to find that I am wrong about this. I love the computer and the support from Matthias and Ralph

[swissdiving]

VPM and Bühlmann.

[heinrichsweikamp]

Thanks for the detailed report, we'll have a look into this. There will be some improvements in the next releases with the VPM (The time for 9m is entered in the time for 6m and then overwritten for the list display) the total ascent time is correct, also the shown "next stop". A bug, yes: But not critical. We'll fix this.

regards,
Matthias

[Divert]

Thanks for the detailed report, we'll have a look into this. There will be some improvements in the next releases with the VPM (The time for 9m is entered in the time for 6m and then overwritten for the list display) the total ascent time is correct, also the shown "next stop". A bug, yes: But not critical. We'll fix this.

regards,
Matthias

Thanks @Matthias What about the disappearing Deco Stop @15m with Bühlmann ZHL-16C as shown in the linked video? Is that a bug on the Simulator or is it a problem with real decompression calculations on the OSTC4?

Please advice if it is safe to dive with.

Regards.

Humberto.

[Divert]

Thanks @Matthias What about the disappearing Deco Stop @15m with Bühlmann ZHL-16C as shown in the linked video? Is that a bug on the Simulator or is it a problem with real decompression calculations on the OSTC4?

Please advice if it is safe to dive with.

Regards.

Humberto.

Any word on this? @Matthias @Ralph ?

[Divert]

Just posting as I am still waiting for an answer.

Any word on this? @Matthias @Ralph ?

[Ralph]

Hi, thanks for mentioning my name as a helpful ressource, let me try to answer to your observations. First of all, i need to say that i do know nearly every line of code of the 2's models, but actually never looked into the model 4 code. Anyhow, i belive your observations relate to fundamental properties and effects of the deco algorithms, and are not specific to the OSTC code (beside the one fault or glitch Matthias already mentioned, of which i have no further information):

VPM

This deco model was originally developed and coded (actually it is defined by some Fortran code) to calculate a dive "offline", i.e. the whole profile needs to be given, then the code calculates a deco schedule. There are known oddities, like when adding a deep bounce dive profile prior to the actual working dive profile, this bounce will shorten the overall deco. So VPM has (like any deco algorithm) its limitations in terms of dive profiles it can handle, and these seem to not be that well (or widely) documented and/or known. On top comes, that VPM was not designed for "realtime" use in a deco computer, as the computer can only know the profile so far, but not the profile of the complete dive, e.g. the part to come yet. So it will take the profile up to "now" and calculate a deco schedule for that dive, and once done it will calculate another deco profile for another dive that is the previous dive plus what happened time & depth wise since the last calc run. So it is a series of deco calculations that is carried out while the dive progresses. This is not different to other algorithms as e.g. Buehlmann, but as shown whith the bounce case, VPM seems not to be so "linear" as other algorithms, but giving abrupt changes in the deco schedule while the dive profile continuous kinda smooth.

Buehlmann with GF

Here the explanation of the disappearing stop lies within the GF extension of the Buehlmann algorithm, along with your extreme GF setting of 30/80. When you look closely, you see that the 15 m stop goes away for a while (it will come back when you extend the bottom time), but not the deco itself: the shallower stops take over the time from the 15 m stop. With GFs, the 1st stop will be placed at the depth where the leading compartment reaches GF low. This depth is then rounded down to the next multiple of 3 meters. All stops further up will have progressively more over saturation, until the last "stop" at the surface which will see GF high as its limit. Every time a new (deeper) stop gets added, it will set a new reference depth for GF low, and hence all existing stops will get some higher supersaturation limit than before assigned (remember: the fixed span GF high - GF low gets evenly distributed between the variable depth span of 1st stop to surface). These increases in allowed super saturation per each stop become the bigger, the bigger the difference between GF low and GF high is. That's why i called 30/80 extreme, because it will allow much more super sat at any stop whenever stops below get added.
The computer will add a stop when a stop is needed for at least 1 second, and this stop time will be rounded up to the next full minute. So the 15 m stop kicks in when computations say that the super sat of GF low will be reached at 12.01 meters. At the same time, the allowed super sat for the 12 m stop (previously the 1st stop) will rise from GF low to GF low + ((GF high - GF low) x 3m / 15m). The computer calculates the off-gassing of the tissues while the ascent, here while going up from 40 m to 15 m. Before, the 1st stop was calculated to be at 12.01 meters (which gave the 15m on the schedule). Now, as some time passed by, it may be at 13 meters when applying the previous super sat limit (still yielding 15m in the displayed schedule), but as the allowed super sat at 12 m has increased, the new stop depth may be 11.5 meters now (just for example), leading to a 1st (displayed) stop at 12 meters (again). The catch is: once a new, deeper stop is found, this depths becomes the reference depth for GF low. This reference depth will only become deeper during the dive with more deco piling up, but will never be reset to a lower depth. If one would do that, the deco calculation would become instable and deliver oscillating results. Once again, also the GF extension was not designed to be used in "realtime" mode...

Does that make sense?

Please allow me to add a personal general note, reflecting many questions that were raised here in the forum over the years: when i made my OWD some 20 years ago, we had to learn the RDP (PADI recreational dive planner) and hence learn the effects of time, depth and surface interval on the no-deco dive time. As said, that was on OWD = basic underwater survival level. Later the RDP was dropped py PADI and the students were told to buy a computer and "let their instructor tell them how to use it". Still OWD here. Today, to get to full trimix, it takes about 4 courses to get there, ~4 times ~1.000 € course fees, and i wonder how many tec instructors do teach their students sound knowledge on how deco calculation (i.e. the algorithms, not just button tapping sequences) works, or know by themselves...

Yours,
Ralph

[chrisrms]

Hi,

I just wanted to let you know, that I did experience the same problem with the OSTC4, that deco stops suddenly disappeared during a dive and the OSTC4 showed 240 minutes. I only dive it with VPM +3. Because this happened, during this same dive, I switched to Bühlmann with GF (30/85) and there, the correct to me stops where still there, I switched back to VPM, but no stops, no deco. The still left deco time, using Bühlmann with GF, was at 6 meters and 3 meters, +/- 10 minutes.

Some further informations: I dive a JJ-CCR and the OSTC4 is the backup computer. The controller on the JJ-CCR did have +/- the same deco time left as what the OSTC4 showed using Bühlmann with GF.
The dive: bottom level was 20 minutes @ 41 meters with TX 20/15, whole dive time 83 minutes.

I never had such, but I hope this was just a one time problem, because +/- 10 minutes missing and deco cleared is shown could IMO be a problem and I'm glad I had another computer which showed the correct deco.

Perhaps my dive can be downloaded from the OSTC4 somehow and analyzed to see what happened?

Best regards
Christoph

[swissdiving]

@Ralp
Thanks for the explanation to VPM (and GF low/ high) That was extremly enlightning
To your last comment...

[Thorsten]

Having Ralphs explanations in mind the next firmware version will have some updates of the VPM function. Not regarding the algorithm itself but with regard to the output data usage.
The planner will freeze the worst case deco table without simulating an ascent => The values provided will look like the tables shown by other offline deco planners.
For the simulator and real dives a VPM table mode will be available as new option. The table mode will keep the deco plan at the time the deco phase is entered. The deco stops will not be completed based on continued deco calculations but by a timer which does a count down while the diver is close to the stop depth. Of course the deco calculation still runs in the background. If the TTS increases for some reasons (bouncing, missed gas switch) the table will be updated to the new (worse) timings.

Summary: An option adding some conversatism for VPM diving and a challenge for those who want to check their buoyancy skills (deco stop area is 1.5m... stay within or stay longer in the water) ;-)

Kind regards,
Thorsten