Why you can't use the PADI RDP table for Multi-Level dives

[L13]

Can you at least agree that if the method aligns or is more conservative than Buhlmann, the origin of the method or the research (or lack thereof) is of less importance?

But if Buhlmann is your metric, why not just use Buhlmann? Why bring in another method of questionable origin?

Or is the goal to have a table/procedure that is a backup to a Buhlmann DC that is at least as conservative?

In that case, wouldn't a backup Buhlmann DC be less error prone and therefore more reliable?

 

[steinbil]

But if Buhlmann is your metric, why not just use Buhlmann? Why bring in another method of questionable origin?

Or is the goal to have a table/procedure that is a backup to a Buhlmann DC that is at least as conservative?

In that case, wouldn't a backup Buhlmann DC be less error prone and therefore more reliable?

The whole point is to test this method ("of questionable origin"). I have been taught this method. I've used it. I know many divers that have used it, some that still do. We can discuss the idea behind having such a method - why learn it, why use it, what's the benefit - but that is kind of a sidetrack. Personally I now use Buhlmann. I would like to see how the computer diving I'm doing now stacks up against the multilevel table diving I did before. I now use the tables (memorized) as a backup, and as a sanity check to see if what my computer is telling me is in the ballpark of what I would expect. So I would like to know just how aggressive or unreliable this method is.

 

[tursiops]

The whole point is to test this method ("of questionable origin"). I have been taught this method. I've used it. I know many divers that have used it, some that still do. We can discuss the idea behind having such a method - why learn it, why use it, what's the benefit - but that is kind of a sidetrack. Personally I now use Buhlmann. I would like to see how the computer diving I'm doing now stacks up against the multilevel table diving I did before. I now use the tables (memorized) as a backup, and as a sanity check to see if what my computer is telling me is in the ballpark of what I would expect. So I would like to know just how aggressive or unreliable this method is.

What flavor Buhlmann do you want to use, and what GFs?

 

[steinbil]

I ran a few simulations...

Software: Subsurface
Algorithm: Buhlmann ZHL-16C
Ascent speed: 9m/min to 50% depth, 3/min to surface

Dive 1a
multilevel dive 30/24m avg 27m - 40 minute bottom time

	depth​	duration​	runtime​
➘	30m	2min	2min
➙	30m	18min	20min
➚	24m	2min	22min
➙	24m	18min	40min
➚	0m	6min	46min

SurfaceGF% 86

Dive 1b
square profile to 27m - 40 minute bottom time

	depth​	duration​	runtime​
➘	27m	3min	3min
➙	27m	37min	40min
➚	0m	6min	46min

SurfaceGF% 89

Dive 2a
multilevel dive 30/24/18m avg 24m - 50 minute bottom time

	depth​	duration​	runtime​
➘	30m	3min	3min
➙	30m	15min	18min
➚	24m	1min	19min
➙	24m	15min	34min
➚	18m	1min	35min
➙	18m	15min	50min
➚	0m	5min	55min

SurfaceGF% 81

Dive 2b
square profile to 24m - 50 minute bottom time

	depth​	duration​	runtime​
➘	24m	2min	2min
➙	24m	48min	50min
➚	0m	5min	55min

SurfaceGF% 89

Dive 3a - continous ascent profile from 30m avg 22.5m (using 24m table) - 50 minute runtime

	depth​	duration​	runtime​
➘	30m	3min	2min
➚	0m	47min	50min

SurfaceGF% 31

Dive 3b
see dive 2b

I planned all the multilevel dives using the GUE method for average depth to the maximum MDL limits of the table. In this tiny sample size it looks like the depth averaging is more conservative than a square profile.

 

[steinbil]

Dive 4a
multilevel dive 24m/12m avg 18m - 100 minute bottom time

	depth​	duration​	runtime​
➘	24m	2min	2min
➙	24m	48min	50min
➚	12m	2min	52min
➙	12m	48min	100min
➚	0m	3min	103min

SurfaceGF% 74

Dive 4b
square profile 18m - 100 minute bottom time

	depth​	duration​	runtime​
➘	18m	1min	1min
➙	18m	99min	100min
➚	0m	4min	104min

SurfaceGF% 92

Dive 5a
multilevel dive 27m/15m/21m avg 21m - 60 minute bottom time

	depth​	duration​	runtime​
➘	27m	2min	2min
➙	27m	18min	20min
➚	15m	2min	22min
➙	15m	18min	40min
➘	21m	2min	42min
➙	21m	18min	60min
➚	0m	5min	65min

SurfaceGF% 80

Dive 5b
square profile 21m - 60 minute bottom time

	depth​	duration​	runtime​
➘	21m	2min	2min
➙	21m	58min	60min
➚	0m	5min	65min

SurfaceGF% 83

Dive 6a
multilevel dive 30m/21m/15m time weighted avg 24m - 50 minute bottom time

	depth​	duration​	runtime​
➘	30m	3min	3min
➙	30m	20min	23min
➚	21m	2min	25min
➙	21m	12min	37min
➚	15m	2min	39min
➙	15m	11min	50min
➚	0m	4min	54min

SurfaceGF% 76

Dive 6b
see dive 2b - SurfaceGF% 89

 

[steinbil]

So far, so good. That some of the square profiles are more aggressive than GF85 shouldn't surprise me, I guess. I have always added some conservative padding when diving tables.

Any ideas for profiles to try that would debunk this method?

Edit to add:
In practice this method will be even more conservative (and less efficient in terms of getting maximum bottom time) as you always round the average depth down. On purpose I used exact averages to match the table and used the maximum MDL times to test the boundaries of the method.

 

[L13]

I ran a few simulations...

Software: Subsurface
Algorithm: Buhlmann ZHL-16C
Ascent speed: 9m/min to 50% depth, 3/min to surface

Dive 1a
multilevel dive 30/24m avg 27m - 40 minute bottom time

	depth​	duration​	runtime​
➘	30m	2min	2min
➙	30m	18min	20min
➚	24m	2min	22min
➙	24m	18min	40min
➚	0m	6min	46min

SurfaceGF% 86

Dive 1b
square profile to 27m - 40 minute bottom time

	depth​	duration​	runtime​
➘	27m	3min	3min
➙	27m	37min	40min
➚	0m	6min	46min

SurfaceGF% 89

Dive 2a
multilevel dive 30/24/18m avg 24m - 50 minute bottom time

	depth​	duration​	runtime​
➘	30m	3min	3min
➙	30m	15min	18min
➚	24m	1min	19min
➙	24m	15min	34min
➚	18m	1min	35min
➙	18m	15min	50min
➚	0m	5min	55min

SurfaceGF% 81

Dive 2b
square profile to 24m - 50 minute bottom time

	depth​	duration​	runtime​
➘	24m	2min	2min
➙	24m	48min	50min
➚	0m	5min	55min

SurfaceGF% 89

Dive 3a - continous ascent profile from 30m avg 22.5m (using 24m table) - 50 minute runtime

	depth​	duration​	runtime​
➘	30m	3min	2min
➚	0m	47min	50min

SurfaceGF% 31

Dive 3b
see dive 2b

I planned all the multilevel dives using the GUE method for average depth to the maximum MDL limits of the table. In this tiny sample size it looks like the depth averaging is more conservative than a square profile.

In general, I would expect this to be true any time the later part of the dive was significantly shallower than the earlier part of the dive. You are spending significant time shallower than the average off-gassing from the early part of the dive. If the dive had portions deeper than the average just prior to the final ascent (possibly a violation of the GUE procedure), you would expect the opposite result, possibly dangerously so.

If you run the dive of 1a with the 30m and 24m segments reversed (same average depth), your Surface GF hits 100% 3min into the 30m segment, and at the end of the dive you have a SurfaceGF of 119%.

 

[inquis]

In this tiny sample size it looks like the depth averaging is more conservative than a square profile.

Yes, since you're mandating the second depth be shallower, the worst-case surfGF for a given average depth occurs when the two depths are equal to the desired average and times are equal (ETA: meaning all 40 mins are at the average).

 

[L13]

So far, so good. That some of the square profiles are more aggressive than GF85 shouldn't surprise me, I guess. I have always added some conservative padding when diving tables.

Any ideas for profiles to try that would debunk this method?

Edit to add:
In practice this method will be even more conservative (and less efficient in terms of getting maximum bottom time) as you always round the average depth down. On purpose I used exact averages to match the table and used the maximum MDL times to test the boundaries of the method.

Does the method strictly require depths to monotonically decrease after the initial decent? If not, it will not always be more conservative!

 

[steinbil]

Does the method strictly require depths to monotonically decrease after the initial decent? If not, it will not always be more conservative!

Yes, one of the requirements of this method is that you start at the deepest point. I included one profile with a twist (27m-15m-21m) to see what the effect would be - it is indeed closer to the square profile, but still a tad more conservative. Of course, it's not as flexible as diving with a computer.