Algorithms

[ianr33]

Dive Rite: "Using a Modified Buhlmann ZH-L16 algorithm, the NiTek Duo can calculate decompression stop times for required stops as deep as 90-feet." I am assuming the Tusa unit is darn close to this.

Yeah but you probably need to do a 500 foot dive to see that

Do 150 for 20 minutes and V-planner will have the first stop at around 80 feet. These computers will have it at 20,maybe 30 feet. I dont plan on ever ascending directly from 150 feet to 20 feet.

 

[cfelliot]

I posted this some time ago. It is my own modeling of Buhlmann ZH-L16 in Excel:

http://www.scubaboard.com/forums/basic-scuba-discussions/229533-excel-dive-computer.html

 

[texdiveguy]

Yeah but you probably need to do a 500 foot dive to see that

Do 150 for 20 minutes and V-planner will have the first stop at around 80 feet. These computers will have it at 20,maybe 30 feet. I dont plan on ever ascending directly from 150 feet to 20 feet.

Yea.... but regardless of the first stop indicated, the units do recommend based on the calculations from entering deco, the ascent time from your present depth to the first indicated stop (say 40ft. for a number), if you ascent at a slow rate and match this suggested ascent time you will be 'flying the curve' through the deeper 1 minute stops (say the 80-70-60-50) till you reached indicated shallower first stop. As in all ascents, and particularly those with mandatory deco ceilings the rate of ascent is the key. So if you follow the indicated ascent time to the first indicated stop you are in fact completing your deeper stops on a controlled fly along the curve. I would not,,, or ever recommend anyone ascending from depth directly to the surface before considering the risks associated.

 

[ianr33]

I hear what you are saying but I dont think its true. Taking my previous example of 150 for 20 on air V-planner gives
DIVE PLAN
Surface interval = 1 day 0 hr 0 min.
Elevation = 0ft
Conservatism = + 2

Dec to 150ft (2) Air 60ft/min descent.
Level 150ft 22:30 (25) Air 1.16 ppO2, 150ft ead
Asc to 80ft (27) Air -30ft/min ascent.
Stop at 80ft 0:40 (28) Air 0.72 ppO2, 80ft ead
Stop at 70ft 2:00 (30) Air 0.65 ppO2, 70ft ead
Stop at 60ft 3:00 (33) Air 0.59 ppO2, 60ft ead
Stop at 50ft 4:00 (37) Air 0.53 ppO2, 50ft ead
Stop at 40ft 5:00 (42) Air 0.46 ppO2, 40ft ead
Stop at 30ft 8:00 (50) Air 0.40 ppO2, 30ft ead
Stop at 20ft 13:00 (63) Air 0.34 ppO2, 20ft ead
Stop at 10ft 23:00 (86) Air 0.27 ppO2, 10ft ead
Surface (86) Air -30ft/min ascent.

Whereas a computer with a Buhlman type algorithm will have the first stop at (maybe) 30 feet.
So following the computer you would be at 30 feet after 24 minutes (30 ft/minute ascent) but following V-planner tables you should not be at 30 feet until 42 minutes. Thats a big difference.

Of course this is purely academic as I know neither of us would ever dive to 150 feet on air

 

[Web Monkey]

This is why I dive a bottom timer and not a computer. I'm not gonna trust some black box that company X will not tell me what it does to keep me safe. With a bottom timer the deco algorithm is in the hands of you and your team.

Bad news. Either way you're still trusting a black box. It's just that with a printed dive plan, the black box is at your house running deco software, or is a geek in a room somewhere with a slide-rule.

The advantage to printed tables isn't where they come from, it's that they're printed. Tables can't crash and say "ERR" during an ascent. OTOH, they also can't tell you the fastest way to the surface if you cut the dive short.

No matter which you choose, it's just a choice with both good and bad points.

Terry

 

[texdiveguy]

Whereas a computer with a Buhlman type algorithm will have the first stop at (maybe) 30 feet.
So following the computer you would be at 30 feet after 24 minutes (30 ft/minute ascent) but following V-planner tables you should not be at 30 feet until 42 minutes. Thats a big difference.

Of course this is purely academic as I know neither of us would ever dive to 150 feet on air

Your right it is 100% academic!!

These computers are not running a true full on 100/100 Buhlman.......they are factoring in as indicated the 'modification' of Lo and Hi GF's which will effect the ascent times/stop depths and curve....so your V-Planner/conserve 2 using a differing algorithm will not match a Duo's calculations. Since we have no way of here running that profile through the Duo for printing we are only speculating as to actual differences. Regardless, the Modified Buhlman algorithm have so far shown to be 'safe' and effective as things in this decompression world of so much unknown goes.

For me the Duo's have a proven success in my dives were I have used them in the profile of say 150ft/20bt/air and proper flying ascent and stop times. Nothing is a sure bet that's for darn sure regardless of the dive wrist computer or desk top planner we so choice to follow....it's all a numbers game.

 

[ArcticDiver]

Being an engineer, I am always interested in the bits and bytes on how things work. Being a noob to SCUBA, I approached it with the same mentality as I do my job. As I started asking questions, I ran in to the same road blocks that you are running into. Each vendor uses a slightly different calculation to provide safe diving times and mitigating as much risk as possible. Comparing two vendor's algorithms is like comparing dive tables from different agencies. Getting the actual algorithm is not likely going to happen, but understanding the tables helped me determine which ones made since for me. Knowing this led me to ask simple questions when I shopped around for a dive computer....Do you want to dive more conservatively or more aggressively with more bottom time?

Is this what you are trying to determine or are you truly interested in the actual calculations?

Being new to SCUBA, I still won't be able to answer your question for you because I have only researched a few vendors, but I think asking the question in a different way may help you get the actual information you are looking for.

What I'm after is an understanding of the algorithms differences so I can make better decisions. To me the terms "aggressive" and "conservative" are meaningless as applied to dive profiles and DCS. I can go into that further, but that is another thread and has been discussed ad infinitum. What does have meaning is the amount of bottom and run time and where ascent stops are called for.

By knowing the differences I can better judge bottom time. I do know that on the limited profiles I've dived with the two computers(one DiveRite with the Buhlman algorithm and the Zeagle with the Boherer algorithm) the Bohrer seems to give less NDL time and shallower stops than the Buhlman. But, I've only done about a dozen dives, none deeper than 130', with the pair so I don't know if those results were the norm for all profiles.

Gage vs. Take Along Computer. An ongoing and never ending discussion. To me usually it makes no sense to use a land based computer to plan a dive when you can take one with you that will account for actual vs. planned dive parameters. HOWEVER, there are exceptions, like the need to put all divers on the same profile, or when doing dives beyond the capability of submersible computers.

So, just what are the differences in the two algorithms? Or, will that have to be a mystery?

 

[NiclasG]

I am just surprised how some thinks that having access the algorithm&code would change anything. Why could you tell anything about the safety by studying the code? What you need is access to the validation data. I do understand that it would be interesting to study the code but the code consists of simple mass balances for different compartments, maybe with a bubble estimator on top, that are adjusted to give reasonable results by diving physiologists.

The only way validate an algorithm is to make time consuming measurements of bubbles post dive for different profiles. Depending on n (the amount of subjects) this costs BIG money and is the reason why e.g. Uwatec will never release code.

Uwatec has spent years on algorithmic development and the end results:
1) don't do the very deep stops (will not give you less bubbles) and
2) a modified Buhlman is as close to the truth(few silent bubbles for most dives) you can come with very simple calculations. Safe diving!

Niclas

 

[Web Monkey]

I am just surprised how some thinks that having access the algorithm&code would change anything. Why could you tell anything about the safety by studying the code?

For one thing, with access to the code, it's possible to run sumulations for all possible dives, map and compare the resulting deco/ascent profiles and look for anomalies. This is how the Pentium III Floating Point Bug was discovered.

For another, it would give a good indication of how clean and well-thought-out the algorithm's implementation is. It's one thing to say "we use deco model 'X'", it's quite another to accurately implement it on an embedded system with no critical bugs.

The only way validate an algorithm is to make time consuming measurements of bubbles post dive for different profiles. Depending on n (the amount of subjects) this costs BIG money and is the reason why e.g. Uwatec will never release code.

You can compare it to the profiles for the same dives done on other computers, which will give you an accurate ranking of how "conservative" they are and where, as well as show any bugs in the implementation.

 

[Gene_Hobbs]

Comparison to a probablistic model would allow for a risk estimation to be drawn from a given profile. You just have to pick the data set you compare to carefully (LE1 could work nicely for this).

example of BVM(3) model:
Gerth, WA; Vann, RD. Probabilistic gas and bubble dynamics models of decompression sickness occurrence in air and nitrogen-oxygen diving. Undersea Hyperb Med. 1997 Winter;24(4):275-92. RRR ID: 2258

With more detail if you want it:
Gerth, WA, Vann, RD. Development of Iso-DCS Risk Air and Nitrox Decompression Tables Using Statistical Bubble Dynamics Models. NOAA, Office of Undersea Research, 1996; Final Report. RRR ID: 4602