Bubble model vs. Gradient Factors redux

Please register or login

Welcome to ScubaBoard, the world's largest scuba diving community. Registration is not required to read the forums, but we encourage you to join. Joining has its benefits and enables you to participate in the discussions.

Benefits of registering include

  • Ability to post and comment on topics and discussions.
  • A Free photo gallery to share your dive photos with the world.
  • You can make this box go away

Joining is quick and easy. Log in or Register now!

Careful now. One particular self-professed deco science guru (/programmer) without academic credentials might come along telling you that if you up the VPM conservatism level too far, that doesn't make sense and isn't a proper algorithm. Even if there's no such limitations in the original algorithm.

A model fits reality within a range of parameters. If someone intimately familiar with the model tells me this parameter is out of range, I'd be inclined to listen. Baker's math does not preclude GF Hi of 108, but I strongly suspect he too would tell you that pushing GF over 100 is not the proper use of his algorithm.
 
Using the NEDU profile again as an example, it is easy to reasonably match the run time of VPM-B+2 with one or more GF pairs and meet the @Storker criteria from post #48. For instance you could test 50 or 60/80 and 75/75. Not quite as easy to match VPM-B+4 but you could test 50/70 and 65/65. The first of each pair look similar to profiles in use, the second of each pair would maximize the difference in deep vs. shallow stops between VPM and Buhlmann
 
My interest in these discussions is twofold: Firstly I am a science nerd so I enjoy these discussions on the purely theoretical level. Secondly, as a diver, I am interested in the practical application, more specifically I want a science-backed answer to the following question:

If I am planning a dive to 70m for 25 minutes BT, which deco algorithm should I use to plan, given that I would like to achieve the shortest runtime within my conservancy limits, as well as keep me as safe as is reasonable during that ascent.

For me, I am not too concerned with actual differences in runtime per se, I accept that different algorithms will by their nature give me different TTS and ascent slope. What I want to know is whether a profile is going to get me out of the water sooner but at a penalty of "safety factor" or whether I can trade some extra time in the water for a decent increase in safety by following a slightly longer TTS that protects my "fast tissues" or limits my supersaturation (heat maps) etc.

Right now, I know that we do not have enough information to make a truly definitive conclusion. However, there are folks who know more about deco theory than I ever will, who treat DCS patients regularly and who have spent large parts of their life studying this topic. I am confident enough to follow their recommendations since none of the science I have seen so far points against it, even if the evidence for is (by necessity) a little circumstantial.

I am personally diving 50/85 for short exposure dives where a chamber is reasonably close and 50/70 for when I need to be more conservative (multi dive days and remote areas). Anecdotally, I do feel as if my "nitrogen tiredness" is much less noticeable post dives but that's worth the paper this is printed on as far as being "evidence" goes. This compared to 20/85 and 30/85, my previous algorithms.
 
This is just pure speculation, but it's how I tend to look at the graph in the interview:
But if venous bubbles have a very coarse correlation with risk, who would you rather be?
This guy?
View attachment 582840
Or this guy?
View attachment 582841

20200426_103217_1-jpg.582840.jpg
20200426_103307-jpg.582841.jpg



I read through the thread, much of this is in the weeds and/or over my head. But these 2 pictures are very interesting. If I understand this bar chart correctly, it seems the goal for safety would be to take whichever deco prescription yields the smallest maximum bubble grade, across all measurements. Squinting at these 2 pictures, it appears that the 20/85 prescription beats out the 50/75 by a small but visible margin assuming a single dive. If doing multiple dives, it would seem that the 50/75 prescription wins out because after the 90 min SI the bubble grade is much smaller. Do I have that right, or have I misunderstood the meaning of these histograms?

I wish there were another data point or two in the [0,30] minute range. I have no idea why this is, but accept that there's probably a good reason not to have them -- maybe logistics, or perhaps the data is too noisy shortly after ascent, who knows. Even better would be data points in the negative range (that is, measurements during ascent). Surely a chamber-based experiment could provide such measurements, but of course gathering these on wet dives may not be possible.

In an ideal experiment I would like to gather data from dives on a whole bunch of deco plans, measuring the bubble grades during the ascent and after the dive. Then look at the bubble grades, and deem whichever one has the smallest high point to be the safest.

Speaking as a random bozo OC rec diver who wandered in here out of curiosity. I could be totally off-base here, so take this with a block of salt.

edit: re-attaching photos from rsingler's post on page 1
 
In regards to the issue of deep stops some posts mentioned bubble quantity (BQ) but no one brought up bubble size (BS). Consider these two scenarios:

Small BS and large BQ (VPMB)
Large BS and small BQ (ZHL16C)

Which scenario is better for dive safety in regards to decompression profiles? While oversimplified I do believe the goals these two scenarios show represent the algorithm associated with them. VPMB favors deep stops (smaller BS) at the expense of BQ while ZHL16C favors a lower BQ at the expense of larger BS. In order to test the validity of deep stops we need to test the two algorithms. We need to take out any deep stops in ZHL16C by using flat GF's (e.g. 75,75, 80/80, 85,85, etc.).

We also need to keep as many variables constant as possible. Design the profiles for say a 5% probability of getting bent. The largest variable is human physiology. All the divers should dive the profiles for both algorithms. Have the divers dive the same profile multiple times on the same algorithm then dive the same profile designed for the other algorithm. Count the number of non-bent dives for all divers of each algorithm. The algorithm with the highest number of dives is the safest. In addition each diver should maintain their lifestyle without changes. Get the same sleep, eat the same number of meals, drink the same amount of water. You get the picture.

We should not tinker with any of the dive parameters for any of the profiles. Here's why. For arguments sake let's choose two profiles: a VPMB and a ZHL16C, both of similar run times. DT = total deco time.

Running the dive plan on my Perdix with VPM-B +2 gave this profile for 170 ft, 30 min on air:
Deco stops (depth/time): 100/1, 90/2, 80/3, 70/4, 60/5, 50/5, 40/7, 30/14, 20/20, 10/34. DT = 95.

ZHL16C with GF's of 85/85 gave this profile for the same depth and BT:
Deco stops (depth/time): 50/1, 40/5, 30/8, 20/19, 10/38. DT = 106 min.

Now, let's shorten the Z profile to match the V profile of 95 minutes DT and run the tests. If more divers get bent on the V profile we can assume the Z profile is safer. Shortening the DT makes the case stronger for Z. What happens if more divers get bent on Z? Houston, we have a problem. Can we conclude the V profile was safer or is Z more dangerous since we shortened the DT? How can we decide?

We have the same problem if we lengthen the V profile to match Z's DT. If more divers get bent on V then we can conclude that deep stops are not good because we added to the DT which should have made it safer. OTOH, if more divers get bent on Z what conclusion can we reach? Is Z more dangerous or did we make V safer by adding to the DT?

By monkeying with the parameters we have removed a variable but we also invalidated the algorithm should the results be inconclusive.
 
VPMB favors deep stops (smaller BS) at the expense of BQ while ZHL16C favors a lower BQ at the expense of larger BS.
Where are the data showing bubble size?

By monkeying with the parameters we have removed a variable but we also invalidated the algorithm should the results be inconclusive.
You have already monkeyed with the parameters by choosing 85/85. What would total deco time be on 90/90, 95/95 or 100/100?
 
Running the dive plan on my Perdix with VPM-B +2 gave this profile for 170 ft, 30 min on air:
Deco stops (depth/time): 100/1, 90/2, 80/3, 70/4, 60/5, 50/5, 40/7, 30/14, 20/20, 10/34. DT = 95.

ZHL16C with GF's of 85/85 gave this profile for the same depth and BT:
Deco stops (depth/time): 50/1, 40/5, 30/8, 20/19, 10/38. DT = 106 min.
How did you get those numbers for total deco time? When I add up the time you would spend at the deco stops, I get 1+2+3+4+5+5+7+14+20+34=95 for your VPM-B+2 plan, but only 1+5+8+19+38=71 for your ZHL16C 85/85 plan.
 
View attachment 583479 View attachment 583480


I read through the thread, much of this is in the weeds and/or over my head. But these 2 pictures are very interesting. If I understand this bar chart correctly, it seems the goal for safety would be to take whichever deco prescription yields the smallest maximum bubble grade, across all measurements.

No that's not the goal..

The goal is to not get bent. The bubbling is a reasonable proxy for decompression stress. Bubbling is a mediocre proxy for DCS. So this graph is 2 steps away from the objective. The presumption is that 30min bubble scores are equated with faster tissues which are more tolerate of over pressure gradients in theory. And the 90min bubbles scores are from slow tissues that are not as tolerate of overpressure gradients. But without any DCS we cant really say which one better addressed the ultimate goal.
 
Where are the data showing bubble size?

I don't believe there are any. The testing would be done on the assumptions of what the models are supposed to do.


You have already monkeyed with the parameters by choosing 85/85. What would total deco time be on 90/90, 95/95 or 100/100?

Parameters was perhaps a poor choice of words. I should have said the output of the algorithm. If we're going to test the veracity of the algorithms we need to follow the output.
 
How did you get those numbers for total deco time? When I add up the time you would spend at the deco stops, I get 1+2+3+4+5+5+7+14+20+34=95 for your VPM-B+2 plan, but only 1+5+8+19+38=71 for your ZHL16C 85/85 plan.

Good catch. I used my spreadsheet for the ZHL profile and just posted the total run time. But, it doesn't matter for the points I was trying to make.
 
http://cavediveflorida.com/Rum_House.htm

Back
Top Bottom