Buhlmann questions

[BlueTrin]

I see what you mean about the strong opinions. I did not know it was the case for deco models and computer as well.

I will add a graph or the option to display the ceilings for all tissues, this will make it easier to either debug or see what's happening.

 

[wedivebc]

perhaps would be more accurate to say up to 16 theoretical tissue compartments, there are 12 and 8 compartment Buhlmann models as well, although the increase in computing power over the years has made those less necessary.

I was referring to Dr Buhlmann's work found in decompression-decompression sickness published in 1983, not the revised adaptations that computer manufacturers use due to lack of computing power.

 

[tursiops]

SB posters worship on the alter of Shearwater.

Bad pun.

 

[greeniguana]

I was referring to Dr Buhlmann's work found in decompression-decompression sickness published in 1983, not the revised adaptations that computer manufacturers use due to lack of computing power.

And for nostalgia sake, this is from my 1992 Beuchat Alladin manual.

 

[KenGordon]

I was referring to Dr Buhlmann's work found in decompression-decompression sickness published in 1983, not the revised adaptations that computer manufacturers use due to lack of computing power.

There is no lack of computing power for this.

it needs of the order of hundreds to thousands of multiplies and adds per second. When I was having my head examined at a shearwater service centre recently I noticed they were using an STM32F series processor. This is a 32 bit arm that runs in the 10s to 100Mhz, it can literally do these calculations in its sleep.

the OSTC whose code is open uses a particularly nasty 8 bit processor and can manage too.

 

[wedivebc]

There is no lack of computing power for this.

it needs of the order of hundreds to thousands of multiplies and adds per second. When I was having my head examined at a shearwater service centre recently I noticed they were using an STM32F series processor. This is a 32 bit arm that runs in the 10s to 100Mhz, it can literally do these calculations in its sleep.

the OSTC whose code is open uses a particularly nasty 8 bit processor and can manage too.

Yeah, I'm not really sure how that relates to my post.

 

[KenGordon]

Yeah, I'm not really sure how that relates to my post.

You were claiming that the choice of Buhlmann version and particularly the number of compartments had to do with computer performance. I am disputing that. It is an extremely simple set of calculations.

 

[dmaziuk]

perhaps would be more accurate to say up to 16 theoretical tissue compartments, there are 12 and 8 compartment Buhlmann models as well, although the increase in computing power over the years has made those less necessary.

It would be more accurate to say that there is a 16-compartment Buhlmann model with 12 sets of a&b coefficients, known as ZH-L12, and it has nothing to do with computing power.

(Never seen a "ZH-L8", know nothing about it.)

 

[rjack321]

You were claiming that the choice of Buhlmann version and particularly the number of compartments had to do with computer performance. I am disputing that. It is an extremely simple set of calculations.

Today sure
There are plenty of older model computers that used 8 compartment for simplicity. most weren't really designed as deco computers

 

[EFX]

I literally have to re-read my own code to answer that. I think that GF hi applies at the surface and am certain that GF low applies at the first stop.

Yes, you are right. The application of GFHi is straightforward. It is simply the percentage of the surfacing m-value (designated M0) of a particular tissue compartment. Baker publishes the M0 values for nitrogen and helium in his paper "Understanding M-values". Baker, in his code, includes a formula for calculating the ceiling for GFLo. This ceiling becomes the first deco stop. What Baker does and what I do in my spreadsheet is to calculate a new GF for each stop so that there is an equal increment of GF's being applied from the first stop all the way to the surface. For example, consider a dive profile for a GF of 50/80. Let's assume that the GFLo of 50 produced a first stop at 30 ft. If the algorithm is based on stops at equal increments of 10 ft then we have stops at 30, 20, 10, and the surface with respective GF's of 50, 60, 70, and 80. (If 20 ft is the last stop then the GF's are 50, 65, 80 for stops of 30, 20, and the surface.) There is also a formula for calculating the GF increments. If you really want a challenge try figuring out a formula for reverse GF's.

Be careful, by the time you are done you will join the group of people that think GF is a horrid hack. This seems much more common amount those that code it up than those that claim it is some kind of open source standard.

I think the GF solution is an elegant "hack" in that it provides a smooth adjustment of the entire offgassing gradient from first stop all the way to the surface. There are others ways, less elegant, that provide conservatism such as adding depth or time at various parts of the dive. It's also fairly easy to understand the application of GF's when looking at plots of depth vs. TC pressure. The GF's as percentages are easily applied to the existing Buhlmann formulas and dive algorithms. The hard part is figuring out how they apply to individual physiology for any given dive profile.