I'm not sure how much this matters but I do believe that there is a genuine risk of blending problems leading to accidents so I'm still posting.
Just as a reminder, what I wrote upthread was:
[D]oubles do not "mix" through the manifold. If the two cylinders in a twinset have somehow ended up with different mixes, they will stay that way and will not "mix" through the manifold even if the isolator is left open and even if a substantial amount of time has gone by.
So, that's what we're talking about here. If you have a twinset, and there are different mixes in each cylinder in the twinset, they will stay that way and will not "mix" even if the isolator valve has left open and even if you wait days, weeks, or months.
3 - The heat exchanges will additional have some equalization.
I'm not sure what you mean, but unless one cylinder is heated more than the other, it won't result in any gas exchange.
But even if you paint one cylinder black and the other white and leave them out in the Texas sun, it's going to take weeks for any substantial blending to take place. The math on that is relatively easy based on the two temperatures, the fill pressure, and the number of cycles. I'll leave it as an exercise.
4 - Dalton's law will take care of the rest.
Dalton's law states that the total pressure is equal to the sum of the partial pressures of each gas. It has zip zap zero to do with diffusion.
5 - This was never a discussion about pure o2/he on one side vs the other or proper PP blending techniques. It still will mix due to Dalton's law, not only when being filled as I believe 2airishuman stated. Yes the higher the gradient, the longer it will take.
This discussion is about whether gases will mix through the manifold. They don't. It doesn't matter what gas blends are in each cylinder. Start with 32% and 36% and after a week you'll still have 32% and 36%.
Or you would pull from both sides equally getting 1/2 breath of 32 from one side and 1/2 breath of 24 from the other side and the mix in the lungs would avg 28%. so its still not a problem.
This is true in most cases but can't be relied upon.
I believe that some of the accidents have historically involved the isolator also being left shut, unintentionally, during the dive.
That aside, the whole point of having an isolator valve is that you can close it when circumstances require. Then you're breathing the unblended gas. (I think the question of whether an isolator valve provide a net improvement in safety, and the question of whether anyone is alive today who would not be but for the fact that they had an isolator valve to close, are both very interesting questions. Another thread, perhaps)
You will also breathe unblended gas if both posts are drawing gas at the same time, for example, during an air share.
There are some other corner cases, such as blockage of one dip tube, or one cylinder being warmer than the other at the start of the dive, so that it cools as the dive begins, which could lead to a period when unblended gas from the initially-cooler cylinder would be delivered to both posts.
@AlexL only problem with blending on the fly is you will be breathing against different gas densities as well as different flow paths. I.e. if you have helium in the right side, and oxygen in the left, equal pressures and breathe off of the right post, then off of the left, you will have two different mixes. You have a longer much more tortuous path through the whole manifold and have a very different gas density. So you have a lower proportion from the opposite side tank as you do from the tank on the breathing side, and the helium is going to have an easier time moving through due to lower density.
I'm not sure I agree with this line of reasoning. The pressure stays the same in both cylinders if the isolator is open. As long as the temperatures of the cylinders are the same and the isolator is open far enough that there isn't a pressure difference between the cylinders, they will both deliver equal volumes of gas.
Does this mean that I should always analyze both sides of my twinset, instead of just analyzing one side and assuming the other is the same?
I mean, how would I know if someone filled it wrong (so that I ended up with different blends on each side), analyzed it and realized it was wrong, but only "fixed" it by adjusting one side. So, if I happened to analyze the same side they "fixed", then I wouldn't realize the other side was wrong.
If you're going to be thorough you would have to close the isolator, analyze each cylinder, then open the isolator.
I confirm that the isolator is open when I drop the twinsets off for fills, and again confirm it is open when I pick them up. That is enough for me, because it guards against reasonable mistakes that might occur when a tech-savvy fill operator is filling a twinset.
Never had I suggested mixing gases in separate tanks and then opening manifold.
I'm not sure what you actually are suggesting, but you seem to disagree with my statement that: "doubles do not mix through the manifold."
Now I can only handle the gases as Ideal only.
The reason they do not mix can be mostly explained while still treating the gases as ideal gases although the possibility of collisions between molecules does make the problem worse.
The random interaction with the vessel wall, which has a surface area probably >3 sq ft. The drip tube has a surface area of less than 1/64 sq in or .0001 sq ft so for every 100,000 interactions a molecule has with the side wall, it will have 3.6 with the [dip] tube.
[...]
The slowness comes from the surface area difference and even then, how many go through the manifold or bounce back.
Right, the problem is compounded by the fact that the geometry of the path through dip tube, valve, crossbar, isolator, valve, and dip tube has a number of reflective surfaces that will tend to send most of the molecules back the way they came. The fact that this is hard to model is what makes the overall problem hard to approach from a theoretical standpoint. The closest reasonable approach would be to have some number of intermediate vessels (3? 5? 7?) connected by orifices to each other, and then calculate the diffusion for each pair.
Being in a Scuba tank does not change physics.
I think we agree on that.
Where I'm at is that, experimentally based on tests other people have performed, the gases don't mix -- not to the extent and on the timescale that matters practically for diving -- and the theory backs that up.