O2 sensors & water/condensation: design differences? real field data?

[justinthedeeps]

While reviewing opinions on causes of rebreather accidents, I saw the suggestion of a dangerous state in which water on sensor membranes blocks the sensors at some arbitrary reading, which could lead to the deadly issues of hypoxia, or uncontrolled oxygen over-injection by a solenoid controller (hyperoxia, uncontrolled ascents)

I am curious to hear people's various insights and observations on this problem.

Have you experienced it?
Which rebreathers (or sensors) are vulnerable, vs. 'protected' from this issue?

I have only experienced this once, when I laid down my rebreather in a vehicle between dives. While doing my surface pre-breathe for the second dive on land, the solenoid went crazy. I promptly "bailed out" to the atmosphere, shut the oxygen valve, disassembled the rebreather, and carefully dried off the cells--soaking up a few drops of moisture using the edge of a clean paper towel. Less than an hour later I was able to complete a normal pre-breathe and complete the dive without incident.

The issue unmistakably was that clean water condensation had drained from the inhale tube back onto my sensors, blocking two of them.

I was glad it didn't happen while diving. In my case, excessive oxygen injection and 1 [good] cell reading oddly high might have prompted me to either bail out immediately and then close the oxygen valve, or vent continuously while closing the oxygen valve, consider a dil flush, and then most likely still bail out. The latter variant is arguably an option with more immediate buoyancy correction. Others free to argue there are additional nuanced options, plugging in offboard oxygen to a MAV port, feathering the O2 valve, etc, but that's not really the point here. Yes there are [sometimes] ways to stay on the loop, or get back onto it, if you really feel the need to do that.

While we are trained in the appropriate responses to low and high oxygen readings, it could be particularly deadly when the sensors give no indication of a real danger before it is too late.

The suggestion I read involved the possible blocking off all cells, with little to no indication of a problem before a life threatening gas mix is breathed.

 

[inquis]

A Fathom blows DIL on the face of cell #1, which tends to keep condensation away. Since condensation can make a cell slow to respond (if at all), I look for similar changes across cells with any addition (Dil or O2).

 

[inquis]

Regarding cell verification (the broader context impacted by condensation), I also verify expected PO2 after larger depth increases. After restoring optimal loop volume, a 9 ft depth increase should raise PO2 by about [0.06 0.05 0.04] after addition of [21 18 15]% diluents.

(The math is simply to increase PO2 by the O2 fraction for every 33 ft or 10 m. This assumes the impact of metabolism is insignificant during the change, which is reasonably true for mCCRs. For an eCCR, it works for "short" timespans, but how short will depend on your metabolism. Perhaps decrease the expected change slightly based on your observations.)

It's an easy addition to your normal routine of cell verification (e.g., a dil flush upon reaching the bottom or an O2 flush at 20 ft). While the above doesn't verify the absolute PO2, it will detect some degree of nonlinearity, definitely current limiting, and I expect the condensation issue.

 

[justinthedeeps]

Thank you for insights
[In addition to design-based discussion]

A small list of sensor signs, likely life-threatening:

1. Solenoid is injecting more than you reckon it should [because cells are blocked & can't meet setpoint value]
2. Solenoid is not injecting at all [the cells are stuck on a meaningless "good" value]
3. Loop volume is decreasing below minimum loop volume, even though the cells read "good" [O2 is being metabolized and not replaced]
4. Readings aren't changing much, or at all--especially with changes in depth
5. Cell(s) voted out for too low or high values--might be the only correct cell(s)
6. ...

Is the "good, but not changing" the most dangerous? Simply because it could so easily be missed.

 

[Brad_Horn]

While reviewing opinions on causes of rebreather accidents, I saw the suggestion of a dangerous state in which water on sensor membranes blocks the sensors at some arbitrary reading, which could lead to the deadly issues of hypoxia, or uncontrolled oxygen over-injection by a solenoid controller (hyperoxia, uncontrolled ascents)

##

While we are trained in the appropriate responses to low and high oxygen readings, it could be particularly deadly when the sensors give no indication of a real danger before it is too late.

The suggestion I read involved the possible blocking off all cells, with little to no indication of a problem before a life threatening gas mix is breathed.

As you've experienced, it's uber easy to recreate and experiment on any rebreather as to exactly how the unit and more importantly if fitted, its electronic oxygen controller, reacts to waterblocked cells; with in most cases by design either a hypoxic or hyperoxic end result.

Caveat emptor however underwater - as to the end result - as any possible mitigation is pre-determined solely based on the quality of the systems of systems engineering that has gone into the units design well before it was sold. Noting as per below report that causations are known to be orientation and whether or not the cell has a cup on the face of it (as specified by the particular rebreathers designers) to trap water/moisture.

The "erratic oxygen cell readings" per this report https://nsscds.org/wp-content/uploads/2024/05/Phantom-2023.10.04.pdf could well be and in all likelyhood, are an example of this exact design issue being openly reported by a diver whom survived. The manufacturer of the unit simply not yet having gotten around to conducting unmanned testing to prove the PPO2 monitoring/control design, is free of this basic cell face waterblocking issue, before it was test dived. Controversal but easily discussed if one knows the specific model # of galvanic cells fitted to the unit and if they are flat faced per below recommendations and are harder to waterblock or have a cup face that traps water orientation dependent.

As to industry acknowledgement of the risk of waterblocked cell faces and exactly how this can be engineered out at the design stage, about the only place you'll see it documented {specifically highlighted in yellow post Wes Skiles fatality} is https://www.opensafetyglobal.com/Safety_files/DV_O2_cell_study_E4_160415.pdf

 

[justinthedeeps]

While I used to flip over and look skyward often for fun on my rebreather without water blocking my cells and killing me yet, I cringe now every time I see someone doing DPV "backflips" on rebreathers

 

[justinthedeeps]

How about backmount vs. sidemount vs. chestmount rebreathers?

The chestmount rebreathers sit below our bodies. Does that create any greater risk for moisture or condensation buildup blocking the cells? Or is that risk designed out in some or all units?

 

[justinthedeeps]

As to industry acknowledgement of the risk of waterblocked cell faces and exactly how this can be engineered out at the design stage, about the only place you'll see it documented ... https://www.opensafetyglobal.com/Safety_files/DV_O2_cell_study_E4_160415.pdf

Just making my way through this quite nicely detailed document/report. Thanks for posting.

10.1.8 The Lethal Danger of Liquid being allowed to pool on O2 Cell Membranes. If water covers part of the O2 cell membrane, a dramatic increase in the response time can occur. If the water blocks the whole membrane, then the sensor output will “freeze” at whatever it was before the block occurred. The output of a blocked sensor will reduce only very slowly over time. For example, if a sensor has water dropped onto the membrane in air, then the output will take around 20 minutes to drop from 0.21 atm to 0.17 atm. ... ...

 

[justinthedeeps]

What does water block look like in the ppO₂ (mV) readings of a dive profile?

We could see the mV readings fixed for a suspiciously long time, however, we could also expect to see that anyway for a finely controlled solenoid or CMF. (Except in cases of depth change.)

It could be all cells, or only certain blocked cells.

Do we have examples?

Do we (or current rebreathers) have any software logic to identify anomalous input and operating states of the system, whether in real time or in retrospect? Might actually be 'easy' with modern data science? This could see patterns in data well beyond just basic mV values. Or did the Poseidons already do all of that.

Effects on oxygen injection (or lack thereof) might also be evident in data logged from "air-integrated" wireless oxygen pressure gauges. (I use one.)

I remember getting full mouthfuls of condensate/saliva mixtures at the end of three hour dives on a rEvo. Many rebreathers are a wet ride.

What are the chances that water block could have affected dives like this?
Is there mV data for all of the cells?

Rob Stewart and Third Dive

BTW, what happens when you strip a person and the corporation he owns of their assets, in order to preclude the assets being siezed as the result of loosing a lawsuit, especially since the person can't file again for personal bankruptcy for several years? Michael You mean like hiding your...

scubaboard.com