In Greek mythology irresistibly seductive female creatures were believed to use enchanted singing to beckon sailors to a watery grave.
Why this myth endured through the centuries is difficult to say. However, my theory is that it helped explain to grieving widows and mothers why ships sometimes inexplicably disappeared, taking their crew with them, never to be seen again. By the reasoning of the time, there must have been some sort of feminine magic involved.
The oxygen sensors in closed-circuit, electronically or computer-controlled rebreathers are a magic device of sorts. They enable a diver to stay underwater for hours, consuming the bare minimum of oxygen required. The only thing better than a rebreather using oxygen sensors would be gills. And in case you wondered, gills for humans are quite impractical, at least for the foreseeable future.
I have written, or helped write three diving accident reports where the final causal event in a rebreather accident chain proved to be faulty oxygen sensors. So for me, the Siren call of this almost magical sensor can, and has, lured divers to their seemingly blissful and quite unexpected death.
Those who use oxygen sensors know that if the sensor fails leading to a hypoxic (low oxygen) state, loss of consciousness comes without warning. If sensor failure results in a hyperoxic state (too high oxygen), seizures can occur, again leading to loss of consciousness, usually without warning. Unless a diver is using a full facemask, loss of consciousness for either reason quickly leads to drowning.
Due to the life-critical nature of oxygen control with sensors, three sensors are typically used, and various “voting” algorithms are used to determine if all the sensors are reliable, or not. Unfortunately, this voting approach is not fail-proof, and the presence of three sensors does not guarantee “triple” redundancy.
In one rebreather accident occurring during the dawn of computer-controlled rebreathers, a Navy developed rebreather cut off the oxygen supply to a diver at the Navy Experimental Diving Unit, and all rebreather alarms failed. The diver went into full cardiopulmonary arrest caused by hypoxia. Fortunately, the NEDU medical staff saved the diver’s life, aided in part by the fact that he was in only 15 feet of water, in a pool.
In two more recent accidents the rebreathers kept feeding oxygen to the diver without his knowledge. One case was fatal, and the other should have been but was not. Why it did not prove fatal can only be explained by the Grace of God.
The two cases were quite different. In one the diver broke a number of safety rules and began a dive with known defective equipment. He chose to assume that his oxygen sensors were in better shape than the rest of his rebreather. If he had been honest with himself, he would have realized they weren’t. If he had been honest with himself, he would still be alive.
The other dive was being run by an organization with a reputation for being extremely safety conscious. Nevertheless, errors of omission were made regarding oxygen sensors which almost cost the experienced diver his life.
In the well-documented Navy case, water from condensation formed over the oxygen sensors, causing them to malfunction. The water barrier shielded the sensors from oxygen in the breathing loop, and as the trapped oxygen on the sensor face was consumed electrochemically the sensor would indicate a declining oxygen level in the rig, regardless of what was actually happening. Depending on how the sensor voting logic operated, and the number of sensors failing, various bad things could happen.
During its accident investigation, when NEDU used a computer simulation to analyze the alarm and sensor logic, it found that if two of the three sensors were to be blocked (locked) by condensed water, the rig could lose oxygen control in either a hypoxic or hyperoxic condition. Based on a random (Monte Carlo) sensor failure simulation, low diver work loads were more often associated with hypoxia than higher work rates, even with one sensor working normally.
We deduce from this result that “triple redundancy” really isn’t.
When the accident rig was tested in the prone (swimming) position at shallow depth, after 2 to 3 hours sensors started locking out, and the rig began adding oxygen continuously. The computer simulation showed that the odds of an alarm being signaled to the diver was only 50%. The diver therefore could not count on being alerted to a sensor problem.
Unfortunately in this near fatal case the rig stopped adding oxygen, the diver became hypoxic and the diver received no alarms at all.
After NEDU’s investigation, the alarm logic was rewritten with a vast improvement in reliability. The orientation of the sensors was also changed to minimize problems with condensation.
Today what is being seen are divers who extend the use of their sensors beyond the recommended replacement date. Like batteries, oxygen sensors have a shelf-life, but they also have a life dependent on use. Heavily used sensors may well be expended long before their shelf-life has expired.
Presumably, the birthing pains of the relatively new underwater technology based on oxygen sensors have now passed. Nevertheless, those who use rebreathers should be intimately familiar with the many ways sensors, and their electronic circuitry, can lead divers ever so gently to their grave.
Like sailors of old, there are ways for divers to resist being lulled to their death by oxygen sensors. First among them is suspicion. When you expect to have a great day of diving, you should be suspicious that your rebreather may have different plans for you. Your responsibility to yourself, your dive buddies and your family is to make sure that the rebreather, like a Siren, does not succeed in ruining your day.
The best way to ward off sensor trouble is through education. To that end, Internet sites like the following are useful. Check with your rebreather manufacturer or instructor for additional reading material.