Improving Rebreather Safety: Part II

Last month, I highlighted a number of issues discussed in May at the Rebreather Forum 3 (RF3), sponsored by PADI, Divers Alert Network and the American Academy of Underwater Scientists.

When talking about diver error, it's common knowledge that most sport divers avoid rebreathers because they seem too technical and complex, and it is indeed the technical side that causes problems. Bruce Partridge, CEO of Shearwater Electronics, summed it up this way, "Divers must interpret the readouts from three roaming oxygen sensors, which are known to be unreliable. They dive with no carbon dioxide gauge, and they don't have good data on the risks, or what is most likely to go wrong."

Most experts agree that current oxygen-sensing systems are the weakest, yet most critical, links on a rebreather. If the PO2 (partial pressure of oxygen) in the loop is too low, the diver will suffer hypoxia, go unconscious and drown; too high, and the diver risks hyperoxia, convulsions and drowning.

But what most divers might not appreciate are the limitations of oxygen-sensing systems, which were made clear at RF3 by Arne Sieber, CEO of Seabear Diving Technology. Sieber explained that the galvanic oxygen sensors, made especially for the biomedical industry, were never designed to be used in diving. In fact, the sensors are meant to be calibrated under the same conditions that they will be used, but that's not how it's done in diving.

"Divers do all the wrong things," Sieber explained. "We calibrate the sensors at 0.2 bar (air) and 1.0 bar (oxygen) at ambient pressure and temperature, then use the sensors at up to 1.6 bar at much hotter temperatures." This leads to increased sensor errors, as well as a decreased lifespan.

Sensors can err because of the gradual consumption of their reactive material and aging, and thus fall out of calibration. Worse is that "transient failures" from a loose electrical connection or condensation cause the sensor to generate erroneous data, and then go back to working correctly when the condition abates. Nigel Jones, principal at RMB Consulting, believes that these "transient failures" are likely behind many unexplained rebreather diver fatalities.

Because of known unreliability of these sensors, the first closed-circuit rebreathers had three oxygen sensors and a voting logic algorithm -- the computer averages the readings from the two sensors whose readings are closest, and uses that average for its oxygen calculations. The idea was that the redundancy of three voting sensors would greatly reduce the risk of sensor failure, and the concept stuck. Today, virtually all rebreathers except the Poseidon use this 50-year-old sensing technology. The problem, says Jones, is that it is simply not reliable. It assumes the failure of sensors is independent; the failure of one sensor does not change the likelihood that others will fail, too. Jones says that's not true; the sensors are dependent because they may have come from the same manufacturers' lot, they experience similar use, share a common environment, suffer common abuse, and use shared measurement and calibration gas. "Having three sensors is barely better than one in some circumstances." Also, risk reduction is eroded further because a diver doesn't know if the reading is correct or incorrect.

During a routine dive, Rich Pyle, database coordinator for natural sciences at the Bishop Museum in Honolulu, found his oxygen sensors read .4, 1.0 and 1.3. At his RF3 presentation, he asked the audience to make the call, "what is the correct PO2 reading?" (The computer's logic would average the 1.0 and 1.3 reading, and call it 1.15). Unfortunately, the majority of the audience got it wrong. The correct answer was 0.4; the system had experienced a double sensor failure. Fortunately, Pyle got it right. If he had ascended at that point in the dive thinking his PO2 was 1.15, he would have risked hypoxia and possible drowning.

Leon Scamahorn, CEO of Inner Space Systems, which makes the Megalodon rebreather, pointed out that "Meg" users could go the "millivolt screen" on their handsets, which actually shows sensor voltage (a linear function of PO2), and with some simple math, determine that the low sensor was correct. This assumes, of course, that the diver was alerted to the problem in time. But I couldn't help wondering if I'd have the presence of mind to do "millivolt math" at 300 feet with the stress of a possible alarm and knowing one or more of my sensors were crapping out. Couldn't a computer do this better than I?

Both Sieber and Jones urged the industry to develop and adopt "active validation" type systems, such as that used in the Poseidon MKVI, which calibrates and test the validity of the oxygen sensors (it uses two) throughout the dive using onboard diluent and oxygen. Sieber added that solid-state sensors, which are currently in prototype form, also hold promise for the future.

However, several rebreather builders I spoke with disagree with Siebers' and Jones' assessment, and said that they overstated the oxygen sensing problem, given improvements in sensor manufacturing, testing, and voting logic software. Nevertheless, in consensus, RF3 strongly endorsed industry initiatives to improve oxygen measurement technologies.

PCO2: The Dark Matter of Rebreather Diving

PCO2 is the term used for the partial pressure of carbon dioxide, and is a measure of how much carbon dioxide is dissolved in the blood. A high PCO2 level (0.03 bar and above) can cause hyperventilation, confusion, mental impairment, unconsciousness and death. It may lower oxygen toxicity thresholds in the central nervous system, and it's believed to be a factor in unexplained rebreather fatalities, hence the moniker "the dark matter." Worse, the diver may not be aware of the problem before a full onset of symptoms occurs.

Divers have two information needs. The first is to monitor the duration of the scrubber canister, which varies with workload, depth and temperature. The second is to detect a carbon dioxide breakthrough because of a spent canister, mechanical failure or channeling. Kevin Gurr, one of the gurus of carbon dioxide sensing, shared data from a recent Internet survey of 323 rebreather divers using 25 different models of rebreathers. The results were surprising. Twenty-three percent of the divers did not know the maximum operating depth of their units, and another 19 percent did not know the manufacturer's stated scrubber duration. Forty-two percent of divers said they experienced symptoms of hypercapnia (elevated PCO2 levels); however, 64 percent of those said they didn't bail out, while 19 percent said they bailed out sometimes. The results suggest better training -- and a cultural shift -- are needed.

Gurr recounted the methods used to monitor scrubber duration, and noted that while thermal sensing, also referred to as the "Temp Stik" (it's used in Ambient Pressure, VR Technology and rEvo rebreathers), is a reasonable predictor of duration, it is slow to react to fast-changing variables like work rate. However, none of the methods is able to detect carbon dioxide breakthrough, e.g., a catastrophic scrubber failure due to spent carbon dioxide-absorbent material, a seal failure or channeling (the creation of a channel through the scrubber bed, which enables exhaled gas to bypass the absorbent.

Dan Warkander from the Navy Experimental Diving Unit reminded us of the days of early scuba, when divers didn't have a pressure gauge, so instead dived with a J-valve. "Wouldn't it be nice to have a gauge for your scrubber to tell you how much time you had left?" Warkander explained that scrubber duration can vary by a factor between 5 and 20, through combined effects of workload, temperature and depth. What's worse, when a scrubber is spent, the threshold between no CO2 and too much can happen in a matter of minutes.

As far as detecting scrubber break-through or a seal failure, VR Technology's Sentinel is currently the only production unit with a gaseous infrared CO2 sensor. Gurr said that we are 80 percent there in fully characterizing a proper CO2 absorption system. The last piece is a mouthpiece sensor that can measure end-tidal CO2, regarded as the "Holy Grail" of CO2 monitoring. Gurr estimated it is still at least three years away.

RF3 acknowledged the poor understanding of operational limits with regard to depth and scrubber duration among trained rebreather divers. Panelists recommended that training agencies do more to emphasize these issues, and manufacturers make data more readily available.

Dive-By-Wire?

Poseidon's latest lovechild, the Poseidon TECH rebreather, is scheduled to ship this November, and features the latest in diving automation. "Our goal is to increase the level of automation by using smart systems that monitor every breath, make adjustments accordingly and interact with the user only when they need to know what's going on," said Poseidon CEO Peter Swartling.

In addition to the many automated features in Poseidon's MKVI recreational rebreather, (such as a wet switch, an auto-checklist that verifies cylinders have the correct gases and that their values are open, and auto-oxygen sensor calibration and validation), the new TECH offers a "Dive-by-Wire" handset that is truly breaking new ground. Smaller than an iPhone, the device provides system information to the diver, letting him control the rebreather to the extent of doing a loop flush or adding oxygen at the touch of a virtual button. The computer, of course, would warn or prevent the diver from taking an action, like adding oxygen if it was ill advised.

This level of automation gave the heebeegeebees to many tech divers I spoke with, but I can't help wondering if this is indeed the future of dive automation. Granted, roughly 15 percent of rebreather divers prefer a strictly manual unit without an electronic solenoid switch for adding oxygen, and other groups, like the "Doing It Right" community, don't even trust dive computers -- not the kind you strap onto your arm, anyway. Ironically, I'm sure most of these people have no trouble trusting their cars' anti-lock braking systems versus feathering the brakes on their own. In fact, their vehicles depend on computer automation, as do the commercial aircraft that flew them to RF3.

Bill Stone, CEO of Stone Aerospace, which builds autonomous vehicles for space exploration (and is working with Poseidon on its new TECH rebreather), addressed the issue by posing the question, "Can we trust automation?" He recounted the development of the autonomous car that can navigate city streets sans driver, and showed video of prototypes in action. Stone said that within five years, you'll be able to buy a car that will drive you home if you had too much to drink, and it will do it as safe as, or safer, than a human driver. Could rebreathers be far behind?

One of the major problems in rebreather (or car, train, plane, spacecraft, etc.) safety is humans' ability -- or rather, inability -- to manage and operate complex machines without incident. Stone's solution, similar to that of VR Technology's soon-to-be released Hollis Explorer, is to simplify the human machine interface by reducing the ways people interact with these systems, and letting the computer do more of the work. "We have to move out of the test pilot era to a new paradigm," he said. Given that Stone's vision 25 years ago helped drive the creation of a consumer rebreather market (he could arguably be considered the godfather of modern rebreathers), his ideas should not lightly be dismissed.

The Final Question

At RF3's closing session, Andrew Fock, head of hyperbaric medicine at the Albert Hospital in Melbourne, Australia, walked up to the mike and put the following question to the community, "Given that fatality rates are five to 10 times that of open-circuit scuba, should we morally offer this technology to the recreational diving community before putting our house in order?"

There was silence as if no one wanted to tackle the question, then another participant took the stand and changed the topic. Eventually, Mark Caney, PADI's vice president of rebreather technologies, worked his way to the mike and addressed his comments to Fock.

"Yes, we should," he said. "Within certain parameters."

Michael Menduno, based in Berkeley, CA, published and edited the monthly magazine aquaCorps: The Journal for Technical Diving (1990-1996), which helped usher technical diving into the mainstream of sport diving. He also organized the first Tek, EuroTek and AsiaTek conferences, as well as Rebreather Forums 1.0 and 2.0.