Zennora was shrink-wrapped last week, just in time for the first New York City snows of the 2016-2017 winter season. It is now protected from the ice and wind and affords me a space to work on the boat regardless of the weather.

Then, late last Friday night, I awoke with an unexplained banging headache, especially unusual as I was entirely sober. But there was something nibbling at me. Could the exhaust from my diesel furnace be filling the enclosure, and so the boat, with deadly carbon monoxide (CO) fumes? Could that be why I had a headache?

Later that night same night, I woke again, still vaguely concerned about the heater. Maybe I had a dream or heard an an unexpected noise, but my reflexive brain told my hand grab a spare CO detector, put the battery in, and plant it near my bunk in the aft cabin.

I went back to sleep. An hour or so later, the new detector sounded. This unit does not have a display and the manual does not clearly explain what the alarm means. I am still not sure if I should have called 911 or just rolled over. The gas/CO detector in the galley did not sound

Anyway, I opened up the hatch, turned the heater down low and the detector was quiet for the rest of night.

On Saturday, it periodically squawked each time I turned the furnace up high. I was increasingly convinced the exhaust was leaking inside the boat or being trapped by the shrink-wrap enclosure. That night, I turned the furnace off for the night and relied on my electric blanket and an oil-filled electric radiator.

The alarm was quiet all night.

The next day, Sunday, I wanted to test if it made any difference how high or low I turned the furnace thermostat to see if that correlated with the squawking of the detector. It kept going off, generally in step with the level of the furnace, but not neatly.

But something was bothering me. The gas and CO detector in the galley had never sounded, even though the new detector was at times sitting on the navigation desk, three feet away from the galley. I grabbed my final spare CO detector and lay it next to the first spare and turned the furnace up again. And it sounded again, but not the galley detector and not the recently added and identical unit.

I moved the two identical units to the aft cabin and again the same unit sounded while the other two remained inert.

Still, something was happening and I resigned myself to twisting into the space behind the transom, where the furnace is installed, to physically check all the hoses and pipes. I had already visually inspected everything and could not see a problem.

I certainly couldn’t risk leaving the furnace running for the night.

Shortly before sunset, however, one of my neighbors lent me one of his CO detectors. Maybe my detector was defective?

His unit did not go off but it had a display and was reading 35 to 40 parts per million of CO on a shelf near my bunk. And it was ticking up.

OSHA says 50 PPM is the average CO threshold for an eight-hour work shift.

The National Fire Protection Association says headaches start at about 100 ppm; dizziness, nausea, fatigue kicks in at 200 ppm. Three hours at 400 ppm can be fatal.

I decided to shut the furnace down. But the CO readings did not go down.  I moved the unit outside and the reading went to zero. I moved it back into the aft cabin. Now it was 50 ppm.

The furnace was off. The stove was off. There was nothing intentionally combusting on boat. I took the detector to bed with me. Now the reading shot up: 70, 80, 90 ppm. The furnace was still off. Nothing made sense.

In my head, I clicked off all the systems that could be causing the release of any gas or was burning something: Me? Yes, I release gas, but was unlikely the cause. The propane stove, engine, furnace? All off. The holding tank? No way. Batteries? A bad battery can get hot and even explode.

Then I found an online firefighters group discussing how they often responded to CO alarms only to find a golf cart in a closed garage with a bad battery connected to a charger. But the batteries were not producing CO, they were venting hydrogen gas.

According to the discussion group, CO detectors can react to hydrogen gas, setting off alarms.

And one of my house battery banks, two 12V 4D batteries in parallel, is directly under my bunk. I quickly pulled off the sheets and folded the cushions back. And as I did so, the readings rose: 100, 110 ppm.

I lifted the battery cover: 130 ppm. I touched the side of the battery. It was hot. I opened one fill cap: 120, 130 ppm. The liquid inside was bubbling.

I removed all the fill caps: 140, 170, 190, 200, 210 ppm.

I immediately switched to the other house bank, turned off the battery charger, disconnected the battery from everything. The bubbling slowed. Within the hour, the detector read zero ppm.

I turned the furnace on — full blast. After another hour, the detector still read zero ppm. The battery had cooled and the bubbling stopped.

As best I can figure the furnace was drawing from the battery bank, which couldn’t efficiently hold the charge. According to the specifications, my Webasto furnace can draw between 15 to 95 watts (up to about 8 amps), depending on the temperature setting. The higher furnace setting, the more the draw, the more the charger pushed, the more hydrogen was released.  But even when the furnace was not running, the charger was still trying to charge and the battery, which continued to release hydrogen, hence the residual reading.

I made my bed and went to sleep. Seven hours later, when I woke, I was hot and the battery still cold. Outside the thermometer read 12 degrees.  I got up, put on the kettle and didn’t bother with a dressing gown or slippers. The cabin temperature was 65 degrees. The detector read zero ppm.

I was alive and once again reminded that causation and correlation are not the same thing. And problems are rarely binary switches.

-30-

Why CO detectors sometimes detect hydrogen gas levels: http://goodforgas.com/why-is-hydrogen-such-a-big-issue-for-co-sensors/

Postscript: The power went out Monday nigh and then there was the joy of having a heater that was independent of shore power.