"Ellie." "Ellie who?" "The Elephant in the living room."
The elephant in our room is NOx, and elephants eat a lot. Then they poop. A lot. We need to watch where we step.
As technicians following the BPI Standards, we test for CO in the flue as well as in the ambient air as part of any audit where there are combustion appliances. Imagine my surprise when my new Testo 327 combustion analyzer gave me flue gas content results that seemed impossibly low . . . in some cases zero. This was something I very rarely – if ever – saw with my trusty Bacharach equipment in the past.
Many of us test our natural gas leak detectors by exhaling out of our mouths directly onto the sensor. The detector should go off, but that doesn't mean that we produce natural gas (at least not by breathing out!). The sensors in our gas leak detectors react to combustible gas or hydrocarbons generally, but they are also reactive to other gases, including VOCs the humidity in our breath and VOCs. VOCs are the reason why pipe dope sometimes sets them off. Still, the hydrocarbon most likely to be present at a joint in a gas line is the one we’re looking for—methane—and there is no good reason to use more sophisticated (and therefore more expensive) gas leak detectors to simple find a leak unless you need to quantify it.
Accuracy is more important when we’re testing flue gas. But many of the CO detectors and combustion analyzers that we use are similarly indiscriminate: They react to Nitric Oxide (NO gas, often called NOx) and other cross-interfering gases, and they include some proportion of the reaction in the displayed CO concentration. The issue is this: Where there is smoke there is fire, and where there is fire there is NOx. This seems like an important consideration to me.
We don’t like NOx because it’s toxic and it’s a greenhouse gas (as is its sulfurous sibling, SOx, another cross-interferrant gas). What’s more, NO converts to NO2 (another toxic gas) shortly after leaving the stack. If the test instrument does not discriminate effectively between CO and NOx, the number we read for CO in the flue is inaccurate by a factor that can be significant.
While this issue came up twice in the past on the RESNET-BPI group on LinkedIn, it did not generate much buzz. I believe that Auditors and Raters should be aware that their measurements might not accurately represent the level of CO in the flue gas being sampled. It is unlikely that the original BPI standards take into account this complicating factor.
The issue is the CO reading in the flue only. NO converts to NO2 about 35 minutes after leaving the stack. NO2 does not interfere with a CO sensor. And its presence is not an issue when sampling ambient CO.
On the good news side, BPI is gathering information on the issue. On the bad news side, BPI (like any standard-setting organization) moves at a pace that would embarrass your average tectonic plate. Since we routinely evaluate the safety of combustion appliances based on action levels for CO in flue gases, and since different people are measuring them with different degrees of accuracy, it only makes sense to get the matter of CO testing more into the public forum.
Many thanks to Bill Spohn of Trutech Tools for making sure that this old English major has the chemistry right.