How to measure SO2, as by-product from the Claus process in refineries, cost-effective and most accurately (crowd sourced)

Question

In my search for air quality sensors there doesn't seem much out there that could be as accurate as governments apparently are monitoring.

See this example from Kanagawa prefecture of Japan with apparent ppb accuracy per hour response time:

http://www.pref.kanagawa.jp/sys/taikikanshi/StationDailyReport/OfficeTimeReport0120501.htm

These government monitoring sites can be limited and do not always represent the reality due to local landscape and wind differences.

Crowd sourcing environmental quality data turned out to be an accepted watchdog of governments already.

See the efforts of Safecast global radiation monitoring for example. Even IAEA is kind of supportive:

http://www.fccj.or.jp/number-1-shimbun/item/330-the-crowdsourcing-genie-escapes-the-bottle/330-the-crowdsourcing-genie-e…

For air quality, specifically SO2 ppb levels can be an important indicator near refineries for potential other pollutive side products like NO, NO2 or NMHC.

You can smell SO2 at relatively low levels and it also has health effect at relatively low levels for people suffering from asthma.

Only this commercial sensor seems to be getting close although the response time is still a bit too slow to get in par with the mentioned Kanagawa example:

http://www.alphasense.com/WEB1213/wp-content/uploads/2013/12/SO2A4.pdf

Anyone at element14 who can shed a light on more accurate alternative solutions?

shabaz · Answer

Hi Fred,

I think you're looking at Figure 4 in the datasheet, but this isn't intended to show the response time, it is intended to show noise at different levels (and they tested at levels of 200, 160, 120, 80ppb and so on).

They are showing noise in the time domain.

The actual response time is the 't90' value on the first page (under Perfomance heading), which shows that from zero to 2ppm (i.e. 2000ppb) the time is less than 20 seconds to reach 90% of the value.

The note in figure 4 indicates that noise is below 15ppb, with the opportunity to reduce that further with averaging algorithms.

Accuracy is not indicated by the way. The datasheet figure 4 suggests that 20ppb is definitely detectable, but accuracy is not given as far as I can see (by the way, I've not used this sensor).

If one needs accuracy from a single sensor to be comparable to government stations, then calibrated devices are presumably needed (I am no expert - but I suspect calibration from a lab is the way to achieve this, which is why the ones used in government stations are expensive).

Indeed, if your aim is to hold governments accountable to specific levels, then they would only accept measurements from calibrated devices, so you have no choice anyway but to pay for expensive calibration regardless of the sensor.

However, in theory there is another way (and you suggest it anyway in your post when you refer to crowd sourcing) - IoT can be empowering, because if you have measurements from multiple users, i.e. multiple sensors, then it becomes far harder to ignore trends if (say) 95% of those sensors in the IoT are suggesting levels are rising (and 5% have failed, for example). Basically stats will help. (I am no math expert in this area though). But I think a single accurate but non-calibrated sensor won't be acceptable, so cheap and multiple sensors are a way forward.

The very innovative 'In the Air' challenge was predominantly established to help develop more and more of these types of IoT solutions by the way, by empowering people to develop and use long-term outdoor wireless sensor nodes and data collection and analysis in the cloud.

There is a lot of information there on battery charging, low power processing, wireless comms, sensors and cloud platforms.

GJZ · Answer

My interest is how this is calibrated and how good the 10bbp is. I had a promotion study for Bsc a long time ago at the measurement labs for the Agriculture University in Wageningen. To be able to test such device you have two options since there is probably no good test laboratory that can help you for a reasonable price. And the companies that have this but pollute are not going to help you. Either get access to a local test station and their data to bench mark your design and their "calibrated" device. Which does not tell you anything how realistic their measurements are. Or get access to something that we have worked on in Wageningen where so called "geurpannels" would try to detect a smell with very small increases of substances in air. When you combine that data with threshold levels then you have something. My nephew happen to be studying in Wageningen but I cannot discuss that here online.

How to measure SO2, as by-product from the Claus process in refineries, cost-effective and most accurately (crowd sourced)

Top Replies