I Interpolation of Graphed, Cutoff Sensor Values

[mishima]

Hi, recently I've been playing with a little CO2 sensor. It is capable of reading ppm (parts per million) of CO2 from near the sensor up to a max of 10,000 ppm. We have a small apparatus for holding various absorbent chemicals, such as lithium hydroxide, and are controlling the release of CO2 from a small tank with a valve. Readings are tabulated automatically from a small micro-controller every 2 seconds, and later this information if plotted in google sheets. Here is an example:

Unfortunately, the sensor's capabilities are beyond our current experimental setup. With even the smallest burst from our supply tank (roughly 60 mL of gas as standard temperature/pressure), the sensor quickly plateaus to its maximum before eventually settling down.again.

I was wondering if there was any mathematical technique able to interpolate/estimate the maximum value from the graphs we've been generating. The value does not need to be exact, even 20% accuracy would be fine for our purposes.

 

[Svein]

You have to do some assumptions. In this case, your setting curve looks like an exponential decay.

 

[mfb]

I don't think you can get a 20% accuracy just from that curve. You can fit something to the slopes, a linear fit to the rising slope and an exponential fit to the falling slope, and see where the curves intersect, but the combination of the two lines is probably not what actually happened in the transition region.

Can you repeat the measurement with the CO2 sensor in a larger gas volume to reduce the peak CO2 concentration? That plus the existing measurement should lead to a more accurate description.

A CO2 sensor with a larger range would be even better, of course.

 

[mishima]

Thanks!

The sensor is placed in the open just below (~5cm) an outlet from the system. Perhaps I could adjust this distance or just try a "Y" type connection with my hoses to drive some CO2 away before it gets there. I was unsure if it was safe to assume that fluid flow through a Y would divide evenly like electrical current, I don't know much about fluid dynamics but that's a separate question (I think so, as long as pressure is the same in both channels?).

Cost likely prohibits a sensor with a larger range, we were unable to find something better than 0-10,000 ppm. I guess a decent rough estimate from these graphs could guide future purchases. We had considered bubbling the outlet CO2 through water and using titration on the resulting carbonic acid...

We were just trying to compare the absorbent effects, and I suppose we could just look at the time the sensor was maxed out as an indicator instead of a peak value.

 

[mfb]

If the sensor is in the open, how can you be sure it will give reliable measurements? Every motion of air in the lab will influence it.

If everything is perfectly symmetric, flow through a Y will be symmetric, but it is impossible to make everything symmetric.
Adding a different gas (e.g. just air) with a controlled known flow ratio would reduce the CO2 content to measurable levels.

We were just trying to compare the absorbent effects, and I suppose we could just look at the time the sensor was maxed out as an indicator instead of a peak value.

That can give a simple comparison ("more", "less") if the timescale of the reactions is the same, but it does not give ratios ("10% more"), as total CO2 won't be proportional to that time.

 

[mishima]

Sorry I misspoke about it being in the open, that was a poor wording choice on my part. The sensor is inside of a 1 liter chemical rinse bottle that has been cut in half. It can not be completely enclosed because of pressure buildup. Currently the gas flows down from above onto the sensor perpendicularly...like a bowl turned upside down on a table (sensor is the table, CO2 outlet is small hole at peak of dome. There is space on the sides because the dome is circular and the sensor is square.

It was oriented vertically to assist with clearing the dome between trials. It only takes about a minute for the sensor reading to return to a (truly) open room reading of around 1500 ppm after a short 60 mL burst from the supply.

It seems as you said increasing the volume of this dome would be a great idea.

 

[mfb]

It would mix the CO2 with more air, reducing the CO2 levels in it.