Correlating Caustic “Air” Measurements with TPO – Theory vs. Practicality

This is the second of a two-part post on caustic “air” measurements.

In my last post I showed how it is possible to have two similar “air” readings in packages, even though the Total Package Oxygen (TPO) was wildly different. It this post we will examine the theoretical basis for the differences and dive a bit more deeply into why the correlation between “air” and TPO is hit-and-miss.

A couple of years ago I worked with a soft drink manufacturer. They wanted to see if they could correlate the dissolved oxygen content of their packages with “air” testing. They collected hundreds of individual data points, taking two different packages off a canning line at a time and running an “air” test on one while measuring dissolved oxygen concentration on the other.  Their results were confusing, so I thought that maybe by plotting the pairs of data – with the “air” results on one axis and the dO2 value on the other — things might make some sense.  What shaped up instead was a fan-shaped blob.

At first I wanted to blame measurement technique or instrumentation malfunction. But the more I tried to make sense of the data, the more I decided there had to be something real I just didn’t understand. It then dawned on me that what the data represented was a combination of “air” trapped in the headspace and “air” coming from the liquid. My next move was to plot how the data would look if 100% of the “air” were originating in the headspace, versus 100% of the “air” originating in the liquid.

The calculations for the headspace were simple. We know that “headspace air” is about 20% O2 and 80% N2. The liquid calculations required a bit more consideration, because if 100% of “air” originates in liquid, then solubility laws will result in air that has about 37% O2 and 63% N2.

Once I had calculated the numbers, I arrived at the following conclusions. At 20 oC, every mL of “headspace air” will contribute about 0.78 ppm (780 ppb) of TPO to a package. At the same temperature, every mL of “dissolved air” will contribute about 1.57 ppm (1570 ppb) of TPO to a package. (Note: although we were measuring “headspace air” and “dissolved oxygen,” at this point I calculated out to TPO because doing that always gives a more accurate picture.) I then plotted this theoretical difference to see how the two lines would fall. Here is the graph:

The area of the graph between the two lines shows all of the theoretical possibilities for valid TPO concentrations based on a given air concentration.  For instance, if the “air” volume is 1.0 mL, the TPO will range between 0.78 ppm and 1.57 ppm.

I wish the reality of these theoretical values could be easily applied to actual package situations, but in my experience the “air” tester rarely picks up all the gases, so it almost always underestimates TPO. The difference seems to depend whether the majority of the “air” is trapped in headspace or is dissolved in the liquid.

My final thought is that “air” measurements can be helpful in certain situations and are always better than not measuring at all, but to seriously zero-in on problems, dissolved O2 measurements (calculated to TPO) are the way to go.

Frivolous Friday Fun – The History of the Pull Tab

If you remember the days of pull-tab necklaces, then you might like this follow-up to last week’s link about a revival of the flat topped can: It’s a link to the history of the pull-tab can end. Thanks to my colleague JP for this idea and last Friday’s fun as well!

Measuring Package “Air” – What Can It Tell You?


This is the first of a two-part post on caustic “air” measurements. We’ll start with a brief historical glimpse at air testing and then talk about both the upside and the limitations of this type of testing.

In 1989 I worked with a team that wanted to quantify the amount of dissolved oxygen in beer. We purchased a Zahm-Nagel “air tester” to see how it compared to using a dissolved O2 analyzer on shaken packages, where the values we obtained were then used to calculate total package oxygen. (An air tester looks at all the gasses in a package that are not CO2.)

The idea was to simply try and correlate “air” to Total Package Oxygen (TPO.) Sometimes the air-to-TPO values were fairly predictable and repeatable, but other times the air readings differed greatly from our expectations. Because of this, and because we knew that TPO was giving us accurate results, we stuck with it and dropped further air testing, but we never really got to the root of the discrepancy.

When I did that air-to-TPO comparison, I used a method in a 1980s edition of the American Society of Brewing Chemists (ASBC) Methods of Analysis (MOA). As I did the research for this post I wanted to look back at that old method, but it had been archived by the ASBC and was no longer available. So I called the Zahm-Nagel Company (they still make air testers) and spoke to a very kind gentleman named Loren, who graciously found “Beer 15” in a 1949 copy of the ASBC MOA. The procedure was apparently adopted in 1946, based on the work of an ASBC subcommittee between 1944 and 1945.

What can an “air” measurement tell you? If you’re sure that excess O2 in a package is an air-in-the-headspace issue, then a simple air measurement may be all you need. But if the issue is dissolved oxygen, then it may not be your best bet. The next part of this story is a practical example of the difference between air and TPO measuring in a troubleshooting situation.

I was in a Latin American country a dozen years back, helping root out the source of occasional high TPO measurements.  The brewers were convinced that the undesirable results were the fault of the dissolved oxygen analyzer, but I couldn’t find anything wrong with the instrument. I proposed that we pull multiple bottles off single-filler valves and look at both the TPO and “air content” of the packages.  Since it was easy to do, we grabbed ten samples from each of three specific filler valves, and ten samples randomly from multiple filler valves.  Below is a table with our results.


Location # of Samples Total Package O2 “Air”


Random 10 83 ± 32 ppb 0.25 cc
Valve A 10 102 ± 106 ppb 0.25 cc
Valve B 10 799 ± 205 ppb 0.35 cc
Valve C 10 84 ± 26 ppb 0.25 cc

As you can see, the TPO data from one valve showed a distinct deviation from the other valves. The “air” reading from the bad valve also showed a slight increase compared to the other valves, but it was not as large as the TPO rise, and did not reflect the extreme discrepancy between valves.  By looking at a few shaken vs. unshaken measurements, we were able to confirm that a faulty seal on the filler head was failing to pull adequate vacuum on the bottles. This data was more confirmation that air results might not correlate very well with TPO, but I never really examined why until a few years later. We’ll talk about that in part two of this post.

For now, this is my final thought: caustic air measurement is a time-tested tool and there may be circumstances when it’s all you need, but if you are having flavor issues due oxidation then TPO will always be a stronger diagnostic tool.




%d bloggers like this: