Purging Oxygen From a Keg: How Much Each Method Actually Removes

Oxygen has its place early on in the brewing process and good wort oxygenation can absolutely improve fermentation speed, yeast health and a number of other fermentation improvements. However, once fermentation is done, it turns into one of the fastest ways to mute freshness and destroy delicate beer character. Oxidation strips hop aroma, darkens the beer, shortens shelf life and brings in stale flavours commonly describe as wet cardboard, paper or sherry. 

It helps to know what the professionals are chasing so we can set ourselves a target. Commercial breweries care enough about this to measure it, working in total package oxygen, or TPO, which is the dissolved oxygen in the beer plus the oxygen left in the headspace. Best practice is a TPO under 50 ppb immediately after packaging(Reducing Total Package Oxygen, BBII). In the real world, plenty of packaged beer lands higher, with package dissolved oxygen often in the 30 to 250 ppb range and TPO anywhere from 50 to 450 ppb. So 50 ppb is the mark to aim at, not the average that gets hit. The point of purging kegs properly and thoroughly is to get the keg's own contribution of DO (dissolved oxygen) low enough that it isn't contributing to staling. 

The good news is that reducing oxygen in your kegs is straightforward, and you don't need ridiculously expensive DO meters like a Cbox to do it properly. Below we walk through four methods in order of effectiveness from least effective to best practice, we’ve also done some math to calculate how much DO each method contributes.

A quick note on the numbers

The figures below are a comparison tool based on physical constants, it gives some pretty good insight into each of the purging methods but real world numbers can vary based on the overall process.

Another limitation is that our calculations estimate the oxygen sitting inside the keg if every bit of it ended up in your beer. That's the absolute worst case and acts as a ceiling rather than the actual DO measurement in the beer, useful for ranking the methods against each other. They don't account for oxygen already dissolved in the beer, or picked up through splashing, leaky disconnects, unpurged transfer lines or bad dry hopping technique. It also doesn’t account for the oxygen displaced during the actual filling of the keg which would actually lower all of these DO numbers significantly. 

For the purge and vent figures we've assumed a keg pressurised to about 15 PSI and completely vented to atmosphere. Each cycle multiplies the leftover oxygen by roughly atmospheric pressure divided by total pressure, which works out to about 0.495 per cycle. If you want to see how I calculated these numbers Ive left my working in the bottom of this post. 

Method 1: Single purge (Bleed and Fill)

1. Seal the empty cleaned and sanitised keg.
2. Pressurise with CO₂ to 15psi.
3. Pull the pressure relief valve to vent the keg.

Results:

• ~138,000ppb estimated remaining oxygen
  
• Uses approx 35g of CO2 to purge

A lot of brewers do this once and call the keg oxygen-free and often quote that CO2 is heavier than air so it sits in the bottom of the keg. Unfortunately gases don't cooperate like that unless the differences in densities are orders of magnitude larger. The incoming CO₂ mixes with the air already in the keg, and when you vent you're letting out a blend, not just the oxygen. It clears out roughly half the oxygen, which sounds good on paper until you look back at our target of 50ppb.

Method 2: Multiple purges (Bleed and fill cycling)

1. Seal the empty cleaned and sanitised keg.
2. Pressurise with CO₂ to 15psi.
3. Pull the pressure relief valve to vent the keg.
4. Pressurise with CO2 to 15 psi.
5. Pull the pressure relief valve to vent the keg.
6. Repeat

Results:

5 purge cycles 

• ~8,260 ppb estimated remaining oxygen
• uses approximately 175g of CO2

10 purge cycles

• ~245 ppb estimated remaining oxygen
• uses approximate 355g of CO2  
  

This follows the same idea as the first method but repeated. Each cycle dilutes what's left, so you're removing oxygen from an already thinned-out mix. The reduction compounds but each purge uses another ~35g of CO2 meaning by the time were at 10 cycles and a relatively low oxygen concentration we’ve used ~355g of CO2 which isn’t very cost effective.

Method 3: Liquid displacement (using sanitiser)

Instead of diluting the air with CO2, you fill the keg completely with liquid and then push that liquid out with CO₂. Because the keg is physically full, there's no room for air to hide and the only mixing taking place is in the tiny headspace.

Use a no-rinse sanitiser solution as your displacement liquid rather than plain water. You'll sanitise the keg and purge it in the same step, which kills two birds with one stone. There's no reason to displace with plain water when sanitiser does the same gas-removal job and leaves you with a ready-to-fill keg. Just ensure you’re not using an oxidiser as your sanitiser like hydrogen peroxide or sodium percarbonate or you’ll be fighting an up hill battle. We would use something like StarSan, StellarSan or Iodine. This method is also very efficient from a CO2 usage point of view since we can use a low pressure of CO2 to displace the liquid and its only one keg full of gas at low pressure not multiple cycles of it.

1. Completely fill the keg no-rinse sanitiser solution.
2. Connect CO₂ to the gas post.
3. Bleed the remaining headspace by pulling the PRV, sanitiser will usually spray from the PRV at this point. We would bleed 5 or 10 times here, because the volume is so small it uses negligible CO2 for the bleed.
4. Push the sanitiser out through the liquid post and through the same transfer hose you'll use for the closed transfer later so the transfer hose is also being purged.

As the liquid leaves, CO₂ takes its place, so you finish with a sanitised keg that's essentially full of carbon dioxide.

Results:

• ~50 ppb estimated oxygen remaining (from oxygen dissolved in the sanitiser solution left over in the keg)
• Uses approx 35g of CO2 if displacing with 15psi of CO2 (less if you’re using lower pressure)
  

One catch worth knowing with this method. Any solution made with tap water isn't oxygen-free. Water holds around 9,000 ppb of dissolved oxygen at room temperature, and a film of that liquid clings to the internal surface as it drains aswell as the tiny bit at the bottom of the keg. So while the gas space ends up very clean, the displacement liquid itself still brings a small oxygen load. 

Method 4: Liquid displacement plus metabisulphite

This is displacement with the dissolved-oxygen problem solved. You displace your sanitiser with some metabisulphite added to it, which doesn't just push the air out, it also reacts with the oxygen in the liquid.

Here's what's happening chemically. Potassium and sodium metabisulphite dissolve in water and hydrolyse to bisulphite ions:

 S₂O₅²⁻ + H₂O → 2 HSO₃⁻

Bisulphite is the main active scavenger. When it meets dissolved oxygen, the sulphur is oxidised from sulphite to sulphate, consuming the oxygen in the process:

2 HSO₃⁻ + O₂ → 2 SO₄²⁻ + 2 H⁺

That reaction is why metabisulphite mops up the dissolved oxygen that plain water or sanitiser would otherwise carry, plus any trace left clinging to the keg walls.

There's one more reason this is our preferred purging method. The CO₂ coming out of your bottle is never perfectly pure. CO₂ itself can carry trace oxygen, which means the very gas you're purging with can introduce oxygen. That sets a floor on how low purging alone can ever get you, because you can't dilute below the oxygen content of your own gas supply. If you’re worried about sulphite ending top in your beer don’t stress, almost all of this sulphite is being pushed out during the purge so the tiny amount left over will be diluted with your full keg of beer. Theres even an argument for better shelf life of the beer from a tiny introduction of sulphite, lager yeast naturally produces sulphur during fermentation that can aid as an antioxidant.

1. Clean the keg.
2. Fill it completely with sanitiser 
3. Add 5g metabisulphite.
4. Bleed the remaining headspace by pulling the PRV, sanitiser will usually spray from the PRV at this point. We would bleed 5 or 10 times here, because the volume is so small it uses negligible CO2 for the bleed.
5. Push out the sanitiser and metabisulphite with CO₂.
6. Fill with keg with beer via a closed transfer.

Results:

• ~5ppb estimated remaining oxygen
• Uses approx 35g of CO2 if displacing with 15psi of CO2 (less if you’re using lower pressure)

The results side by side

That CO₂ column is worth looking at, because it really shows the efficiency of liquid displacement. Chasing low oxygen with purge cycles alone burns through your gas, roughly ten times what a liquid displacement uses, and still only gets you to ~245 ppb. Liquid displacement uses about the same gas as a single purge, around one keg volume, but lands you in safe dissolved oxygen territory. So displacement isn't just the better result, it's also the more efficient use of gas. If you find yourself doing eight or ten purge cycles on every keg, displacement will usually save you CO₂ and lend itself to lower DO.

Our recommendation:

Given how efficient and easy it is to perform a liquid displacement we don’t really see any reason not to. After all the hard work you put into brewing a delicious beer why would you stumble at the last hurdle. It might take an extra 5 minutes to do it properly but given the benefits to shelf life and freshness we really see it as a no brainer. If you are allergic to sulphites stick with the liquid displacement with sanitiser, if you’re not we’d use liquid displacement plus metabisulphite.

Got questions about closed transfers, displacement setups or which fittings you need? Come have a chat in store or drop us a message. We're always happy to help.

The calculations used:

The numbers behind every figure in this piece come from basic gas law and some assumptions and nothing beyond high-school chemistry.

Assumptions:

• A 19 L corny keg, completely full of air when empty
• Air is 20.9% oxygen by volume
• Room temperature, 20°C (293 K), at atmospheric pressure (101.3 kPa)
• The keg is filled with 19L (close enough to 19 kg) of beer
• Purge cycles go to 15 PSI gauge and vent back to atmosphere
• Worst case: every bit of oxygen left in the keg ends up dissolved in the beer

That last assumption is deliberately pessimistic. In reality, a lot of the air gets pushed out as beer fills, and not all of the oxygen that remains actually dissolves. It makes these numbers a ceiling, this is the best comparison point I could come up with.

Step 1: How much oxygen is in a keg full of air

Start with the ideal gas law to find the total gas in the keg:

n = PV / RT = (101,300 Pa × 0.019 m³) / (8.314 × 293 K) ≈ 0.79 mol of gas

Oxygen is 20.9% of that:

0.209 × 0.79 ≈ 0.165 mol O₂

Convert to mass (O₂ = 32 g/mol):

0.165 mol × 32 g/mol ≈ 5.28 g O₂

Expressed as a concentration in 19 kg of beer:

5.28 g ÷ 19 kg ≈ 0.278 g/kg = 278,000 ppb

That's the baseline reading for an empty keg that we can compare our other readings to.

Step 2: What one purge does

Pressurise the empty keg with pure CO₂, then vent. Pressurising only adds CO₂, so the oxygen already in the keg doesn't change, but the total gas roughly doubles. Venting releases an air and CO2 mixture, so you carry off a share of the oxygen with it when the keg is bled.

The fraction of oxygen left after one cycle is atmospheric pressure divided by total (absolute) pressure:

r = 14.7 / (14.7 + 15) = 14.7 / 29.7 ≈ 0.495

So one purge leaves about half the oxygen:

278,000 × 0.495 ≈ 138,000 ppb

Step 3: Repeated purges

Each cycle multiplies what's left by the same 0.495, so after n cycles:

Oxygen remaining = 278,000 × 0.495ⁿ

Five cycles: 0.495⁵ ≈ 0.0297 → 278,000 × 0.0297 ≈ 8,260 ppb

Ten cycles: 0.495¹⁰ ≈ 0.00088 → 278,000 × 0.00088 ≈ 245 ppb

One caveat for this equation this assumes the CO₂ you're purging with is pure. It isn't. Real bottled CO₂ carries a little oxygen of its own, so in practice you can't dilute below whatever your gas supply contains, no matter how many cycles you run. Given how much oxygen remains in the keg with even 10 purges this innaccuracy is only a marginal rounding error.

Step 4: Liquid displacement

Here the keg ends up full of liquid, then CO₂, so there's no real headspace of air left to calculate the way we did in Step 1. Any gas pocket that survives a thorough push is negligible after bleeding the tiny headspace before displacement.

The actual floor comes from somewhere else entirely: the oxygen already dissolved in the thin film of liquid that clings to the keg's internal walls, dip tube, and posts after displacement, since none of that gets pushed out with the rest. Air-saturated water holds around 9,000 ppb of dissolved oxygen at room temperature. A residual film of somewhere between 20 and 100 mL spread across the keg's internal surfaces and below the liquid dip tube works out to:

20 mL: (9,000 ppb × 0.02 kg) ÷ 19 kg ≈ 9.5 ppb
100 mL: (9,000 ppb × 0.1 kg) ÷ 19 kg ≈ 47 ppb

Which is where the ~10–50 ppb range comes from. It's the leftover liquid doing the damage here, not any trapped gas.

There is also dissolved oxygen in the sanitiser in the keg that will come out of solution and end up in the keg but there's no accurate way to calculate this so we will exclude this from the numbers too.

Step 5: Liquid displacement with metabisulphite

Same setup as Step 4, but now the sulphite in solution scavenges the dissolved oxygen sitting in that residual film leftover and the sanitiser at the start, converting it to inert sulphate before it has a chance to reach the beer. With the film's oxygen mopped up chemically rather than left to dissolve the contribution drops to roughly 5 ppb, limited mostly by how completely the sulphite reaction runs to completion in the time available, not by how much liquid is left behind. A side note on this is that acid based sanitisers will slow down the reduction of oxygen as sulphite exists as different species based on different liquid pH, bisulphite is not the main species at the pH that stellarsan work at.

Step 6: How much CO₂ each method uses

Filling a 19 L keg at atmospheric pressure and 20°C takes about 0.79 mol of gas, which as CO₂ (44 g/mol) is roughly 35g per keg volume.

Purge cycles: Each cycle raises the keg from atmospheric to 15 PSI gauge, so the CO₂ added is the gas needed for that pressure rise:

n = (Pg × V) / (R × T) = (103,400 Pa × 0.019 m³) / (8.314 × 293 K) ≈ 0.81 mol
0.81 mol × 44 g/mol ≈ 35 g per cycle

Fifteen PSI is almost exactly one extra atmosphere, which is why one purge works out to roughly one keg volume of gas. Stack the cycles up:

1 cycle ≈ 35 g
5 cycles ≈ 175 g
10 cycles ≈ 355 g

Liquid displacement: You push one keg volume of liquid out and CO₂ takes its place, so you use about one keg volume of gas, ~35 g, finishing near atmospheric pressure. Push harder and you'll use a bit more, since the keg ends up holding gas at whatever pressure you stopped at. Metabisulphite displacement is identical, so the same ~35 g.

A single purge and a full liquid displacement cost about the same gas (~35 g), but displacement gets you roughly three to four orders of magnitude lower on oxygen. Heavy purge cycling is the only method that actually burns a lot of CO₂, and it still can't match displacement in effectiveness.