On Friday, February 24, 2017 at 4:45:05 PM UTC, Adam Funk wrote:
On 2017-02-24, Roger Mills wrote:

On 24/02/2017 11:32, Adam Funk wrote:
I keep finding air in the tops of the same two radiators. This made
sense in the old vented system, because air could dissolve into the
water at the cold surface in the cistern, then come out of solution
later in the hot parts. But we have a sealed system (with a
combination boiler) now.

I'd also expect to get a bit of dissolved air (to come out later) in
the mains water when I add more to the system. But usually I bleed
the two radiators, check the pressure gauge on the boiler,& find I
don't need to add more water, but still find some air a week or two
later. Where is it coming from in that case? Just more air from the
previous top-up coming out a while later?

As others have said, it could be hydrogen - caused by corrosion. Is
there any corrosion inhibitor in your system?

The plumbers (who I think are very good) drained & power-flushed the
whole system in spring 2014 when they installed the new boiler, &
added corrosion inhibitor. They may have added inhibitor in spring
2015 when they serviced the boiler, & definitely added it when they
moved & replaced a radiator & serviced the boiler in spring 2016.

Proper answer, some details mentioned by others.

Most probably hydrogen from electolytic/bi-metallic corrosion. That produces H2 & O2. The oxygen combines with the inside surface of your radiators to form magnetite sludge. Igniting it at the air vent is most unlikely to cause an explosion (I've never heard of that happening), but the recommended procedure is to bleed the gas into an upturned glass/bottle (H2 floats upwards) and then ignite that. You may otherwise have difficulty in closing the air vent after it has turned into a blow-torch (see Youtube).

If hydrogen, the gas in the upturned glass will light with a pop and burn upwards.

The important thing is that you'll usually only get the electrolytic/bi-metallic corrosion if the water is acidic. The acid usually gets in to the system as the residues from active flux (contains ammonium chloride, forms HCl on heating ISTR) that hasn't been thoroughly flushed out.

The French braze copper heating pipes (no flux needed); they have virtually no black sludge problems, there is no market for power-flushing in France.. Go figure.

It's been power-flushed. They will have used an acidic cleaner (HCl); that has to be flushed, neutralized (with caustic soda), flushed, checked with litmus paper (did you see that done?) and flushed again for good measure.

It might be air (nitrogen); that can get drawn into the system on the suction side of the pump, as mentioned elsewhere. The oxygen forms magnetite, as above. The expansion vessel connection (neutral point, PONPC) should be on the pump inlet side.

Most good domestic inhibitors (Sentinel, Fernox, etc) contain sodium molybdate; ISTR that the mixture is slightly acidic. I like to run a flushed system with clean water for a few hours, check it's not acidic, then add inhibitor.