I have heard from somebody who did a flight test on this. It was done in an old TB20 and in the south of Europe (Turkey actually, and they had trouble finding cold enough air) but the engine installation has never changed on the type.
With alt air off, the delta-T from OAT to the RSA fuel servo inlet is 10K.
With alt air on, the delta-T from OAT to the RSA fuel servo inlet is 30K.
So, with alt air off, you will get -5C (the worst case structural icing temp) at the fuel servo with an OAT of -15C. That confirms my own experience exactly, and confirms my earlier written belief that the compression heating in the air duct raises the air temp by about 10K.
And, with alt air on, you will get -5C (the worst case structural icing temp) at the fuel servo with an OAT of -35C.
This means that the POH advice of always using alt air on in icing conditions is right only down to about -30C (0C at the servo inlet). Below -30C the fuel servo will freeze up and you will have no way out.
Obviously the above risk is only a risk if in IMC.
The above is just a quick report. I have no more info. I am waiting for the detailed data.
It does suggest that at temps below about -30C (-25C to be sure) your only defence is the spray coming off the prop TKS system.
Aircraft which don’t have prop TKS, or get the engine air via a NACA inlet on the side of the cowling (e.g. the SR22 – maybe this one) have no solution, and will get an engine stoppage at such temperatures.
Carburetors should be mostly unaffected. I have a carb temp gauge and even in -35°C, the temperature inside is positive. Also the carbs don’t have small orifices like the fuel servo and are equipped with a heated air path.
Is there any point along the path where the temperature is actually above 0? If the outside air contains only ice, no supercooled liquid, that ice isn’t going to turn into supercooled liquid unless you warm it up past the melting point first. And if it stays ice, I don’t see how it’ll stick.
Is there any point along the path where the temperature is actually above 0? If the outside air contains only ice, no supercooled liquid, that ice isn’t going to turn into supercooled liquid unless you warm it up past the melting point first. And if it stays ice, I don’t see how it’ll stick.
Good point. I don’t know. If you warm up SLDs from -35C (at which temp they certainly won’t stick to anything) to -5C, will they stick?
But I can tell you that flying in -15C, in IMC, you sure as hell can’t open the alternate air door. It’s frozen pretty well. I did open it, with a big pull. So that proves that whatever water existed at -15C did accrete on the inside of the air duct. But there was zero icing on the aircraft exterior (which would be as expected for -15C) so clearly SLDs which won’t stick at -15C do most definitely stick at -5C.
Does anything significant happen to SLDs from -35C to -15C, which affects whether they will stick to stuff when warmed to -5C?
This clearly has a huge impact on aircraft with automatic alternate air doors.
Is there any point along the path where the temperature is actually above 0?
I would very much doubt it. The gas compression (the cross-section reduction) in the TB duct is monotonic. In some other systems it isn’t…
Or you can use a NACA inlet which avoids ice / water droplets being absorbed.
Could the alternate air door have been frozen from water that got in there earlier in the flight, at higher temps, and then froze as the temps dropped?
Supercooled liquid sticking to a surface should not depend on temperature (well, as long as the surface is below freezing…). Whether the droplets impact the surface does depend on the drop size and the flow geometry, hence the importance of stagnation points.
Does anything significant happen to SLDs from -35C to -15C, which affects whether they will stick to stuff when warmed to -5C?
Yes, they all freeze. 
By the time you get to -40, they’re all gone (due to homogeneous freezing, i.e., without the need for an ice nucleus). At -35, you’d probably have to be in a CB to find them.
Or you can use a NACA inlet which avoids ice / water droplets being absorbed.
How does it do that?
Could the alternate air door have been frozen from water that got in there earlier in the flight, at higher temps, and then froze as the temps dropped?
Possible… however there are drain holes at the base of the duct, right behind the air filter and below the level of the alt air door.
Yes, they all freeze. By the time you get to -40, they’re all gone (due to homogeneous freezing, i.e., without the need for an ice nucleus). At -35, you’d probably have to be in a CB to find them.
So you can be 100% certain that any moisture found in IMC colder than say -25C will never stick to anything, no matter how warmed up it gets, provided that it never rises above 0C?
Peter wrote:
So you can be 100% certain that any moisture found in IMC colder than say -25C will never stick to anything, no matter how warmed up it gets, provided that it never rises above 0C?
Below -40 OAT, you can be absolutely certain. At -25C, there could still be liquid drops around. But if there are, they should also show up on the wings. And they should definitely block the filter (I assume that’s why the POH says alternate air in IMC).
Is there a way to explain these events with ice particles? What is the diameter of the servo openings? Can they be clogged by ice particles (~100 micron-sized)? But how did the ice particles get through the air filter?
The servo inlet hole is about 60mm dia and the openings in the four sensing tubes are about 3mm dia.
I don’t think ice particles can get through the filter, but liquid water definitely can – evidently so.
All this only relates (supposedly) to the TB20. There is no evidence that any other aircraft exhibits this behaviour.
