Merlin negative G cutout too quick?

[Viper2000]

Ok guys, several things here.

Firstly, the dive was only posted for the g history; there would have been no risk of cutout because the aeroplane was a PR.XI, which would have had a 2 stage engine, and all the 2 stage engines had later carburettors.

The point was just that you can get the nose down quite smartly without recourse to negative g, which is probably why people didn't immediately realise that a negative g capability was required for a military engine.

Secondly, I've gone digging through my archives and have found some relevant RRHT books.

This means that I can now hopefully shed some light both on the engine ratings and FTHs (which are slightly more conservative than given in some Pilots Notes) and also the negative g cut.

Appendix VI
The Merlin S.U. Carburettor Under Negative-G
This Appendix expands the comments made about the effects of negative-g on the S.U. carburettor on page 46 of the third edition of "The Merlin in Perspective - the combat years".

A characteristic of the Merlin which has led to some comment was the cut that occurred if negative-g was applied when entering a dive. The momentary power loss was in two stages; an initial loss of power which some pilots regarded as only a very brief hesitation, followed by a rich cut of about 1½ secs duration. Full power returned once the negative-g was removed. The tactical disadvantage it caused could be reduced by pilot technique, such as entering the dive from one wingtip. Wing commander 'Roly' Beaumont, apart from undertaking the first flights of the Canberra and Lightning fighter, flew Hurricanes in the air fighting in 1940 prior to and during the Battle of Britain, in No. 87 Squadron. I am grateful to him for allowing me to use his quote at the Hurricane 50th Anniversary evening, which while dealing with the abilities of the aeroplane encompasses the engine.

"Another thing we did was to devise a manoeuvre which was aimed at getting us out of a difficult corner if we ever got into one. This may sound very extraordinary, probably, to practising pilots today, but it consisted of putting everything into the left-hand front corner of the cockpit. If you saw a 109 on your tail, and it hadn't shot you down at that point, you put on full throttle, fine pitch, full left rudder, full left stick and full forward stick. This resulted in a horrible manoeuvre, which was in fact a negative-g spiral dive. But you would come out of it with no Me 109 on your tail and your aeroplane still intact."

Comparisons have been drawn with the direct injected DB600 series engines in the Me 109, which did not suffer in this way. The problem was known before the war, but it was thought that British fighters would be operating as bomber destroyers and the situation would not arise, which shows how wrong one can be. The Daimler-Benz engine was good, but direct injection provided no charge cooling due to the latent heat of evaporation of fuel, which was worth 25ºC reduction in charge temperature. Like all engines, it had its own, problems, including failures of injector pipes. Direct injection had been considered by Rolls-Royce, and had been turned down in favour of the carburettor. An example of its use by the Company was the direct injected sleeve valve diesel Kestrel, powering Captain George Eyston's Flying Spray, which had for a time held long distance records and the speed record for diesel cars.

Most of the credit for reducing the negative-g problem to negligible proportions goes to the late Miss Shilling (Mrs Naylor) - Tilly to her friends, who was in charge of the carburettor section of RAE at Farnborough. This remarkable woman saw further than most through a barn door, recognising that the engine fuel pump was part of the problem. It had been designed as two separate gear pumps with individual quill drives, so if one pump were to fail the other would keep the engine running. Each pump was capable of meeting maximum engine demand plus 20%, a relief valve controlling outlet pressure. Miss Shilling saw that if the float needle was not controlling, both pumps could suddenly put a large excess of fuel into the carburettor. She developed the RAE restrictor, which was fitted in the fuel line to the carburettor and which limited the flow to just above the maximum engine demand. It seems unlikely that any were fitted during the Battle, except for a trial quantity of six. Later retrospective action was taken and it was a useful palliative. The scheme finally adopted was for an anti-g version of the S.U. carburettor developed by her section, which allowed the existing carburettor to be modified. The weak cut, which was caused by the fuel in the float chamber rising to the top and starving the jets when negative-g was applied, was cured by fitting jet shrouds which continued to draw fuel from halfway up the chambers. Ball valves were added to prevent the fuel from flooding out of the air vents under this condition. The ensuing rich cut was prevented by forming a restrictor on the end of the float needle, so that when forced fully open by the fuel pressure, the floats then having lost control, the restrictor, like that fitted earlier limited the flow to slightly above maximum engine demand. This carburettor continued to be fitted to various marks of Merlin until production ceased. The changes to the S.U. carburettor to give it its anti-g characteristics are shown in the illustration.[attached]

As mentioned in the main text a negative-g version of the S.U. carburettor was developed, in parallel with the anti-g version, to the point of undergoing service trials, but was abandoned as its mechanism could not sense zero g. On certain marks of engine the RAE restrictor was retained, leaving the carburettor unmodified, but where the anti-g features were embodied the RAE restrictor was removed. Two-stage engines in the range of 60 to 100 series, depending upon mark number, could have either an S.U. or Bendix carburettor, the latter not being subject to g effects. The Bendix was fitted to the 65, 66, 70, 76 and 85. Above 100 series single point S.U. injection was featured and again was not sensitive to g effects. Packard engine were fitted with Bendix carburettors, with the exception of some 100 series equivalents built at the end of the war with Bendix units and Simmonds power control units.

Harvey-Bailey (1995).
I have preserved the original formatting as far as possible, which means that the paragraphs are rather long. I have therefore highlighted some important sections in red so that they aren't missed.

The illustration is of interest because if you remove the modifications then you're back to the standard S.U. carburettor fitted during the Battle.

You can see that the lean cut was caused by the exhaustion of the fuel in the small chamber; whilst the rich cut was caused by the big chamber filling up with fuel and subsequently flooding this small chamber, as well as by fuel leaking through the air line which the mod protected using the ball valve.

Recovery from the rich cut requires that the engine consume the excess fuel in the big chamber. We know that this took about 1½ seconds.

This allows us to make several observations about the behaviour of the system.

1. The maximum rate of fuel pump delivery was about 2.4 times the rate at which the engine could consume fuel.
2. The amount of time taken to flood the big chamber would therefore be about 1.5/2.4 = 0.625 seconds in the worst case.
3. The small chamber is something like half the size of the normal space above the fuel level in the big chamber. We can therefore infer that full power engine demand would empty the small chamber in something like 0.75 seconds.
4. However, under negative g, fuel can also escape the small chamber through the holes that admit it under positive g. Since these holes were at least big enough to supply full engine demand, this would at least double the rate at which the small chamber could empty; therefore the lean cut would be expected within 0.3 seconds of negative g onset.

This means that we can probably safely say that when reduced or negative g is applied, nothing much will happen for at least say ¼ a second, because there is certainly enough fuel in the small chamber to supply the engine for that amount of time even if it's flowing out of the entry holes. And under reduced positive, close to zero g, it will take more like ½ a second.

Once the fuel in the small chamber is exhausted, the engine will start to suffer a lean cut.

However, we have already calculated that the big chamber is likely to be flooded in about the 0.6 seconds.

Once the big chamber is full of fuel, an unregulated supply of fuel will be forced into the small chamber under pressure.

It will take perhaps another ¼ second or so to fill the small chamber, at which point it will then proceed rapidly into the carburettor and cause the rich cut.

So the sequence of events was probably:

ZERO G Onset:

1. t = 0 s, pilot pushes to zero g
2. t = 0.6 s, small chamber empties; engine suffers weak cut
3. t = 0.75 s, fuel flow into small chamber resumes
4. t = 1 s, rich cut begins

NEGATIVE G Onset:

1. t = 0 s, pilot pushes to negative g;
2. t = 0.3 s, small chamber empties; engine suffers weak cut
3. t = 0.75 s, fuel flow into small chamber resumes
4. t = 1 s, rich cut begins

Once the rich cut has started, both chambers are full of fuel; therefore the recovery from negative g would be identical to the recovery from reduced positive g.

Recovery:

1. t = 0, pilot returns positive g
2. t = 1.5 s, excess fuel in float chamber consumed; float resumes controlling

The final piece of the jigsaw is the fact that it was felt tactically advantageous to roll the aeroplane into a dive rather than to suffer the cut.

If you look at the roll rate diagrams here you can see that the worst-case for high speed roll rate would have been about 60º/s, and therefore it would take about 3 seconds to roll inverted for dive entry.

The alternative, of pushing through the cut would result in the engine going on strike for roughly the push time plus 1¼ seconds (because the engine would keep supplying full power for roughly the first ¼ second of the manoeuvre which is therefore subtracted from the 1½ second rich cut recovery after positive g is restored).

If we assume -20 m/s^2 acceleration and constant 150 m/s TAS, since
a = v^2/r,
r = v^2/a = 1125 m
The turn circumference is therefore about 7 km, and so the time to execute a complete outside loop would be about 47 seconds; the time taken to push through 90º would therefore be something like 11.75 seconds, and so the total duration of the loss of power would be about 13 seconds.

So it's fairly obvious that in this sort of situation you'd win by rolling and pulling rather than pushing, because you'd lose a lot of distance in 13 seconds.

The critical case would be a pitch change of about 20º, because at -2 g you'd get there in about 2.5 seconds, for a total cut duration of 3.75 seconds or so, which is of the same order as the amount of time lost in the roll.

This all seems pretty reasonable to me.

The cut duration lines up with the current reports, and the calculated 0.3 second grace period explains the lack of misbehaviour in turbulence.

But what about the reduced positive case?
Well, that's been puzzling me for a while, because the float position is still defined by the float position, and therefore it's not immediately obvious why there would be a problem.

But if you look at the diagram, you'll see that the fuel has to flow down through the holes in the bottom of the big chamber into the small chamber in order to supply the jet. The driving force for this is the head of fuel in the big chamber, which is of course gz.

So when g reduces close to zero, the force driving the fuel through the holes into the small chamber is dramatically reduced, and therefore it follows that the flow rate reduces.

If the flow rate is less than engine demand then the small chamber will gradually empty and starve the jet. This explains the fact that it takes such a long time for reduced positive g to induce a cut. Indeed, it implies that if you waited long enough at say 0.75 g you'd probably get a lean cut eventually; it's just that in reality this never happens because people don't fly like that.

Of course, under reduced positive g the float is still controlling. However, because the flow rate through the holes into the small chamber is less than would normally be the case, whilst the pump delivery rate remains normal, the big chamber starts to over-fill. The float moves up and reduces the rate of fuel supply, but the equilibrium under reduced positive g will be a higher float position such that the progressive reduction in fuel flow into the float chamber balances the reduced rate at which fuel leaves to enter the small chamber.

This means that when 1 g flight is restored, there is going to be too much fuel in the system until equilibrium is restored.

This will happen somewhat more quickly than in the zero or negative g cases because the equilibrium point for reduced positive g is reached when the float chamber is only partially (albeit still excessively) filled rather than totally filled.

Therefore the duration of the rich cut recovery time should be expected to progressively increase towards the 1½ second maximum as g tends to zero.

Finally, what about g onset rate?
I've been thinking about this, and I suspect that it wouldn't make a lot of difference. If the sudden negative g was applied then the float would rise at the same rate as the surface of the fuel in the float chamber; this would momentarily cut off the fuel flow into the float chamber. However, as soon as the fuel hits the top of the float chamber, the float will instantly float downwards, re-opening the valve and admitting fuel at the full pump delivery rate. It won't bounce around because buoyancy would just peg it to its stop.

Therefore any misbehaviour is likely to simply be a function of g and duration.

Reference:
Harvey-Bailey, A. 1995. The Merlin in Perspective - the combat years. Derby: Rolls-Royce Heritage Trust.