Power from fuel flow - looking for data

[AlexPeterson]

Kevin, nice work in progress!

A couple things - it does seem that those with flow transducers before the fuel pump wander around a bit. I have mine after the fuel servo, and is usually will only flicker between tenths of a gallon/hr. Maybe those reporting data to you can also give you an idea of tank to tank reliability of the reported fuel burn vs actual.

I would think the variation in mixtures between cylinders will be difficult to handle - the few carb'd planes here that we have taken data for have variations of 1 to 2 gph between first and last peaks. It is more like four engines connected together. Perhaps you have a method for dealing with that.

Good luck - we will follow this one with interest!

[frazitl]

might be another variable you want to look at. I'd at least get information on the prop used so you can evaluate if this is an issue...

This flight testing stuff reminds me of a vehicle dynamics class I took at UofI way back. Real hardware and real data. Fun stuff!

[Kevin Horton]

Originally Posted by frazitl

might be another variable you want to look at. I'd at least get information on the prop used so you can evaluate if this is an issue...

Yeah, prop efficiency is definitely a variable. It is a very difficult variable to deal with though, as it is difficult to get prop efficiency maps from the manufacturers, and even if you get them you need to take them with a grain of salt, as they are theoretical predictions, which assume a generic engine cowling shape. According to Les Doud from Hartzell, the RV cowling shape is favourable compared to the generic one they assume. And even if you could get prop efficiency maps, they are a real PITA to deal with.

Fortunately, over the small range of powers and speeds that I am looking at for each test, the prop efficiency won't change much at all. So I assume it is constant. The error that this introduces is certainly much smaller than the errors in IAS and fuel flow.

[Kevin Horton]

I've got the first two data sets. Thanks Webb.

These first results are very encouraging. I plotted calculated power vs IAS, and the ROP and LOP test points match up very nicely. If I assume that each IAS could have up to one kt of error (which is really quite a small error, if you consider the small variations you see in IAS even in smooth air), all test points fall within 1 hp of a line of perfect fit.

That suggests that this method might actually work, and it also suggests that our intrepid RV test pilot did a pretty good job gathering the data. Congrats to Webb.

I'll hold off posting the data plots until I get his permission. He may not want Bob Axsom to know just how fast his RV is.

[Kevin Horton]

I created fixed pitch prop test cards in Excel (and also PDF format). The difference with fixed pitch prop is that the rpm will change when you change the mixture. So, we need to also record rpm at each mixture, and I need to attempt to deal with the impact on the power calculation. I think I have a way to deal with that complication, but we'll see how well it works once I have some data to analyze.

I suspect the data from aircraft with fixed pitch props will have more noise than data from aircraft with constant speed props. That is because the rpm will vary during the small IAS variations you get even in the best attempt at level flight. These rpm variations cause the power to vary, which makes the IAS variations even worse. The use of an autopilot with altitude hold, if available, will help immensely.

People with constant speed props can get test cards in Excel (and also PDF format).

[Kevin Horton]

Webb Willmott sent me data from 2400 rpm, full throttle, at 8,500 and 10,500 ft.

The first two plots are IAS * calculated power vs IAS^4. In a perfect world, with perfect data, zero error in the airspeed system, constant prop efficiency, and a perfect way to calculate power, all points would fall on a straight line. The real world is not perfect, so we see some noise in the data. But the ROP and LOP points line up nicely, which suggests that this method of calculating power may be able to account for different mixture settings.



The next two plots are calculated power vs IAS. You can see that if we assume any of the points could have an error of one kt (which is a very small error, given the normal variations in speed) that all points fall within one hp of the fit line. Very encouraging.


Thanks Webb.

[Kevin Horton]

A sharp-eyed RVer pointed out that I had an error in the fixed pitch prop test cards test card (PDF format). I had built the fixed pitch one starting from the constant speed card, and hadn't changed the text that talked about keeping rpm constant when you changed the mixture. I knew that was impossible to do, but forgot to change the text.

Now I have changed the procedure description to say to keep the throttle position constant while you change the mixture. The updated cards are at the above links. Thanks Howard.

People with constant speed props can get test cards in Excel (and also PDF format).

[Kevin Horton]

I updated all the test cards, yet again, as I realized it would be useful to also record manifold pressure (if the aircraft has an MP gauge). That will provide an answer to the inevitable question of "how does the power calculated by this method compare to that from the Lycoming power charts?"

See earlier posts for links to the test cards.

[Kevin Horton]

I got some more power from fuel flow data today - this data was from Hevansrv7a's RV-7A with "Superior IO-360 Plus (roller lifters) 180 HP", two mags and a Catto three blade fixed pitch prop. He got me four different fuel flows at about 2500 rpm at 7600 ft - one pretty close to peak power, and one at peak EGT and two more LOP. I ran the data through the Lycoming power from fuel flow method, with a hypothetical addition to the method to make it work for fixed pitch prop. The plots certainly look good. The variation of speed vs calculated power looks very good, which suggests that this power calculation method provides powers that vary correctly with fuel flow.

The basic method as described by Lycoming assumes the rpm is the same with mixture for peak EGT as it is at the fuel flow of interest. Thus the method is only workable for a constant speed prop, as the rpm will vary with fuel flow if you have a fixed pitch prop. I created a hypothetical extension to Lycoming's method, that assumes that fuel flow will vary linearly with rpm for small changes of rpm at a constant MP, altitude and mixture control position. So, for example, if we have 2500 rpm and 10 gph at peak EGT, and 9 gph and 2400 rpm at the fuel flow of interest, I assume that if you could keep rpm constant, the fuel flow at peak EGT at 2400 rpm would be = 10 * 2400/2500 = 9.6 gph. Given that this calculation method is much less sensitive to errors in fuel flow at peak EGT as it is to fuel flows at the test condition, I think this approach will work for fixed pitch props.

It is too early to declare victory yet, but the first two results are encouraging. Note: I don't yet have a very good idea as to whether this method accurately calculates power. At the moment my investigation has focused on whether it provides a calculated power that varies correctly as the mixture varies. The question of whether it provides the correct power value is a more difficult one to answer. I hope to eventually have enough data to be able to compare the power values from the classical Lycoming power charts. But even that will not provide a definitive answer, as there are so few of our engines actually match a configuration that was delivered by Lycoming, so it is hard to find a good power chart to use for every engine.

Thanks Howard.



Keep the data coming guys.

[ghatch]

Back in my Cirrus days we used a formula to calculate % Power which is directly proportional to Fuel Flow while LOP.

Formula: FF X 14.95 = HP, HP/Max HP = % Power

7.7gpm x 14.95 = 115.12HP, Assuming a 180HP IO-360, 115.12/180 = 64% Power. Sound about right?

I use that formula for an 310HP IO-550 but I believe the constant is constant for our internal combustions engines but, only while running LOP. Walter Atkins might chime in to verify.

Gerry