By Sam Cooper
For those of you who were able to attend the March 2, 2021 Chapter Meeting presentation on Aircraft Engine Management, I hope that you found the material interesting and informative. One of the things I have found frustrating is that very little hard data was presented on four cylinder aircraft engines, let alone a carbureted aircraft engine. So, instead of unkind adjectives I will show you some data on a typical Experimental/Amateur Built (E/AB) aircraft engine.
In the middle of April I was generously invited by fellow member Norm Biron to ride along in his Glastar while we tested the air/fuel (A/F) mixture distribution on his Superior XP-O-360 with an updraft carburetor and a constant speed propeller. After departing KDTO, we headed north, cleared the Class B airspace and climbed up to 8,000 feet. Once we were level and stable, Norm set the engine RPM, went full throttle and full rich on the mixture. Norm’s Dynon FlightDEK-D180 did the hard work of logging the flight and engine data. Once the temperatures were stable, Norm started the mixture sweep by reducing the fuel flow by about one gallon per hour. We let everything stabilize, and then reduced the fuel flow again. We used eleven different fuel flow settings until the engine vibration at low fuel flow (i.e. lean A/F ratios) led us to stop the mixture sweep. The vibration was not horrible, but not where one would choose to run the engine. Returning to a richer mixture, we descended and landed back at KDTO.
Norm sent me the Dynon log and I spent some time extracting and analyzing the mixture sweep data. Norm has graciously allowed me to share the following data with the rest of the Chapter.
Figure 1 shows the primary result of our mixture sweep test, the variation of Exhaust Gas Temperature (EGT) and Cylinder Head Temperature (CHT) with fuel flow for Norm’s Superior XP-O-360 carbureted engine. The response of the EGTs and CHTs matched the response I was expecting and I was happy to see that we were able to get all of the EGTs to peak before we stopped the test. All of the CHTs generally peaked about 40 °F Rich of Peak (ROP) before its cylinder hit Peak EGT. There are text boxes in Figure 1 noting when the Peak EGTs and the Peak CHTs were reached. The two rear cylinders (4 and 3) peaked in sequence, then both front cylinders peaked roughly together. The total fuel flow spread between the first and last cylinder to peak was 1.2 gallons per hour (gal/hr).
For the CHTs, all of the cylinders peaked within a 10 °F temperature range. And with a maximum CHT of 385 °F on Cylinder 4, we had left a nice margin to the cylinder temperature limits. The flight conditions are noted in Figure 1. At 8,000 feet indicated we ran at a roughly 66% power ROP setting. The outside air temperature (OAT) at altitude (34-37 °F) was 4-7 °F above Standard Day temperatures.
I have seen curves for how an engine’s horsepower output will vary with fuel mixture. Some were conceptual, and some had hard numbers, but only a single curve for the engine’s total output. That is what is more important to a pilot. However, a lot more detailed data is measured for serious engine development, but not necessarily shared. Like most E/AB aircraft, Norm’s Glastar does not measure the engine’s horsepower output. Norm’s Dynon EFIS and EIS estimate engine power percentage, which is very useful to Norm when he flies the Glastar.
I was curious to see if there might be a way to estimate the effect of the A/F mixture distribution on each cylinder’s horsepower output and the cumulative effect on the engine’s total output. I had access to Superior IO-360-B1A2 (injected version of Norm’s engine configuration) EGT and Horsepower fuel mixture curves for a 23/2300 (about 69% power ROP) condition. Using that data, I was able to estimate the individual cylinder power percentage and the total engine power percentage relative to the peak EGT power of the XP-O-360 during the mixture sweep. Those first order power estimates are shown in Figure 2 along with the EGT measurements versus the fuel flow.
If you look at the Engine Estimated Power line (purple dashed line) in Figure 2 you are probably wondering how can that be above 100%? The only engine performance landmark we directly measured with Norm’s EIS is Peak EGT for each cylinder. That Peak EGT fuel flow establishes where that cylinder had a stoichiometric (i.e. chemically balanced) A/F ratio. From this I estimated individual cylinder A/F ratios and fuel flow percentages relative to cylinder Peak EGT. Then, cylinder power was estimated relative to cylinder Peak EGT. Superior’s mixture curves show Peak HP at 50–80 °F ROP at 4-5% more horsepower than at Peak EGT. My estimated engine power reached a maximum just below 104% of the Peak EGT horsepower level at 9.2 gal/hr fuel flow, which was 10 °F ROP for cylinder four, which peaked first. The variation of the horsepower percentage is what is important here, not its absolute value. The horsepower percentage peaks above 100% because the reference is Peak EGT, which is not where maximum horsepower is produced.
The A/F mixture distribution causes the timing of power loss relative to fuel flow on individual cylinders to be different from each other. This can be seen in the lower left corner of Figure 2. The result is different power levels (vertical spread between the cylinder estimated power curves) from the individual cylinders once the mixture is lean enough. The power pulse differences every 180° of crankshaft rotation cause vibrations that the pilot usually choses to avoid if they are great enough. The A/F mixture distribution also flattened out and spread the engine’s cylinder horsepower peaks around so that Peak engine HP is very close, 10 °F ROP for cylinder four, to the first cylinder to hit Peak EGT. The net effect is to slightly reduce the maximum engine horsepower that is available, to broaden and flatten the top of the horsepower peak, while moving that peak horsepower closer to where the first cylinder will hit Peak EGT.
After 800+ hours of flying the Glastar, where does Norm fly it when he wants to cruise efficiently? He sets the power level, manifold pressure and RPM, to reasonable values and then leans till cylinder 4 just goes past Peak EGT and cylinder 4 is just Lean of Peak. The engine vibration is fine at that condition, leaner fuel flows just make more vibration and quickly result in lower airspeed. To me, it looks like this data set and analysis supports what Norm has learned through many years of flying his Glastar.
Learning from the Process
The newsletter version of this article focused on the air/fuel mixture distribution results from the mixture sweep flight test. This expanded version covers additional topics that may be of interest to the readers.
For the mixture sweep flight the Dynon FlightDEK-D180 was configured to log data at ten second intervals, which is Norm’s usual setting that allows him to log several flights before his older data is overwritten. The FlightDEK-D180 can log as frequently as once a second, but older data is then overwritten in 30 minutes of operation. The longer logging interval for our flight was not ideal, but did not substantially affect the results.
Figure 3 shows the as logged EGT and Fuel Flow values. The start of the mixture sweep test itself was about 16:16, when the mixture was set to full rich from a much leaner condition. We ended the test at 17:13. The fuel flow trace is the least smooth. The EGT traces all had similar smoothness, but I will note that they were logged in 1 °F intervals. I prefer the 1 °F resolution of the data and can smooth out the “noise” as will be shown shortly. Yes, EGT1 sure appears to be reading low, which it has done for the last few months. Norm has already replaced his EGT probes at least once, the probes are consumables. But, EGT1 had the same time history profile as the other sensors, just with a smaller vertical scaling factor.
Figure 4 shows the smoothed EGT and Fuel Flow values that I used to select the extracted data values. For post test data smoothing I like to use a centered, moving average. A centered, three point moving average for a row is just the average of the row’s reading, the previous row’s reading, and the next row’s reading. A centered, five point moving average is just the average of the row’s reading, the previous two readings, and the next two readings. (This would not be possible if live data is smoothed since the “future” readings have not been made yet.) The gray vertical dashed lines in Figure 4 show the location of the time stamped data rows that I selected to extract the mixture sweep data points from. For example, the first row extracted was at 16:19:10, the fuel flow was 14.1 gal/hr and EGT1 to EGT4 were 968, 1035, 1101, 1137 respectively.
I tried to extract the data points from locations were the fuel flow had been steady for a while so that the temperatures would be closer to a steady state condition. So, most of the gray vertical dashed lines are at the right end of a plateau in the fuel flow data. Due to our inexperience with this type of testing, we were a little quick to adjust our fuel flow setting, especially at the beginning of the mixture sweep. This compromised some of the first four sets of readings, but we were more patient after that. We also transitioned nearly straight through the EGT4 peak temperature, which occurred at 16:43:00. I used that data point since it was the best we had for Peak EGT4 temperature.
Figure 5 shows the smoothed CHT and Fuel Flow values that were used to select the extracted data values. I did not have to apply any smoothing to CHT2 and CHT4, these are the as logged values. In contrast, the as logged CHT1 and CHT3 values had more “noise”, the cause of which is not know at this time. So, I applied a centered three point moving average to CHT1 and CHT3 for the traces shown in Figure 5. The gray vertical dashed lines in Figure 5 show the location of the time stamped data rows, the same rows as in Figure 4, that were used to extract the CHT temperatures from.
During the Chapter Meeting presentation on Aircraft Engine Management, Technical Counselor Mel Asberry made the comment that CHT readings are very slow to respond to changes in conditions. Prior to this test flight, I did not know exactly how slow that response can be. It is very obvious in Figure 5, if you look at the slow response of CHT4 before the last three extracted data values. If one of the goals of the mixture sweep is to have stabilized CHT values, there will need to be significant dwell time at each individual fuel flow settings.
If the end portion of the mixture sweep time history in Figures 4 and 5 from about 17:00 onwards is inspected, some of the responses look out of line with previous responses to fuel flow reductions. This was when the engine vibration was at its worst. Three of the EGTs show a dip and then a rise even though the fuel flow was still decreasing. EGT4 became a lot noisier, even with the smoothing. CHT4 had more “normal” appearing drops to plateaus. But, CHT1, CHT2 and CHT3 got hotter in loose synchronization with their EGTs. I do not know the cause of this. Given the engine vibration at those fuel flows, most pilots will not choose to operate this way.