The method of studying exhaust system
design using the EPG (exhaust pressure graph) was described in the January 1996 issue of
Sport Aviation in the first part of this series of articles. In this Part II report, we
will focus on some comparative results obtained from the hundreds of test runs made at the
CAFE Foundation test facility. Primary header pipe length and diameter, collector length
and diameter, a stock muffled system, a "Tri-Y" crossover system and a look at
exhaust wave behavior will be included here. METHODS
The system used to obtain the data presented here was described in the Part I article.
Some of these EPG's were created by combining the data recorded on separate full throttle
runs of similar RPM.
To eliminate any fluctuations in static RPM which we thought might be caused by the
prop governor, we converted the constant speed propeller on the test engine to a fixed
pitch con- figuration. After doing so, we continued to witness the same fluctuations
during full throttle static RPM tests. We then realized that even mildly different wind
vectors into the propeller disc area were the source of these fluctuations. We obtained a
wind velocity and direction indicator to attempt to control for this variable. The best
method, however, is to perform the EPG testing in dead calm air, and the results presented
here were selected for such test conditions.
Soft silicone fuel hose of 3/16" diameter O.D. x 4" long attached the
pressure transducers to the 1/8" x 18" copper tube sampling ports which are
clamped onto the exhaust pipes. These soft tubes probably serve as filtering
"balloons," absorbing some resonance and noise in the sampling ports.
To maintain an accurate zero reference, all of these recordings were made by testing
the pressure readings from each transducer before starting the engine. Representative EPG
data can be captured in just 10 seconds of run time so that the full power runs can be
brief. However, the engine was allowed to warm up before taking the data. The capture is
made only after the full, stable static RPM has been reached. This typically requires
about 10 seconds and allows the prop to induce its own stable inflow field.
The lengths of pipe cited here are the centerline length of the pipe. Collector lengths
cited are the length of the constant diameter portion of the collector, excluding the
merge zone. The merge zone was 4"-5" long for most collectors. To afford
adequate ground clearance, all collectors used except file 009 included a 90* bend in
their constant diameter portion about 2" beyond the end of the merge zone. The
sampling port for collector pressure was consistently placed 10.5" downstream from
the end of the merge zone.
THEORY
The most important moment for achieving low backpressure is during overlap TDC when
both intake and exhaust valves are open. Exhaust systems which achieve this, though they
may have high backpressures during other parts of their EPG, can show more horsepower than
systems having lower overall average backpressures. In the EPG's presented here, special
attention should be given to the P wave amplitude as it crosses the overlap TDC mark.
Also important is the backpressure at the moment the exhaust valve opens, the very
beginning of the P wave. The lower this pressure, the more readily the "slug" of
exhaust gas jumps out of the combustion chamber.
HEADER LENGTH
Figure 1 shows the results obtained when the length of all 4 header pipes was varied
from 36" to 42". Changing the length within this range did not make much
difference in backpressure. A more prominent change is observed by holding the header
length constant and substantially lengthening the collector pipe. All of the 4 into 1
collector exhaust designs used in Figure 1 consistently produced higher static RPM and
fuel flow levels, implying more horsepower, than the independent headers with no collector
shown in Figure 2. The systems in Figure 1 showed lower average backpressures than the one
in Figure 2.
Note that the R (reflected) waves are prominent in Figure 1 regardless of pipe length.
The energy of these waves is the source for some of the backpressure reduction achieved in
these sys- tems.
HEADER DIAMETER
Figure 2 shows, as might have been predicted, that the fattest header pipe demonstrates
the lowest backpressure.
Figure 2 also shows very small R waves, seen on each trace's downslope. These represent
the return waves from the open end of the pipe, unamplified by the other cylinder exhaust
pulses. The lack of energy in these R waves casts some doubt on whether adjusting the
lengths of independent pipes can produce enough wave tuning to lower the backpressure. It
may be that much longer lengths are needed than those tested here, but such long pipes
become unworkable in aircraft applications.
None of the independent pipes in Figure 2 ever achieves a sub-zero or negative
backpressure. In many other EPGs, the addition of a collector to independent pipes was
repeatedly found to lower the backpressure. The addition of a collector also tends to
reduce noise compared to independent header pipe systems. However, collectors create
problems in bulk, weight and space requirements. They also must be carefully suspended and
vibration-isolated from the engine.
A comparison of the climb and cruise performance of an aircraft using independent pipes
versus one with those same pipes coupled to a collector is planned for a future study by
the CAFE Foundation.
COLLECTOR DIAMETER
Figure 3 shows that the effect of substituting a larger diameter collector is to lower
the backpressure and to increase the power produced. The presence of four rather than the
usual three separate R waves in the top trace is unusual and indicates some interference.
COLLECTOR LENGTH
Figure 4 shows a the EPG's obtained from collectors of different lengths (see photo).
All collectors were of 2.5" diameter and had a 5" long parallel merge zone. A
pyramid shaped "goilet" spike1 was present inside the merge zone to smooth the
transition from header to collector. The CAFE Foundation has learned that the internal
details of the collector merge are very important in the performance of the system.2 The
goal here is to avoid any abrupt increase in cross-sectional area as the header
transitions into the collector.
In most cases, the change in collector length was accomplished by simply hacksawing off
a portion of the collector, making no change in the detailed anatomy of the merge.
Figure 4 shows that the 26.5" collector gave lower backpressure and higher static
RPM than the 14" collector. The pressure traces from the col- lector sampling ports
show that the shorter collector has exhaust pulses of low amplitude, approximating ambient
pressure. The longer collector's trace shows repeating waves, labeled "C",
indicating the preservation of the energy of each cylinder's exhaust pulse. These C waves
are reflected into the header to produce the R waves shown.
In our tests, increasing the collector length beyond about 36" only seemed to
increase the backpressure.
The addition of a megaphone to the collector was more beneficial in reducing
backpressure than was a change within collector length alone. This has been a consistent
finding in our tests of the 4 into 1 exhaust systems.
The relative trend of increased noise as the collector length was shortened was not
surprising. The noise quality was markedly changed by the addition of the megaphone to the
collector, being more objectionable even though the sound meter indicated relatively
little change in the noise level.
WAVE VELOCITY
Figure 5 shows the very high velocity of the pressure wave traveling down the header
into the collector. The collector trace shows a pulse for each and every cylinder's
firing, and those pulses can be seen to quickly reflect back up into the header trace. The
wave velocity measured here traveled 24.6" in 1.43 msec. This would be 1433 fps in
the header primary. If we use the collector arrival time, the computation is 49.4" in
3.04 msec giving a speed of 1354 fps. The wave velocity slows as the gas cools enroute, so
that the speed measured would be reduced by longer headers, a fatter collector or a
sampling port further downstream on the collector.
A STOCK MUFFLER
Figure 6 shows the EPG's of an all stainless steel system of the type commonly used in
production aircraft. It incorporates four 1.75" diameter x .035" wall header
pipes of 22.5"-23.5" length converging into a 5" diameter by 10" long
cylindrical muffler can. The 3" diameter tailpipe exits the can at an approximately
30* angle and includes one 70* bend along its 24" length. (see the adjacent photo).
The two most striking findings in Figure 6 were the consistently high backpressure and the
large amount of "cross-talk" detected in each pipe. The small muffler can's lack
of plenum effect probably accounts for the high level of cross-talk observed. It is
presumed that the cross-talk produces chaotic destructive interference in the wave energy
and this would be expected to quiet the exhaust. However, this interference also robs wave
energy which could be used to reduce back pressure.
There was a significant reduction in full throttle static RPM and fuel flow with this
muffler. Most surprising of all was the relatively loud 110.6 dBA cabin noise level
recorded during this run, which is not significantly quieter than systems where a
non-muffled collector was used.
This stock muffler system appears to actually exceed the 2" Hg. average
backpressure required by the FAA certification standard. One SAE study3 states:
"There is a 1% power loss for each 1" Hg increase in exhaust backpressure."
If this holds true, then about 7% more horsepower can be obtained over the stock muffler
system by using the best tuned system thus far tested here. That equates to 14 bhp.
TRI-Y SYSTEM
The "Tri-Y" exhaust system is one in which the four individual primary header
pipes merge into two secondary pipes which, in turn, merge into one tertiary common
collector. It has been postulated that this system can produce a substantial reduction in
backpressure if the 4 into 2 merge is made by joining headers from cylinders whose firings
occur 360 crank degrees apart. The exhaust pulse or "P" wave in one header would
presumably "crossover" and travel up the paired header pipe and reflect off of
the closed exhaust valve of that cylinder. The resulting reflected negative wave would
then, presumably, travel back upstream to the original cylinder and deliver suction to
that cylinder at just the right time during its overlap stroke. This theory is the basis
of the commonly used "crossover exhaust system" popular on many homebuilt
aircraft and is heavily based upon the "sonic" or wave theory of exhaust system
behavior.
The Tri-Y system shown in the adjacent photo produced the EPG shown in Figure 7. There
is some negative back pressure occurring during the early part of the exhaust stroke. The
very large R waves come mainly from the exhaust of the paired cylinder. At the header wave
velocity derived from Figure 5, and the Tri-Y's 37" prima-ries, calculation predicts
that cylinder #1's P wave would bounce off of cylinder #2's closed exhaust valve and
return in 8.6 milliseconds. At 2700 RPM, the IO-360 exhaust valve stays open for 15
milliseconds. A key unknown is the duration of the negative wave influence at the valve,
i.e., its wave length.
Because the RPM and fuel flow level of the Tri-Y system shows some promise for improved
horsepower, the CAFE Foundation plans extensive further EPG testing to find the optimum
combination of diameters, and lengths for primary, secondary and tertiary pipes in the
Tri-Y system.
SHARE YOUR IDEAS
A number of very knowledgeable EAAers have written and emailed further information to
the CAFE Foundation about exhaust systems. We welcome this input and look forward to much
more in the future.
A word about noise: the CAFE Foundation EPG study is focusing on reductions in
backpressure as the main goal, since this is the path to the highest efficiency.3 We feel
that once we have matured our understanding of how to reduce backpressure, then we can
"trade away" some backpressure by adding noise reducing components to the
exhaust.
In our Part I article, the EPG in Figure 2 showed a persistently sub-zero backpressure,
causing some readers to question whether that is possible. We are investigating this
promising finding and will attempt to include those results in our Part III article.
Bibliography
- Seeley, Brien, Tuned Aircraft Exhaust Systems, Sport Aviation, Vol. 39, No. 11,
November, 1980.
- Schalk, Paul, personal communication. Ricardo North America. Engi- neer experienced in
OEM automotive exhaust design by flow bench. Phone 313-397-6666.
- Adams, Tim G., Effect of Exhaust System Design on Engine Performance. SAE Technical
Paper Series. Paper #800319, presented at Cobo Hall, Detroit, February 25-29, 1980.
