Characteristics of Combustion in a Miniature Four-Stroke

Engine

doi:10.6125/JoAAA.200606_38(2).01
Journal of Aeronautics, Astronautics and Aviation. Series A, 38(2), 2006
航空太空及民航學刊, 38(2), 2006
作者/Author： Joseph Papac;Dunn-Rankin Derek

頁數/Page： 77-87
出版日期/Publication Date：2006/06
引用本篇文獻時，請提供DOI資訊，並透過DOI永久網址取得最正確的書目資訊。
To cite this Article, please include the DOI name in your reference data.

請使用本篇文獻DOI永久網址進行連結:
To link to this Article:

http://dx.doi.org/10.6125/JoAAA.200606_38(2).01

DOI是數位物件識別碼（Digital Object Identifier, DOI）的簡稱，

是這篇文章在網路上的唯一識別碼，
用於永久連結及引用該篇文章。

若想得知更多DOI使用資訊，
請參考 http://doi.airiti.com

For more information,

Please see: http://doi.airiti.com
 Journal of Aeronautics, Astronautics and Aviation, Series A, Vol.38, No.2, pp.077 - 088 (2006) 77

Invited Paper

Characteristics of Combustion in a Miniature Four-Stroke

Engine*

Joseph Papac and Derek Dunn-Rankin**

Department of Mechanical and Aerospace Engineering

University of California Irvine
U.S.A.

ABSTRACT

Liquid hydrocarbon fuels are an excellent source of chemical energy

per unit weight and volume. However, an efficient process for converting
this energy into useful work has thus far been elusive for portable power
systems. This work characterizes the combustion process in a 5-cc
displacement miniature four-stroke engine with a peak specific power
measured to be approximately 300 W/kg. Although this value is
significantly higher than the 100 W/kg achieved by most electrochemical
devices, the peak measured fuel conversion efficiency (9.3%) is low
compared to full-size engines, and the value falls below 5% over much of
the engine’s operating range. The efficiency is even worse when
considering the possibility that some of the lubricating castor oil might also
be acting as a fuel. Cylinder pressure measurements show that friction and
pumping losses are substantial. Based on a semi-theoretical analysis, the
turbulence level and burning velocity in the combustion chamber of the
miniature engine are not fundamentally different from those in a full-size
engine. However, energy lost through heat transfer to the cylinder walls,
lack of sufficient residence time for combustion, and the effects of crevice
volume are amplified at small scales because of high
surface-area-to-volume ratios. The experimental results provide the basis
for some redesign recommendations.

Keywords: Portable power, Mesoscale combustion, Internal combustion

engine

I. INTRODUCTION examples of such potential applications.

The power requirements of autonomous systems can
One of the limiting constraints in the development be described in terms of two principal components:
of portable electronic and electro-mechanical systems is specific energy density (defined as the energy stored per
the power source. This power limitation forces design unit mass; the specific energy requirement is driven by
tradeoffs between operating duration and device mass the desired operating duration without refueling), and
(and overall size). Historically, batteries have been the specific power (defined as the power delivered per unit
standard for powering small devices and they will mass of the system; the specific power demanded is
continue to be well-suited for such low power devices as driven by the maximum power needed for the
cell phones and PDAs. For portable and autonomous application). The inherent tradeoff between these two
devices in the size range of a few centimeters to a parameters for a variety of power sources can be seen
decimeter, however, onboard combustion becomes a graphically in Figure 1.
practical alternative [1]. Miniature autonomous The power sources shown represent the actual
vehicles and clusters of portable computers are two performance of commercially available electrochemical

*
Manuscript received, Jun. 3, 2006, final revision, Jun. 9, 2006
**
To whom correspondence should be addressed, E-mail: [email protected]
78 Joseph Papac Derek Dunn-Rankin

and thermo-chemical power devices (as distinct from the small scale engines to provide high specific power in tiny
projected performance of research systems, as these packages [2, 3]. After considerable study and effort [4-8],
values can be misleading) carrying their own fuel. The it appears that the smallest practical combustion engines
diagonal lines on the plot represent the operating duration. will be in the range of a centimeter or so in their critical
The engine curves have two asymptotes controlled by the dimension. A combination of surface-to-volume ratio
ratio of fuel mass to overall system mass. When the effects (primarily thermal management, friction, and
system is carrying very little fuel, its mass is dominated combustion reaction time) and the realization that
by the power conversion structure. This scenario results electrochemical devices are often reasonable alternatives
in maximum specific power but very low energy density. for tiny power outputs has pushed attention to engines in
This appears as a horizontal asymptote in the log-log the range of 30–1,000 Watts. Although some of the
scales of the figure. At the opposite extreme, nearly all of challenges facing engines at the microscale are avoided at
the system mass is fuel, so the energy density is the centimeter size, many remain. The current work
constrained primarily by the system's thermal efficiency, examines some of these challenges through the
appearing as a vertical asymptote. Also noted on the experimental evaluation of a commercially available
curve is the very high energy density of a typical liquid small-scale engine. In particular, we examine a
hydrocarbon fuel at approximately 12-13 kW-hr/kg, four-stroke internal combustion engine designed for
which highlights the need for an efficient system to radio-controlled airplanes.
convert this chemical energy to useful work. While miniature engines have experienced many
years of refinement for their use in the model airplane
and hobbyist industry, there are several features that
make them unsuitable for implementation in portable
power generating devices. For example, they generate
excessive noise and vibration. In addition, the design of
the lubrication and cooling systems are sufficient for
model vehicle propulsion, but they rely on vehicle
motion; they are inadequate for use in stationary devices.
The tradeoffs made by the model engine industry to
reduce the cost and the complexity of these miniature
engines result in high power performance and at the
expense of lower efficiency and greater pollutant
production than is acceptable for portable power
generation.
Figure 1 Ragone plot of source power density vs.
stored energy density of power devices: II. RELATED MINIATURE ENGINE
fuel cell (solid curve), batteries (dotted RESEARCH
curves), and combustion engines (dashed
curves). A primary concern of miniature engine research is
identifying the dominant paths of energy losses so that
engines may be designed to reduce these losses for
The figure also shows the challenge of using maximum efficiency. Annen et al. [9] performed a
electrochemical devices (batteries and fuel cells) for scaling analysis showing the effect of leakage,
applications that demand high specific power. For any combustion quenching, and heat transfer as a function of
application that requires greater than 100 W/kg of size for miniature internal combustion engines. Their
specific power (for example all aircraft and most ground analysis confirmed that leakage and flame quenching are
vehicles), the only choices with acceptable operating potential losses, however the most significant loss
duration are liquid hydrocarbon-fueled engines. There are mechanisms for engines in the 100 W range are frictional
expectations that fuel cells and advanced battery designs losses, pumping losses, and heat transfer to the cylinder
(nickel-metal hydride and lithium-based) will eventually walls. A miniature two-stroke model airplane engine was
achieve sufficient specific power to be considered for dynamometer tested by Cadou et al. [10, 11]. They
such applications, but many difficulties remain. For reported power output of approximately 0.5 brake
example, since the energy release in these horsepower, a value less than one-third of the
electrochemical systems occurs at a surface interface, manufacturer's claimed maximum brake power. The
power production in these devices is governed by surface measured efficiency was near 3.5%, which is much too
area. In addition, the rapid transport of electrons and low for a practical power generation system. The first
ions from deep within a storage matrix produces resistive cylinder pressure measurements of a miniature model
losses. Hence, as the power demands increase, energy airplane engine were performed by Raine and Thorwarth
density capabilities and efficiency decrease. Volumetric [12]. They also used a two-stroke engine as the basis of
energy release through a process like combustion will be their experiments. While their cylinder pressure
needed for maximum specific power. measurements were qualitative, the authors were able to
It was the recognition of the energy considerations conclude that the glow ignition engine shows greater
discussed above that generated an enthusiastic search for cycle-by-cycle variability than a two-stroke spark
 Characteristics of Combustion in a Miniature Four-Stroke Engine 79

ignition engine. They also showed that the type of glow atmosphere and its low vapor pressure results in
plug used greatly affects engine performance and that deposition on surfaces in the immediate surroundings of
good engine performance requires that the air-fuel ratio the engine. Model engines do not have active cooling
be adjusted carefully for each combination of fuel systems. They rely on convective and radiative cooling of
composition and glow plug type. the engine case. A finned, aluminum heat sink engine
A miniature rotary engine that operates on glow fuel head provides a high surface area to facilitate heat
is the focus of ongoing research at UC Berkeley [13-15]. transfer. Furthermore, the relatively high enthalpy of
Their research is showing promise for power generation vaporization of the methanol fuel component acts as an
by making incremental improvements to their engine in-cylinder cooling mechanism.
design. Some of the challenges that they are facing Fuels for model engines are composed of methanol,
include internal sealing, ignition, and fuel delivery issues. CH3OH, nitromethane, CH3NO2, and oil. Typical
The present work examines the operating behavior mixtures are 10-20% (by volume) oil, 0-50%
of a commercially available engine in some detail in nitromethane, and a balance of methanol. Synthetic based
order to identify the opportunities for design changes that oils, castor-based oils, and mixtures of the two are used
could lead to superior portable power performance. exclusively for lubrication. Because of their proprietary
nature, it is difficult to find any useful information about
III. DESCRIPTION OF THE ENGINE synthetic-based oils. Therefore, all of our experimental
results have been obtained using castor oil as the sole
The O.S. Engines FS-30-S was chosen for this study lubricant. Castor oil on the engine surfaces will form
because it is a representative example of a modern complex polymers as it is heated so it can continue to
mass-produced model engine design. The engine is a provide lubrication even after heating. Polymerization
single cylinder, 4.89 cc displacement four-stroke design, results in the formation of high molecular weight
with single intake and exhaust valves driven by pushrods. compounds that tend to create a coke or varnish.
The piston has a single piston ring. According to the Castor oil that is mixed into the combustion chamber,
manufacturer's specifications, the practical rpm range is however, can burn. Because the heat of combustion of
2,500—13,000, the weight of the engine is 278 grams, castor oil is 39.5 MJ/kg, it can contribute greatly to the
and the maximum power output is 0.5 brake horsepower heating value of the fuel mixture. In fact, castor oil can
(373 watts), which occurs at 10,000 rpm. Based on be used to produce biodiesel fuel [16]. The issue of how
these specifications, the specific power is 1300 W/kg much of the castor oil polymerizes and how much of it
(neglecting the weight of the fuel). burns is an important one for calculating the fuel
Fuel is delivered to the engine by a needle valve conversion efficiency and the equivalence ratio. This
carburetor. The fuel tank is pressurized by the pressure in complication will be addressed in the Results Section.
the exhaust pipe. Hence, the pressure driving the fuel The spontaneous ignition temperatures of methanol
through the carburetor is influenced both by the and nitromethane are 574 °C and 419 °C respectively
hydrostatic pressure of the liquid fuel and by the pressure [17]. Glow plug temperatures are typically between 700
of the exhausting products. While this design has the °C and 800 °C. Therefore, while pure methanol fuel
advantage of simplifying the balance of plant by will ignite, fuel mixtures with higher concentrations of
eliminating the need for a fuel pump, it contaminates the nitromethane will ignite earlier in the engine stroke. The
fuel supply with exhaust products. Also, because the lower heating value, QLHV, of methanol is 20.0 MJ/kg and
exhaust pressure fluctuates with the operating conditions the QLHV of nitromethane is 11.3 MJ/kg. It is apparent
and the fuel level of the storage tank, the fuel-air mixture that the energy density of a fuel mixture decreases as the
delivered to the engine fluctuates in time (even with a nitromethane concentration increases (relative to
constant mixture setting on the carburetor and at a methanol concentration). The advantage of adding
constant throttle setting). nitromethane to the fuel mixture is that the ratio of
Ignition is initiated by a resistively heated glow plug chemical energy to volume of the stoichiometric reactant
wire with a platinum catalyst. Once the engine has mixture increases. Nitromethane acts as a superior
reached a steady-state operating temperature, electrical oxygen carrier, requiring less air (and therefore less inert
energy to the glow plug is no longer necessary. The nitrogen) for combustion than methanol, thereby
heat retained in the glow plug between cycles continues allowing a greater mass of combustible mixture into the
to provide a catalytic hot spot for ignition in the engine same size combustion chamber, which results in greater
cylinder. Because of this design, indirect methods of specific power. The stoichiometric molar air-fuel ratios,
ignition control, such as adjusting the compression ratio (A / F)s, of pure methanol and nitromethane are 7.14 and
of the engine and changing the glow plug and fuel 3.57 respectively and the volumetric energy density of a
properties, must be employed. The compression ratio of nitromethane-air mixture is nearly double that of a
the engine is varied by adjusting the number of spacers methanol-air mixture (6.73 versus 3.51 J/cm3 gas mixture
underneath the engine head. Glow plugs of varying at standard temperature and pressure).
temperature range and length are manufactured, as are
fuels of varying composition. IV. EXPERIMENTAL APPARATUS
Lubrication of the engine parts is accomplished by
premixing the fuel with oil. This simplifies the engine In order to measure the specific power and the fuel
design, but all of the unburned oil is exhausted into the conversion efficiency of the O.S. FS-30-S miniature
80 Joseph Papac Derek Dunn-Rankin

engine, an electric motor engine dynamometer was

constructed. The angular velocity of the engine shaft is
measured optically by determining the frequency of the
reflections of a helium-neon laser. The light is reflected
off of a marking on the engine shaft once per crankshaft
revolution and focused onto a fast photodiode, which
outputs a periodic waveform corresponding to the
intensity of the reflected light. The frequency of this
waveform is a measure of the angular velocity of the
engine. The torque of the engine shaft is measured by
applying a known load to the engine via an electric motor
interfaced by a belt and pulley drive train. Typical belt
efficiencies are 95—98%, therefore we assume that
negligible losses occur in the pulley coupling system. Figure 2 Block diagram schematic of the engine
The electric motor acting as the dynamometer is of test stand.
the permanent magnet (PM) DC-type (Ever Motor
ERS-380PM-3270). Due to the interaction between the
stator field and the armature current, the torque generated
V. MEASURED POWER AND SPECIFIC
by a PM DC electric motor is directly proportional to the
armature current [18]. POWER

The engine was tested with three fuel combinations,

τ = kT ⋅ i (1)
each with 18% castor oil lubricant, which was the
manufacturer's minimum recommended value. The
The value of this torque constant, kT = 5.16 engine was operated over a range of torques (8—60
milli-Newton-meters/amp, is specified by the motor milli-Newton-meters) and engine speeds (3,500—13,500
manufacturer. The electric motor is connected in parallel RPM). A summary of the findings is shown in Table 1.
to a high power resistor bank. The external resistance, The data used in the tables and figures below are
Rext, can be varied between 0.6 Ω and 10.0 Ω. An averaged values from 5 data sets.
increase in external resistance reduces the applied load by
decreasing the current provided to the motor windings.
The voltage, V, between the electric motor's positive and Table 1 Summary of measured brake power,
negative terminals is measured with a digital multimeter. specific power, and fuel conversion
From this voltage measurement, the torque of the engine, efficiency of the OS Engines FS-30-S
τ, can be calculated directly. D2 and D1 are the diameters tested with three fuel mixtures.
of the pulleys attached to the engine and the electric
motor, respectively. Maximum
Specific
Maximum Maximum
Fuel Volume Brake Efficiency Efficiency
Fuel Species Power
Mixture Fraction Power (including (omitting
D2 kT V (W/kg)
τ= ⋅ (2) (W) castor oil) castor oil)
D1 Rext methanol 79%
A nitromethane 3% 48.0 173 3.5% 5.6%
castor oil 18%
The fuel flow rate measurement is determined by taking a methanol 72%
B nitromethane 10% 82.9 298 5.8% 9.3%
series of measurements of the mass of the fuel in the fuel castor oil 18%
tank. The fuel mass flow rate is then calculated from a methanol 62%
least squares fit of the data. The intake air volumetric C nitromethane 20% 48.3 173 3.5% 5.7%
castor oil 18%
flowrate is measured by a rotameter. To prevent thermal
failure of the engine components, a 77 ft3/min fan
provides external cooling. Steady state operation is
achieved over a two-minute warm-up period before data The maximum steady-state brake power was
is acquired. A fixed load setting is applied to the engine measured to be 82.9 W, which is a value significantly less
and a fixed angular velocity is maintained for than the manufacturer’s specified maximum power of 0.5
approximately ten minutes, while measurements are brake horsepower (373 W). The measured power output
logged. A schematic of the experiment is shown in Figure corresponds to a specific power rating of 298 W/kg, as
2. opposed to the 1300 W/kg that would result from the
The sound amplitude was measured with an Extech manufacturer’s value. Nevertheless, this is an impressive
Model 407736 sound level meter mounted 5 cm away result, considering that most electrochemical devices
from the engine. The ambient sound level in the room manage far less than 100 W/kg. The maximum power
was 55 dB. The sound level was fairly constant was obtained with fuel mixture B. From the earlier
throughout the experiments with an output of discussion regarding the volumetric energy density of
approximately 80 dB. nitromethane as compared to methanol, maximum power
should occur with maximum nitromethane content. It
 Characteristics of Combustion in a Miniature Four-Stroke Engine 81

appears, therefore, that ignition timing (which in this 70 16000

engine is compression controlled) plays an important role 14000
60
in the power performance as well. That is, the lower

Angular Velocity (RPM)

50 12000

Torque (mN-m)
ignition temperature of nitromethane would suggest 10000
ignition earlier in the compression stroke than would 40
Mix A, Torque
8000
occur for a mixture with less nitromethane. In addition, 30 Mix B, Torque
6000
Mix C, Torque
the latent heat of methanol is substantially higher than 20 Mix A, Angular Velocity 4000
that of nitromethane, which may lead to a slightly cooler 10 Mix B Angular Velocity
2000
charge in the cylinder, thereby further delaying the onset 0
Mix C, Angular Velocity
0
of ignition. The management of ignition timing by this 0 20 40 60 80 100
mechanism is difficult to control over a broad operating Brake Power, W
range. Non-ideal ignition timing is likely the reason that
we did not achieve maximum power with fuel mixture C.
Figure 3 Plot of brake power versus torque and
Figure 3 is a plot of brake power versus torque and
angular velocity.
angular velocity. There appears to be a stronger
dependence of brake power on torque. One explanation
for this phenomenon is that frictional losses become
The stoichiometric air-fuel mass ratios for fuel mixtures
greater at higher speeds. An experimental study on
A, B, and C are 6.11, 5.43, and 4.62 respectively. It is
frictional losses of over 300 four-stroke engines ranging
standard practice to describe engine efficiency by the fuel
in size from 50 cc to 1,100 cc produced an empirical
conversion efficiency, ηf [22].
correlation showing that the total engine friction,
characterized as the frictional mean effective pressure,
Wb
increases in proportion to the square of the engine speed ηf =
m f (Ymethanol QLHV , methanol + Ynitromethane QLHV , nitromethane + Ycastoroil QLHV ,castoroil

[19, 20]. This finding should be of importance when
considering the parameters under which a miniature (4)
power generation device will be designed to operate.
Specific to the target application, an operating duration Wb is the measured brake power, m f is the fuel mass
and maximum power consumption rate must be defined. flow rate, Yi is the mass fraction of species i, and QLHV,i is
The operating duration of the device is constrained by the the lower heating value of species i.
fuel conversion efficiency of the engine and the amount Assuming the limit of zero castor oil combustion,
of fuel carried with the device. Similarly, the maximum the maximum fuel conversion efficiency of ηf = 9.3% is
power consumption rate is constrained by the maximum achieved with fuel mixture B. Because both maximum
brake power of the engine and conversion efficiency of efficiency and power are achieved with mixture B, it
either the power train transmission (for propulsion appears that the ignition timing is more suited to this fuel
devices such as miniature aircraft) or the electrical combination. Fuel conversion efficiencies for typical
generator (for portable electronic devices). The miniature full-size engines can reach 30%. This measurement
engine should be optimized to operate at the torque and highlights the potential for improvement of miniature
angular velocity that corresponds to the specifications of engine performance.
the target application. For devices that require long Figure 4 is a plot of the fuel conversion efficiency
operating duration, it may be beneficial to operate versus equivalence ratio. Note that the optimal engine
miniature engines at high torque and relatively low operation occurs lean of stoichiometric conditions, φ < 1
angular velocity so that frictional losses are minimized. if castor oil is not included, but operation is rich when all
This is the approach taken by Honda to improve fuel
the oil burns. For zero castor oil fuel rich operation, φ > 1,
economy for small motorcycles [21]. For devices that
the efficiency and power output are very low. In contrast,
require very high power, the engine should be operated at
adequate efficiency and power production are achieved
high torque and high speed, with the unfortunate side
under fuel lean conditions to a limit of φ ≈ 0.6, after
effect of lowering the fuel conversion efficiency.
which stable operation was not achieved. If castor oil is
not burning, the engine operates in a very narrow window
VI. MEASURED EFFICIENCY AND of lean, but near-stoichiometric air-fuel ratios under all
EQUIVALENCE RATIO conditions. This result is in contrast to the generally held
view that model engines operate very fuel-rich, and so it
The equivalence ratio, φ, is calculated under the is almost certain that castor oil is burning as well. It is
assumptions that air is composed of 21% O2 (by volume) also possible that the amount of castor oil burning may
and 79% N2, and that the castor oil lubricant is either an depend on engine operating temperature which could
inert species or a fully combustible fuel. essentially broaden the effective equivalence ratio
beyond the narrow stoichiometric zone noted above. In
(m methanol + m nitromethane + m castoroil ) / m air addition, these engines do not operate as premixed
φ=
(m methanol + m nitromethane + m castoroil ) / m air , stoichiometric systems. Because some of the fuel-oil mixture is used as
an in-cylinder cooling mechanism, the engine exhausts a
(3)
large amount of this material unburned. The fuel-laden
exhaust that gives the impression of rich operation does
82 Joseph Papac Derek Dunn-Rankin

not necessarily result from an overall rich combustion, Calculated from the measured power and rotational
but from a combination of lean or stoichiometric speed data, the maximum brake mean effective pressure,
combustion and a partially unburned wall film. The high bmep, is 147 kPa. Assuming a peak cylinder pressure of
levels of unburned fuel in the exhaust made accurate twice the maximum bmep, the peak temperature resulting
emission measurements problematic because of the from an isentropic compression stroke from room
narrow operating range of standard gas analyzers. temperature is roughly 410 K, a value less than the
spontaneous ignition temperatures of methanol and
nitromethane, which are 574 ºC and 419 ºC respectively.
It is apparent, therefore, that when the engine is cold the
reaction must be initiated at the platinum catalyst of the
glow plug. Once the engine achieves its normal operating
temperatures, which we measured to be between 150 ºC
and 250 ºC, the incoming charge is heated somewhat and
the compression heating can then reach ignition
temperatures.

VII. CYLINDER PRESSURE

MEASUREMENTS

Cylinder pressure measurements were obtained with

a fiber optic diaphragm pressure transducer. This
pressure sensor operates on the principle of light
reflection from a metal diaphragm flexing under the
Figure 4 Plot of fuel conversion efficiency versus effect of pressure, as described by Wlodarczyk et al. [23].
equivalence ratio. The bars represent The transducer is mounted flush in the top of the
the range of equivalence ratio that is combustion chamber, adjacent to the intake and exhaust
possible if different amounts of the valves. The cylinder pressure was measured as a function
lubricating castor oil are assumed to burn. of time for both motored (no combustion) and fired
engine cycles.
Figure 6 shows cylinder pressure versus time for a
Figure 5 summarizes a key result of the experiments. complete motored (no combustion) engine cycle. The
It is a plot of measured brake power versus fuel peak pressure obtained is approximately 110 psia. This
conversion efficiency for the miniature engine operating corresponds to 7.5 atmospheres. The peak pressure
on the three fuel mixtures, using the best case assumption remains fairly constant from cycle-to-cycle during
that none of the castor oil acts as fuel. The figure shows motoring. The large pumping losses appear as the
an essentially monotonic increase of power with secondary peak. This effect is measured by motoring the
efficiency for all of the fuel blends tested. Hence, engine with the carburetor butterfly valve partially
maximum power is achieved coincident with maximum closed.
efficiency. This result suggests that better fuel
evaporation and mixing would further improve the
8
engine performance, especially at low speed where
turbulent mixing is less influential. 7

6
Pressure (atm)

0
4.5 4.52 4.54 4.56 4.58 4.6 4.62 4.64
Time (sec)

Figure 6 Plot of cylinder pressure versus time

during a motored (no combustion) engine
cycle.

Figure 7 shows pressure in the cylinder during firing

Figure 5 Plot of brake power versus fuel conversion cycles. There are significant cycle-by-cycle fluctuations
efficiency in the cylinder pressure. This is consistent with the
 Characteristics of Combustion in a Miniature Four-Stroke Engine 83

observations of Raine et al. [12] in a two-stroke model Wi = Wc ,in ⋅ ω / 2 (5)

engine. All of the peaks have the same shape, but the
peak pressure fluctuates between 190 and 260 psia. In
full-size engines, three factors have been found to For the cycle shown in Figure 8, Wi = 29.3 W. The
influence this cycle-by-cycle variation [22]: (1) the brake power was simultaneously measured for this
variation in gas motion in the cylinder during combustion, engine cycle and is approximately half of the indicated
(2) the variation in the amounts of fuel, air, and residual power, Wb = 15.0 W. This analysis shows that pumping
exhaust gas supplied to the cylinder for each cycle, and and frictional losses can make up a significant amount of
(3) variations in mixture composition within the cylinder the total losses in the miniature engine.
each cycle due to variations in mixing between air, fuel,
and residual gas. Figure 7 also shows an expanded view
of one complete four-stroke cycle.
16

14 compression
expansion
12 exhaust
intake

Pressure (atm)
10 Poly. (expansion)
Poly. (compression)
Poly. (intake)
8 Poly. (exhaust)

0
0 1 2 3 4 5 6
Volume (cc)

Figure 8 Plot of cylinder pressure versus cylinder

volume
Figure 7 Plot of cylinder pressure versus time
during fired cycles, including an expanded
view of one cycle. VIII. ANALYSIS OF MINIATURE ENGINE
COMBUSTION

The data from the engine cycle shown in Figure 7 Experimental studies have shown that the
was used to construct a pressure-volume indicator compression and expansion processes in spark-ignited
diagram. Because we did not have a shaft encoder of internal combustion engines are well fitted by a
sufficient resolution to obtain crank-angle degree polytropic relation [22]. The value of the polytropic
measurements, a few assumptions were necessary in exponent for typical fuels is n = 1.3 ± 0.05. While the
order to obtain pressure as a function of cylinder volume method of ignition differs in a glow-ignited IC engine,
from our time domain pressure data. First, we assumed the compression and expansion processes are the same.
that the peak pressure measured over one complete cycle Therefore, we assume that the polytropic relation holds.
corresponds to the instant at which the piston is at For simplicity, we further assume that the incoming
top-dead-center (TDC). Similarly, the instant of fuel-air mixture is a fully premixed ideal gas composed
minimum pressure corresponds to the time at which the of methanol vapor and pure air (21% by volume O2 and
piston is at bottom-dead-center (BDC). Lastly, we 79% N2) in stoichiometric proportions at an initial
assumed that the velocity of the piston remains temperature T1 = 350 K (slightly preheated) and pressure
approximately constant as it moves from TDC to BDC P1 = 1 atm. The temperature, T2, and pressure, P2, of the
and vice-versa. More recent measurements on the same mixture at TDC prior to combustion are then calculated.
engine have confirmed the validity of these assumptions
[24, 25]. The P-V diagram for the fired engine cycle T2 ⎛ Vc ⎞
n −1

under these assumptions is shown in Figure 8. A sixth =⎜ ⎟ (6)

order polynomial curve was fit to the data and integrated T1 ⎝ Vd = Vc ⎠
over the cycle to obtain the gross indicated work per
cycle, Wc,in = 0.498 Joules. The pumping work per n
P2 ⎛ Vc ⎞
cycle was calculated by integration over the intake and =⎜ ⎟ (7)
exhaust strokes, Wp = 91.6 mJ. The pumping work per P1 ⎝ Vd = Vc ⎠
cycle is 15.5 % of the Wc,in. The indicated power, Wi ,
may be obtained from the indicated work and the angular Vc = 0.81 cc is the clearance volume and Vd = 4.89 cc is
velocity. the displaced volume of the engine. With the
assumption of n = 1.3, the temperature and pressure of
84 Joseph Papac Derek Dunn-Rankin

the fuel-air mixture at TDC prior to ignition are T2 = 629

K and P2 = 12.7 atm. 2α
Abraham et al. [26] suggest that turbulence δL = (11)
SL
intensities, v'rms, in spark-ignition engines can be
approximated by v'rms ≈ vp/2, at the time of spark, where
vp is the piston velocity. Also, the integral scale, lo, can be α = 0.241 cm2/s is the thermal diffusivity. The value of
approximated by lo ≈ h/2, where h is the instantaneous α used is that of air at the mean temperature of Tavg = (T
clearance height between the top of the piston and the + Tf)/2, where T = 629 K, Tf = 2,200 K and is at P = 12.7
cylinder head in disk-shaped combustion chambers. From atm. The flame thickness is δL = 40.4 µm.
kinematic analysis, the instantaneous piston speed in m/s The Dahmköhler number, Da, measures the
is related to the angular velocity, ω (rev/s), the crank importance of chemical reaction rates relative to
angle after TDC, θ, the ratio of the connecting rod length convective time scales. For the miniature engine, we
to crank radius, R*, and the engine stroke, S. calculated Da = 19.1. Figure 9 is a plot of the
Dahmköhler number versus the turbulence Reynolds
number. The values of Da = 19.1 and Relo =1048 are
⎛ ⎞
π ⎜ cos θ ⎟ consistent with those typical of full size IC engines. They
v p = 2 Sω sin θ ⎜1 + 1/ 2 ⎟
(8) lie near the boundary between the wrinkled laminar flame
2 ⎜
⎝
*2

(
R − sin 2 θ ) ⎟
⎠ and the flamelets-in-eddies regimes of turbulent
combustion [29]. Based on this analysis, the effects of
The instantaneous clearance height, h, is given by: turbulent flow in the centimeter scale would appear to be
similar to those effects in full-size internal combustion
engines.
h
(
= 1 + 0.5 ( rc − 1) ⎡⎢ R* − 1 − cos θ − R* − sin 2 θ ) ⎤
2 1/ 2

hTDC ⎣ ⎥⎦
(9)

where rc is the compression ratio of the engine.

The dimensions of the O.S. engine were measured
and are: hTDC = 2.67 mm, rc = 7.05, S = 16.4 mm, and
R* = 3.49. From the equation above, the piston velocity
reaches a maximum near the middle of the stroke. The
values of the turbulence intensity and the integral scale
based upon this approximation are v'rms = 2.68 m/s, and
lo = 1.73 mm. These values are calculated for ω = 166.7
rev/s (10,000 rpm) and θ = 30º before TDC, which is a
typical value for spark timing in automotive engines. The
integral scale represents the mean size of the large eddies,
those of low frequencies and long wavelengths. It is of
the same order of magnitude as the characteristic length
of the flow. The turbulence Reynolds number based upon
the integral scale is Relo = 1048.
The dynamic viscosity, µ = 3.15 x 10-5 N⋅s/m2, and
the density, ρ = 7.11 kg/m3, of the working fluid are
assumed to be equal to that of air at T2 = 629 K and P2 =
12.7 atm. The laminar burning velocity, SL, of the
stoichiometric methanol-air mixture at 629 K and 12.7
atm is calculated from the empirical correlation
developed by Metghalchi and Keck [27].

ε β
⎛T ⎞ ⎛ P ⎞
S L = S L ,o ⎜ 2 ⎟ ⎜ 2 ⎟ (10) Figure 9 Chart of Damköler number versus
⎝ To ⎠ ⎝ Po ⎠ turbulence Reynolds number (based on
the integral scale).
The values of SL,o, ε, and β are constants that depend
on the fuel type and equivalence ratio. The values of T0 =
298 K and P0 = 1 atm are the reference temperature and With the previously stated assumptions, a turbulent
pressure corresponding to standard atmospheric burning velocity, ST, can be calculated using the wrinkled
conditions. The calculated value of the laminar burning laminar flame correlation developed by Clavin and
velocity in the miniature engine is SL = 119 cm/s. Williams [30].
The flame thickness, δL, is calculated from
Spalding's [28] laminar 1-D premixed flame approach.
 Characteristics of Combustion in a Miniature Four-Stroke Engine 85

{ }
1/ 2 ring, and cylinder wall and the gaps in the glow plug and
ST / S L = 0.5 ⎡⎢1 + (1 + 8Cvrms ) ⎥⎦
' 2 1/ 2 ⎤
(12) glow plug threads. These concerns are amplified in
⎣
miniature engines because the crevice volumes can
become a larger percentage of the total displaced volume.
C is a constant with a value near unity. For the
These effects result in a reduction of the cylinder pressure
miniature engine, the value of the turbulent burning
during the latter stages of compression, during
velocity at the operating conditions described above is ST
combustion, and during expansion. Also, local cooling of
= 230 cm/s. With the turbulent burning velocity known,
the high temperature combustible mixture near crevice
the time required for flame propagation across the
volumes could amplify heterogeneous combustion and
cylinder may be estimated by assuming that the flame
slow the reaction rates and lead to flame quenching.
kernel ignites at the center of the combustion chamber
and that the flame propagates radially outward to the
cylinder walls at the turbulent burning velocity. Then the
IX. SUMMARY AND DISCUSSION
time required for complete combustion is 4.2
milliseconds. The time is actually somewhat shorter than There is great potential to improve the performance
this because the hot product gases expand on the inside of of current production model engines for portable power
the flame ball and contribute to the flame dynamics. At applications. The measured power density of the O.S.
an operating condition with ω = 10,000 rpm, which is FS-30-S is very impressive when compared to the current
near the upper limit of stable combustion, each electrochemical devices on the market, but the
crankshaft revolution occurs over 6 milliseconds, which performance of these miniature engines leaves much
is on the same order as the burning time. Hence, it is room for improvement. The maximum brake power
likely that part of the combustion inefficiency results measured, 80 Watts, was substantially less than the
from the lack of residence time. claimed power output of the manufacturer, a result
An important concern for scaling engines to smaller consistent with the observations of Cadou et al. [11].
size, whether nano, micro, or miniature, is that the Even when making extreme assumptions regarding the
surface-area-to-volume ratio of the combustion chamber role of lubricating oil as fuel, the highest fuel conversion
becomes larger as the size of the engine decreases. Since efficiency measured was less than 10%, a value that is
combustion is a volumetric phenomenon and heat unacceptably low when compared to full-size engines.
transfer is a surface phenomenon, there exists a critical Although several modifications to the engine design
combustion chamber size at which the heat transfer to the would make it easier to analyze, the results obtained give
cylinder walls begins to dominate the heat production. If a fairly clear picture of the dominant phenomena in small
the engine size is smaller than this critical dimension, the reciprocating engines. Ignition in these engines is
regime changes from one where external cooling is initiated by a glow plug making it difficult to control. As
necessary (in order to protect the engine components in HCCI engines, one must resort to indirect methods i.e.,
from mechanical failure due to thermal stresses) to one in those which alter the compression process or the fuel
which insulation of the combustion chamber is necessary oxidation kinetics, to control the ignition timing.
in order to maintain high enough reaction rates so that the The combustion process in these miniature engines
chemical reactions can proceed within the allocated appears to operate as some combination of glow plug
residence time. initiated diffusion flame, a diesel mode, and a
While this limit of minimum combustion chamber partially-premixed burning mode. The
size has been reached in some nano-scale and micro-scale partially-premixed component burns the lean fuel-air
engines, miniature engines of the centimeter scale operate mixture prepared as the liquid fuel bleeds into the air
in the regime where sufficient heat is produced in the stream through the carburetor. In addition, a wall film of
combustion chamber to necessitate the use of external unevaporated fuel and lubricant forms on the cylinder
cooling. For our experimental tests, the 77 CFM surfaces. Fuel evaporation from this film produces a
cooling fan was necessary and at times insufficient to relatively rich mixture near the walls that might sustain a
maintain engine temperatures below the recommended diffusion flame or diesel-like combustion process, where
maximum running temperature of 150 ºC. the fuel vapor comes not from fuel spray but from a wall
A rough estimate of the convective heat transfer film.
from the engine assuming a transitional Reynolds number, It is interesting that film fed combustion may occur
surface temperature of 500 K, and surface area of 10 cm2 in this centimeter scale engine since recent results have
gives 10-20 W of heat loss. While significant in shown that for small devices, a thin liquid wall film has
comparison to the mechanical power production, this higher surface-to-volume ratio than would droplets
value is only a few percent of the total chemical energy generated in standard ways [31, 32]. The recent papers on
of the fuel. Furthermore, we observed that the maximum continuous small fuel-film combustors demonstrate this
measured brake power did not occur at the maximum phenomenon. Also, the existence of liquid films in
cylinder head temperature as might be expected for an full-size automotive engines is problematic because it can
engine in which heat losses dominate heat production. lead to lower fuel economy, higher soot emissions, and
In addition to the surface-area-to-volume concerns, increased deposit formation [33, 34].
crevice volume becomes more important at small scales. As identified by others, the limitations that arise
As the cylinder pressure increases, gas flows into when scaling full-size engines down to the centimeter
crevices such as the regions between the piston, piston scale can be attributed primarily to the increase of the
86 Joseph Papac Derek Dunn-Rankin

surface-area-to-volume ratio of the combustion chamber. of California, Irvine’s Center for Portable Power
The energy lost by heat transfer to the cylinder walls is Systems.
amplified when the size of the engine is reduced.
Furthermore, local quenching of the reaction near crevice REFERENCES
volumes becomes more significant due to the smaller
displaced volume. Also of reasonable concern in [1] Dunn-Rankin, D., Leal, E., and Walther, D.,
miniature engines is the lack of sufficient residence time “Personal Power Systems,” Progress in Energy and
for combustion, which can be attributed to high engine Combustion Science, 31, 2006, pp. 422-465.
speeds and relatively low levels of turbulence inside the [2] Fernandez-Pello, A. C., “Micropower generation
cylinder. using combustion: Issues and approaches,” Proc.
When considering design modifications to current Combust. Instit., 29, 2002, pp. 883-889.
engines in order to adapt them for use in portable power [3] Whalen, S., Thompson, M., Bahr, D., Richards, C.,
generation devices, the most important issues that should and Richards, R., “Design, fabrication and testing of
be addressed are the lubrication system and fuel-air the p3 micro heat engine,” Sensors and Actuators A,
delivery system. The lubrication system must be 104: 290–298, 2003.
modified before these engines can be used for portable [4] Cadou, C. P. and Leach, T. T., “Combustion
power generation applications because the addition of processes in micro-scale engines,” Proc. of the
inert oil into the combustion chamber reduces the power Western States Section of the Combustion Institute
density of the system by lowering the reactant Fall Meeting, Los Angeles, CA, 2003.
concentration/volume ratio. It also causes excess [5] Petersen, R. B. “Size limits for regenerative heat
emissions of unburned hydrocarbon oils. Furthermore, engines,” Microscale Thermophysical Engineering 2,
the addition of inert oil into the fuel storage tank reduces 1998, 121–131.
the extractable energy density of the fuel (the overall [6] Kyritsis, D. C., Guerrero-Arias, I., Roychondury, S.,
energy density is higher, but the oil is not fully and Gomez, A., “Mesoscale power generation by a
combusted). If the heating value of the oil is considered catalytic combustor using electrosprayed liquid
in the fuel conversion efficiency calculation, then the hydrocarbons,” Proc. Combust. Instit., 29, 2002, pp.
measured peak efficiency is reduced, for example, from 965–972.
9.3% to 5.8%. The design of the fuel-air delivery system [7] Aichlmayr, H. T., Kittelson, D. B., and Zachariah, M.
should be improved to facilitate mixing of the fuel/air R., “Miniature free-piston homogeneous charge
mixture prior to ignition. The carburetor design does not compression ignition engine-compressor
sufficiently atomize the reactants, resulting in incomplete concept—part i: performance estimation and design
combustion. It may be desirable to add a swirling considerations unique to small dimensions,”
mechanism to the engine intake. Furthermore, the needle Chemical Engineering Science, 57, 2002, pp.
valve adjustment on the carburetor is very sensitive and 4161–4171.
requires user intervention. For practical application the [8] Aichlmayr, H. T., Kittelson, D. B., and Zachariah, M.
reactant delivery process could utilize a feedback R., “Miniature free-piston homogeneous charge
controller requiring no end-user input. compression ignition engine-compressor
The control of ignition timing in these miniature concept—part ii: Modeling hcci combustion at small
engines must be improved in order to increase thermal scales with detailed homogeneous gas phase
efficiencies. The experimental results have shown the chemical kinetics,” Chemical Engineering Science,
importance of optimizing ignition timing in order to 57, 2002, pp. 4173–4186.
achieve maximum power production and fuel conversion [9] Annen, K., Stickler, D., and Woodroffe, J.,
efficiency. This is also consistent with the experimental “Linearly-oscillating miniature internal combustion
observations of Raine et al. [12]. The O.S. Engines engine (mice) for portable electric power,” Proc. of
FS-30-S that we tested achieved maximum performance the American Institute of Aeronautics and
with the fuel mixture containing 62% methanol, 18% Astronautics Conference, Reno, NV, 2003.
castor oil, and 10% nitromethane. We expected that [10] Cadou, C., Sookdeo, T., Moulton, N., and Leach,
engine performance would increase with higher T., “Performance scaling and measurement for
concentrations of nitromethane, however this was not the hydrocarbon-fueled engines with mass less than 1
case. The engine ignition timing depends on the glow kg,” Proc. of the American Institute of Aeronautics
plug temperature and the pressure of the reactants and Astronautics Conference, Reno, NV, 2002.
(controlled by the compression ratio). If the glow plug [11] Cadou, C., Moulton, N., and Menon, S.,
ignition method is used, these parameters should be “Performance measurement and scaling in small
optimized for the specific fuel mixture that the engine is internal combustion engines,” Proc. of the American
designed to operate with. Institute of Aeronautics and Astronautics Conference,
Reno, NV, 2003.
ACKNOWLEDGEMENT [12] Raine, R. R. and Thorwarth, H., “Performance and
combustion characteristics of a glow ignition
This work was supported in part by a grant from the two-stroke engine,” SAE Paper 2004-01-1407,
National Science Foundation (CTS-0212163) under the 2004.
supervision of Dr. Linda Blevins and from the University [13] Fu, K., Knobloch, A., Martinez, F., Walther, D.,
 Characteristics of Combustion in a Miniature Four-Stroke Engine 87

Fernandez-Pello, C., Pisano, A., Liepmann, D., Germany, 1999.

Miyaska, K., and Maruta, K., “Design and [24] Pompa, J., Karnani, S., Mulroney, A., Leal, A., and
experimental results of small-scale rotary engines,” Dunn-Rankin, D., “Effects of Hydrogen Addition on
Proc. of the International Mechanical Engineering Combustion in a Miniature Four Stroke Engine,”
Congress and Exposition, New York, NY, 2001. Western States Section/The Combustion Institute
[14] Swanger, M., Walther, D. C., Fernandez-Pello, A. C., Paper 05-F27, Fall Meeting, Stanford, CA, October
and Pisano, A. P., “Small-scale rotary engine power 17-18, 2005.
system development status,” Proc. of the Western [25] Pompa, J., Karnani, S., and Dunn-Rankin, D.,
States Section of the Combustion Institute Spring “Effects of Cooling Techniques on Combustion in a
Meeting, Davis, CA, 2004. Centimeter Scale 4-stroke Engine,” Western States
[15] Sprague, B., Tsuji, Y., Walther, D. C., Pisano, A. P., Section/The Combustion Institute Paper 06S-24,
and Fernandez-Pello, A. C., “Observations of flame Spring Meeting, Boise, ID, March 27-28, 2006.
speed and shape in small combustion chambers,” [26] Abraham, J., Williams, F. A., and Bracco, F. V., “A
Proc. of the Western States Section of the discussion of turbulent flame structure in premixed
Combustion Institute Spring Meeting, Davis, CA, charges,” SAE Paper 850345, 1987.
2004. [27] Metghalchi, M. and Keck, J. C., “Burning velocities
[16] De Oliveira, D., Di Luccio, M., Faccio, C., Dalla of mixtures of air with methanol, isooctane, and
Rosa, C., Bender, J. P., Lipke, N., Menoncin, S., indolene at high pressure and temperature,”
Amroginski, C., and Oliveira, J. V., “Optimization Combustion and Flame, 48, 1982, pp. 191-210.
of enzymatic production of biodiesel from castor oil [28] Spalding, D. B., “Combustion and Mass Transfer,”
in organic solvent medium,” Applied Biochemistry Pergamon Press, 1979.
and Biotechnology, 113-16, 2004, pp. 771–780. [29] Turns, S. R., “An Introduction to Combustion
[17] Mullins, H. P., Spontaneous ignition of liquid fuels, Concepts and Applications,” McGraw-Hill Inc,
AGARD, Butterworths Scientific Publications, 1996.
1955. [30] Clavin, P. and Williams, F.A., “Effects of Molecular
[18] Histand, M. B. and Alciatore, D. G., Introduction to Diffusion and of Thermal Expansion on the
Mechatronics and Measurement Systems, Structure and Dynamics of premixed Flames in
WCB/McGraw-Hill, 1999. Turbulent Flows of Large Scale and Low-Intensity,”
[19] Yagi, S., Ishibashi, Y., and Sono, H., “Experimental Journal of Fluid Mechanics, 116, 1982, pp.
analysis of total engine friction in four stroke s.i. 251-282.
engines,” SAE Paper 900223, 1990. [31] Sirignano, W. A., “Analysis of a miniature
[20] Yagi, S., Fujiware, K, Kuroki, N., and Maeda, Y., liquid-film combustor,” Proc. of the Eastern States
“Estimate of total engine loss and engine output in Section of the Combustion Institute, Fall Meeting,
four stroke s.i. engines,” SAE Paper 910347, 1991. Hilton Head, SC, 2001.
[21] Suzuki, M., Ando, R., and Ishibashi, Y., [32] Sirignano, W. A., Pham, T. K., and Dunn-Rankin, D.,
“Paradoxical approach to improve fuel economy for Proc. Combust. Instit., 29, 2002, pp. 925-931.
small practical motorcycles,” SAE Paper [33] Zhao, F., Lai, M. C., and Harrington, D. L.,
2004-01-0989, 2004. “Automotive spark-ignited direct-injection gasoline
[22] Heywood, J. B. “Internal Combustion Engine engines,” Progress in Energy and Combustion
Fundamentals.” McGraw-Hill Inc, 1988. Science, 25, pp. 437-562.
[23] Wlodarczyk, M. T., Poorman, T., Arnold, J., and [34] Solomon, A., Anderson, R. W., Najt, P. M., and
Coleman, T., “Long-Life Fiber-Optic Pressure Zhao, F., “Direct fuel injection for gasoline
Sensors for Harsh Environment Applications,” 9th engines,” SAE Special Publication PT-80, 2000.
Trade Fair and Conference Sensor 99, Nuremberg,