Vous êtes sur la page 1sur 20

Paper # E24 Topic: Engine

5th US Combustion Meeting


Organized by the Western States Section of the Combustion Institute
and Hosted by the University of California at San Diego
March 25-28, 2007.

Experimental and computational investigations of small internal


combustion engine performance.
Shyam Menon and Christopher Cadou
Department of Aerospace Engineering, University of Maryland at College Park,
College Park, Maryland 20742, USA

A numerical simulation of the performance of a 2.5 cc, two-stroke model aircraft engine is
developed and compared to actual engine performance measurements. The engine cycle is
simulated by solving the mass and energy transfer equations coupled with volume change (forcing
function) and a detailed heat release mechanism. Two separate volumes are considered in the
simulation, a cylinder volume and a crankcase volume. Mass and energy are transferred across the
boundaries between the cylinder and the crankcase. Similar interactions take place between each
volume and the external environment during the exhaust and intake processes. The simulation
assumes zero dimensional processes and each volume is considered to be a perfectly mixed
reactor. The coupled system of ordinary differential equations (ODE’s) is integrated in time using
a fifth order runge-kutta scheme. A detailed chemical reaction mechanism for
nitromethane/methanol chemistry is used to generate a heat release profile. The chemical kinetic
equations are solved using CANTERA whose output is coupled to the main engine cycle code.
Heat transfer coefficients measured experimentally are used to include a heat loss mechanism to
the cylinder wall. The model is tuned for the engine by adjusting the loss coefficient in the
carburetor venturi so that the volumetric efficiency predicted by the simulation matches that
measured experimentally. Contours of power, torque and overall efficiency are plotted by running
the simulation for different engine speeds and are compared to experimental measurements. The
results show that improved models for frictional loss, heat loss, and ignition are required in order
to achieve good correspondence between the experimental measurements and the simulations.
Notation
P Pressure ω Angular rotation rate
m Charge mass Yi Mass fraction of the ith species
V Component Volume. MWi Molecular weight of the ith species
T Temperature (K) ωɺ i Production rate of the ith species
R Gas constant (J/kgK) Cd Loss coefficient
Ru Universal gas constant (J/mol K) A Reference area
Cv Specific heat at const. volume(J/kgK) γ Ratio of specific heats
u Internal energy (J)
q Net heat addition/loss (J)
γ u Ratio of specific heats of unburnt gas
htrans Heat transfer due to mass convection γ b Ratio of specific heats of burnt gas
hloss Heat loss P1,P2 Upstream & downstream pressures
B Cylinder bore hgas Gas heat transfer coefficient
S Piston stroke c1,c2 Empirical constants
lc Connecting rod length k Gas thermal conductivity
θ Crank angle υ Kinematic viscosity

1
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

Re Reynolds number Vdisp Cylinder displacement


mɺ f Fuel flow rate C1 Empirical coefficient for friction
QR Fuel heating value model
a1-7,b1-10 Empirical coefficients from Ceviz
& Kaymaz model Sp Mean piston speed (m/s)
N Number of species considered
RPM Engine speed

1. Introduction

The increased use of small (<10 kg) Unmanned Air Vehicles (UAV’s) for civilian and military
applications is driving a need for compact propulsion systems that are inexpensive to build and
operate and consume high energy density liquid hydrocarbon fuels. The keys to maximizing the
range or endurance of these vehicles are maximizing the thermal efficiency (or brake specific
fuel consumption) of the powerplant [1] and ensuring that the engine-propeller combination is
optimized. Therefore, knowledge of the complete operating map of the engine is required [2].
Unfortunately, aside from the efforts of one research group [3,4] and some simple measurements
reported by designers of small UAVs [5], this type of detailed performance information is not
available in the scientific literature. Recently, more comprehensive performance information has
become available for a small (~150 g) piston engine called the ‘Yellowjacket’ which is
manufactured by AP Engines’ [6]. The key scale-dependent processes that govern the engine’s
performance include heat transfer, friction, leakage, fuel-oxidizer mixing, and
ignition/combustion. Many studies have focused on parts of the scaling problem. Some focused
on thermal losses using system-level heat transfer analyses [7-9] while others have focused on
the effects of friction and leakage [10]. The objective of the work reported here is to use the
findings from these previous efforts to develop a simple thermodynamic model of the
‘Yellowjacket’ engine can be used to improve our understanding of the factors limiting its
performance. More importantly, the model can be used as a tool by which to predict how the
performance of small piston engines scales with engine size. Of particular interest is how heat
loss and charge leakage from the cylinder affect the efficiency of the combustion process in these
small engines.

2. Model Description.

The AP Yellowjacket engine is a two-stroke design where the fuel/oxidizer/oil mixture is drawn
into the crankcase and pressurized before being injected through a transfer port into the cylinder.
Therefore, the numerical simulation solves the conservation, energy, entropy, and state equations
for both the crankcase and cylinder. The timings of the intake, transfer, and discharge processes
are controlled by the positions of a rotary valve and of the transfer and exhaust ports in the
cylinder. A schematic diagram of the model used to represent the engine is presented in Figure
1. The formulation and solution of the governing equations in the portion of the cycle where
combustion is not involved follows the general approach outlined in Ferguson [11]. In this
portion of the cycle, the changes in mixture gas constant and specific heat ratios are accounted
for using empirical relations by Ceviz and Kaymaz [12]. These relations and their empirical
constants are presented in Appendix A. The compression, combustion, and expansion portions of

2
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

the cycle are simulated using CANTERA [13] and a multi-step chemical model for the reaction
of methanol/nitromethane mixtures with air [14].

Exhaust Port
Qloss

Exhaust Transfer
Vcyl Port
Vexhaust

Vintake
Fuel & Air
Vcase

Rotary Valve θ

Figure 1. Schematic diagram of two-stroke engine volumes modeled in the simulation.

Overall Governing equations.


1. State.
∂V ∂P ∂T ∂m ∂R
P +V = mtot R + RT tot + mtotT (1)
∂θ ∂θ ∂θ ∂θ ∂θ
2. Conservation of mass.
∂m f v (θ , ∆P )
= (2)
∂θ ω
The function f v (θ , ∆P ) represents the opening and closing of the valves and
ports connecting the crankcase and cylinder to each-other and the outside
environment. This function will be described in more detail later.
3. Conservation of energy (first law of thermodynamics).
∂T ∂m ∂C ∂htransfer ∂hloss ∂V
mCv + C vT + mT v = − −P (3)
∂θ ∂θ ∂θ ∂θ ∂θ ∂θ
The methods used to compute the net heat transfer and heat loss
( f h ω = ∂htransfer ∂θ − ∂hloss ∂θ ) will be described in more detail below.
4. Volume change (forcing function) of cylinder and crankcase volumes.
  1
2 − 
 
∂V π B 2   S *sin θ    1  2  S *sin θ    2    S 2 *sin 2 θ    (4)
= *   −   *  lc −     *  −2* *cos θ   
∂θ 4  2    2   2     4 
   
5. Second law of thermodynamics.

2
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

∂S 1 dq
=− (5)
∂θ T ∂θ
6. Conservation of species.
∂Yi wɺ
= i (6a)
∂θ ω m
∂mi ∂m
mtot − mi tot
∂Yi ∂θ ∂θ
= (6b)
∂θ mtot 2

Equation 6a applies when chemical reaction is occurring and equation 6b applies


at all other times.
7. Change of gas constant.
∂R  1 ∂Yi 
= Ru ∑   (7)
∂θ  MWi ∂θ 
8. Change in ratio of specific heats.
∂γ u ∂T
= (a2 + 2a3T + 3a4T 2 + 4a5T 3 + 5a6T 4 )
∂θ ∂θ
(8)
∂γ b ∂T b b 2b T
= (b2 + 2b4T + 6 + 3b7T 2 + 92 + 10 )
∂θ ∂θ φ φ φ
9. Change in specific heat at constant volume.
∂R ∂ (γ − 1)
(γ − 1) −R
∂Cv ∂θ ∂θ
= (9)
∂θ (γ − 1)2
Equations 8 and 9 only apply when chemical reaction is not occurring.
Taken together, equations 1-9 form a set of ordinary differential equations (ODE’s) that describe
changes in the thermodynamic properties and composition of the gases in the cylinder and
crankcase.
Mass inflow and outflow
Mass flow into and out of the crankcase and plenum are computed using equation 10 below. The
first column gives the expression used to calculate the mass flow rate and the second column
gives the constraints on the use of the equation. The first expression corresponds to the situation
where mass leaves the component of interest because the pressure inside is greater than the
pressure outside and the flow is choked. The second corresponds to the situation where mass
leaves the component of interest but the pressure ratio is not sufficient to choke the flow. The
third and fourth expressions correspond to choked and un-choked flow entering the component
of interest. A(θ) is the valve opening area which is a function of crank angle. Table I shows the
crank angles corresponding to the opening and closing of the various ports and valves.

3
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

γ +1
 γ (γ −1)
 γ  2  2 (γ −1) P1  γ + 1 
− C d A(θ ) P1   ≥  ; P1 > P2
 RT1  γ + 1  P2  2 
 γ −1 1 2
 1γ   γ (γ −1)
− C A(θ ) P2   2γ 1 1 − P2   P1  γ + 1 
γ

 P   γ − 1 RT <  ; P1 > P2
 d  1  1  P1   P2  2 

f v (θ , P1 , P2 ) =    (10)
γ +1
γ (γ −1)
 γ  2  2 (γ −1) P2  γ + 1 
 C d A(θ ) P2   ≥  ; P2 > P1
 RT 2  γ + 1  P1  2 
 1γ 
γ −1 1 2
 γ (γ −1)
  P1   2γ 1  P2  γ  P2  γ + 1 
 C d A(θ )   1 −  <  ; P2 > P1
  P2   γ − 1 RT2  P1   P1  2 
  

Table I Timing of engine port events.


Event Crank Angle
Bottom Dead Center (BDC) 180
Intake Valve Open 213
Transfer Valve Close 235
Exhaust Valve Close 255
Top Dead Center (TDC) 360
Intake Valve Close 415
Exhaust Valve Open 465
Transfer Valve Open 485
Bottom Dead Center 540

Gas mixture composition


The composition of the mixture entering the crankcase is set by the equivalence ratio for the case
being considered. The fuel is assumed to consist of 80% methanol and 20% nitromethane. The
oil component of the fuel is assumed to be inert and therefore is not included in the calculation.
Table II gives the composition and density of the fuel mixture as well as an estimate of the
mixture’s overall heating value based on the heating values of the individual constituents. The
balanced chemical reaction for the complete combustion of the fuel with air is given by:
4 A + 3B (11)
( ACH 3OH + B CH 3 NO2 ) + f (0.21O2 + 0.79 N 2 ) → ( A + B)CO2 + H 2O + 0.79 x N 2
2

4
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

where A = χ CH 3OH ρ CH 3OH MWCH 3OH , B = χ CH 3 NO2 ρ CH 3 NO2 MWCH 3 NO2 .CANTERA is used to
determine the species consumption/production rate at each integration step based on the gas
pressure, temperature, and composition along with a multi-step mechanism for the reaction of
methanol-nitromethane mixtures with air [14]. The mechanism incorporates 77 species and 484
reactions and was initially developed to investigate HCCI combustion at small scales.
Table II Properties of glow fuel
constituents and mixture.
Component Χ Ρ Qr
3
(g/cm ) (MJ/kg)
CH3OH 0.8 0.81 21.12

CH3NO2 0.2 1.137 11.6

Mixture 1.0 0.87 18.65

The rates of change of the species mass fractions in the crankcase and in the cylinder when the
mixture is not reacting are computed using the following equations:
∂mi ∂m
mtot − mi tot
fi 1 ∂Yi 1 ∂θ ∂θ
= =
ω ω ∂θ ω mtot 2

mi = Yi × mtot (12)
∂mi
= (Yi )cyl × mɺ exhaust + (Yi )CC × mɺ transfer + (Yi )environment × mɺ int ake
∂θ
In these equations, mɺ exhaust , mɺ transfer and mɺ int ake , are functions of θ and ∆P and correspond to the
mass flow rates through the exhaust port, transfer port, and intake port respectively computed
using equation 10. Note that mɺ int ake = 0 when considering the cylinder. When the mixture is
reacting, the species consumption/production rates are computed directly using Cantera.

Scavenging
Scavenging is very important in two stroke engines and refers to the process of fresh incoming
charge from the crankcase displacing the burnt combustion products from the cylinder. The
effectiveness of the scavenging process determines the amount of burnt gases that remain in the
cylinder volume. This, in turn, limits the power output of the engine. A “complete mixing” [15]
scavenging model is assumed here, where the mass of the fresh charge entering the cylinder from
the crankcase is assumed to mix instantaneously with the cylinder charge resulting in a uniform
mixture in the cylinder volume. In an actual engine, there is both mixing of the charge and short-
circuiting where the fresh charge escapes directly through the exhaust port. In this work, the
‘burned’ gas corresponds to everything left in the cylinder after the combustion process
The masses of the burned and unburned fractions are tracked separately in order to determine the
scavenging efficiency of the engine but must sum to the total mass in the system:

5
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

mtot = mub + mb (13)


Similarly,
∂mtot ∂mub ∂mb
= + (14)
∂θ ∂θ ∂θ
The changes in the burned and unburned masses are given by:

∂mb f (θ , ∆P ) 1   1  
 + mɺ transfer  1
 
= b = mɺ exhaust  1+ k  + mɺ int ake  (15)
∂θ ω ω  1+ k  
 cyl   cc 
and

∂mub f (θ , ∆P ) 1   k cyl  
 + mɺ transfer  k cc
 
= ub = mɺ exhaust  1+ k  + mɺ int ake  (16)
∂θ ω ω  1+ k  
 cyl   cc 
Again, mɺ exhaust , mɺ transfer and mɺ int ake , are functions of θ and ∆P and correspond to the mass flow
rates through the exhaust port, transfer port, and intake port respectively computed using
equation 10. mɺ int ake = 0 when considering the cylinder. kcyl and kcc, the ratios of unburned to
burned mass in the cylinder and crankcase, are defined as follows:
m 
kcyl =  ub 
 mb cyl
(17)
m 
kCC =  ub 
 mb CC
These are updated at every angular step in the integration and used in equations 15 and 16 to
estimate the mass flow rate.

Heat loss
Figure 2 is a schematic diagram illustrating the process by which heat is lost from the cylinder to
the environment. Heat is transferred convectively from the combustion chamber to the cylinder
wall, transported via conduction through the cylinder wall, and finally via forced convection to
the environment. Anand’s Nusselt number correlation (equation 18) [16] is used to determine
the heat transfer coefficient between the combustion gases and the cylinder wall.
k
hgas = c1 Rec2 ; c1 = 0.76, c2 = 0.64
B
Sp × B
Re = (18)
υ
hloss = hgas × Aheat transfer × (Tgas − Tw,i )

6
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

The heat transfer coefficient for heat transfer from the outer wall of the cylinder to the
environment is determined using a Nusselt number correlation derived from measurements in a
model aircraft engine [17]. The correlation is given by
Nu r = C r Re 0.618 Pr 1 / 3 (19)
where Re is the Reynolds number of the flow over the cylinder head (taken to be 55000 here)
and Pr is 0.7, the Prandtl number for air. The total change in heat content of the mixture is the
sum of the heat additions and subtractions via mass transfer plus the heat losses to the
environment.
fh
ω
=
1
ω
[mɺ exhaust C P ,cyl Tcyk + mɺ transfer C P ,cc Tcc + mɺ int ake C P ,env Tenv − hloss ] (20)

Note that mɺ exhaust = 0 when considering the crankcase and mɺ int ake = 0 when considering the
cylinder.

qɺenv
Tg

Tw,i
Tw,o

Piston
Tenv
Cylinder
Wall

Figure 2. Schematic diagram of the heat transfer process from the cylinder to the environment.

Friction Loss
Engine frictional losses arise from pumping losses, friction between piston and cylinder surfaces,
and crankshaft bearing friction. The total frictional loss is modeled using a correlation for an
indirect injection engine [18]. The correlation is expressed as a drop in mean effective pressure
and is converted to friction power before subtracting it from the power output computed from the
model results.
 RPM  2
Motoring mep (kPa ) = C1 + 48   + 0.4 S p
 1000  (21)
Friction Power = Motoring mep *Vdisp * RPM

7
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

3. Method of Solution.

Numerical Method: Two types of ODE systems are solved during the engine cycle. The first
applies to the combustion process in the cylinder and is represented by the matrix Equation 22.
The system of equations is solved using MATLAB and its stiff ordinary differential equation
solver ode15s.
1 0 0 0 0 0 0 0 ⋯ 0  mtot   0 
0 m   
 1 0 0 0 0 0 0 ⋯ 0   ub   0 
0 0 1 0 0 0 0 0 ⋯ 0  mb   0 
     
0 0 0 1 0 0 0 0 ⋯ 0  h   hloss 
0 0 0 0 1 0 0 0 ⋯ 0 d V   V (θ ) 
   =  (22)
0 0 0 −1 P 1 0 0 ⋯ 0  dθ u   0 
0 0 0 −1 T 0 0 1 0 ⋯ 0  S   0 
     
0 0 0 0 0 0 0 1 ⋯ 0  Y1   ( f i )Cantera / ω 
0 ⋱ 0    ⋮ 

0 0 0 0 0 0 0
  ⋮   
0 0 0 0 0 0 0 0 ⋯ 1   YN  ( f N )Cantera / ω 

The source terms ( f i )Cantera ⋯ ( f N )Cantera are computed using Cantera function calls. The pressure
and temperature during this part of the cycle are computed directly from Cantera function calls
and therefore are not included in equation 22. The total mass in the component volume is defined
as the sum of burnt and unburnt masses. At the beginning of the combustion process the existing
ratio of unburnt to burnt gases in the cylinder is used to set the composition of the charge mass.
At the end of the combustion process all the mass in the cylinder is converted to burnt mass and
the mass of the unburnt gas is set to zero.
V − RT 0 0 0 P − mtot R 0 − mtotT 0 0 0 ⋯ 0  P   0 
0  m   f (θ , ∆P) / ω 
 1 0 0 0 0 0 0 0 0 0 ⋮ ⋯ ⋮   tot   tot 
0 0 1 0 0 0 0 0 0 0 0 ⋮ ⋯ ⋮  mub   fub (θ , ∆P ) ω 
     
0 0 0 1 0 0 0 0 0 0 0 ⋮ ⋯ ⋮  mb   fb (θ , ∆P ) ω 
0 0 0 0 1 0 0 0 0 0 0 ⋮ ⋯ ⋮  h   fh / ω 
     
0 0 0 0 0 1 0 0 0 0 0 ⋮ ⋯ ⋮ V   V (θ )  (23)
0 CvT 0 0 −1 P mtot Cv 0 0 0 mtotT ⋮ ⋯ ⋮ d  T   0 
   = 
0 0 0 0 −1 T 0 0 1 0 0 0 ⋮ ⋯ ⋮  dθ S   0 
0    
0 0 0 0 0 0 0 1 0 0 ⋮ ⋯ ⋮ R  
f 
  R

0 0 0 0 0 0 0 0 0 1 0 ⋮ ⋯ ⋮ γ   fγ 
     
0 0 0 0 0 0 0 0 −(γ − 1) R −(γ − 1) 2 ⋮ ⋯ ⋮  Cv   0 
0 0 0 0 0 0 0 0 0 0 0 1 ⋯ ⋮ Y   f / ω 
   1   i

0 0 0 0 0 0 0 0 0 0 0 0 ⋱ ⋮  ⋮   ⋮ 
0 0 0 0 0 0 0 0 0 0 0 0 ⋯ 1  Y   fN / ω 
  N   

8
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

The matrix presented in equation 23 shows the system of equations that is solved to determine
gas properties in the cylinder volume during the non-combustion part of the cycle and in the
crankcase during all parts of the cycle. This set of ODE’s is solved using MATLAB and its
regular ode solver, ode45. V(θ) represents the derivative of the volume forcing function given in
equation 4.

Convergence Criterion: Equations 22 and 23 are integrated forward in theta (time) until the
cyclic variation in properties reaches a steady state. Steady state is assumed to be achieved when
the difference between the net mass inflow and outflow falls below 0.1%. This usually occurs
within 15 cycles. The cycle work, power, torque, efficiency, etc. are computed using the property
variations through the converged cycle.

Adjustment of volumetric efficiency: The ability of the engine to ingest air limits its
performance. This ability is described using the ‘volumetric efficiency’ [19] which is the ratio of
the mass of air in the cylinder to the mass of air in the displaced volume at atmospheric
conditions (Equation 24).
ma
ηvol = (24)
ρ aVd
The loss coefficient Cd in equation 10 is selected so that the curves of volumetric efficiency as a
function of engine speed produced by the simulation match those measured experimentally.
Figure 3 shows this comparison.

80

70
Volumetric Efficiency (%)

60

50

40

Cd = 1
30
Cd = 0.5
20 Cd = 0.2
Cd = 0.1
10 Cd = 0.06
Experimental Results
0
0 5000 10000 15000 20000 25000 30000 35000 40000
Engine Speed (RPM)

Figure 3. Comparison of volumetric efficiency predicted by the simulation to that measured


experimentally. The best correspondence is achieved when Cd=0.06.

9
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

Ignition.
The glow plug plays a key role in initiating the combustion reactions in the cylinder volume,
however the catalytic surface reactions underlying its operation are not incorporated in the
simulation. Instead, ignition is achieved by adding heat to the fuel mixture from 5 degrees after
BTDC until the gas ignites. An equivalent amount of heat is removed during the exhaust process
so as not to affect the net power produced by the engine. The problem with this approach is that
it has the effect of fixing the timing of the ignition event. Unfortunately, this is all that we can
do at the present time as the authors are not aware of an appropriate ignition model for this fuel.

Computation of engine performance from simulation results: The useful work produced over
one cycle is determined by integrating the pressure-volume traces as indicated by equation 25.
Work = ∫ ( PdV ) − ∫ ( PdV ) (25)
cylinder CrankCase

Equation 17 shows that the work required to compress the charge in the crankcase reduces the
net work output of the system. Equations 26 – 29 show how power, torque, and efficiency are
computed.
Work
Gross Power = (26)
60 / RPM
Net Power = Gross Power − Friction Power (27)
Net Power
Torque = (28)
(2 * π * RPM ) / 60
Net Power
ηthermal = (29)
mɺ f * QR

4. Experimental Setup and Test Engine Details.

Table III Major characteristics of test engine.


Wt. Cycle Stroke Bore Compression Rated Speed Displacement
(gm) (mm) (mm) Ratio (kRPM) (cm3)
150.2 2 12 15.5 5 18 2.83

The test article is the small model aircraft engine made by AP engines (Yellowjacket, 2.45 cm3,
150 g) illustrated in Figure 4. The key attributes of the engine are presented in table III. The
engine’s performance is measured using a specialized dynamometer system specifically
developed to test small engines. The details of the measurement system are presented elsewhere
[6] and only a brief synopsis is presented here.

10
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

Figure 4: Photograph of the AP Engines Yellowjacket.

Figure 5 is a photograph of the dynamometer illustrating the major components. The engine to
be tested is mounted in a cradle that is supported on precision low-breakaway torque bearings
about an axis coincident with the engine’s axis of rotation. The cradle is prevented from rotating
by a load cell that anchors it to the cradle support. The absorber is a Magtrol double hysteresis
brake. It is connected to the engine through a gear system and provides a continuously variable
load to the engine. As the load is applied, the engine reacts against the cradle causing a load to
be exerted on the load cell. The product of the load cell force and the moment arm length gives
the engine torque.

Fuel Tank Cyl. Head T Throttle Servo

Cooling Duct Scale Exhaust T

Engine

Speed Sensor Absorber Cradle Moment Arm Load Cell

Fig. 5 Photograph of dynamometer system showing the important components.

11
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

The engine is operated on ‘glow fuel’ that is a mixture of 10% nitromethane, 20% castor oil, and
70% methanol. The heating value of the mixture was measured by a commercial fuel testing
laboratory using standard procedures and found to be 21.82 MJ/kg. The fuel flow rate is
measured using a nutating microflowmeter. Two flexible hoses connect the tank to the engine on
the cradle. One carries the fuel to the needle valve inlet. The other extends from a fitting on the
muffler back to the fuel tank so that pressure in the exhaust manifold drives fuel from the tank
into the engine. The forces exerted by the hoses on the cradle do impact the torque measurement
but this is accounted for by calibrations performed before and after every run. The fuel pressure
in the tank is monitored using a differential pressure transducer. Air flow rate is measured using
a mass flowmeter. Since the flow sensor only works properly under steady flow conditions, a
plenum is placed between the flowmeter and the engine in order to damp out fluctuations. A
flexible tube connects the plenum to the carburetor in order to minimize the impact on the torque
measurements.

Engine speed is measured using an ‘off the shelf’ system. It consists of a magnetic ‘pulser’ disc
containing sixteen alternating poles that is attached to the absorber shaft, a proximity sensor that
detects the passage of these poles, and a signal conditioner that outputs an analog voltage
proportional to frequency. A model aircraft propeller nose cone attached to the end of the
absorber shaft permits the engine to be started using a conventional model aircraft starter.
Cooling air for the engine is provided by a specially constructed duct that directs air from a
blower around the moving parts of the cradle and directly across the cylinder head. The engine
will overheat rapidly without the cooling air.

The throttle is controlled remotely using a Futaba servo, FM receiver, and battery all of which
are attached to the cradle. Cylinder head temperature is measured using a stainless steel sheathed
thermocouple held in contact with one of the cylinder head bolts. A similar thermocouple
inserted into the muffler exit measures exhaust gas temperature.

5. Discussion of Results

Analysis of a single cycle


Figure 6 shows pressure-volume diagrams in the cylinder and crankcase for a single cycle at
10400 RPM and an equivalence ratio of 3. These traces are integrated per equation 18 to
determine the net work output of the cycle. The overall pressure ratio of the cycle is about 15
whereas the geometric compression ratio of the engine is 5. Figure 7 shows the cylinder pressure
as a function of crank angle. Ignition occurs at about 20 degrees after TDC. This can be seen in
Figure 8 which shows that the concentrations of the reactant molecules go to zero at
approximately 380 degrees.

12
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

5
x 10
15
Cylinder
Crankcase

Pressure (N/m )
2
10

0
0 1 2 3 4 5 6
x 10 3 -6
Volume (m )
Figure 6. Pressure-Volume diagram for a single engine cycle.
5
x 10
15
Cylinder
Crankcase
Pressure (N/m2)

10

0
100 200 300 400 500 600
Crank Angle (deg)
Figure 7. Pressure trace as a function of crank angle for a single engine cycle.

Figure 9 shows the scavenging process occurring in the cylinder at 10400 RPM. The blue curve
corresponds to the unburned gas while the red curve corresponds to the burned gas. The masses
of the components are constant during the compression process from 255 to 465 degrees. At the
end of the combustion process, the entire mass in the cylinder is re-classified as burned and a
fraction of it is displaced as fresh unburned mixture enters the cylinder. However, the figure
shows that a significant fraction of the burned gases remains in the cylinder for the next cycle
thereby diluting the fresh charge.

13
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

0.6
O2
0.5 CH3OH
CH3NO2
CO2
0.4
Mole Fraction
H2O

0.3

0.2

0.1

-0.1
250 300 350 400 450 500
Crank Angle (deg)
Figure 8. Time history of reactant and product mole fractions in the cylinder volume during the
combustion process.

-6
x 10
1.8

1.6 Unburnt gas


Burnt gas
Cylinder mass (kg)

1.4

1.2

0.8

0.6

0.4

0.2

0
100 200 300 400 500 600
Crank Angle (deg)
Figure 9. Time history of burned and unburned mass in the cylinder over a single engine cycle at
10400 RPM showing the scavenging process.
Figure 10 shows the total mass of the gas mixtures in the cylinder and crankcase volumes during
one engine cycle. It clearly shows that the crankcase fills with reactants during the compression
stroke and then rapidly discharges into the cylinder as the exhaust valve opens. Figure 11 shows
the temperature of the gases in the cylinder and crankcase during one engine cycle. The
temperature of the gas mixture rises slowly during compression until the ignition point at which
point the temperature rises rapidly to a peak value of approximately 1080K.

14
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

-6
x 10
7
Cylinder
6 Crankcase

5
Mass (kg)
4

0
100 200 300 400 500 600
Crank Angle (deg)
Figure 10. Total mass in the cylinder and crankcase volumes as a function of crank angle at 10400
RPM.

1100

1000 Cylinder
Crankcase
900
Temperature (K)

800

700

600

500

400

300

200
100 200 300 400 500 600
Crank Angle (deg)
Figure 11. Cylinder and crankcase temperature as a function of crank angle at 10400 RPM and
Φ=3.
Engine performance results and comparison to experiments
Figures 12 and 13 compare predictions of engine power and efficiency as a function of engine
speed to the values measured using the small dynamometer. The power and efficiency are
computed from the converged engine cycle using the equations 26-29 mentioned previously.
Since the equivalence ratio varies with engine speed, the equivalence ratio in the simulation is
adjusted to match the experimentally measured value at each engine speed.

15
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

The solid circles in Figure 12 shows the experimentally measured power and the open circles
show the corresponding predictions of the baseline simulation. The simulation captures the
power output of the engine at speeds below 10,000 RPM to within 15%. However, it breaks
down beyond 10,000 RPM where the simulation predicts continued increases in power output
with RPM whereas the measured power output begins to decline. While the reasons for this
deviation remain unclear, two prime suspects are the models used to represent the frictional and
thermal losses both of which were developed for much larger engines and may not be at all
appropriate for engines of this scale. The open squares in the figure show the effect of changing
the multiplier on the last term in equation 21 from 0.4 to 2.5. Similarly, the open triangles in the
figure show the effect of changing the value of C1 in equation 18 from 0.76 to 5. Neither
modification produces a peak in the power curve but both shift it significantly indicating their
important influence on the power output of the engine.
The solid circles in Figure 13 show the experimentally measured efficiency whereas the open
circles show the efficiencies predicted by the simulation. While the measurements and the
simulation results agree within 25% and show a general decrease in efficiency with increasing
engine speed, the trend is much weaker in the simulations indicating that losses do not seem to
be accounted for correctly. The open squares and triangles show the effects of the same changes
to the frictional and heat loss models applied to equation 12. As with the power results, the
effects of these changes are significant.
160
Baseline Simulation
Experimental Uncertainty
Experiment
140 Modified friction
Net power output (W)

Modified heat loss


120

100

80

60

40
7000 8000 9000 10000 11000 12000
Engine Speed (RPM)
Figure 12. Comparison of measured and predicted power output as a function of engine speed.

16
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

6
Experiment Uncertainty

5
Efficiency (%)

2 Baseline Simulation
Experiment
1 Modified friction
Modified heat loss
0
7000 8000 9000 10000 11000 12000
Engine Speed (RPM)
Figure 13. Comparison of measured and predicted overall efficiency.

6. Concluding Remarks

In spite of the care taken to simulate chemical reaction using multi-step chemistry and to
incorporate models for thermal and frictional losses based on measurements in other engines, the
simulation predictions do not match the experimental measurements well. While the reasons for
this discrepancy remain unclear, a strong contributor is undoubtedly the models used to represent
thermal and frictional losses. These models were developed for much larger engines and may
not be appropriate for engines of this small size. Another important factor whose effects were
not measured is the effect of ignition timing. In the present simulation, ignition is achieved by
adding heat at a certain point in the cycle. However, this is not how ignition is achieved in
practice. Taken together, the results demonstrate the need for models capable of accurately
representing friction, heat loss, and ignition in miniature engines with high surface/volume
ratios.

Acknowledgments
The authors would like to thank the Army Research Office for supporting this work through the MAV MURI
Program (Grant No. ARMY-W911NF0410176), Technical Monitor Dr. Thomas Doligalski.

References
[1] Hill, P. G., Peterson, C. R., Mechanics and Thermodynamics of Propulsion, Addison Wesley, 1965, p.145.
[2] Lowry, J. T., ‘Fixed-Ptch Propeller/Piston Aircraft Operations at Partial Throttle’, AIAA Journal of Propulsion
and Power, 15, 4, pp. 497-503, 1999

17
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

[3] Papac, J., and Dunn-Rankin, D., ‘Combustion in a Centimeter-Scale Four-Stroke Engine’, paper 04S-11,
Western States Section of the Combustion Institute, 2004
[4] Papac, J. and Dunn-Rankin, D., ‘In Cylinder Pressure and Combustion Measurements in a Miniature
Reciprocating Engine’, 4th Joint Meeting of the U.S. Sections of the Combustion Institute, 2005
[5] Rais-Rohani, M. and Hicks, G., ‘Multidisciplinary Design and Prototype Development of a Micro Air
Vehicle’, AIAA Journal of Aircraft, 36, 1, pp. 227-234, 1999
[6] Menon, S. and Cadou, C. P., “Development of a Dynamometer for Measuring Small Internal-Combustion
Engine Performance”, AIAA Journal of Propulsion and Power, Vol.23, No.1, Jan-Feb 200
[7] Peterson, R. B., “Size Limits for Regenerative Heat Engines”, Microscale Thermophysical Engineering, 2, pp
121-131, 1998.
[8] Cadou, C. P., “Reactive Processes in Micro-scale Combustion Systems”, Invited paper, Western States
Section of the Combustion Institute, October 20-21, Los Angeles, CA 2003
[9] H. T. Aichlmayr, D. B. Kittelson, D.B., M. R. Zachariah, Chem. Eng. Sci. 57 (19) (Oct 21 2002), 4161-4171
[10] Annen, K. D., Stickler, D. B., and Woodroffe, J., ‚Linearly-Oscillating Miniature Internal Combustion Engine
(MICE) for Portable Electric Power’, 41st Aerospace Sciences Meeting and Exhibit, Jan 6-9, 2003
[11] Ferguson, C. R., Internal Combustion Engines, John Wiley and Sons, 1986
[12] Ceviz, M.A., Kaymaz I., “Temperature and air-fuel ratio dependent specific heat ratio functions for lean
burned and unburned mixtures”, Energy Conversion and Management, Vol.46, 2005, pp 2387-2404
[13] Goodwin, D. G., www.cantera.org
[14] Aichlmayr, H.T., Kittelson, D.B. and Zachariah, M.R., “Micro-HCCI combustion: Experimental
Characterization and development of a detailed chemical kinetic model with coupled piston motion”,
Combustion and Flame, 135, 2003, pg 227-248.
[15] Heywood, J. B., Internal Combustion Engine Fundamentals, McGraw-Hill Book Company, 1988.
[16] W.J.D. Annand, “Heat transfer in the cylinders of Reciprocating Internal Combustion Engines”,
Proc.I.Mech.E., Vol.177,p193,1963.
[17] Bubert, T., and Cadou, C., “Empirical Observation of Forced Convective Cooling for Small Internal
Combustion Engines”, student paper at the AIAA student section meeting, 2006.
[18] Heywood, J. B., Internal Combustion Engine Fundamentals, McGraw-Hill Book Company, 1988, pg.724.
[19] Heywood, J. B., Internal Combustion Engine Fundamentals, McGraw-Hill Book Company, 1988, pg.54.

Appendix A.

Empirical expressions for ratio of specific heats from Ceviz and Kaymaz.

a7
γ u = a1 + a2T + a3T 2 + a4T 3 + a5T 4 + a6T 5 +
φ
b3 b5 b6T b8 b9T b10T 2
γ b = b1 + b2T + + b4T +
2
+ + b7T +
3
+ +
φ φ2 φ φ3 φ2 φ

Table IV Empirical coefficients for Ceviz and Kaymaz model.


Coefficients Values Coefficients Values
a1 1.4642 b1 1.4981

a2 -0.0001 b2 -0.0001

a3 -7.3485e-8 b3 -0.2668

a4 1.5572e-10 b4 4.0364e-8

18
th
5 US Combustion Meeting – Paper # E24 Topic: Engine

a5 -7.6951e-14 b5 0.2734

a6 1.1953e-17 b6 5.7462e-5

a7 -0.0631 b7 -7.2026e-12

b8 -0.0821

b9 -1.3029e-5

b10 2.3573e-8

19

Vous aimerez peut-être aussi