Applied Energy 159 (2015) 117131

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

High load performance and combustion analysis of a four-valve direct

injection gasoline engine running in the two-stroke cycle
Macklini Dalla Nora , Hua Zhao
Brunel University London, Centre for Advanced Powertrain and Fuels Research (CAPF), Kingston Lane, Uxbridge, Middlesex UB8 3PH, United Kingdom

h i g h l i g h t s

The two-stroke cycle was achieved in a four-valve highly boosted gasoline engine.

High charging efficiencies were realised by means of valve timing optimisation.

The engine achieved 2.4 MPa equivalent IMEP with 7 MPa in-cylinder pressure.

Scavenging inefficiencies and poor airfuel mixing limited the high speed operation.

a r t i c l e i n f o a b s t r a c t

Article history: With the introduction of CO2 emissions legislation or fuel economy standards in Europe and many coun-
Received 25 March 2015 tries, significant effort is being made to improve spark ignition gasoline engines because of their domi-
Received in revised form 23 August 2015 nant market share in passenger cars and potential for better fuel economy. Amongst several
Accepted 27 August 2015
approaches, the engine downsizing technology has been adopted by the automotive companies as one
of the most effective methods to reduce fuel consumption of gasoline engines. However, aggressive
engine downsizing is constrained by excessive thermal and mechanical loads as well as knocking com-
Keywords:
bustion and low speed pre-ignition (also known as super-knock). In order to overcome such difficulties,
Two-stroke poppet valve engine
Gasoline direct injection
a gasoline direct injection single cylinder engine was modified to run under the two-stroke cycle by oper-
Engine downsizing ating the intake and exhaust valves around bottom dead centre (BDC) at every crankshaft revolution. The
Controlled auto-ignition combustion products were scavenged by means of a reversed tumble flow of compressed air during the
positive valve overlap period at BDC. The engine output was determined by the charging and trapping
efficiencies, which were directly influenced by the intake and exhaust valve timings and boost pressures.
In this research a valve timing optimisation study was performed using a fully flexible valve train unit,
where the intake and exhaust valve timings were advanced and retarded independently at several speeds
and loads. A supercharger was used to vary the load by increasing the intake pressure. The effects of valve
timing and boost pressure in this two-stroke poppet valve engine were investigated by a detailed analysis
of the gas exchange process and combustion heat release. Gaseous and smoke emissions were measured
and analysed. The results confirmed that the two-stroke cycle operation enabled the indicated mean
effective pressure to reach 1.2 MPa (equivalent to 2.4 MPa in a four-stroke cycle) with an in-cylinder
pressure below 7 MPa at an engine speed as low as 800 rpm. The engine operation was limited by scav-
enging inefficiencies and short time available for proper airfuel mixing at high speeds using the current
fuel injector. The large amounts of hot residual gas trapped induced controlled auto-ignition combustion
at high speeds, and thus the abrupt heat release limited higher loads.
2015 Elsevier Ltd. All rights reserved.

Abbreviations: ATDC, after top dead centre; BSFC, brake specific fuel consumption; CA, crank angle; CAI, controlled auto-ignition; CO, carbon monoxide; COVIMEP,
covariance of the indicated mean effective pressure; dP/dh, rate of pressure rise; ECR, effective compression ratio; EER, effective expansion ratio; EGR, exhaust gas recycling;
EVC, exhaust valve closing; EVO, exhaust valve opening; GDI, gasoline direct injection; UHC, unburned hydrocarbon; IMEP, indicated mean effective pressure; ISCO, indicated
specific carbon monoxide; ISFC, indicated specific fuel consumption; ISUHC, indicated specific unburned hydrocarbon; ISNOx, indicated specific oxides of nitrogen; IVC,
intake valve closing; IVO, intake valve opening; KLS, knock limited spark advance; LHV, lower heating value; MBT, minimum spark advance for maximum break torque; NOx,
oxides of nitrogen; rpm, revolutions per minute; SI, spark ignition.
Corresponding author. Tel.: +44 (0)74630 95392.
E-mail address: macklini.dallanora@brunel.ac.uk (M. Dalla Nora).

http://dx.doi.org/10.1016/j.apenergy.2015.08.122
0306-2619/ 2015 Elsevier Ltd. All rights reserved.
118 M. Dalla Nora, H. Zhao / Applied Energy 159 (2015) 117131
1. Introduction
Two-stroke engines are well known for their superior power
density and reduced weight compared to equivalent four-stroke
units and are employed to power handheld tools to large marine
engines [1,2]. Their use for high performance purposes is widely
spread for motorbikes, snowmobiles and outboard vehicles, with
claimed power densities above 220 kW/L [3]. However, these
advantages, mainly related to crank-case scavenged two-stroke
engines, are often offset by drawbacks regarding gaseous emis-
sions, thermal efficiency and engine components durability [4].
On the subject of emissions, the fuel short-circuiting in mixture
scavenged two-stroke engines results in significant unburned
hydrocarbon (UHC) emissions. The lubricant added to the fuel
has much less effect on emissions from crank-case scavenged
two-stroke engines according to [3], as modern units use propor-
tions as low as 1% of oil in the fuel. Regarding the thermal effi-
ciency, conventional two-stroke engines usually lose expansion
work in favour of enhanced scavenging through early exhaust port
opening. This procedure uses the exhaust blow-down phase to
reduce the levels of residual gas trapped prior to the intake pro-
cess, ensuring higher degrees of charge purity [5]. Lastly, the
reduced components durability (piston, rings and liner) of ported
two-stroke engines can be attributed to uneven thermal loads
and reduced lubricant oil film when UHC emissions is a concern
[6]. It is important to keep in mind that all these disadvantages
are related to cross-scavenged and loop-scavenged two-stroke
engines with intake and exhaust ports, where the crank-case is
employed as a pump for the air or air/fuel mixture and therefore
lubricant oil needs to be added to the air stream. Such problems
can be avoided by the uniflow two-stroke engine concept, in which
externally compressed air is supplied through ports at bottom dead
centre (BDC) and the exhaust gas is forced out through conven-
tional poppet valves in the cylinder head. Greater charging efficien-
cies can be achieved with such designs [1], but production
complexity and packaging restrictions have limited its application
to large marine diesel engines so far, though some attempts have
been made to adopt such an engine design for vehicular applica-
tions [7].
In the beginning of 1990 a new concept of two-stroke operation
was proposed as a possible solution to overcome the problems
related to conventional ported two-stroke engines. Based on the
design of modern four-stroke engines, the two-stroke scavenging
process was achieved through the overlap period of overhead
intake and exhaust valves around BDC at every engine revolution
[2,8]. Because of the use of poppet valves higher power outputs
could be achieved with the same engine durability as four-stroke
engines. The high levels of UHC emissions due to fuel short-
circuiting had been addressed by direct fuel injection and air-
assisted fuel injection [5]. The lubricant oil consumption, charac-
teristic of crank-case scavenged engines, had been eliminated by
using wet sump and external scavenge pump, mostly roots blower
superchargers. When applying this concept to Diesel engines, 40%
higher torque and reduced combustion noise compared to an
equivalent four-stroke model was demonstrated by Toyota [6].
A reported problem of gasoline direct injection (GDI) two-
stroke poppet valve engines was the insufficient mixing between
fuel and air in the cylinder, mainly attributed to the shorter time
available and the relatively lower injection pressures used at the
time [8]. This poor airfuel mixing results in stratified charging
and incomplete combustion, with large emissions of carbon
monoxide (CO), UHC and soot [9]. The boosted air supplier, essen-
tial for this type of two-stroke engine, has been also reported as
responsible for hindering fuel consumption due to isentropic and
mechanical inefficiencies [10]. Nevertheless, significant advances
have been made in high pressure fuel direct injection systems
and high efficiency boosting devices (superchargers, turbochargers
and e-boosters). In addition, flexible variable valve actuation sys-
tems, particularly fast variable cam devices, have been developed
to enhance air management for production engines [11]. Such
technological improvements have prompted renewed interest in
developing two-stroke poppet valve gasoline engines [12,13] and
diesel engines [10,14], considering their potential for engine down-
sizing with lower in-cylinder pressures and less structural stresses
than downsized four-stroke engines [15,16]. Furthermore, the two-
stroke poppet valve engine shares nearly all components from the
contemporary four-stroke engine architecture and hence can be
produced from the same manufacturing process.
The majority of studies with two-stroke poppet valve engines
has taken advantage of the inherent residual gas trapped at
reduced charging efficiencies to achieve high in-cylinder thermal
conditions, and hence induce auto-ignition combustion in both
spark-ignition [12,13,17] and compression-ignition engines
[18,19]. In the auto-ignition combustion high thermal efficiency
and near zero oxides of nitrogen (NOx) emissions can be achieved
across the part load engine operation range. However, the low
charging efficiency, usually 70% at most, results in large amounts
of hot residual gas trapped and increased combustion noise at high
loads. The violent combustion in this case limits the engine load to
values of indicated mean effective pressure (IMEP) from 0.5 to
0.7 MPa using commercial gasoline.
Considering this, the present study aims at investigating the
two-stroke poppet valve engine behaviour under charging efficien-
cies above 90% at low speeds, showing the full potential of this
type of engine for aggressive downsizing and downspeeding com-
pared to four-stroke engines. Such level of charging efficiency
could be only achieved with high values of boost pressure and a
valve timing optimisation study, where the intake and exhaust
valves were operated independently aiming at reducing the levels
of residual gas trapped to a minimum. The effect of the scavenging
process on the engine performance was evaluated at several
speeds and loads, whilst measurements of gaseous and smoke
emissions, as well as a heat release analysis, were performed to
study the airfuel mixing and combustion process. High values of
IMEP with low in-cylinder pressures are expected in this two-
stroke direct injection gasoline poppet valve engine.
2. Experiments
2.1. Experimental facilities
All the experiments were conducted on a single cylinder
research engine mounted on a transient test bed. The engine was
equipped with an electro-hydraulic fully variable valve train unit
capable of independent control over the timings and lifts of each
of the four valves, enabling both two-stroke and four-stroke cycles
operation [20]. The valve control unit operated under closed loop
control over the oil pressure, oil temperature and valve timing/lift,
ensuring precise valve operation up to 3000 rpm in the two-stroke
cycle. A Ricardo rCube engine control unit was used to manage the
throttle position, injection pulse width, spark timing and valve
parameters. An AC dynamometer enabled both motored and fired
operations whilst an external cooling system provided fully auto-
mated control over the engine oil and coolant temperatures. Com-
mercial gasoline (95 RON) was directly injected into the
combustion chamber through a Denso solenoid double-slit type
injector [21]. The instantaneous fuel mass flow rate was measured
 M. Dalla Nora, H. Zhao / Applied Energy 159 (2015) 117131 119

by an Endress + Hauser Promass 83A Coriolis flow meter, with a

maximum error of 0.2% for the flow range studied. The boosted
air was supplied by an AVL 515 sliding vanes compressor unit with
closed loop control over the pressure. The air mass flow rate was
measured by a Hasting HFM-200 laminar flow meter with a max-
imum error of 1%. The intake and exhaust pressures were mea-
sured by two Kistler piezo-resistive transducers installed in the
intake plenum (4007BA20F) and in the exhaust port (4007BA5F),
with a maximum error of 0.1%. The in-cylinder pressure was mea-
sured by a Kistler 6061B piezo-electric sensor, with a maximum
measurement error of 0.5%. To record the crank angle position a
LeineLinde incremental encoder with a resolution of 720 pulses
per revolution was employed. Averaged temperatures were mea-
sured at the intake plenum, exhaust runner, oil gallery, coolant
jacket and fuel rail by using K-type thermocouples with an accu-
racy of 1%. An AVL 415SE smoke meter was used to measure
the smoke levels, with repeatability better than 3% of the measured
Fig. 2. Cylinder head details.
values. Gaseous emissions were analysed by a Horiba MEXA
7170DEGR using the non-dispersive infrared principle for CO, a
heated flame ionization detector for UHC, a paramagnetic detector intake ports, as shown by a small recession in the engine geometry.
for oxygen (O2) and a heated chemiluminescence detector for NOx. The unconventional intake ports arrangement aimed at improving
The overall error attained to each gas measurement was smaller the reverse tumble flow necessary to scavenge the burned gases,
than 2%. The location of all instruments described above can be while keeping acceptable values of discharge coefficients.
found in Fig. 1, as well as the temperature and pressure measure- The reverse tumble flow was improved by masking the cylinder
ment points labelled as T and P, respectively. head region around both intake valves, so the intake charge was
A National Instruments 6353 USB X card was used for data prevented from going straight into the exhaust system in the so
acquisition (DAQ) and an in-house software was employed for called air short-circuiting. The majority of the flow was then direc-
combustion analysis. ted downwards in the cylinder, as seen in Fig. 3, and the scaveng-
ing process could be improved.
2.2. Engine details
2.3. Test procedures
The prototype engine had an 81.6 mm bore and 66.9 mm stroke,
which resulted in a swept volume of 350 cm3 and an oversquared The two-stroke cycle was achieved by opening both the intake
bore-to-stroke ratio of 1.22. A geometrical compression ratio of and exhaust valves around BDC as presented in Fig. 4. The long
11.8:1 was achieved with a dome-in-piston and a pent roof com- valve overlap period allowed the inlet boosted air to scavenge
bustion chamber. The intake valves were 28 mm in diameter while the combustion products. The start of fuel injection (SOI) occurred
the exhaust valves were 30 mm, which is peculiar compared to after all the valves were closed to avoid fuel short-circuiting to the
conventional four-stroke engines where the intake valves are usu- exhaust. In addition, the SOI after intake valve closing (IVC) pre-
ally bigger. In the two-stroke cycle, although, the scavenging pro- vented fuel from entering into the intake ports through backflow,
cess at high engine speeds is more dependent on the exhaust which may occur if the in-cylinder pressure becomes higher than
flow characteristics and hence the exhaust flow area should be the intake port pressure. The fuel entrained in the intake port could
greater as seen in ported two-stroke engines. The engine had two then be carried back into the cylinder and pass directly to the
conventional side exhaust ports and two individual upright- exhaust port in the following cycle, contributing to increased
straight intake ports shown in blue and green, respectively, in UHC emissions.
Fig. 2. The spark plug was located in the centre of the combustion At each of the engine speeds studied, i.e. 800, 1500, 2200 and
chamber while the fuel injector was side mounted between the 3000 5 rpm, five intake pressure levels were applied (where

Fig. 1. Research engine and test cell facilities.

120 M. Dalla Nora, H. Zhao / Applied Energy 159 (2015) 117131

4 IVO 130 IVO 135

IVO 140 IVO 145
IVO 150 EVO 130
3

Valve lift (mm)

2

0
110 150 190 230 270
Crank angle ()

Fig. 5. Intake valve timing optimisation.

Fig. 3. Cross section of cylinder head at the valves plane.

EVO 120 EVO 125

4
possible) as a way to control the engine load. By increasing the EVO 130 EVO 135
boost pressures from 120 2 kPa to 280 3 kPa the delivery ratio EVO 140 IVO 140
increased and less residual gas was trapped, resulting in greater 3

Valve lift (mm)

air mass in the cylinder and higher engine power output. At some
operation points stable combustion was not achieved as the
covariance of the indicated mean effective pressure (COVIMEP)
reached a limit of 10%. This value seems high for four-stroke
engines where a value around 5% is usually considered [22].
However, bearing in mind the doubled firing frequency of
two-stroke engines the torque variation is reduced and the levels
of vibration and harshness are attenuated. In a previous study,
stable operation in a two-stroke poppet valve engine was claimed
at COVIMEP values as high as 35% [5].
At each engine speed, nine different combinations of intake and
exhaust opening/closing timings were tested as shown in Figs. 5
and 6. The intake and exhaust valve durations were kept constant
at 100 CA (crank angle) and 120 CA, respectively. At each engine
speed and a given boost pressure, the exhaust valve timing was
kept fixed and firstly the intake valve opening (IVO) was varied
from 130 CA to 150 CA after top dead centre (ATDC), in steps of
5 CA. Then, the intake valve timing was fixed and the exhaust
valve opening (EVO) was varied from 120 CA to 140 CA ATDC,
also in steps of 5 CA.
In the previous research at part-load conditions [23] the
exhaust valve closing (EVC) took place before the IVC to increase
the residual gas trapped for controlled auto-ignition (CAI) combus-
tion. In this study, however, the EVC was delayed to obtain higher
charging efficiencies. In addition, the exhaust valve opened earlier
during the expansion process to increase the exhaust blow-down
period. During the experiments the maximum advance of IVO
was set to the EVO.
The engine coolant and oil temperatures were kept at 353 3 K
for all cases studied. The intake air temperature was in the range
4 Intake valves Exhaust valves
Valve lift (mm)
2
1
0
110 150 190 230 270
Crank angle ()
Fig. 6. Exhaust valve timing optimisation.
from 295 K to 305 K, except for the maximum intake pressure of
280 kPa at 800 rpm when it reached 325 K due to insufficient air
cooling.
The ignition timing was set to minimum spark advance for
maximum brake torque (MBT) or knock limited spark (KLS) at con-
ditions when knocking combustion occurred. A knocking combus-
tion threshold of 1 MPa/CA was set for the maximum rate of
pressure rise (dP/dh).
To ensure the same airfuel mixing conditions for all the valve
timings studied the SOI was set to 260 CA ATDC, which was the
latest EVC timing tested. The fuel injection pressure was set to
15.0 0.5 MPa and its temperature kept at 293 5 K.
At each intake pressure and engine speed tested, the fuelling
rate was increased until a fuel rich in-cylinder charge was
obtained. This fuelling rate was then kept constant as the intake
and exhaust valve timings were varied, so the effect of the gas
exchanging process could be solely evaluated. Furthermore, as
the engine speed changed, the fuel flow rate was also varied
accordingly to ensure a fuel rich in-cylinder charge as seen in
Table 1. The reason why lean mixtures could not be employed
relies on the method used in this research to calculate the gas

3
exchanging parameters, particularly the air trapping efficiency.
2 By evaluating the exhaust gas composition, the technique reported
SOI
by [4,24] considers that all free oxygen in the exhaust stream is
1
resulted from inefficiencies during the scavenging process. In this
case a sufficiently rich mixture was required during the combus-
tion process to ensure the minimum possible oxygen remaining
0
0 90 180 270 360 within the cylinder at EVO, otherwise the air trapping efficiency
Crank angle ( ) would be underestimated. The equation used to calculate the trap-
ping efficiency based on this presumption is presented in the fol-
Fig. 4. Two-stroke cycle operation principle. lowing section.
 M. Dalla Nora, H. Zhao / Applied Energy 159 (2015) 117131 121

Table 1 0:5yCOCO CO2 

Fuel flow rates at different engine speeds and intake pressures. H2  6
CO 3CO2 
Fuel flow rate (mg/cycle) Intake pressure (kPa)
where [H2] is the exhaust concentration of hydrogen, y is the hydro-
120 160 200 240 280 gen to carbon ratio of the fuel (considered 1.87), [CO] is the exhaust
Engine speed (rpm) 800 17.1 20.4 24.8 27.1 30.8 concentration of carbon monoxide and [CO2] is the exhaust concen-
1500 12.3 20.5 25.5 34.3 tration of carbon dioxide. All exhaust gas concentrations in ppm
2200 8.86 16.8
(volumetric basis).
3000 6.01
The air trapping efficiency, defined as the ratio of in-cylinder
trapped air mass to the total intake air mass, was calculated based
on the analytical method developed for fuel rich combustion in
2.4. Data analysis two-stroke engines [24]:

Based on the in-cylinder pressure and crank-angle measure-
ments, the mass fraction burnt was calculated by integrating the
net heat release rate as presented in [25]:
dQ net c dV 1 dp
p V
dh c  1 dh c  1 dh
where Qnet is the net heat release in J/CA, h is the crank angle in CA,
c is the ratio of specific heats (considered constant and equal to
1.33), p is the in-cylinder pressure in Pa and V is the in-cylinder vol-
ume in m3.
The effective compression and expansion ratios were calculated
from the instantaneous in-cylinder volumes at EVC and EVO,
respectively.
V EVO V clr
EER
V clr
V EVC V clr
ECR
V clr
where EER is the effective expansion ratio, ECR is the effective com-
pression ratio, VEVO is the instantaneous in-cylinder volume at EVO,
VEVC is the instantaneous in-cylinder volume at EVC and Vclr is the
clearance volume (32.4 cm3).
Exhaust emissions were converted from parts per million (ppm)
to g/kW h based on the UN Regulation number 49 [26]:
ugas cgas kh q_ exh
ISgas
Pi
where ISgas is the indicated specific gas emission (ISCO, ISUHC or
ISNOx) in g/kW h, ugas is the dimensionless specific gas constant
(CO = 0.000966, UHC = 0.000499 and NOx = 0.001587) for gasoline
fuelled engines, cgas is the gas concentration in the exhaust stream
in ppm (volumetric basis), kh is the dimensionless correction factor
to convert gaseous measurements of CO and NOx from a dry basis to
wet basis (see Ref. [26] for more details), q_ exh is the exhaust mas
flow rate (g/h) and Pi is the indicated power output (kW).
The combustion efficiency was calculated based on the emis-
sions products not fully oxidized during the combustion. The CO,
UHC and H2 mass flow rates were acquired by multiplying their
respective indicated specific emissions by the indicated power, so
the values in g/h could be obtained.
m _ UHC LHVUHC m
_ CO LHVCO m _ H2 LHVH2
gc 1  _ fuel LHVfuel
m
where gc is the combustion efficiency, m _ CO is the mass flow rate of
CO, LHVCO is the lower heating value (LHV) of CO (10.1 MJ/kg), m _ UHC
is the mass flow rate of UHC, LHVUHC is the LHV of UHC (44 MJ/kg),
m_ H2 is the mass flow rate of H2, LHVH2 is the LHV of H2 (120 MJ/kg),
m_ fuel is the fuel mass flow rate and LHVfuel is the LHV of the fuel
(44 MJ/kg).
Emissions of hydrogen (H2) were estimated based on the mea-
surements of CO and CO2 according to [27]:
Trapeff A
 
yKCO2 
0:5CO CO2  0:25 COKCO 
CO CO 2  0:5NOx
 2

yKCO2 
0:5CO CO2  O2  0:25 COKCO CO CO2  0:5NOx
2
7
1
where Trapeff(A) is the air trapping efficiency, [CO] is the exhaust
concentration of carbon monoxide, [CO2] is the exhaust concentra-
tion of carbon dioxide, y is hydrogen to carbon ratio of the fuel (con-
sidered 1.87), K is the watergas reaction equilibrium constant
(considered 3.5), [NOx] is the exhaust concentration of oxides of
nitrogen and [O2] is the exhaust concentration of oxygen. All
exhaust gas concentrations in ppm (volumetric basis).
Due to the air short-circuiting from the intake to the exhaust
during the valve overlap period, the measured exhaust lambda
2 value differed from the in-cylinder lambda. The in-cylinder lambda
was then calculated based on the air trapping efficiency and fuel
trapping efficiency [24]:
3
Trapeff A
kin-cylinder kexhaust 8
Trapeff F
where kin-cylinder is the in-cylinder lambda, kexhaust is the exhaust
lambda, Trapeff(A) is the air trapping efficiency and Trapeff(F) is the
fuel trapping efficiency.
The fuel trapping efficiency (defined as the ratio of in-cylinder
trapped fuel mass to the total injected fuel mass) in a GDI engine
is expected to be 100%, where no fuel short-circuiting is supposed
4 to happen. However, due to the high levels of fuel stratification
resulted from the short time available for airfuel mixing at high
speeds and loads, some of the fuel could not take part in the
combustion process and left the cylinder unburned. Thus, the fuel
trapping efficiency was introduced to take into account the
short-circuited fuel from the previous cycle, similar to that used
in conventional ported two-stroke engines [24]:
CO CO2 
Trapeff F 9
CO CO2  UHC
where Trapeff(F) is the fuel trapping efficiency, [CO] is the exhaust
concentration of carbon monoxide, [CO2] is the exhaust concentra-
tion of carbon dioxide and [UHC] is the exhaust concentration of
unburned hydrocarbons. All exhaust gas concentrations in ppm
(volumetric basis).
The charging efficiency, described as the ratio of delivered air
5 mass retained in the cylinder charge to the total in-cylinder charge
at intake conditions, was used to indicate how efficiently the
burned gases were displaced during the scavenging process. It
can be calculated based on the air trapping efficiency and delivery
ratio, as follows:
 
mair
Charg eff Trapeff A 10
Swv ol qair
where Chargeff is the charging efficiency, Trapeff(A) is the air trapping
efficiency, mair is the delivered air mass per cycle in kg, Swvol is the
122 M. Dalla Nora, H. Zhao / Applied Energy 159 (2015) 117131

engine swept volume in m3 and qair is the air density at intake con- When the fuelling rate was reduced to avoid excessive heat release
ditions in kg/m3. The term between brackets in Eq. (10) is the deliv- rate at higher boost pressures, unstable combustion occurred as
ery ratio, which compares the current delivered air mass per cycle measured by the higher COVIMEP values. On the other hand, when
to the reference air mass in an ideal charging process. the fuelling rate was increased to avoid combustion instabilities,
the dP/dh rose above the knock limit. The occurrence of violent
combustion or unstable combustion was likely related to the large
3. Results and discussion
amount of hot residual gas trapped, resulted from insufficient time
available for scavenging at higher engine speeds [28]. The presence
The results presented here are averaged over 100 consecutive
of hot residual gas raised the charge temperature and accelerated
cycles and plotted as a function of valve timings at given engine
the occurrence of auto-ignition combustion in the unburned mix-
speeds and intake pressures. The nomenclature of the different
ture, resulting in rapid and violent heat release rate. In addition,
valve timings studied consists of the IVO and the EVO timings in
since the SOI took place as late as 260 CA ATDC (similar to that
CA ATDC. The Y-axis is further divided into four parts according
used in stratified charge four-stroke GDI engines), significant fuel
to the engine speed.
stratification was present resulting in increased mixture reactivity.
At 800 rpm all the boosting levels could be tested throughout
3.1. Performance and combustion analysis the valve timings studied except for the latest IVO (150 CA) and
earliest EVO (120 CA), when combustion became excessively
Fig. 7 shows the indicated mean effective pressure (IMEP) unstable. From the left to the middle point along the x-axis the
results at different engine speeds and boost pressures. It is noted IVO was retarded from 130 to 150 CA ATDC at a constant EVO
that highly boosted operation was not possible at higher speeds of 130 CA ATDC. At the lowest boost pressure of 120 kPa the IMEP
(2200 rpm and 3000 rpm) due to violent and unstable combustion. values varied little with IVO. When the boost pressure was higher
than 160 kPa, the IMEP increased with the retarded IVO and
reached its peak at IVO 150 CA ATDC. It is noted that the higher
0.26 3000 RPM the boost pressure the more pronounced is the change in IMEP
with IVO. This can be explained by an increase in the charging effi-
ciency as presented in Fig. 8, resulted from higher pressure differ-
0.23
ence between the intake and exhaust ports. When the IVO was
retarded, a more effective blow-down event without intake air
0.20 contamination was allowed. Such effect would be even more pro-
nounced at higher boost pressures. At 1500 rpm the IVO and EVO
2200 RPM
0.59 sweeps had similar effects on the IMEP, but no stable combustion
could be achieved at the boost pressure of 280 kPa.
0.47 From the right to the middle along the x-axis in Figs. 7 and 8,
when the EVO was advanced from 140 to 120 CA ATDC and the
0.35 IVO kept at 140 CA ATDC, the charging efficiency (and therefore
the IMEP) changed little at lower boost pressures but rose steadily
0.23 to reach its peak at the middle of the graph. This behaviour mir-
rored the left part of the curve and can be explained by the
1500 RPM increased blow-down period and higher pressure ratio across the
IMEP (MPa)

1.20
exhaust valves at an earlier EVO. In addition, the difference
between the intake air pressure and the in-cylinder burned gases
1.00
was greater at the same IVO as the in-cylinder pressure had
dropped to a lower value due to extended exhaust blow-down.
0.80
At 800 rpm the peak IMEP of 1.2 MPa was achieved at an intake
pressure of 280 kPa, producing a specific torque of 195 N m/L with
0.60
the in-cylinder peak pressure as low as 6.8 MPa. To produce the
same torque at the same speed in a four-stroke engine of the same
0.40
displacement, the engine would need to be operated at 2.4 MPa
800 RPM IMEP. The in-cylinder pressure in this case would be expected to
1.23
be nearly twice as high (13.6 MPa), resulting in great structural
1.03
stresses and thermal load issues. Such operation condition could
only be achieved in a highly downsized engine, assuming the
0.83 engine would not be limited by knocking combustion and/or low
speed pre-ignition (LSPI) inducing super-knock [29]. This high
0.63 value of torque at low speeds is comparable to extremely boosted
modern diesel engines under the concept of downspeeding, where
0.43 120 kPa 160 kPa 200 kPa the engine operation region is shifted towards lower speeds with
240 kPa 280 kPa reduced friction and gas exchange losses [30]. As the engine speed
0.23 increased, although, the charging efficiency dropped and therefore
IVO 130, EVO 130

IVO 135, EVO 130

IVO 140, EVO 130

IVO 145, EVO 130

IVO 150, EVO 130

IVO 140, EVO 120

IVO 140, EVO 125

IVO 140, EVO 135

IVO 140, EVO 140

lower loads could be realised at reduced values of charge purity. At

2200 rpm, for instance, the maximum specific torque dropped to
about 88 N m/L and resulted in a modest specific power output
of about 20 kW/L. This result is still less than half of that found
in an equivalent downsized four-stroke engine operating at the
same speed, where specific power values around 54 kW/L are
Fig. 7. Indicated mean effective pressure. reported [31]. The linear trend of specific torque and in-cylinder
 M. Dalla Nora, H. Zhao / Applied Energy 159 (2015) 117131 123

20 250 7.5
3000 RPM

In-cylinder pressure (MPa)

18 200 6.0

Specific torque (Nm/litre)

16
150 4.5

44
2200 RPM
100 3.0
36

28 50 1.5
Specific torque

20
In-cylinder pressure
0 0.0
1500 RPM 500 1250 2000 2750 3500
90
Engine speed (rpm)
efficiency (%)
Charging

75 Fig. 9. Maximum specific torque and corresponding in-cylinder pressure achieved

at the speeds tested.

60

It can be seen from the PV diagram that the largest amount of

45 useful work was achieved with the earliest EVO (IVO 140, EVO 120)
and the latest IVO (IVO 150, EVO 130), when the scavenging pro-
30 cess was optimised and less residual gas was trapped. As the valve
timing was moved towards IVO 130, EVO 130, the greater charge
95 800 RPM dilution promoted by the internal EGR reduced the heat release
rate and hence the peak pressure. It can be seen that in this case
80
as the intake and exhaust valves opened at the same time, part
of the burned gases mixed with the intake charge and thus com-
65
promised the in-cylinder charge purity during the next cycle
50 [32]. The valve timing IVO 140, EVO 140 was characterised with
even lower in-cylinder peak pressure as a result of greater amounts
35 120 kPa 160 kPa 200 kPa of residual gas trapped, as shown by the lower charging efficiency
240 kPa 280 kPa (Fig. 8). As presented by the zoomed part of the PV diagram in
20 Fig. 7, in this case the EVO was the most retarded and the
IVO 135, EVO 130

IVO 140, EVO 130

IVO 145, EVO 130

IVO 150, EVO 130

IVO 140, EVO 120

IVO 140, EVO 125

IVO 140, EVO 135

IVO 140, EVO 140

IVO 130, EVO 130

expansion loop was the longest amongst those shown. These

two extreme valve timings also showed the highest in-cylinder
pressures around BDC, which caused the poor scavenging as the
pressure drop across the intake valves decreased. Moreover, the
in-cylinder pressure at the end of the compression phase for these
two cases was about 50% lower than that for IVO 150, EVO 130
Fig. 8. Charging efficiency. and IVO 140, EVO 120, resulted from less trapped fresh air mass
and higher levels of residual gas with larger heat capacity.
The two valve timings with the highest in-cylinder pressures in
pressure with the engine speed is presented in Fig. 9, for the Fig. 10, i.e.: IVO 150, EVO 130 and IVO 140, EVO 120, presented
selected valve timing IVO 145, EVO 130. similar peak pressures (less than 4% of difference), although the
The results in Fig. 8 illustrate that the maximum IMEP was a early EVO case had reduced useful work and hence 2% lower IMEP.
direct result of the most completed scavenging process achieved At this speed it is possible to confirm that the exhaust blown-down
at the latest IVO (IVO 150, EVO 130) and earliest EVO (IVO 140, phase can be partially replaced by a later EVO (130) with
EVO 120). Because the fuelling rate was kept constant at a given improved expansion work, without compromising the purity of
intake pressure, it would have been possible to achieve even higher the charge. For this two valve timings the difference in charging
engine power outputs by increasing the fuelling rate at these valve efficiency was less than 0.5% (Fig. 8), whilst the IMEP increased
timings at 800 rpm. However, it would have been at the expense of by 2% with later EVO (Fig. 7).
poorer combustion efficiency and higher fuel consumption. At any The gas exchange process in this two-stroke poppet valve
given IVO and EVO timings the charging efficiency dropped with engine was also affected by the actuation speed of the hydraulic
the increased engine speed because of the reduced time available valve train. As shown in Fig. 11, the valve opening and closing
for gas exchanging. For instance, at 2200 rpm and 120 kPa the slopes became less steep as the engine speed increased, resulting
residual gas level was found around 75%, whilst at 3000 rpm it in reduced effective flow area. Such limitation of the camless sys-
reached 82%. Furthermore, at each engine speed the charging effi- tem can be overcome by using a conventional camshaft of higher
ciency decreased from the middle to the both sides of the x-axis, lift driven by the crankshaft at the same speed.
reaching a minimum when the valves opened at the same time, i. Whilst the charging efficiency measured the effectiveness of the
e. IVO 130, EVO 130 and IVO 140, EVO 140. In order to better removal of burned gases, the air trapping efficiency was calculated
understand the scavenging results, the pressurevolume (PV) dia- to determine the air short-circuiting rate. As shown in Fig. 12, the
grams of four valve timings at 800 rpm and 200 kPa are plotted in trapping efficiency rose steadily with the engine speed as a result
Fig. 10. of shorter time available for gas exchanging and consequent lower
124 M. Dalla Nora, H. Zhao / Applied Energy 159 (2015) 117131

6.4 IVO 130, EVO 130 0.6

IVO 150, EVO 130
5.6 IVO 140, EVO 120 0.5

In-cylinder pressure (MPa)

IVO 140, EVO 140
4.8 0.4

4.0 0.3

3.2 0.2

2.4 0.1
270 290 310 330 350
1.6

0.8

0.0
25 50 75 100 125 150 175 200 225 250 275 300 325 350
In-cylinder volume (cm3)

Fig. 10. Pressurevolume diagram for selected valve timings at 800 rpm and 200 kPa intake pressure.

800 rpm 1500 rpm

4
2200 rpm 3000 rpm 3000 RPM
72
3
Valve lift (mm)

66

2
60

1
2200 RPM
68

0 64
110 140 170 200 230 260
Crank angle () 60

Fig. 11. Effect of engine speed on valve opening and closing durations. 56

70 1500 RPM
air short-circuiting rate. Higher trapping efficiencies were found
efficiency (%)
Air trapping

for earlier EVO and hence earlier EVC, particularly at 2200 rpm 55
and 3000 rpm, when the overlap period was reduced.
It is noted that when the intake air pressure was set at 120 kPa 40
the air trapping efficiency at 800 rpm and 1500 rpm exhibited dif-
ferent trends from the other pressures. This different pattern may
25
be attributed to a transition from a displacement dominated scav-
enging process to a mixing dominated scavenging process, as ide-
10
alised by the Benson-Brandham two-part model for gas
exchanging in two-stroke engines [33]. According to this theory 120 kPa 160 kPa 800 RPM
50
the scavenging was firstly dominated by a displacement process 200 kPa 240 kPa
until it reached a certain value of delivery ratio, which in this case 280 kPa
42
was around 1.5 at 800 rpm and 0.6 at 1500 rpm. After this point
the fresh air and the burned gases were more prone to mix until 34
the end of the scavenging process.
The combustion duration, calculated from 10% to 90% of the 26
mass fraction burned (MFB), is presented in two parts according
to the intake pressures: the first part for 120/160 kPa (Fig. 13) 18
and the second part for 200/240/280 kPa (Fig. 14).
At 800 rpm it is noted that the combustion durations decreased 10
IVO 135, EVO 130

IVO 145, EVO 130

IVO 150, EVO 130

IVO 140, EVO 125

IVO 140, EVO 135

IVO 140, EVO 140

IVO 130, EVO 130

IVO 140, EVO 130

IVO 140, EVO 120

slightly as the boost pressure and load increased because of the

higher charge temperatures and pressures. In addition, it can be
seen from Figs. 13 and 14 that the combustion duration was
between 13 CA and 19 CA at 800 rpm, which is much shorter than
that of spark ignition (SI) combustion in four-stroke engines. This
suggests that the heat release process might have taken place in
the form of a spark ignited flame around the spark plug and Fig. 12. Air trapping efficiency.
 M. Dalla Nora, H. Zhao / Applied Energy 159 (2015) 117131 125

3000 RPM 50% greater than that at 800 rpm as a result of poorer charging effi-
24 ciencies at higher speeds. In addition, it can be seen that the most
retarded KLS occurred at the earliest EVO because of the minimum
12 residual gas concentration as evidenced by the highest charging
efficiency (Fig. 8). For the same reason, the KLS timing became
0 more retarded when the IVO was moved from 130 to 150 CA ATDC
and less residual gas was trapped. When the boost pressure was set
2200 RPM to 120 kPa, MBT could be achieved for all the valve timings and
35
more advanced MBT timings were realised near the middle of
25 the x-axis, when both the charging efficiency and trapping effi-
ciency were maximized.
At 800 rpm and 1500 rpm the valve timing IVO 150, EVO 130
15
presented the overall best indicated specific fuel consumption
(ISFC) seen in Fig. 17. At 800 rpm the ISFC of 252 g/kW h was found
5
at 1.1 MPa IMEP, while 264 g/kW h was achieved at 0.67 MPa. At
31 1500 RPM 1500 rpm the ISFC increased at all loads and the fuel consumption
deteriorated to about 316 g/kW h at the same 1.1 MPa IMEP. At
duration ( CA)
Combustion

2200 rpm and 3000 rpm the engine operation benefited from ear-
28
lier intake and exhaust valve opening as given by the valve timing
IVO 140, EVO 120. At these speeds the burning duration short-
25 ened due to the occurrence of CAI combustion and hence the
expansion work was not hindered by early EVO. The ISFC was
22 found at 264 g/kW h and 246 g/kW h in the range of loads 0.24
0.33 MPa IMEP at 2200 rpm and 3000 rpm, respectively.
19 Compared to downsized four-stroke engines operating at simi-
lar loads and speeds, the ISFC values were found in the same order
800 RPM of magnitude as the brake specific fuel consumption (BSFC) values
21
presented by [16] for a 850 cm3 two-cylinder GDI engine. Although
19 the two-stroke poppet valve engine does not experience pumping
losses, the power consumed by a supercharger should be consid-
17 ered alongside the friction losses when making direct comparisons
to BSFC results. In another study conducted in a highly boosted sin-
15
gle cylinder 400 cm3 engine running at stoichiometric airfuel
13 ratio [36], the ISFC of 253 g/kW h was registered at 1.6 MPa IMEP
120 kPa 160 kPa and 2000 rpm. For the sake of comparison, the poppet valve
11
IVO 130, EVO 130

IVO 135, EVO 130

IVO 140, EVO 130

IVO 145, EVO 130

IVO 150, EVO 130

IVO 140, EVO 120

IVO 140, EVO 125

IVO 140, EVO 135

IVO 140, EVO 140

31 1500 RPM

28

25
Fig. 13. Combustion duration for 120 kPa and 160 kPa intake pressures.
22
duration (CA)
Combustion

auto-ignition combustion of some premixed charge in the end-gas

19
[34,35]. As the engine speed went up to 1500 rpm the combustion
duration increased in terms of crank angles, but decreased slightly 800 RPM
20
in absolute time as the flame speed was accelerated by the higher
flow turbulence and the auto-ignition combustion was favoured by
18
the hotter residual gas.
At 2200 and 3000 rpm stable engine operation was mainly lim- 16
ited to the boost pressure of 120 kPa. During such operation it was
found that the spark timing had little effect on the combustion 14
phasing and auto-ignition combustion became the dominant heat
release process as evidenced by the very short combustion dura- 12
tions. The combustion duration remained nearly independent of 200 kPa 240 kPa 280 kPa
10
the IVO variation when the EVO was set to 130 CA ATDC. In com-
IVO 135, EVO 130

IVO 140, EVO 130

IVO 145, EVO 130

IVO 150, EVO 130

IVO 140, EVO 120

IVO 140, EVO 125

IVO 140, EVO 135

IVO 140, EVO 140

IVO 130, EVO 130

parison, the EVO had a more pronounced effect on the combustion

duration as shown by the earliest EVO (120 CA ATDC) producing
the shortest burning duration.
Figs. 15 and 16 show the spark timings set for MBT (coloured
symbols) or KLS (grey symbols) at 800 rpm and 1500 rpm, above
which CAI combustion took place and the spark timing had little
effect. It is noted that the presence of KLS at 1500 rpm was about Fig. 14. Combustion duration for 200 kPa, 240 kPa and 280 kPa intake pressures.
126 M. Dalla Nora, H. Zhao / Applied Energy 159 (2015) 117131

-10 340
1500 RPM 3000 RPM
300
-13
260
-16
220

-19 2200 RPM

420
Spark timing
(CA ATDC)

-22 360

800 RPM
-2 300

-6 240

-10 1500 RPM

ISFC (g/kWh)
450
-14
400
-18
120 kPa 160 kPa 350
-22
IVO 130, EVO 130

IVO 140, EVO 130

IVO 140, EVO 125

IVO 140, EVO 135

IVO 135, EVO 130

IVO 145, EVO 130

IVO 150, EVO 130

IVO 140, EVO 120

IVO 140, EVO 140

300

250

120 kPa 160 kPa 800 RPM

440
200 kPa 240 kPa
Fig. 15. Spark timings set for MBT (coloured symbols) or KLS (grey symbols) for 400 280 kPa
120 kPa and 160 kPa intake pressures. (For interpretation of the references to colour
in this figure legend, the reader is referred to the web version of this article.) 360

320

280

1500 RPM 240

-2
IVO 130, EVO 130

IVO 135, EVO 130

IVO 140, EVO 130

IVO 145, EVO 130

IVO 150, EVO 130

IVO 140, EVO 120

IVO 140, EVO 125

IVO 140, EVO 135

IVO 140, EVO 140

-7

-12

-17
Fig. 17. Indicated specific fuel consumption.
Spark timing
(CA ATDC)

-22

800 RPM
-2
11
-6
Effective compression/
expansion ratio

-10 10

-14
9
-18
200 kPa 240 kPa 280 kPa
8 Effective compression ratio
-22
IVO 130, EVO 130

IVO 135, EVO 130

IVO 140, EVO 130

IVO 145, EVO 130

IVO 150, EVO 130

IVO 140, EVO 120

IVO 140, EVO 125

IVO 140, EVO 135

IVO 140, EVO 140

Effective expansion ratio

7
IVO 130, EVO 130

IVO 135, EVO 130

IVO 140, EVO 130

IVO 145, EVO 130

IVO 150, EVO 130

IVO 140, EVO 120

IVO 140, EVO 125

IVO 140, EVO 135

IVO 140, EVO 140

Fig. 16. Spark timings set for MBT (coloured symbols) or KLS (grey symbols) for
200 kPa, 240 kPa and 280 kPa intake pressures. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this
article.) Fig. 18. Effective compression and expansion ratios.
 M. Dalla Nora, H. Zhao / Applied Energy 159 (2015) 117131 127

two-stroke engine achieved 276 g/kW h at 1500 rpm and 0.8 MPa work by the higher EER did not result in improved ISFC as a result
IMEP (1.6 MPa equivalent in the four-stroke cycle). This repre- of the lowest ECR amongst all points. Similarly, the case IVO 140,
sented 9% higher fuel consumption for the particularly optimised EVO 120, which presented the highest ECR, could not achieve the
valve timing IVO 150, EVO 130. highest efficiency at 800 and 1500 rpm due to retarded spark tim-
It could be observed that the ISFC was intrinsically linked to the ing and combustion phasing. At higher speeds, when CAI combus-
expansion work, charging efficiency and in-cylinder mixture com- tion prevailed, this increment in effective compression ratio
position and combustion. Therefore, there was a trade-off between ensured the lowest ISFC of 246 g/kW h.
higher scavenging rates through exhaust blow-down with early The most significant cause for the change in ISFC as a function of
EVO and higher expansion works achieved with late EVO. This valve timings was found from the combustion efficiency plots in
effect was demonstrated in the PV diagram in Fig. 10, though it Fig. 19. It can be seen that the combustion efficiency results mir-
is clearly seen in Fig. 18 where the effective compression ratio rored those of ISFC presented in Fig. 17. The highest combustion
(ECR) and effective expansion ratio (EER) were plotted as a func- efficiencies and lowest ISFCs occurred in the middle of the graphs
tion of the valve timings. around IVO 150, EVO 130 and IVO 140, EVO 120 at 800 rpm.
Fig. 18 shows that for a given exhaust valve timing both the The combustion efficiency decreased with higher engine speeds
effective compression and expansion ratios were constant and at the same boost pressure, and it dropped with higher boost pres-
the EER was higher than the ECR by about one unit. The effective sures at each particular engine speed. Besides the fuel rich mixture
compression and expansion ratios matched each other at IVO at all operation points, which already reduced the overall combus-
140, EVO 120, and when the EVO was retarded from 120 to tion efficiency, the main reason attributed to the low values
140 CA ATDC the EER increased and the ECR was reduced. The achieved at higher speeds and loads relied on the time available
highest EER and hence highest expansion work was achieved with for mixture formation with a SOI as late as 260 CA ATDC. For
the most retarded EVO. However, such an increase in the useful instance, at 120 kPa intake pressure the combustion efficiency for

100 0.98
3000 rpm 3000 rpm

85 0.94

70 0.90

2200 rpm 2200 rpm

95 0.93

80 0.87

65 0.81

50 0.75
1500 rpm
In - cylinder lambda (-)

94
1500 rpm
0.96
efficiency (%)
Combustion

78
0.90

62 0.84

46 0.78

30 0.72

800 rpm 800 rpm

95
0.99

83 0.93

71 0.87

59 0.81
120 kPa 160 kPa 200 kPa 120 kPa 160 kPa 200 kPa
240 kPa 280 kPa 240 kPa 280 kPa
47 0.75
IVO 130, EVO 130

IVO 140, EVO 130

IVO 150, EVO 130

IVO 140, EVO 125

IVO 140, EVO 140

IVO 135, EVO 130

IVO 140, EVO 130

IVO 145, EVO 130

IVO 150, EVO 130

IVO 140, EVO 120

IVO 140, EVO 125

IVO 140, EVO 135

IVO 140, EVO 140

IVO 130, EVO 130

IVO 135, EVO 130

IVO 145, EVO 130

IVO 140, EVO 120

IVO 140, EVO 135

Fig. 19. Combustion efficiency. Fig. 20. In-cylinder lambda.

128 M. Dalla Nora, H. Zhao / Applied Energy 159 (2015) 117131

the best valve timing remained around 90% at all speeds. However, have been hindered. In this case the gas exchanging analysis
at 160 kPa it continually decreased from 94% at 800 rpm to only required the adoption of fuel rich mixtures to evaluate the air trap-
70% at 1500 rpm even at in-cylinder lambda values around 0.94, ping efficiency, so lean mixtures would have underestimated it due
as seen in Fig. 20 at IVO 140, EVO 120. to the presence of free oxygen in the exhaust.
As shown by the in-cylinder lambda values in Fig. 20, the
change in combustion efficiency with valve timings could be 3.2. Emission analysis
attributed to the variation of in-cylinder air/fuel mixture with
the gas exchanging process. The higher the relative air/fuel ratio As shown in Fig. 21, CO emission increased significantly as the
(lambda) the more complete the combustion became. The leanest mixture became richer with more advanced IVO or retarded EVO
mixture of near stoichiometric air/fuel ratio was reached at at each engine speed. Fig. 21 shows that negligible CO emission
800 rpm and resulted in a combustion efficiency of 94%. As the was produced at 800 rpm when the charging efficiency and lambda
engine speed was increased from 800 rpm to 2200 rpm, the were maximised. Based on the estimated in-cylinder lambda
decreased charging efficiencies led to richer air/fuel mixtures and results in Fig. 20, some noticeable CO emission was expected by
lower combustion efficiencies. At the lowest boost pressure of combustion of overall fuel rich mixture. The lower than expected
120 kPa, the combustion efficiency became higher at 3000 rpm CO emission could be caused by the oxidation of CO to CO2 by
(90%) than 2200 rpm (88%) mainly because of the leaner mixture the short-circuited air mixed with the burned gas during the scav-
and faster heat release rate (Fig. 13). The extremely low results enging process. As the engine speed was increased from 800 rpm
for combustion efficiency at both ends of the valve timings studied to 2200 rpm, the poorer scavenging and combustion of richer mix-
were then justified by the low in-cylinder lambda values. These tures resulted in significant increase in CO and UHC emissions at
results could have been greatly improved by reducing the fuelling higher engine speeds. In addition, the mixture was less homoge-
rate, although the solely effect of valve timing alterations would neous at higher engine speed because of the reduced time available

100 3000 RPM 3000 RPM

40

65 20

30 0

2200 RPM 2200 RPM

90
290

60
170
30

50 0

1500 RPM 1500 RPM

ISCO (g/kWh)

ISUHC (g/kWh)

270
200

180 150

100
90
50

0 0
120 kPa 160 kPa 800 RPM 120 kPa 160 kPa 800 RPM
240 200 kPa 240 kPa 100 200 kPa 240 kPa
280 kPa 280 kPa
180 75

120 50

60 25

0 0
IVO 130, EVO 130
IVO 130, EVO 130

IVO 135, EVO 130

IVO 140, EVO 130

IVO 145, EVO 130

IVO 150, EVO 130

IVO 140, EVO 120

IVO 140, EVO 125

IVO 140, EVO 135

IVO 140, EVO 140

IVO 135, EVO 130

IVO 140, EVO 130

IVO 145, EVO 130

IVO 150, EVO 130

IVO 140, EVO 120

IVO 140, EVO 125

IVO 140, EVO 135

IVO 140, EVO 140

Fig. 21. Indicated specific carbon monoxide emissions. Fig. 22. Indicated specific unburned hydrocarbon emissions.
 M. Dalla Nora, H. Zhao / Applied Energy 159 (2015) 117131 129

between the end of injection and the beginning of combustion. smoke emission was noticeably more affected by the engine load
This could have contributed to the very rapid rise in CO emissions and speed than by the valve timing itself, as the fuel impingement
when the engine speed was changed from 800 rpm to 1500 rpm, as increased with longer injection durations at higher loads.
in the case of the valve timing IVO 150, EVO 130 when this The overall excessive smoke emission presented in Fig. 23
emission increased from 4 g/kW h to 43 g/kW h in the intake pres- reflected the level of charge stratification resulted from late fuel
sure range 120200 kPa. injections, besides the global fuel rich mixtures and the relatively
As shown in Fig. 22, UHC emission showed less dependency on low injection pressure used (15 MPa). Even at 800 rpm and
valve timings and lower correlation with the charging efficiency 0.67 MPa IMEP the smoke level remained around 0.55 FSN at the
and in-cylinder lambda. As late injections were employed, most best valve timing tested, while it increased by about four-times
UHC emissions were likely produced by the fuel rich combustion when the engine speed reached 1500 rpm. This was certainly a
as well as fuel impingement due to retarded injection. The UHC drawback of the two-stroke poppet valve engine, particularly when
emissions are not only dependent on the overall air/fuel ratio but comparing it to the values achieved by four-stroke GDI engines.
also on its homogeneity. As injection took place after 260 CA Tests performed by [16] in a two-cylinder downsized engine
ATDC, there was limited time available for a homogeneous mixture demonstrated FSN values around 0.15 when running at 1200 rpm
to form and very rich mixtures could be present in some regions and loads between 0.7 and 1.8 MPa IMEP. Nevertheless, such study
producing UHC emissions. In addition, at higher loads and boost employed a multi-hole centrally mounted injector working at
pressures, the end of injection could be as late as 290 CA ATDC, 20 MPa, and the time available for airfuel mixing was about three
when the piston was only at about 25 mm from the cylinder head. times longer than those realised in the present study.
Thus, the fan shaped spray impinged onto the piston and formed From the NOx emission presented in Fig. 24, more residual gas
pool fires on its top. For the same reasons, high smoke emission was trapped (lower charging efficiencies) when moving along the
was observed as seen in Fig. 23. Compared to UHC emissions, the

3000 RPM 1.0

3000 RPM
3.0
0.5
1.5

0.0 0.0

2200 RPM 2200 RPM

7.5
4.2

5.0 2.8

2.5 1.4

0.0 0.0

1500 RPM 1500 RPM

Smoke (FSN)

ISNOx (g/kWh)

5.0 11.0

4.0 8.0

3.0 5.0

2.0 2.0
120 kPa 160 kPa 800 RPM 120 kPa 160 kPa 800 RPM
2.6 200 kPa 240 kPa 200 kPa 240 kPa
280 kPa 37
280 kPa
29
1.8

21
1.0
13

0.2 5
IVO 145, EVO 130
IVO 130, EVO 130

IVO 135, EVO 130

IVO 140, EVO 130

IVO 150, EVO 130

IVO 140, EVO 120

IVO 140, EVO 125

IVO 140, EVO 135

IVO 140, EVO 140

IVO 135, EVO 130

IVO 140, EVO 130

IVO 145, EVO 130

IVO 150, EVO 130

IVO 140, EVO 120

IVO 140, EVO 125

IVO 140, EVO 135

IVO 140, EVO 140

IVO 130, EVO 130

Fig. 23. Smoke emissions. Fig. 24. Indicated specific oxides of nitrogen emissions.
130 M. Dalla Nora, H. Zhao / Applied Energy 159 (2015) 117131
x-axis from the middle to the both sides. Because of the increased
heat capacity of CO2 and reduced oxygen availability by the pres-
ence of recycled burned gases, the combustion temperature and
hence NOx formation were significantly reduced [37].
At 800 rpm the early EVO raised the charge oxygen availability
increasing NOx emissions to levels of downsized four-stroke engi-
nes operating at similar conditions [22]. As the speed increased
from 800 to 3000 rpm, the combustion mode progressed from
SI towards CAI as a result of higher levels of hot residual gas
trapped. Consequently, the NOx emissions progressively
decreased thanks to the higher charge dilution and lower com-
bustion temperature.
From Fig. 24 it is also noted that the NOx emissions were more
sensitive to the valve timings studied than to the load itself, espe-
cially at 800 rpm. At this speed the emission of oxides of nitrogen
increased by 20% as the boost pressure was changed from 120 to
280 kPa (0.661.22 MPa IMEP). In comparison, by retarding the
IVO in 10 CA from 130 to 140 CA ATDC the NOx emissions nearly
doubled. The spark timing also played an important role in NOx
emissions, as shown by the point IVO 140, EVO 120 at 200 kPa
boost. The ignition timing in this case had to be retarded to avoid
knocking combustion (Fig. 16), reducing the in-cylinder peak tem-
perature and NOx production.
At 2200 and 3000 rpm and intake pressure of 120 kPa, pure CAI
combustion took place. At 2200 rpm the NOx emissions rose
rapidly as the boost pressure was increased from 120 kPa to
160 kPa, as a result of both reduced residual gas concentration
and presence of high temperature flame in the spark assisted CAI
combustion.
4. Conclusions
In this study, a four-valve direct injection gasoline engine was
operated in the two-stroke cycle mode by opening both the intake
and exhaust valves around BDC. The exhaust gas was scavenged by
compressed air during the valve overlap period. At each engine
speed and boost pressure, the engine output was measured as a
function of intake and exhaust valve timings. The results can be
summarised as follows:

At 800 rpm the peak IMEP of 1.2 MPa was achieved at an intake
pressure of 280 kPa, producing a specific torque of 195 N m/L
with the in-cylinder peak pressure as low as 6.8 MPa. At each
engine speed, the maximum IMEP was obtained with the high-
est charging efficiency. As the engine speed was increased, the
maximum output was limited by the scavenging process and
violent heat release rate. At 2200 rpm and 3000 rpm the maxi-
mum loads were limited to 0.55 and 0.25 MPa IMEP, respec-
tively, which is equivalent to 1.1 and 0.5 MPa IMEP in the
four-stroke cycle.

For the given valve opening duration and valve lift, the maxi-
mum charging efficiency of 95% could be achieved at 800 rpm.
At any given IVO and EVO timings the charging efficiency
dropped with the increased engine speed due to the reduced
time available for gas exchanging. At 2200 rpm the charging
efficiency was limited to 45%, while at 3000 rpm it dropped to
about 20%.

At 120 kPa intake pressure, the trapping efficiency increased
from about 35% to 70% with higher engine speeds as the air
short-circuiting rate was reduced. At low speeds and long valve
overlaps, although, the trapping efficiency was reduced to val-
ues as low as 15%.

At 800 rpm and 1500 rpm the heat release process was domi-
nated by spark ignited flame propagation combustion. At higher
engine speeds, CAI combustion took place and the spark timing
had little effect.

The ISFC was primarily determined by the combustion effi-
ciency, which was directly related to the in-cylinder air/fuel
ratio. The relative air/fuel ratios of the in-cylinder mixture could
be increased by optimisation of the valve timings for maximum
charging efficiency. Furthermore, by reducing the fuelling rate
better ISFC would be expected, although it would not be possi-
ble to calculate the gas exchanging parameters using the
selected approach.

Compared to four-stroke downsized engines operating at simi-
lar loads and speeds, the overall two-stroke poppet valve engine
efficiency was found below values normally achieved by those
engines. Improved and competitive results are possible to be
achieved by leaner combustion and engine operation at those
valve timings which demonstrated better charging efficiencies
i.e. IVO 150, EVO 130 and IVO 140, EVO 120.

The CO emissions were directly affected by the in-cylinder
lambda. At 800 rpm, negligible CO emission was measured with
optimised valve timings, although at higher speeds these values
increased due to the lower temperature combustion.

Besides the overall fuel rich combustion, uHC and smoke emis-
sions were found also affected by fuel impingement considering
the late SOI adopted. At 1500 rpm and 120/160/200 kPa, these
emissions had a lighter variation than that found for CO under
constant fuelling rate and regardless the valve timing used.

NOx emission was found very low at higher engine speeds
when there was high residual gas concentration and CAI com-
bustion. At high loads and low speeds, although, these values
increased due to the greater charge oxygen concentration and
higher combustion temperatures.
The above results have demonstrated that the scavenging pro-
cess and fuel preparation are the two most important issues affect-
ing the two-stroke poppet valve engines performance and
emissions. The scavenging process can be further optimised by
additional experiments with different valve opening durations,
particularly at higher engine speeds.
After the acquisition of all gas exchanging parameters under
fuel rich conditions, additional tests can be performed with lean
combustion so that improved ISFC results can be achieved. In this
case the delivery ratio needs to be kept constant so the in-cylinder
lambda can be evaluated at the same air and fuel trapping efficien-
cies. At fuel lean conditions the mixture preparation process can be
further improved by higher injection pressures and/or split injec-
tions. A more robust stratified charge combustion system design,
such as a centrally mounted fast DI injector, would be also
desirable.
Acknowledgement
The first author would like to acknowledge the Brazilian council
for scientific and technological development (CNPq Brasil) for
supporting his PhD at Brunel University London.
References
[1] Hooper PR, Al-Shemmeri T, Goodwin MJ. Advanced modern low-emission two-
stroke cycle engines. Proc Instit Mech Eng, Part D: J Automob Eng
2011;225:153143. http://dx.doi.org/10.1177/0954407011408649.
[2] Hundleby G. Development of a poppet-valved two-stroke engine the flagship
concept. SAE Technical Paper, 1990; 900802. http://dx.doi.org/10.4271/
900802.
[3] Kenny RG. Developments in two-stroke cycle engine exhaust emissions. Proc
Instit Mech Eng, Part D: J Automob Eng 1992;206:93106. http://dx.doi.org/
10.1243/PIME_PROC_1992_206_165_02.
[4] Blair G. Design and simulation of two-stroke engines. Warrendale: Society of
Automotive Engineers; 1996, ISBN 978-1-56091-685-7.
[5] Stokes J, Hundleby G, Lake T, Christie M. Development experience of a poppet-
valved two-stroke flagship engine. SAE Technical Paper, 1992; 920778. http://
dx.doi.org/10.4271/920778.
 M. Dalla Nora, H. Zhao / Applied Energy 159 (2015) 117131
[6] Nomura K, Nakamura N. Development of a new two-stroke engine with
poppet-valves: Toyota S-2 engine, a new generation of two-stroke engines for
the future? Proceedings of the IFP international seminar, Paris, France, 1993. p.
5362.
[7] Knoll R. AVL Two-Stroke Diesel Engine. SAE Technical Paper, 1998; 981038.
http://dx.doi.org/10.4271/981038.
[8] Nakano M, Sato K, Ukawa H. A two-stroke cycle gasoline engine with poppet
valves on the cylinder head. SAE Technical Paper 1990; 901664, http://dx.doi.
org/10.4271/901664.
[9] Sementa P, Vaglieco BM, Catapano F. Thermodynamic and optical
characterizations of a high performance GDI engine operating in
homogeneous and stratified charge mixture conditions fuelled with gasoline
and bio-ethanol. Fuel 2012;96:20419. http://dx.doi.org/10.1016/
j.fuel.2011.12.068.
[10] Pohorelsky L, Brynych P, Macek J, Vallaude P, et al. Air system conception for a
downsized two-stroke diesel engine. SAE Technical Paper, 2012; 2012-01-
0831. http://dx.doi.org/10.4271/2012-01-0831.
[11] Cairns A, Zhao H, Todd A, Aleiferis P. A study of mechanical variable valve
operation with gasolinealcohol fuels in a spark ignition engine. Fuel
2013;106:80213. http://dx.doi.org/10.1016/j.fuel.2012.10.041.
[12] Osborne R, Li G, Sapsford S, Stokes J. et al. Evaluation of HCCI for future
gasoline powertrains. SAE Technical Paper, 2003; 2003-01-0750. http://dx.doi.
org/10.4271/2003-01-0750.
[13] Zhang Y, Zhao H. Investigation of combustion, performance and emission
characteristics of 2-stroke and 4-stroke spark ignition and CAI/HCCI operations
in a DI gasoline. Appl Energy 2014;130:24455. http://dx.doi.org/10.1016/j.
apenergy.2014.05.036.
[14] Tribotte P, Ravet F, Dugue V, Obernesser P, et al. Two strokes diesel engine
promising solution to reduce CO2 emissions. Procedia Soc. Behav. Sci.
2012;48:2295314. http://dx.doi.org/10.1016/j.sbspro.2012.06.1202.
[15] Martin S, Beidl C, Mueller R. Responsiveness of a 30 bar BMEP 3-cylinder
engine: opportunities and limits of turbocharged downsizing. SAE Technical
Paper, 2014; 2014-01-1646. http://dx.doi.org/10.4271/2014-01-1646.
[16] Eichhorn A, Lejsek D, Hettinger A, Kufferath A. Challenge determining a
combustion system concept for downsized SI-engines comparison and
evaluation of several options for a boosted 2-cylinder SI-engine. SAE Technical
Paper, 2013; 201301-1730. http://dx.doi.org/10.4271/2013-01-1730.
[17] Zhang Y, Zhao H, Ojapah M, Cairns A. CAI combustion of gasoline and its
mixture with ethanol in a 2-stroke poppet valve DI gasoline engine. Fuel
2013;109:6618. http://dx.doi.org/10.1016/j.fuel.2013.03.002.
[18] Benajes J, Molina S, Novella R, De Lima D. Implementation of the partially
premixed combustion concept in a 2-stroke HSDI diesel engine fuelled with
gasoline. Appl Energy 2014;122:94111. http://dx.doi.org/10.1016/j.
apenergy.2014.02.013.
[19] Benajes J, Novella R, De Lima D, Tribott P. Analysis of combustion concepts in
a newly designed two-stroke high-speed direct injection compression ignition
engine. Int J Engine Res 2015;16:5267. http://dx.doi.org/10.1177/
1468087414562867.
[20] Osborne R, Stokes J, Lake T, Carden P, et al. Development of a two-stroke/four-
stroke switching gasoline engine the 2/4SIGHT Concept. SAE Technical Paper
2005; 2005-01-1137. http://dx.doi.org/10.4271/2005-01-1137.
[21] Sadakane S, Sugiyama M, Kishi H, Abe S, et al. Development of a New V-6 high
performance stoichiometric gasoline direct injection engine. SAE Technical
Paper, 2005; 2005-01-1152. http://dx.doi.org/10.4271/2005-01-1152.
131
[22] Zhao H. Advanced direct injection combustion engine technologies and
development, vol. 1. Cambridge: Woodhead Publishing; 2010, ISBN
9781845693893.
[23] Zhang Y, Ojapah M, Cairns A, Zhao H. 2-Stroke CAI Combustion Operation in a
GDI Engine with Poppet Valves. SAE Technical Paper, 2012; 2012-01-1118.
http://dx.doi.org/10.4271/2012-01-1118.
[24] Douglas R. AFR and emissions calculations for two-stroke cycle engines. SAE
Technical Paper, 1990; 901599. http://dx.doi.org/10.4271/901599.
[25] Heywood JB. Internal combustion engines fundamentals. New York: McGraw-
Hill; 1988, ISBN 978-0071004992.
[26] United Nations Regulation 49. Uniform provisions concerning the measures to
be taken against the emission of gaseous and particulate pollutants from
compression-ignition engines and positive ignition engines for use in vehicles.
Off J Eur Union, 2013. <http://eur-lex.europa.eu/legal-content/EN/TXT/?uri=
CELEX:42013X0624(01)>.
[27] Xu R. A Convenient technique for determining two-stroke emissions
measurements accuracy and A/F ratio. SAE Technical Paper, 1996; 961804.
http://dx.doi.org/10.4271/961804.
[28] Andwari AM, Aziz AA, Said MFM, Latiff ZA. Experimental investigation of the
influence of internal and external EGR on the combustion characteristics of a
controlled auto-ignition two-stroke cycle engine. Appl Energy 2014;134:110.
http://dx.doi.org/10.1016/j.apenergy.2014.08.006.
[29] Wang Z, Qi Y, He X, Wang J, Shuai S, Law CK. Analysis of pre-ignition to super-
knock: Hotspot-induced deflagration to detonation. Fuel 2015;144:2227.
http://dx.doi.org/10.1016/j.fuel.2014.12.061.
[30] Wetzel P. Downspeeding a light duty diesel passenger car with a combined
supercharger and turbocharger boosting system to improve vehicle drive cycle
fuel economy. SAE Technical Paper, 2013; 2013-01-0932. http://dx.doi.org/10.
4271/2013-01-0932.
[31] Lumsden G, OudeNijeweme D, Fraser N, Blaxill H. Development of a
turbocharged direct injection downsizing demonstrator engine. SAE Int J
Engine 2009;2(1):142032. http://dx.doi.org/10.4271/2009-01-1503.
[32] Li N, Xie H, Chen T, Li L, Zhao H. The effects of intake backflow on in-cylinder
situation and auto ignition in a gasoline controlled auto ignition engine. Appl
Energy 2013;101:75664. http://dx.doi.org/10.1016/j.apenergy.2012.07.050.
[33] Benson RS, Brandham PT. A method for obtaining a quantitative assessment of
the influence of charging efficiency on two-stroke engine performance. Int J
Mech Sci 1969;11:30312. http://dx.doi.org/10.1016/0020-7403(69)90048-4.
[34] Xie H, Li L, Chen T, Yu W, Wang X, Zhao H. Study on spark assisted compression
ignition (SACI) combustion with positive valve overlap at mediumhigh load.
Appl Energy 2013;101:62233. http://dx.doi.org/10.1016/j.apenergy.2012.
07.015.
[35] Ortiz-Soto EA, Lavoie GA, Martz JB, Wooldridge MS, Assanis DN. Enhanced heat
release analysis for advanced multi-mode combustion engine experiments.
Appl Energy 2014;136:46579. http://dx.doi.org/10.1016/j.apenergy.2014.
09.038.
[36] Li Y, Zhao H, Stansfield P, Freeland, P. Synergy between boost and valve
timings in a highly boosted direct injection gasoline engine operating with
miller cycle. SAE Technical Paper, 2015; 2015-01-1262. http://dx.doi.org/10.
4271/2015-01-1262.
[37] Zhao H. HCCI and CAI engines for the automotive industry. Cambridge:
Woodhead Publishing; 2007, ISBN 978-1-84569-128-8.