Towards zero emission engines through the adoption of combustion lead engine design realised through a split cycle topology

Full text

THIESEL 2018 Conference on Thermo- and Fluid Dynamic Processes in Direct Injection Engines
Towards zero emission engines through the adoption of combustion-
lead engine design realised using a split cycle topology
R. Morgan1, F. Khalid1, A. Atkins2 S. Harvey1, Firmansyah1, D. Mason1, K. Vogiatzaki1 and M. Heikal1
1AEC – of Brighton, Lewes Road, Brighton, BN24GJ, United Kingdom.
E-mail: r.morgan2@brighton.ac.uk
Telephone: +(44) 1273 641950
2 Ricardo Innovations, Shoreham by Sea, BN43 5FG, United Kingdom
E-mail: andrew.atkins@ricardo.com
Abstract.
Although electrification is expected to dominate city transportation, chemically fuelled propulsion sys-
tems will remain the technology of choice for intercity freight transportation. This is due to the high
journey energy requirements that cannot be covered by batteries, even at the predicted future power
density levels. Tailpipe emissions will still need to approach zero for the vehicle to enter emissions-
controlled areas such as the centre of major cities. It is well known that to achieve zero emissions,
conditions in the combustion chamber must be lean enough to avoid particulate matter formation and
cool enough to avoid the formation of NOx. Many concepts have been proposed with varying degrees
of success. Issues, such as high reaction rates resulting in knock, high cycle-cycle variability, misfire
and high pumping work have been reported. The palliatives required to address these issues have
inevitably compromised the proposed solutions, resulting in complexity, poor power density and low
thermal efficiency.
A common feature of most solutions proposed for achieving zero emissions combustion is the adoption
of conventional engine architectures. In this paper, the potential of a new combustion concept is ex-
plored that has been derived by taking an alternate design approach where the ideal thermodynamic
conditions in the chamber to achieve zero emissions at maximum thermodynamic efficiency are first
defined. Following, having defined the ideal thermodynamic conditions, a novel fuel and hardware con-
figuration to achieve these conditions is suggested instead of simply adapting conventional architec-
tures.
The paper is structured as following: A fundamental kinetic analysis using CHEMKIN is described first
in order to define suitable starting conditions for the reaction process considering variables such as
temperature, mixture strength (i.e., air-fuel equivalence ratio (φ)) and fuel chemistry. Based on this anal-
ysis, suitable operating zones for the combustion system are defined that fulfil the criteria of complete
combustion at zero NOx and particulate emissions. One of the main findings of the analysis is that pre-
cise control of temperature at the start of combustion was critical to delivering practical zero emissions
combustion. A split cycle engine, where the compression stroke is performed in a separate cylinder and
the charge air is ‘injected’ with the fuel into the combustion chamber, is presented as optimum hardware
configuration that can give the necessary control of the initial conditions and so avoid the formation of
emissions.
Notations
ATDC Degree crank after TDC
°CA Degree crank angles
CH4 Methane
C8H18 Iso-Octane
C12H26 N-Dodecane
EGR Exhaust Gas Recirculation
h Heat transfer co-efficient
IMEP Indicated mean effective pressure
ISFC Indicated specific fuel consumption
n Index of expansion
N2 Nitrogen
NOx Nitrogen oxides
O2 Oxygen
P Pressure
RI Ringing Intensity
T Temperature
TDC Top dead centre
t Simulation time
V Volume
Φ Equivalence ratio
ηt Thermal efficiency

THIESEL 2018 Conference on Thermo- and Fluid Dynamic Processes in Direct Injection Engines
1. Introduction
The internal combustion engine (ICE) has been the dominant propulsion system for small and me-
dium scale applications for over 100 years. The adoption of the ICE occurred in an urban environment
that had severe air pollution problems that in a single event in London in 1876 saw 1150 people die in
three days. This pollution was in part due to horse-drawn transportation and coal burnt for heating and
cooking. Today, whilst air quality is much better than in the late 19th century, there remains a pressing
need to deliver further improvements to the transport sector in order to safeguard air quality in the cities.
In conjunction, there is pressing need to de-carbonise human activity to minimise Climate Change. Elec-
trification of the powertrain, with batteries providing the on-board energy storage, is set to dominate
personal transportation as battery technologies mature rapidly becoming more affordable and power
grids transition to renewable energy sources. However, for long range high load applications such as
intercity freight transportation, it is unlikely electrification will be viable for the foreseeable future [21].
Various fuel cell technologies have been proposed but cost and the manufacture, distribution and on
vehicle storage of hydrogen remains challenging [1].
The internal combustion engine has many attractive attributes, including high power density, low cost
and the ability to consume a wide range of energy dense ‘fuels’. Renewable synthetic fuel technology
is continuing to advance with the goal of achieving a sustainable energy supply in the near future. A
transition to renewable fuels is therefore to be expected but these fuels will inevitably be more expensive
than fossil fuels [3], meaning powertrain efficiency will become increasing important. The main disad-
vantage of the ICE is the toxic emissions produced during combustion, principally oxides of nitrogen and
carbon based particulate matter. Significant improvements have been made, with current Euro VI com-
mercial vehicles emitting a fraction of the emissions compared to the vehicles prior to the introduction
of legislation, but this is still not enough to deliver clean air in large cities. It is well known that the
formation of emissions can be prevented by avoiding fuel rich areas and high temperatures in the com-
bustion chamber [18]. Near zero emissions have been demonstrated using homogenous charge com-
pression ignition (HCCI) strategies. Although HCCI can deliver the required conditions for zero emis-
sions, problems with controlling the combustion process remain. The charge conditions at the point of
ignition are critical if misfire and knock are to be avoided. In an HCCI engine, the intake charge condi-
tions must be controlled so precisely that the concept becomes impractical in a conventional Otto cycle
engine with the load and speed flexibility required for on road transportation. An evolution of the HCCI
engine is the Reactivity Controlled Compression Ignition (RCCI) engine, where ignition is controlled by
the injection of a second high reactivity fuel into a homogenous mixture of the primary low reactivity fuel
[22]. Significant progress has been made in developing the RCCI concept, but the maximum load and
efficiency of the engine is still compromised by the need for high levels of charge dilution with exhaust
gases [2].
In the work reported in this paper, an alternative investigative approach was taken. The starting point
of the study was unbounded by the constraints imposed by classical Diesel and Otto cycle engine con-
figurations. Fundamental analysis of the thermal – fluid – chemical kinetic processes in a closed volume
chamber was used to define the boundary conditions required for zero emissions combustion. This
study confirmed previous scientific discoveries and provided additional insights into how the combustion
process dynamics needed to be conditioned to achieve complete, stable and importantly clean combus-
tion. Having defined these requirements, a novel mechanical architecture was then suggested to deliver
the required conditions, following the ‘inside to outside’ approach.
The paper first describes a zero-dimensional study of combustion in a well stirred reactor using the
CHEMKIN kinetic code. The study mapped out the operating space of the new engine and provided
boundary conditions for the second phase of the work. In the second phase of the work, the CHEMKIN
model was extended to include the expansion process to refine the boundary conditions of the ideal
engine. The analysis aimed to define the essential requirements of an ideal cycle rather than provide
an absolute prediction of performance and emissions. In the final stage of the research, a new engine
architecture was defined based on the recuperated split cycle engine concept [7] which delivered the
required conditions. The split cycle engine has been in development for many years as a means of
achieving ultra-high efficiency. The research reported here aimed to define the fundamental architecture
requirements for low emissions using a ‘clean sheet’ approach. The results showed the split cycle ar-
chitecture has the required characteristics for low emissions as well as high efficiency.
The split cycle engine concept is introduced, and preliminary results are presented to illustrate the
potential of this concept in achieving simultaneously high efficiency and zero emissions.

Towards zero emission engines using a combustion lead engine design realised via a split cycle topology 3
2. Fundamental Analysis of Combustion Process in a Well-Stirred Reactor
In the present work, single zone combustion modelling was undertaken using the CHEMKIN software.
CHEMKIN models heat and mass transfer in chemically reacting flows [20] such as those found inside
the combustion chamber of an internal combustion engine. A suite of detailed-kinetics reactor models
are used to describe the chemical reactions at a molecular level during combustion of a fuel inside a
combustion volume. For the current research, a zero-dimensional closed homogeneous perfectly stirred
reactor (PSR) CHEMKIN model with a constant volume was used. This means the rate of reaction is
governed by the chemical reaction kinetics and not by mixing. Three different hydrocarbon fuels; Me-
thane, Iso-Octane and N-Dodecane were considered. Iso-octane and n-dodecane were used to repre-
sent gasoline and diesel. Clearly this simplifying assumption ignores the complexity of real fuels in
particular for the diesel fuel. However this assumption allowed us to simplify the analysis and reduce
the computational times. Key parameters and boundary conditions are summarised in Table 1. Table 2
gives details of the assumptions made to model the three fuels.
Table 1. PSR model – Key Parameters
Parameter
Value
Volume, V
0.567l
Pressure, P
60 atm
Equivalence ratio (φ)
0.4, 0.5 and 0.67(Lean);
1.0 (Stoic.); 1.5, 2.0 & 2.5 (Rich)
Simulation time, t
5ms sec
(equivalent to 60°CA @ 2000 rpm)
Instantaneous heat transfer coefficient, h
30W/m2K
Table 2. No. of Reaction Kinetics Mechanisms and Species used in the Reaction Kinetics file for the 3 fuels
Fuel
Kinetics Mechanisms
Simulating Fuel
CH4 (Methane)
17 species, 58 reactions[10]
Methane
C8H18 (Iso-Octane)
874 species, 3796 reac-
tions[15]
Gasoline
C12H26 (N-Dodecane)
163 species, 887 reac-
tions[25]
(reduced mechanism)
Diesel
2.1Results
Results from the PSR modeling for the three fuels are presented in Fig. 1, showing the initial and final
temperature against equivalence ratio, φ at 60 atm pressure. The reason for using 60 atm pressure as
one of the model initial conditions is that it represents the cylinder pressure at the end of the compression
stroke in a compression-ignition engine and is the typical air induction pressure for a split cycle engine.
The threshold temperature where NOx emissions would be expected to form in significant quantities was
assumed equal to 2100K. Although, the maximum NOx temperature limit that has been reported in the
wider literature such as Kitamura et al. [13], Kamimoto [12] and Dec, et al. [5] showing the regions of
NOx and Soot formation using φ-T maps is 2200K. But because of the combustion taking place in a
constant volume, the NOx temperature limit has been further reduced by 100K to 2100K. For reference,
the maximum NOx temperature limit is shown with a red solid-line on the plot in the Fig. 1. The initial
temperature at each equivalence ratio was set to the minimum temperature at which the fuel would
ignite, and the reaction would be at least 99.9% complete at the end of the simulation (5ms). This time
threshold was set as the typical amount of time available for the combustion process to run to completion
in a typical internal combustion engine operating at 2000rpm. As such, the minimum temperature on

4 R. Morgan, F. Khalid, A. Atkins, S. Harvey, F. Firmansyah, D. Mason, K. Vogiatzaki and M Heikal
Fig. 1 defines the equivalence ratio – temperature limit for auto-ignition. The final temperature, in the
case of the PSR also represents the peak temperature as the heat losses from the chamber are minimal
in this case and no work is extracted from the combustion products.
Considering first the results for methane (CH4), a temperature of over 1450K was required to initiate
the auto-ignition reactions. This is expected given the stability of the methane molecule. The final
temperature peaks at the stoichiometric equivalence ratio, as expected. An important conclusion from
this analysis is that in all cases, the peak temperature exceeds the threshold for the formation of NOx
emissions. This means that the conditions for achieving (a) auto-ignition and (b) zero NOx emissions
cannot be achieved with methane without recourse to other measures such as the addition of a diluent
such as EGR or a means of initiating the ignition of the fuel.
Considering the longer chained molecules, the temperature required to initiate combustion lowers
significantly for isooctane (C8H18) and further still for n-dodecane (C12H26). This is due to the longer
chain hydrocarbons that are easier to ignite due to compression. What is more important is the end of
combustion temperature falls because of starting the combustion process at a lower temperature. The
same trend as observed with methane with equivalence ratio is observed, with the temperature peaking
at the stoichiometric air fuel ratio. However, with lean mixtures, the final temperatures are below the
threshold temperature for NOx formation. In both cases, this gives a narrow operating window where
the fuel can be (a) compression ignited and (b) burn cool enough not to initiate the formation of signifi-
cant NOx emissions. As observed by [24], a slight dip for iso-octane initial temperatures between 700-
800K reflects low temperature reactivity because of negative temperature co-efficient (NTC) behavior
which is mainly attributed to these high octane-number fuels around this temperature range. Above
800K, NTC behavior is not observed [24].
Fig. 1. Initial vs Final Temperatures for different equivalence ratios for three fuels at 60atm pressure (with red
solid-line showing threshold NOx formation temperature of 2100K [12] [6])
To investigate this finding further, the initial temperature was increased in stages from the previously
defined minimum at a range of fixed equivalence ratios. The results are presented in Fig. 2 for the three
fuels studied, where the temperature at the end of the simulation, equivalent to the maximum tempera-
ture during the combustion process is plotted against the start temperature for a number of equivalence
ratios. For each fuel, a quadrilateral area is defined. The left-hand side of the region represents the
ignition limit for the fuel, where further lowering the temperature would result in misfire or incomplete
combustion. As the initial temperature is increased at each equivalence ratio, the end of reaction tem-
perature also increases. The resulting area defined for methane in all cases exceeds the threshold

Towards zero emission engines using a combustion lead engine design realised via a split cycle topology 5
temperature of 2100K for the formation of NOx emissions. In the case of isooctane, the first data point
to reach 2100K threshold temperature occurs at an equivalence ratio of 0.4 with an initial temperature
of 920K. As the equivalence ratio is reduced from 0.4 (i.e. leaner), the end temperature data points will
fall below the NOx threshold limit but with an increased reaction start temperatures as shown by the
green-dashed extrapolated lines on the Fig. 2 to achieve ignition. For the given equivalence ratio φ=0.4,
no data points fall below 2100K temperature. For a practical engine, this is likely to result in control
issues and possibly limitations on the viable engine load and speed range. However, with n-dodecane,
a wider operating range is possible. The equivalence ratio range was found to be from 0.4 to 0.5 and a
130K inlet temperature range was observed. This operating envelope is much more promising for im-
plementation in reciprocating engine with the flexibility required for road transport applications.
From this initial study, it was clear that long chain fuel molecules are generally favourable for zero
emissions combustion system as ignition can be initiated at a lower temperature, and precise control of
the start of temperature at the start of combustion was essential. Clearly the ignition mechanism is more
complex when isomers are considered but the general trend of longer chains being favoured gives a
useful guide to the types of fuel that should be considered. Zero emissions combustion for shorter
molecules may still be possible but requires alternative ignition mechanisms.
Fig. 2. Start temperature, end temperature and equivalence ratio comparison for N-Dodecane, Iso-Octane and
CH4
3. Fundamental Analysis of Combustion Process in a Variable Volume Combus-
tion Chamber
The 0-D CHEMKIN combustion model was extended to the engine environment using the FORTE
CFD software [9]. The mixture was still assumed to be perfectly mixed and quiescent, but now work is
extracted during the combustion event due to the increase in volume of the chamber due to the motion
of the piston. The FORTE package combines the advanced chemistry solver module (CHEMKIN-PRO)
with a spatially resolved mass, momentum and heat transfer model improving the accuracy of the sim-
ulation compared to a real engine. Based on the 0-D modelling results presented in section 2 above,
the 3-D analysis was restricted to N-Dodecane case only as it showed a wider range of equivalence
ratio and ignition temperatures were the NOx formation reactions are supressed. The boundary condi-
tions for the 3-D modelling were also derived based on the 0-D CHEMKIN PSR results. An N-Dodecane

6 R. Morgan, F. Khalid, A. Atkins, S. Harvey, F. Firmansyah, D. Mason, K. Vogiatzaki and M Heikal
skeletal mechanism comprising of 105 species and 420 reactions, developed by Luo et al. [14] was used
as the chemistry model. The simplified chemistry model resulted in reduction of total computational
times for different simulation scenarios. A skeletal mechanism retains the main features of a detailed
mechanism including the reactions of polycyclic aromatic hydrocarbons for diesel fuel. The mechanism
was also validated using constant volume ignition delay data and engine combustion experiments using
N-Dodecane as surrogate for diesel fuels [14].
3.1Model Set Up
A cross-section of the engine piston-bowl geometry is shown in Fig. 3 and is based on a medium duty
engine. It should be noted that at this stage of the research, the combustion chamber has not been fully
optimized. Other key model parameters are listed in Table 3. All the simulations cases were run be-
tween 5° before top dead center (TDC) and 45° after TDC in order to reduce the total simulation time
for each particular scenario. Swirl was modeled by imposing solid body rotation on the trapped air
centered on the center line of the piston.
Fig. 3. FORTE’ engine model – A cross-section of the piston-bowl geometry
Table 3. FORTE’ Engine Model Specifications
Engine parameter
Value
No. of cylinders
1
Bore
105mm
Stroke
130mm
Displacement
1.125l
Connecting rod length
265mm
Compression ratio
16.5:1
Combustion chamber
Open shallow bowl
Swirl ratio
0.5
Local Turbulent Kinetic energy
(cm2/sec2)
92,000
Turbulent length scale (cm)
11.2mm
Engine Cycle
2 stroke
Three cases were considered at three different start temperatures imposed at the start of the simulation.
The three temperatures were derived using 0-D PSR modelling for a complete combustion of fuel-air
reaction mixture. The medium temperature (764K) is the baseline temperature whereas the other two

Towards zero emission engines using a combustion lead engine design realised via a split cycle topology 7
are the lower and upper extremes of the baseline i.e., 689K (lower) and 814K (higher) reaction start
temperatures. The boundary conditions for the cases studied are summarised in Table 4.
Table 4. Model initial conditions for 3 different start temperatures
Start Temperature
764 K
689 K
814 K
Start Pressure
60 atm
Mixture Fraction
(by mass)
O2 = 0.23 %
N2 = 0.74%
nC12H26 = 0.03%
Engine Speed
800 rpm
Equivalence ratio
0.52
3.2 Results
Plots of cylinder pressure, temperature, NOx and particulate concentrations are shown in Fig. 4 for
the three different starting temperatures. Key engine results are also tabulated in Table 5. Referring to
the cylinder pressure and temperature plots (Fig. 4a and b), the 814K and 764K results are similar as
the change in ignition delay between the two start temperatures is small. However, as the temperature
is further reduced to 689K, a retardation of the main ignition event is observed, by 17° relative to the
814K case. It should be noted that in all cases the fuel is fully mixed, and combustion is complete so
the retardation of the main heat release event is due to a slowing of the initial cool flame reactions. As
the fuel is pre-mixed, once the temperature has risen due to the early reactions, the main heat release
event in all cases is rapid, controlled by the chemical reaction kinetics and not by the flame speed unlike
in a conventional Otto cycle engine or the mixing of air and fuel as in a Diesel cycle engine. Interestingly,
the peak chamber temperature is only reduced by 100K despite 17° of retard of the heat release.
Referring to the NOx plot (4c), in all cases the NOx concentrations at the end of the combustion pro-
cess are low especially at the 689K case where a concentration of 6ppm is predicted. In this case, the
peak gas temperature only just reaches 2155K and so the formation of thermal NOx is suppressed
resulting in a low overall NOx value at the end of combustion
Referring to the particulate plot (4d), some particulate matter is generated during the combustion
event but is burnt as the reactions run to completion. Figure 4d gives an approximation of mass of the
soot particulate species formed using the two-step semi-empirical soot model built-in FORTE’ simula-
tions. The two-step model consists of soot formation and soot oxidation steps. The governing equations
for both of these steps are given by [11] and [17] respectively. The soot model uses acetylene (C2H2)
as a soot-pre-cursor species approximating the mass and particle size of the soot formed based on the
local concentrations of C2H2 species formed during each simulation. The reduced mechanism devel-
oped by Luo et al. [14] has already got C2H2 species added into it. Looking at the Fig. 4d it can be
observed that essentially no particulate emissions are produced from the engine emissions as would be
expected from a lean combustion system. The chemical reaction in all cases was complete and so
hydrocarbon and CO emissions are also essentially zero.
A widely reported issue with HCCI and RCCI engines is ‘ringing’ leading to high noise from the high
rate of pressure rise [5]. The three CA-Pressure plots have been normalized to start at the 2% angle to
facilitate comparison of the shape and peak rate of pressure rise in Fig. 5. From these peak pressure
values a value for ringing intensity (RI) was calculated as shown in Table 5. These calculated values
show that ringing is a potential problem that would need to be addressed in the future investigation,
since all the values lie above the typical RI limit of 5MW/m2 [5]. The addition of a diluent could be con-
sidered to reduce the rate of pressure rise.
The 689K case presents the most interesting result, giving very low NOx emissions. However, the
late combustion in this case would compromise the efficiency of a conventional Otto or Diesel cycle
engine as the combustion process is retarded from the optimal. For conventional engine cycles, the
optimal 50% mass fraction burnt angle is 8-10° ATDC where as in this case, the main heat release only
starts at 17° ATDC. Therefore this combustion strategy into a conventional engine cycle would compro-
mise the thermal efficiency of the engine. The closest strategy applied to a conventional engine is the
late injection low temperature combustion strategy [18]. Here, the fuel injection in a conventional diesel

8 R. Morgan, F. Khalid, A. Atkins, S. Harvey, F. Firmansyah, D. Mason, K. Vogiatzaki and M Heikal
engine is retarded resulting in more mixing of the fuel and air before ignition. Low NOx emissions are
reported, but also a reduction in thermal efficiency. The challenge remains to achieve simultaneously
low NOx and high thermal efficiency.
Fig. 4. Results from the CFD analysis (a) Pressure vs °CA, (b) Temperature vs °CA, (c) NOx vs °CA and (d) Soot
vs °CA
Table 5. Comparison of engine summary results of key parameters for the 3 start temperatures
Parameter
Unit
T = 764K
T = 814K
T = 689K
Peak Cylinder Pressure
Atm
185
176
166
Peak Cylinder Temperature
K
2245
2292
2155
Gross Indicated Power
kW
33.18
31.62
31.28
Max. Pressure Rise Rate
Bar / deg
76.0
96.4
71.3
Ringing Intensity
MW/m2
10.3
17.5
9.85
Gross IMEP
Bar
20.44
19.5
19.2
Fuel mass used
g / cycle
0.219
0.205
0.24
Wall heat transfer loss
kJ
1.069
0.967
0.922
50% heat release burn angle
deg ATDC
1.04
1.98 BTDC
17.03

Towards zero emission engines using a combustion lead engine design realised via a split cycle topology 9
NOx
ppm
21.11
35.13
4.89
Fig. 5. Cylinder pressure traces normalized at the 2% burn angle for the 3 different start of reaction temperatures
4. Design requirements for the zero emissions engine
The analysis presented in sections 2 and 3 shows that if the thermodynamic conditions in the com-
bustion chamber can be controlled with the right fuel chemistry, ultra-low emissions are achievable.
Longer chain fuel molecules are preferred, requiring lower temperatures for auto-ignition compared to
shorter chain molecules and for the remaining analysis, dodecane will be assumed as the fuel. Three
essential design requirements can be derived from the analysis:
1. The charge temperature at the start of ignition must be controlled in the range of 690 to 820K to
avoid misfire and NOx formation.
2. The fuel must be well mixed with the charge air to avoid particulate formation.
3. Phasing the combustion to occur later in expansion stroke widens the stable operating window,
between the misfire and knock limits and reduces the rate of pressure rise avoiding ringing.
It is possible to engineer a conventional engine architecture to meet these conditions, but with signif-
icant compromises in engine load, thermal efficiency and complexity. The challenges in delivering the
optimal in-cylinder conditions are primarily the delivery of precise charge temperatures within the re-
quired range described earlier. In a conventional engine, this means either the compression ratio must
be dropped to levels that would compromise efficiency or levels of charge dilution used which in turn will
compromise the breathing of the engine. To illustrate the challenge, in a conventional diesel engine,
the end of compression temperature is typically more than 900K, well in excess of the maximum target
temperature of 820 K. The end of compression temperature is also determined by the temperature,
pressure and trapped mass during the induction event. Any variability from cycle to cycle in the start of
compression conditions will inevitably result in variations in the temperature at the point of ignition, re-
sulting in cycle to cycle variability when the thermodynamic conditions are close to the ignition limit.
The other observation from the analysis was that if the combustion process was phased to occur later
in the expansion stroke, the rate of pressure rise could be tempered without the need for high levels of

10 R. Morgan, F. Khalid, A. Atkins, S. Harvey, F. Firmansyah, D. Mason, K. Vogiatzaki and M Heikal
charge dilution. However, in a conventional Otto or Diesel cycle, retarding the combustion process is
detrimental to thermal efficiency.
Fig. 6 (a) Diagram of the recuperated split cycle engine and (b) Temperature enthalpy diagram [16]
To achieve precise control of the charge temperature at the start of combustion, an alternative engine
architecture is proposed in which the compression and combustion processes are separated into differ-
ent chambers (Fig. 6 (a)). A more detailed explanation for all the processes of Fig. 6 has been given by
Dong et al. [7]. The so called split cycle engine has several variants including Scuderi engine [19] and
more recently by Morgan et al [16] in which a recuperator was added between the compression and
combustion cylinders. In the recuperated split cycle engine, the charge air is first compressed isother-
mally (1-2) and then heated using energy recovered from the exhaust gases (2-3) prior to induction to a
combustion cylinder. Fuel is added and burnt in the combustion cylinder (3-4) and expanded (4-5) to
generate work. Heat is then recovered from the exhaust gases in the recuperator (5-6). This means
the temperature at the start of combustion can be precisely controlled independent of the compression
ratio of the engine by the control of the quantity of pre-heating of the air. This feature of the recuperated
split cycle engine readily achieves design requirement 1 enabling precise control of the charge temper-
ature at a lower temperature (enabled by isothermal compression of the charge) than would be achiev-
able in a conventional engine architecture.
The second design requirement for the air and fuel to be well mixed is more challenging as the fuel
cannot be pre-mixed with the charge air before induction as it would inevitably ignite in the intake system.
However, another feature of the split cycle engine is that the charge air induction process occurs with
high pressure differences across the combustion cylinder inlet valves, of typically over 25:1 at the start
of induction. The charge flow across the valves is choked and very high in-cylinder velocities are gen-
erated into which the fuel is sprayed, as illustrated in preliminary CFD analysis shown in Fig. 7. In this
analysis, the flow field into a large chamber is modeled using the Fluent software package. In this case,
the geometry is representative of the research engine reported in [16], but the valve and piston were
static and so the analysis essentially represents a flow bench. The analysis therefore shows the flow
field at the start of the induction event before the pressure between the inlet port and combustion cham-
ber has equalized. High velocities, in excess of 400m/s are observed in a curtain around the valves,
resulting in two counter rotating ‘tumbling’ structures developing. High velocities are in particular noticed
in the region around the injector tip. The analysis work is currently being extended to include moving
geometries and the fuel spray, but does illustrate the high mixing potential of a split cycle engine. A
conceptual model of the air-fuel jet interactions is presented in Fig. 8. It is proposed the air and fuel jets
will strongly interact, potentially promoting the rapid mixing of the air and fuel.
This feature of the split cycle engine was also observed by Rossi [23] and proposed as an enabling
technology for HCCI combustion. In the case of Rossi’s work, the fuel was introduced into the transfer
port between the compressor and combustion cylinders and utilized the mixing energy generated during
the transfer event between the compressor and transfer port, but the basic concept is similar to the
concept proposed in the current work. Single cylinder testing, reported in [16] also showed high rates
of pressure rise in a split cycle combustion test rig and it was suggested that this was due to a high
contribution from pre-mixed combustion. The split cycle engine therefore offers a promising engine
architecture in achieving the second essential design requirement.

Towards zero emission engines using a combustion lead engine design realised via a split cycle topology 11
High pressure rise rates can cause excessive ringing in conventional engine architectures but as
ringing intensity is a function of temperature, the recuperated split cycle engine can counter this problem
with adequate control of the intake air temperature.
The final design requirement is for the engine to be efficient even with retarded combustion. In the
split cycle engine, the air is inducted into the combustion chamber around top dead center, resulting
naturally in late combustion around 15° after top dead center. In a conventional engine, these timings
would compromise the thermal efficiency of the engine. However, as exhaust heat is recovered pre-
combustion, the impact of late timings is significantly reduced, and the recuperated split cycle engine is
still predicted to achieve greater than 55% and approaching 60% break thermal efficiency [16]. Thus,
the recuperated split cycle engine achieves two of the three essential design requirements and shows
promise in achieving the third requirement.
Fig 7. CFD analysis of the split cycle engine geometry with a static piston and valves, illustrating the
flow field at the start of air induction.
Fig 8. Conceptual model proposed for the air-fuel interaction at the start of induction
5. Discussion
The adoption of an alternative engine and thermodynamic cycle design that is compatible with the
requirements for zero emissions avoids the issues reported earlier with conventional Otto and Diesel
cycle engines. The split cycle engine therefore offers a potential route to simultaneously achieving high
efficiency and zero emissions in a reciprocating engine.
To check the combustion chamber analysis reported in this work is compatible with the overall cycle
requirements and to produce meaningful whole cycle results, two additional investigations and correc-
tions must be made:
Tumble flow structure
High air velocity around
injector

12 R. Morgan, F. Khalid, A. Atkins, S. Harvey, F. Firmansyah, D. Mason, K. Vogiatzaki and M Heikal
1. Confirmation the exhaust temperature at the end of expansion is higher than the target pre-
combustion chamber charge temperature, in order to guarantee that there is sufficient exhaust
energy to pre-heat the air and the cycle energy balance is maintained.
2. Subtraction of the ideal isothermal compression work from the gross indicated work to pro-
duce meaningful specific emissions, based on the true indicated work from the cycle.
The exhaust temperature was calculated by taking the end of combustion conditions and assuming
an adiabatic expansion process to the end of the expansion stroke. By assuming a polytrophic index of
1.3, the end of expansion temperature can then be calculated. The results for the three cases presented
in section 3 are shown in Table 6.
To account for the compression work, the ideal isothermal work was calculated using:
𝑊 = 𝑚̇ 𝑅𝑇1𝑙𝑛 𝑃2𝑃1
⁄
Where P1 and P2 are the start and final pressures, T1 is the temperature at the start of the compression
process, R the gas constant and m the charge air mass flow. This analysis does not include all the
parasitic losses, as the isothermal compression will not be ideal and other losses such as valve and
recuperator pressure losses are not included, but gives a reasonable indication of the cycle gross work.
Dong previously reported the impact of the isothermal compression process on the overall efficiency of
the split cycle engine [8]. From Dong’s work, it is clear the performance of the isothermal compressor
is critical to the success of the split cycle. However, work reported by Coney [4] on an early variant of
the recuperated split cycle engine showed a near isothermal compression could be achieved and so the
isothermal assumption used hear will give a reasonable assessment of the gross cycle work.
The results for the three cases presented in section 3 are shown in Table 6 below.
Table 6. Gross cycle work and engine emissions produced for all the 3 engine cases
Case 1
Case 2
Case 3
Pressure at start of
combustion
60atm
60atm
60atm
Temperature at the
start of combustion
764K
814K
689K
End of expansion tem-
perature
975K
1006K
960K
Gross Combustor Cyl-
inder power
33.2kW
31.6kW
31.3kW
Gross cycle work
25.2kW
23.7kW
23.3kW
GINOx
0.32g/kWh
0.52g/kWh
0.085 g/kWh
GIPm
1.56E-10g/kWh
4.79E-13g/kWh
1.04E-07g/kWh
Finally, the power density of the split cycle engine was estimated. This is important as an obvious
criticism of the concept is the fact two cylinders are now required to do the job of one in a conventional
engine. The combustion chamber volume for this analysis was 1.25l, which was the same as the single
cylinder research engine reported in [16]. To calculate a representative power density, the compression
cylinder must also be included in the calculation. In a split cycle engine, the ratio of the combustion
chamber volume to the compression chamber volume is approximately 2.5:1, due to the density differ-
ence between the cold charge air in the compression cylinder and hot charge air in the combustor. This
means that for the purposes of calculating power density, the effective swept volume in this case is
including both the compression and combustion cylinders 1.75 l, giving a power density at the lowest

Towards zero emission engines using a combustion lead engine design realised via a split cycle topology 13
NOx case of 13.3kW/l. If the parasitic losses due to friction and ancillaries were assumed to remain the
same as for a conventional heavy duty diesel engine, at approximate 5%, the resulting power density is
12.6kW/l. For a 13l 6cylinder engine, this would give a power at this engine speed of 164kW, compa-
rable to the Volvo D13K Euro VI engine which develops 147kW at 800rpm. The power density of the
split cycle engine is achieved through the 2 stroke cycle which compensates for the additional compres-
sor cylinder. There remains a question as to how the split cycle engine combustion system would
perform at higher engine speeds, which at the current stage of the research is unknown.
6. Conclusions
A fundamental chemical and combustion process focused analysis was undertaken to determine the
ideal thermodynamic conditions of an ultra-low emissions internal combustion engine. The analysis was
performed using the CHEMKIN software, first assuming a constant volume perfectly stirred reactor and
then of an engine. From the analysis, it was clear longer chain hydrocarbons are easier to compression
ignite and react below the NOx formation threshold and three essential design requirements were de-
rived:
1. The charge temperature at the start of ignition must be controlled in the range of 690K to 820K
to avoid misfire and NOx formation.
2. The fuel must be well mixed with the charge air to avoid particulate formation.
3. Control phasing of the combustion in order to occur later in expansion stroke in order to widen
the stable operating window, between the misfire and knock limits and reduces the rate of
pressure rise avoiding ringing.
The recuperated split cycle engine was proposed as an architecture that fulfills the design require-
ments. The split cycle engine architecture facilitates the precise control of the charge temperature at
the point of ignition through the separation of the compression and expansion processes into different
chambers. Through the intra-cycle waste heat recovery process, the split cycle engine also effectively
recovers waste exhaust energy and thus maintains high thermal efficiencies even when the combustion
process occurs during the expansion stroke.
Additional work is required in several areas to further this interesting concept:
1. Investigation of the air-fuel mixing processes to ensure the fuel and air are well mixed before
ignition. Preliminary experimental and modelling results are encouraging but more work is re-
quired to provide further understanding of the mixing process and degree of pre-mixing.
2. Extension of the work to cover a wider range of loads and speeds typical of the target vehicle
application
In conclusion, the recuperated split cycle engine shows promise in delivering both high thermal effi-
ciency and ultra-low emissions from an internal combustion engine cycle.
Acknowledgements
The work reported in this paper was funded by Innovate UK as part of the Cryopower programme,
EPSRC under grant awards EM/M009424/1 Ultra Efficient Engines and Fuels, EP/P012744/1 Unified
modelling framework of sub- and super- critical injection dynamics and the Institutional Sponsorship
2017 - University of Brighton EP/R512977/1. The authors would like to thank Delphi for the supply of
the DFI 1.5 fuel system and subsequent support and LMS for the provision of the AMESim software
under their academic support programme. The authors would also like to thank the directors of Ricardo
for their permission in publishing this paper.
References
[1] Ball, M. and Wietschel, M. (2009) The future of hydrogen – opportunities and challenges,
International Journal of Hydrogen Energy, vol. 34, pp. 615-627.
[2] Benajes, J., García, A., Monsalve-Serrano, J., Balloul, I., and Pradel, G. (2016) An assessment
of the dual-mode reactivity controlled compression ignition/conventional diesel combustion

14 R. Morgan, F. Khalid, A. Atkins, S. Harvey, F. Firmansyah, D. Mason, K. Vogiatzaki and M Heikal
capabilities in a EURO VI medium-duty diesel engine fueled with an intermediate ethanol-
gasoline blend and biodiesel, Energy Conversion and Management, vol. 123, pp. 381-391.
[3] Christensen, A. and Petrenko, C. (2017) CO2-Based Synthetic Fuel: Assessment of Potential
European Capacity and Environmental Performance, European Climate Foundation and the
International Council on Clean Transportation.
[4] Coney, M. W., Stephenson, P., Malmgren, A., Linnemann, C., Morgan, R., Richards, R., et al.
(2002) Development of a reciprocating compressor using water injection to achieve quasi-
isothermal compression, Proceedings of the 16th International Compressor Engineering
Conference at Purdue, July 20th.
[5] Dec, J. E., Yang, Y., and Dronniou, N. (2011) Boosted HCCI - Controlling Pressure-Rise Rates
for Performance Improvements using Partial Fuel Stratification with Conventional Gasoline,
SAE International Journal of Engines, vol. 4, pp. 1169-1189.
[6] Dec, J. E. (2009) Advanced compression-ignition engines—understanding the in-cylinder
processes, Proceedings of the Combustion Institute, vol. 32, pp. 2727-2742.
[7] Dong, G., Morgan, R. E., and Heikal, M. R. (2016) Thermodynamic analysis and system design
of a novel split cycle engine concept, Energy, vol. 102, pp. 576-585.
[8] Dong, G., Morgan, R., and Heikal, M. R. (2015) A novel split cycle internal combustion engine
with integral waste heat recovery, Applied Energy, vol. 157, pp. 744-753.
[9] FORTE' Theory Manual (2017) FORTE' Theory Manual (Release 18.2), August 2017. ANSYS
INC. Southpointe, 2600 ANSYS Drive Canonsburg, PA 15317. http://www.ansys.com.
[10] GRI-Mech 3.0 (1999) GRI-Mech Version 3.0 7/30/99 CHEMKIN format, WorldWideWeb home
page http://www.me.berkeley.edu/gri_mech/.
[11] Hiroyasu, H. and Kadota, T., "Models for Combustion and Formation of Nitric Oxide and Soot
in Direct Injection Diesel Engines," 1976.
[12] Kamimoto, T. and Bae, M. (1988) High Combustion Temperature for the Reduction of
Particulate in Diesel Engines, SAE paper 880423.
[13] Kitamura, T., Ito, T., Kitamura, Y., Ueda, M., Senda, J., and Fujimoto, H. (2003) Soot Kinetic
Modeling and Empirical Validation on Smokeless Diesel Combustion with Oxygenated Fuels.
[14] Luo, Z., Som, S., Sarathy, S. M., Plomer, M., Pitz, W. J., Longman, D. E., et al. (2014)
Development and validation of an n-dodecane skeletal mechanism for spray combustion
applications, Combustion Theory and Modelling, vol. 18, pp. 187-203.
[15] Mehl, M., Curran, H., Pitz, W., and Westbrook, C. (2009) Chemical kinetic modeling of
component mixtures relevant to gasoline, Lawrence Livermore National Laboratory (LLNL),
Livermore, CA, No. LLNL-CONF-410968.
[16] Morgan, R. E., Jackson, N., Atkins, A., dong, G., Heikal, M., and Lenartowicz, C. (2017) The
Recuperated Split Cycle - Experimental Combustion Data from a Single Cylinder Test Rig, SAE
International Journal of Engines, vol. 10, pp. 2596-2605.
[17] Nagle, J. and Strickland-Constable, R. F., "Oxidation of Carbon between 1000-2000℃," in
Proceeding of the 5th Conference on Carbon, 1982, 1982.
[18] P.B. Musculus, M., C. Miles, P., and M. Pickett, L. (2013) Conceptual models for partially
premixed low-temperature diesel combustion, Progress in Energy and Combustion Science,
vol. 39, pp. 246–283.

Towards zero emission engines using a combustion lead engine design realised via a split cycle topology 15
[19] Phillips, F., Gilbert, I., Pirault, J.-P., and Megel, M. (2011) Scuderi Split Cycle Research Engine:
Overview, Architecture and Operation, SAE International Journal of Engines, vol. 4, pp. 450-
466.
[20] Reaction Design (2013) CHEMKIN Product Brochure 2013,
http://www.reactiondesign.com/products/chemkin/.
[21] Regnart, J. (2017). Roadmaps Explored: Thermal propulsion systems will still play a part in the
automotive sector. Available: http://www.apcuk.co.uk/2017/12/roadmaps-explored-thermal-
propulsion-systems-will-still-play-a-part-in-the-automotive-sector
[22] Reitz, R. D. and Duraisamy, G. (2015) Review of high efficiency and clean reactivity controlled
compression ignition (RCCI) combustion in internal combustion engines, Progress in Energy
and Combustion Science, vol. 46, pp. 12-71.
[23] Rossi, R., Musu, E., Frigo, S., Gentili, R., and Reitz, R. D. (2013) Simultaneous Reduction of
Soot and NOX Emissions by Means of the HCPC Concept: Complying with the Heavy Duty
EURO 6 Limits without Aftertreatment System, SAE paper 2013-24-0093.
[24] Westbrook, C., Pitz, W., and Mehl, M. (2015) A Chemical Kinetic Study of Octane Sensitivity,
Lawrence Livermore National Laboratory (LLNL), Livermore, CA, No. LLNL-CONF-668405.
[25] Westbrook, C. K., Pitz, W. J., Herbinet, O., Curran, H. J., and Silke, E. J. (2009) A
comprehensive detailed chemical kinetic reaction mechanism for combustion of n-alkane
hydrocarbons from n-octane to n-hexadecane, Combustion and Flame, vol. 156, pp. 181-199.