JSAE 20119326

SAE 2011-01-2025

A Novel Two-Stroke SI Design for NOx Reduction in

Natural Gas RI Species Enhanced Engine
David A. Blank
HCRI Technologies International, LLC

Copyright © 2011 Society of Automotive Engineers of Japan, Inc.

ABSTRACT paraffin hydrocarbon fuels.2 Thus, its use typically

results in a trade-off penalty of higher than desired
This study examines the novel use of a strong NOx emissions.
presence of radical ignition (RI) species to augment
flame front propagation in a two-stroke spark ignition
(SI) engine. Periphery mounted secondary chambers
enable the generation of these RI species in one cycle
for use in the next cycle. These chambers are outfitted
to enable fuel-insertion and rapid heat addition. The
new technology examined in the study employs the
chemistry of homogeneous combustion radical
ignition (HCRI) for the RI species enhancement of
pre-mixed charge (PC) SI. The aim is to see if this
chemistry can increase the lean burn threshold of this
2-stroke engine with natural gas (NG). The analysis
uses experimental data together with a full
chemical-kinetics simulation formulation that also
accounts for thermo-chemical and hydro-dynamic
exchanges that are both between the chambers and
with the environment. The mechanism for the
chemical kinetics consists of 97 chemical reactions
involving 33 species. The results suggest the potential
for extending the experimentally successful 4-stroke
RIS(A)-PCSI technology to 2-stroke SI engines for
NOx reduction.

INTRODUCTION

Until now the power-to-weight advantage of the

2-stroke engine has been offset by its lower fuel
efficiencies. However, technologies that enable
operations with leaner fuel-to-air ratios coupled with
fuels that increase power densities, can together
partially compensate for this offset.
Natural gas (NG), one of the world’s most abundant
fossil fuels, is capable of the power densities of diesel
fuels1 and is a viable alternative to gasoline in spark
injection (SI) ICE (internal combustion engine)
mobility applications. As a result, its use for mobility
has been increasing. In India for example, it is
becoming the fuel of choice for urban mass transit in
both large buses and in smaller three-wheeled
vehicles. However, NG normally requires larger
fuel-to-air ratios when used in spark ignition (SI)
engines, resulting in higher operating temperatures.
This is due to methane (the major constituent of
natural gas) being one of the least ignitable of the
Figure 1: RIS(A)-PCSI 2-Stroke Engine Design,
general features
Herein, the potential of an example RIS(A)-PCSI
(radical ignition species (augmented) – premixed
charge spark ignition) 2-stroke engine design (Figures
1-3) for reducing this NOx trade-off penalty is
considered. This innovation takes advantage of the
pre-ignition and ignition chemistries of HCRI
3
(homogeneous combustion radical ignition). Though
the main chamber combustion in the RIS(A)-PCSI
technology is not homogeneous, the controlled
pre-placement of a regulated presence of radical
ignition (RI) species ahead of the moving flame front

1
causes the front’s chemical-kinetics to be altered to
include the HCRI chemistry. This approach has been
experimentally shown to reduce NOx in a 4-stroke
RIS(A)-PCSI engine operating on NG.4 In general,
the use of RI species has been established as a
natural way to extend the combustion fuel-lean limit of
3-7 8
a number of hydrocarbon and alcohol fuels and in
many applications has been shown to increase fuel
ignitability over entire engine operating regimes.3-8
Figure 2: RIS(A)-PCSI 2-Stroke Engine Design, head
details
EARLIER 2-STROKE RADICAL ENGINES
The first 2-stroke engine to make purposed adjunct
use of select radical species in its combustion process
was developed by Noguchi, et. al.9 Their resultant new
self-ignition combustion process was called
Toyota-Soken (TS) combustion. The intake-exhaust
port design of the horizontally opposed-piston TS
engine created an overlap between the leaving
burned gases and the entering fresh charge. This
allowed for the transfer of heat and also the carry-over
of some chemically active products to the cold fresh
charge, enabling low-temperature self-ignition.
High-speed photographs of flame propagation
showed ignition occurring spontaneously at numerous
points around the center of the combustion chamber
and then spreading rapidly in all directions. It was
initially believed that this result was due to the early
presence of radicals. Optical measurements detected
CHO, HO2 and O prior to ignition, followed by CH, C2
and H, and finally the OH radical (where in
conventional spark-ignition combustion, all these
radicals first occur at almost the same time).
However, prior to the onset of ignition (before 25o
BTDC), none of the radicals CHO, HO2, O, CH, etc.
were detected. Thus, it was considered unlikely that
these radicals were being obtained directly from the
exhaust gases, but were originating instead from other
2
sources. Also, there was no experimental evidence for
the presence of the dominant RI species
formaldehyde (CH2O) and hydrogen peroxide (H2O2)
(two of the prime RI species facilitating the HCRI
chemistry). Therefore, TS was eventually deemed
not to be a RI homogeneous combustion engine, but
rather a TI (thermal ignition) homogeneous
combustion engine with some level of radical activity.
TS has since come to be properly classified together
with the family of homogeneous-charge
compression-ignition (HCCI) engines.
Independently, Onishi, et. al., developed a two-stroke
homogeneous-charge lean burn gasoline engine
(called the Active-Thermo-Atmosphere Combustion
(ATAC) engine) operated in the compression-ignition
mode. In spite of the different name, ATAC involves
essentially the same TI phenomena as TS
10
combustion.
The first example of use of a strong presence of the RI
species typical of the HCRI chemistry to augment
self-ignition and combustion in a direct injection (DI)
2-stroke engine is reported in reference [11].
Autoignition in this DI engine was accomplished via RI
species to enable ignition under the conditions of
homogeneous combustion. For this DI-HCRI engine
there were five main means available to control the
start of (SOC): (1) fuel injection quantity management
within the mini-chambers; (2) temperature
management of the mini-chambers; (3) management
of the fuel injection timings within the main chamber;
(4) management of the boost pressure provided by
either a turbocharger or a supercharger; and (5) the
use of controlled cool EGR to recycle RI species.
The fourth of these means, boost pressure
management, was shown fully capable of unilaterally
controlling the timing of the SOC for homogeneous
combustion in this 2-stroke engine for single ignition
mode (RI) operations across the entire operating
11
regime of the engine.
STUDY OBJECTIVES
Earlier, the study of the HCRI chemistry was focused
on how both to enhance and to better control
self-ignition in a variety of engine types (2-stroke,
7-8,11
4-stroke and rotary) with various fuels. However,
more recently, there has also been keen interest in
using this chemistry to extend the fuel-leanness limit
for which an SI flame front can be made to stably
propagate. To this end, over the past several years,
the affects of the HCRI chemistry on flame fronts has
been examined experimentally using the
constant-speed premixed-charge single-cylinder
4-stroke SI natural gas test engine of the Argonne
National Laboratory (ANL).
The ANL natural gas test engine could not be
significantly modified for the series of experimental
tests with the RIS(A)-PCSI (RIS(A)-premixed charge
SI) engine technology. As a consequence, only
control “passive” mini-chambers could be configured
into the head of this engine, with no means for
regulating either the production or transfer (to the
main chamber) of the RI species (generated therein).
No other changes were possible. Consequently, the
ANL test series represented an extreme challenge for
an ignition that is dependent on the HCRI chemistry.
The main aim of this ANL work was to study the
potential of the HCRI chemistry for NOx reduction in
SI applications. The design of the prototype for these
studies was hastened by simulation. Thus, not
surprisingly, with only the second design iteration the
high load ANL tests at a φ (equivalence ratio) value
of 0.65 resulted in simultaneous NOx and CO
emissions reductions (of 32% and 4%, respectively).
During the testing of another design iteration, the
low-load fuel-lean limit was extended (from baseline)
by 4.4% to a φ value of 0.63 (from 0.66), resulting in
a 62% reduction in NOx emissions (from baseline).
With this same other ANL RIS(A) prototype design
iteration, at high load the fuel-lean limit was instead
extended from a φ value of 0.65 to 0.62 (i.e., by
4.8%) and the NOx emissions were reduced by 71%
(from baseline) . The only other means by which ANL
has ever been able to come near to this fuel-lean limit
at high load on their natural gas test engine has been
via the use of a laser ignition plug (though this newer
RIS(A) high load result is still slightly leaner and its
accompanying NOx reduction is significantly
greater).12 The results from this ANL physical study
have since been corroborated by simulation.4
The encouraging results from this ANL exercise has
prompted further research interests in the
RIS(A)-PCSI technology.13 One of these interests has
been in examining the potential of this technology in
2-stroke engines with the use of control “active”
mini-chambers (for potentially better results). Further,
and in line with this, the main aim of the current study
is to help determine how the presence of radical
ignition species can be better used to facilitate low
NOx operations in the 2-Stroke RIS(A)-PCSI engine.
A second aim is to study the characteristics of the
natural gas RIS(A) SI combustion process at a typical
SI compression ratio (CR) in a turbocharged 2-stroke
engine cycle. A third aim is to test the use of a new
active RI species transfer control procedure to help
reduce the total volume of M-Cs needed to sustain
RIS(A) SI operations in 2-stroke engines. One
additional aim is to develop a means for more rapidly
designing future 2-Stroke RIS(A)-PCSI engine
variants.
THE 2-STROKE RIS(A)-PCSI ENGINE
A simple schematic of a sample 2-stroke RIS(A)-PCSI
engine is given via Figures 1-3. Figure 1 shows a
schematic depiction for one “cylinder” of this engine.
This is not a true cross section, though it is cast in the
axial plane that passes perpendicularly through line
D-D of Figure 2. As is apparent from a comparison
with Figures 2 and 3, this figure (Figure 1) presents
information from multiple planes. Figure 2 is a
functional schematic of this engine’s head
arrangement. Figure 3 is a schematic (again not a
cross section) of the engine’s intake port and exhaust
port arrangement looking down on the cylinder when
the piston is at BDC. These figures are thus
3
intended as functional representations only. In
addition to the particular arrangement shown
schematically through Figures 1-3, many other design
variations are possible for the 2-stroke RIS(A)-PCSI
engine. Use can be made of one or more sets of
secondary chambers, called mini-chambers (M-Cs),
typically located in the head or/and upper portion of
the cylinder liner.
Figure 3: RIS(A)-PCSI 2-Stroke Engine Design,
inlet-exhaust details with piston at BDC
As shown through Figures 1 and 2, in this study’s
sample design, two M-Cs (mini-chambers) emanate
from the center of the cylinder head. A control fluid
inserter (insertion device) feeds these M-Cs with
control fluid. This device is designed to insert much
smaller metered quantities of either fuel or air (or
another fluid) into the mini-chambers through M-C
insertion orifices. In the current study the control
fluid employed is the fuel. In Figure 1 a single simple
vent per M-C is shown, though there may be more
than one co-located vent per M-C. These vents are
of appropriate lengths and diameters to ensure both
“flame quenching” of the combustion flames in the
main chamber (so that they do not reach the M-C and
cause it to autoignite)14 and throttled flow between the
main and mini chambers. These vents are designed
(aimed) to ensure that the outflow mass jets (that at
times discharge from their mini-chambers) are
directed to enable their mixing with the inlet gases.
Also shown in Figure 2 is the M-C ignition device for
facilitating the active RI species transfer control
procedure. Finally, as can be seen in Figure 3, four
inlet ports and two exhaust ports are symmetrically
located in the cylinder liner.
The specifications and operating conditions used for
the example engine design and simulation analysis
are provided in Table 1. These values are designed to
be reasonable 2-stroke equivalents of those used in
the earlier 4-stroke RIS(A)-PCSI engine experimental
study conducted at the Argonne National Laboratory
(ANL).4 The purpose of creating such coordinated
specifications is to define a 2-stroke RIS(A)-PCSI species dominated flame front propagation for both
“simulation state” that is plausibly close to the 4-stroke the mini and main chambers.
ANL RIS(A)-PCSI prototype specifications used
during its experimental high-load lean-limit run. To accommodate these critical requirements, while
also accommodating the analysis objectives for this
Table 1: 2-Stroke RIS(A)-PCSI Engine Specifications 2-stroke RIS(A)-PCSI engine combustion study, the
and Operation Conditions simulation makes combined but modified use of two
established and corroborated HCRI coding
Effective Compression Ratio 11.0:1 formulations. The first of these is the overall HCRI
Bore (cm) 13.0 combustion coding architecture for HCRI without
3
Stroke (cm) 13.5 flame fronts and modified for 2-stroke engine
11
analysis. The second of these is the advanced
Connecting Rod (cm) 25.5
coding architecture used in the 4-stroke RIS(A)-DISI
Charge Equivalence Ratio ( φ overall ) 0.62 13
flame front studies and in the main chamber 4-stroke
4
Engine Speed (RPM) 1800 RIS(A)-PCSI flame front studies.
Power at high load (kW) 66
Air-Inlet Pressure (Atm.) 2.601
Exhaust Manifold Pressure (Atm,) 2.340
Volume One M-C, Volm-c (cm3) 1.5
Surface Area One M-C (cm2) 10.41
Number of M-Cs 2
Air-Inlet Temperature (K) 313.15
Top Cylinder Temperature (oC) 307.0
Side Cylinder Temperature (oC) 285.0
Piston Surface Temperature (oC) 260.0
M-C Surface Temperature (oC) 250.0
Fuel (by mass) 95% CH4+5% C2H6
M-C Fuel Injection Temperature (oC) 325.
M-C Fuel Injection Pressure (bar) 349.
M-C Fuel Injection Orifice Dia. (cm) 0.0287
No. of M-C Fuel Injection Orifices 1
M-C Fuel Injection Start (deg CA) 100 BTDC Figure 4: Sample 2-stroke RIS(A)-PCSI schematic
M-C Fuel Injection Duration (deg CA) 45 showing main chamber simulation zones
(b -burned gas, flame front, u - unburned gas)
M-C Flame Front Start (deg CA) 165 BTDC
M-C Flame Front Duration (deg CA) 6
Total (all 4) Intake Port Width (cm) 9.4
Intake Port Max Height (cm) 2.5
Intake Port CD 0.90
Total (of 2) Exhaust Port Width (cm) 8.4
Exhaust Port Max Height (cm) 2.5
Exhaust Port CD 0.75
Height of Mid-Piston Rise (cm) 1.0

SIMULATION MODEL

Critical to simulating the chemical-kinetics and

thermo-fluid processes occurring in this new
combustion engine are: (a) a set of open-system
energy and species conservation equations for the
main and mini chambers; (b) empirical formulations
for the heat transfer taking place between the gases
and all surfaces of main and mini chambers as well as
the surfaces of the connecting vent walls; (c) Figure 5: Sample 2-stroke RIS(A)-PCSI schematic
momentum equations for the exchange between the showing mini-chamber simulation zones
engine manifold (through the ports) and the mini and (b -burned gas, flame front, u - unburned gas)
main chambers (through the connecting vents); (d) an
accounting for the conservation of mass and species In addition, details of an advanced solver architecture,
during mass and momentum exchanges between that is also the foundation of the new 2-Stroke
systems (between the manifold and the main chamber RIS(A)-PCSI code, are given in reference [15]. This
and between the mini and main chambers); (e) an architecture is primarily based on a multidimensional
equation of state for the gases of the mini and main secant method16 for the solution of the temperatures
chambers; (f) a chemical-kinetics mechanism for and species mass fractions of the various chambers.
CH4-H2-NOx-Air(O2-N2-AR); and an accounting for RI This solution also incorporates a second-order

4
prediction procedure15 to help speed convergence.
Combined with this, a different third-order prediction
15,17
method is used to project the next half time step
solution for the pressures of both main and mini
chambers, as well as the value of the mass flow rates
in the conduits. Finally, a vent flame quenching
14,18
model, is used to project the next half time step
bulk temperature of the gases coming into one system
(chamber) from the other (via the connecting
conduits/vents). This projection is also of third order
15
accuracy. Together with the species concentration
values, this bulk temperature value is in turn used to
determine the incoming enthalpy of the gases being
transferred from one chamber to the other (sometimes
at very high rates). Later within each time-step
computation the half-time step vent bulk-temperature
value and the two half time-step chamber-pressure
15
values are corrected to third-order accuracy for the
start of the next time-step calculations.
The newer features of the present work stem mainly
from the adaptation of both the main chamber 3-zone
4
multiple solution region solution approach (Figure 4)
and the new M-C 3-zone solution multiple solution
region approach13 (Figure 5)) into the original 2-stroke
HCRI solution code11. It is these adaptations that
make possible the insertion of experimental ANL
flame front data into the current engine design under
compatible operational conditions (the conditions of
Table 1). An outline of the novel 2-stroke
RIS(A)-PCSI code is given in the Appendix.
Figure 6: Semenov’s general ignition limits19
RESULTS AND CYCLE DISCUSSION
As explained, the numerical results presented are for
the 2-stroke RIS(A)-PCSI configuration needed for
this new engine concept to undergo flame propagation
combustion under design conditions that are similar to
those of the earlier ANL 4-stroke RIS(A)-PCSI engine
4
high-load lean-limit tests. Thus the results presented
herein are for the engine under the high-load
conditions of Table 1. The constituency of the natural
gas fuel for these numerical calculations is considered
to be a mixture of 95.0% methane and 5% ethane by
mass. Via these results, it will be shown how the
5
2-stroke RIS(A)-PCSI cycle entails a succession of
individual but “connected” sub-processes occurring
“within” and “between” the mini and main chambers,
as well as “with” the intake and exhaust ports.
This succession can be better understood with the
help of Figure 6, applicable to the general ignition
limits of most hydrocarbon fuels19 (including
methane20-21, see Appendix Figure A1). However, for
NG, the extent of the domain of the cool-flames region
can vary with different factors. For example, the
introduction of even small quantities of formaldehyde
(CH2O) into a slow methane-oxygen reaction will induce a
21
cool flame. Significant quantities of CH2O are
recurrently maintained in both the main and mini
chambers during the RIS(A)-PCSI cycle, making the
cool-flame domains in these chambers significant and
constantly varying.3
It is also helpful to first examine the cycle
temperatures of the new 2-stroke NG RIS(A)-PCSI
cycle in some detail in order to gain an overview
understanding of this process succession. In Figure 7
the cycle history of the global temperatures for both
chambers is presented. These are the “effective”
(overall) temperatures for these chambers found from
maintaining a strict conservation of energy for each
chamber during the cycle (see the Appendix).
Figure 7: Cycle Global (Effective) Temperatures,
Main (1) and Mini (2) Chambers
RI species are “carried over” within the M-C from the
previous cycle. Then, starting at 165 deg CA BTDC,
the fuel-air mixture in the M-C is rapidly ignited, driving
some of the RI species resident in the M-C (in the
unburned mass region (u) of the M-C) into the main
chamber. This new active RI species transfer control
procedure consists of metering desired amounts of
fuel DI into the M-C during the compression event of
the previous cycle (to adjust the M-C RI species
generation) and then “time regulating” this new rapid
M-C ignition procedure (as per the example timings in
Table 1).
This M-C ignition sub-process can be better visualized
with the help of Figure 5. As the flame front
propagates across the mini-chamber, the chemistry
occurring within the mass sliver of gases at the
interface of the flame front (where the unburned “u”
and burned “b” masses of the leading sub-zone [b1 or
b2] meet in Figure 5) follows the same pre-ignition
chemistry found in main chamber autoignition with NG
in the DI-HCRI engine.3 The state of the M-C gases
ahead of the flame front is typically within the
slow-combustion region (A) (or sometimes in the
cool-flames region) of Figure 6. Found dominant
during this pre-ignition chemistry is the reaction set:
Fi + HO2 = Ri,,n + H2O2 (1)
where Fi is the fuel molecule (primarily CH4), Ri,n is a
radical species with one less H than the fuel molecule
Fi (primarily the CH3 radical). The subscript “n” is used
with Ri because, for the larger molecule components
of the NG fuel (ethane, etc.) there are more that one
species resulting from this reaction set.
The effect of this ignition sub-process on the early
M-C temperatures (T2) is clearly visible (via the large
spike in these temperatures starting at 165 deg of CA
BTDC in Figure 7). Shortly afterward, the temperature
profile of the mini-chambers resume their more
3,7-8,11,22
general HCRI shapes.
In Figure 8 the histories of the vent and two port mass
flow rates ( m& v , m& in _ port and m& ex _ port ) are given.
Again, the effect of the procedure for boosting the
presence of main chamber RI species can be seen on
m& v near the beginning of compression. Because of
the effect of charge dilution (with mass exchange), the
use of Figure 8 is needed for properly interpreting the
mass fraction data plotted in Figures 9 and 10.

Then, as the advancing high temperature flame front

physically encompasses more of this same mass
sliver of unburned gas (which is still at a much lower
temperature) within the flame interface, the following
reaction set begins to dominate at lower
temperatures:
H2O2+M = OH+ OH+M (2)
Fi + OH = Ri,n+ H2O (3)
where M is an arbitrary third body species, Fi is a fuel
molecule (again primarily CH4 for NG), Ri,n is a
radical species with one less H than the fuel molecule
Fi (again primarily the CH3 radical for NG). The CH3
radical is highly reactive and helps set in motion a
series of other reactions.

Reaction Equation (2) serves as chain initiation for

this ignition mechanism sub-process and reaction
Equation (3) serves initially as the chain propagation
reaction. The state of the M-C gases during the time
these reactions are operating will typically transition
from region A to region B of Figure 6.
Figure 8: Cycle Vent (Mdot_Vent) and Port (Mdot_In
and Mdot_Ex) Mass Flow Rates
(Positive flow is into the Main Chamber)

As soon as the temperature becomes high enough
within this mass sliver to support direct oxidation, the
ignition mechanism in this portion of the flame front
begins to shift to more conventional oxidation based
flame front chemical-kinetics. For NG conventional
chain initiation is via the reaction set:
Fi + O2 = Ri,n+ HO2 (4)
The mass behind the flame front is at much higher
temperatures than the unburned mass ahead of the
flame front. Thus, behind the flame front a mass
quantity has more volume after it has burned than the
same mass quantity possessed beforehand.
Therefore, as the flame front moves through the M-C,
creating a growing zone of burned gases, it is the
expansion of these burned gases that is forcing the
unburned gases (u) ahead of the flame front (Figure 5)
into the main chamber.
At the start of the compression stroke all of the M-C
gases are in the unburned zone of Figure 5. As the
flame front starts propagating, the solution shifts to a
three-zone approach (b, flame front, u) and a short
while afterward the special conditions of sub-zone b1
start being computed. With time, as the flame further
propagates, the special computations in the burned
mass zone (b) further include sub-region b2.
Shown in Figure 9 are the key M-C RI species mass
fractions of methane in the unburned zone (ahead of
the flame front). The CH2O and H2O2 are significantly
diminished during M-C ignition. As can be further seen
in Figure 9, after the conclusion of flame front
propagation, the solution of the two zones (b and u)
merges back into one solution zone (using a weighting
procedure for the regions b1, b2 and u). Due to the
extremely high turbulence levels that keep the M-Cs
well stirred while mass is flowing into the

6
mini-chambers, this numerical merger is most realistic
physically. Further, the effect of charge dilution on the
M-C RI species CH2O and HO2 mass fractions during
compression can be clearly seen in Figure 9. While
the H2O2 is similarly affected, its mass fractions in
both chambers are closer in magnitude. Similarly, the
dramatic growth in this mass fraction, that starts
before the end of the compression event (along with
the acceleration in growth of the mass fractions of
these other two RI species), is in large measure due
to a concurrent main chamber sub-process.
Figure 9: Mini-Chamber Unburned Zone Mass
Fractions of Select RI Species
Figure 10: Main Chamber Unburned Zone Mass
Fractions of Select RI Species
7
Thus, at this juncture in the cycle, the focus shifts to
the events occurring in the main chamber during
compression. As the temperature of the gas mixture in
the main chamber increases during compression, the
state of the gases are in transition up and to the right
through region A of Figure 6. During this time,
because of the already significant (strong) presence of
the RI species in the main chamber, a pre-chemistry
sequence occurs that is not typical of conventional SI.
The path of this main chamber pre-chemistry is very
similar to the path of the pre-chemistry occurring in the
mini-chambers, with reaction Equation (1) again
dominating at first and with reaction Equations (2) and
(3) growing in significance. These reaction sets result
in the production of Ri,n radicals like CH3, which can
then react very easily with oxygen and with many of
the other species present. One result of the
chemical-kinetics sub-mechanism established during
this time is the further growth in the mass fractions of
the RI species themselves (as seen in Figure 10). The
sub-mechanism for the creation of the RI species from
the CH3 radical (and from the other Ri,n ) is highly
complex.
Then, after the initiation of spark ignition in the main
chamber, as the flame front progressively reaches
new hemispherical slivers of mass, the same flame
front ignition sub-process occurs as in flame
propagation in the M-Cs. Initially again, as the flame
front starts influencing a particular mass sliver,
reaction Equation (1) serves as chain initiation. As the
flame front encompasses more of this mass sliver,
reaction Equation (2) serves as chain initiation and
reaction Equation (3) serves as chain propagation.
Eventually, as the heat of the flame front engulfs the
mass sliver, reaction Equation (4) takes over the role
of chain initiation and the chemical-kinetics follows
more of the chemistry of conventional SI.
Akin to the approach used in the M-C, as the flame
front starts propagating through the main chamber,
the solution thus again shifts to a three zone approach
(b, flame front, u) to accommodate this ignition
sub-process. With time, as the flame propagates,
the burned mass zone further evolves into multiple
regions (b1, b2 and b3) (Figure 4).
In parallel, the simulation considers the individual
chemistry and physics of each mass sliver. It does this
by computing and tracking their temperatures for use
in computing the overall temperature of the unburned
region ( Tb ) (see Appendix). As can be further seen in
Figure 10, after the conclusion of flame front
propagation, the solution of the two large zones (b and
u) merges back into one solution zone (using a
weighting procedure for the regions b1, b2 , b3 and u).
Due to the extremely high turbulence levels being
created by the mass jets leaving the M-Cs through the
vents at the time of this re-convergence of solution
zones (Figure 8), this numerical merger is also
physically well justified.
As mentioned, during the middle portion of the
compression event, a regulated amount of fuel was
injected into the M-Cs. Also, fuel is conveyed into
the M-C from the main chamber during the
compression event (and during the first portion of
main chamber SI) by the high mass transfer into the
M-Cs shown in Figure 8. The entrained amount of fuel
is deterministic and is used together with the
regulated amount to help control RI generation in the
M-C. Under the states (pressure, temperature and
chemical conditions) in the vicinity of area "A" (in
Figure 6) and in the presence of the moderate but
ever growing presence of RI species concentrations,
the "entrained" and “injected” portions of the fuel are
then decomposed in M-C in a non-explosive
luminescent process.
Much of the OH needed for starting this RI species
generation sub-process is obtained via the thermal
decomposition of portions of the H2O2 being
transferred to M-C from the main chamber when the
piston is in the vicinity of TDC. This
thermal-decomposition reaction is also exothermal,
and can serve as a chain initiation reaction for the RI
species generation sub-process:
H2O2+M = OH+ OH+M (2)
Figure 11: Chamber Heat Release Rates, Main
Chamber (1) and Mini-Chamber (2)
One result of this sub-process is more RI species. The
general chemical-kinetics for this M-C RI species
generation sub-process is given in some detail for
methane in [3]. Depending on the design of the
RIS(A)-PCSI cycle (and not in this case), for a time
some quantities of OH may also be directly
transferred to the M-Cs from the main chamber via the
pressure driven mass jets. During main chamber
combustion the concentration of OH in the burned
mass zone (b) spikes up (and remains high as long as
the temperatures are high). These jets strengthen
significantly after spark ignition. Regardless of the
source, it is this presence of OH that acts as a type of
8
catalyst to rapidly convert the resident M-C fuel into
select “radical ignition species.” This is because, in
the ensuing mini-chamber RI species generation
process, the most dominant of the reaction sets found
in this complex chemical-kinetics mechanism is again
3,7
the reaction set:
Fi + OH = Ri,n+ H2O (3)
This non-explosive exothermic OH-driven M-C RI
species generation sub-process occurs primarily
during the expansion phase of the main chamber
cycle and results in the creation of significant new
quantities of RI species in the M-Cs. With the NG
being primarily methane, the main RI species
generated are CH2O, H2O2 and HO2. Later in the
expansion phase, with the decline of pressure and
temperature in the M-Cs (due to main chamber
expansion), the RI species CH2O and H2O2 become
non-reactive. Also, the RI specie HO2 becomes far
less reactive. This occurs after the state of the mixture
in the M-Cs has transitioned down through the cool
flames region and is well to the left and down in the
slow combustion region shown in Figure 6. By the
time the exhaust ports open these species have
become essentially reactively inert (Figure 9). The
non-reactive chemical condition of these RI species is
referred to as “frozen equilibrium”.23 Also later in the
expansion phase, as the temperature in the main
chamber drops, the fuel and RI species in the mass
jets coming out of the M-Cs react with the OH still
resident in the main chamber to produce additional RI
3
species (Figure 10).
Combustion heat release rates for both of the
chambers are given in Figure 11. Again the effect of
the M-C ignition procedure is most visible in this figure,
as well as in Figure 12 for the cycle pressure history
for both chambers.
Figure 12: Pressures, Main (1) and Mini (2)
Chambers
Experimental and simulation results for the 4-stroke
RIS(A)-PCSI engine are reported in [4]. By making
use of these earlier results, comparisons between
these two and four stroke “specification-compatible”
variants of the RIS(A)-PCSI engine can be analyzed.
In this earlier physical 4-stroke engine prototype,
coolant-channel related head space limitations made
the use of control passive M-Cs necessary. Thus the
procedure used herein to push RI species out the
M-Cs could not be employed. This necessitated that
the total volume of the M-Cs be substantially greater
in the ANL RIS(A)-PCSI 4-stroke variant than the
volume used herein. These differences are a part of
what has been accounted for in determining the
“simulation state” for the current 2-stroke RIS(A)-PCSI
engine specifications (Table 1) required to attempt to
effectively replicate the combustion characteristics of
this earlier RIS(A) ANL engine prototype.
Figure 13: RIS(A)-PCSI 2-Stroke (left) & 4-Stroke
(right) Main Chamber Unburned Zone RI Species
Mass Fractions
In Figure 13 the main chamber unburned zone mass
fractions of key RI species for the 2-stroke design are
compared with the earlier premixed charge 4-stroke
engine RI species mass fraction simulation results.
From this it can be seen that, even with the smaller
total volume of M-Cs in use, the 2-stroke engine has a
substantially stronger presence of RI species (than
the 4-stroke engine) in the main chamber after the
time of the “effective SOC” at the approximate point in
time that the flame front actually starts to physically
propagate (15 deg of CA BTDC) in each. This
suggests that even smaller total M-C volumes can be
used for 2-stroke RIS(A)-PCSI designs.
Further, a comparison of simulated NOx production
for the 2-stroke RIS(A)-PCSI and ANL 4-stroke
RIS(A)-PCSI4 engines is provided in Table 2. The
4-stroke results are based on revised simulations. It
can be seen in Table 2 that simulation predicts less
NOx production per cycle with this new 2-stroke
RIS(A)-PCSI technology variant.
9
Table 2: Simulated 4-Stroke and 2-Stroke
RIS(A)-PCSI NOx results at high-load lean limit
Engine φ NO NO2 NOx
g/kW-hr
RIS(A)-PCSI g/cycle g/cycle
-5 -4
4-stroke .62 1.53x10 2.30x10 0.401
-5 -5
2-stroke .62 1.03x10 8.30x10 0.153
Many of the above aspects of flame front chemistry of
the 2-stroke RIS(A)-PCSI engine technology with
natural gas were also observed earlier during the ANL
experiments. One of these observations concerns the
effect of the RI species as the engine approaches its
fuel-lean limit at a given load. In one conventional
(baseline) engine run at low load, as the ANL baseline
engine approached its lean limit, the COVIMEP
(Coefficient of Variation for the Indicated Mean
Effective Pressure) was found to increase rapidly. In
contrast, the COVIMEP of the RIS(A) PCSI engine was
experimentally found to hardly increase as its lean
limit was approached, despite the fact that in this
earlier run series the RIS(A) lean limit was 4.4%
“leaner” than that of ANL baseline engine at the same
low-load condition. This earlier RIS(A)-PCSI
experimental study thus suggests that the presence of
the RI species was enabling greater combustion
stability, with the combustion characteristics of the
engine remaining more consistent from cycle to cycle.
The current simulation study sheds a bit more light on
these earlier experimental findings. As observed in
Figure 10, the pre-ignition chemistry in the unburned
mass ahead of the flame front results in an
impressively rapid build up of the RI species CH2O,
H2O2 and HO2. Thus hydrogen peroxide is readily
available (in much greater quantities than in the
conventional SI case) to participate in the early portion
of the flame front chemistry as the advancing flame
front starts to consume a particular new hemispherical
mass sliver. At this particular juncture in this process
(flame front encompassing a new mass sliver), the
observed improvement in the combustion stability and
the additionally observed ability of the general
RIS(A)-PCSI engine for leaner combustion are
partially explained by comparing the relative speeds of
chain initiation for the HCRI chemistry (Equation (2))
with chain initiation for conventional flame front
chemistry with methane (Equation (4)). With Fi as CH4
and with Ri,n as CH3, at a temperature of 1050 K
(well above the temperature required for this reaction
to occur) the forward reaction rate of reaction
Equation (2) is 1.095x104 times faster than forward
reaction rate of conventional SI reaction Equation (4)
at a temperature of 1200 K (a typical temperature
required for its start).24 Thus, along the leading edge
of the flame front, the combustion of the mass sliver is
able to begin at a temperature that is much lower
than that required for Equation (4) (for conventional
combustion) and, as the temperature of a given mass
sliver section rises, the rates of Equations (2) and
(3) grow faster. Ignition of the mass sliver under
leaner conditions is assured, and thus also the
facilitation of greater combustion stability.
The rationale for the decrease in the NOx production
at given φ values also goes back (in part) to this
new chemical-kinetics caused by the much stronger
than normal (conventional) presence of RI species.
From both the current simulation study and the earlier
ANL experimental work, it is now clear that there are
two reasons why the NOx is being reduced. The first
follows from the ability to enable leaner combustion,
as explained in the previous paragraph. With NOx
production mostly a thermally driven process, fuel
leaner combustion means lower local and global main
chamber temperatures and less NOx. However, even
when the φ values in the ANL baseline and ANL
RIS(A) engines were the same, in one high-load run
with adequate RI species present the NOx was
significantly reduced with an accompanied reduction
in CO. This is because the flame front propagation
speed is influenced by the presence of the RI species
in the unburned gases ahead of the flame front. As the
flame front propagates, ignition chain initiation of each
given hemispherical mass sliver can start at a much
lower temperature (via reaction Equations (2) and (3)).
Thus, with the presence of RI species ahead of the
flame front, while the flame front propagation remains
more stable (even at smaller [i.e., leaner] φ values),
the flame front propagation speed is decreasing.4
Consequently, the rate of main chamber pressure rise
is slower, as the pressure rise itself is being sustained
while being distributed over a longer period of time.
This in turn results both in the lowering of main
chamber temperatures and in a “pressure production
rate” that is more compatible with reciprocating piston
motion during the early to mid expansion portion of the
engine cycle.
CONCLUSIONS
This study has set out to examine the viability of
placing the new concept 4-stroke RIS(A)-PCSI engine
technology into a 2-stroke engine application. For this,
the characteristics of the natural gas RIS(A) SI
combustion process in a turbocharged 2-stroke
engine cycle with an 11:1 CR have been studied. The
design employed for this study uses only 1/3 the M-C
volume employed earlier in an engine equivalent
4
4-stroke ICE ANL experimental study. Never-the-less,
the simulation predictions show this newer 2-stroke
RIS(A)-PCSI design actually maintaining more RI
species ahead of the flame front in the main chamber
than were found being maintained in the older engine
(Figure 13), resulting in comparative NOx reduction.
This improved comparative result was partially
facilitated by the use of control active mini-chambers
and by the employment of the new active RI species
transfer control procedure.
10
From the results of this novel exercise it can now be
suggested that implementation of the RIS(A)-PCSI
technology into two-stroke NG engine applications
should result in NOx reductions on a scale that is
similar to or better that found in equivalently
configured 4-stroke variants of the same technology.
If this conjecture is correct, the same experimentally
corroborated reduction percentages of the earlier ANL
study4 should also be obtainable. However, until
dedicated experimental tests are conducted to
replicate the current simulation findings within a
physical 2-stroke RIS(A)-PCSI engine, the validity of
this conjecture remains unsubstantiated. Thus,
encouraged by the findings herein, the next step of
developing a physical prototype of this new concept
engine is already underway, with the results of the
current study providing a roadmap for the rapid design
of this new physical prototype.
ACKNOWLEDGMENTS
This work is highly indebted to the excellent support
provided by Thomas Wallner, (Ph.D.) of the Center for
Transportation Research at the Argonne National
Laboratory, U.S.A.
REFERENCES
1. Rubas, P.J., Paul, M.A. et.al., “Methane Jet
Penetration in a Direct-Injection Natural Gas
Engine”, SAE Paper 980143, 1998.
2. Lovell, W.G., “Knocking Characteristics of
Hydrocarbons”, Industrial and Engineering
Chemistry, Vol. 40, No. 12, Dec., 1948.
3. Blank, D. A., “CNG / Methane-Combustion in
Homogeneous-Combustion Radical-Ignition D.I.
Diesel Engines”, SAE Paper No. 2007-01-0047,
2007.
4. Blank, D. A., “NOx Reduction in Natural Gas RI
Species Augmented Large Bore Four-Stroke SI
Engines”, SAE Paper No. 2011-26-0005, 2011.
5. Blank, D. A., “CNG / Methane-Combustion
Kinetics (without N2) and Frozen Equilibrium in
Radical-Initiated Low Compression Ratio D.I.
Diesel Engines Using Pistons with
Micro-Chambers”, 2004 SAE Transactions, Vol.
113, Journal of Fuels and Lubricants, Section 4,
pages 724-763, 2004 (also SAE Paper No.
2004-01-1677).
6. Blank, D. A. and Pouring, A. A., “Radical
Controlled Autoignition at Reduced Compression
Ratios in a Hydrogen D.I. Diesel Engine with
Piston Micro-Chambers”, 2004 SAE Transactions,
Vol. 113, Journal of Engines, Section 3, pages
1185-1182, 2004 (also SAE Paper No.
2004-1846).
7. Blank, D. A., "Methane/CNG Combustion in a DI
Radical-Ignition Rotary Engine with Low-Heat
Rejection", JSAE Paper 20077210 and SAE
Paper 2007-01-1912, 2007.
8. Blank, D. A., "Lean Combustion
Chemical-Kinetics Studies of and Ethanol
Four-Stroke Radical-Ignition DI-Engine", 2004
SAE Transactions, Vol. 116, Journal of Engines,
 Section 3, pages 444-462 (also SAE Paper No. of Methane” Chemical Engineering Science, Vol.
2007-01-0623). 48, No. 14. pp. 2643-2661, 1993.
9. Noguchi, M, Tanaka, Y., Tanaka, T. and Takeuchi, 25. https://trac.hpc.usu.edu/projects/hauserlab/brows
Y., "A Study on Gasoline Engine Combustion by er/thauser/Programs/OpenFOAM-1.2/applications
Observation of Intermediate Reactive Products /test/readCHEMKINIII/
During Combustion", SAE Paper 790840, 1979. CHEMKINdata/chem.inp_c8h18_egr.
10. Onishi, S., Jo, S.H., Shoda, K., Jo, P.D. and Kato, 26. Heywood, J. B., Internal Combustion Engine
S., "Active-Thermo-Atmosphere Combustion Fundamentals, McGraw-Hill, New York, 1988.
(ATAC) - A New Combustion Process for Internal
Combustion Engines", SAE Paper 790501, 1979. CONTACT
11. Blank, D. A., “Radical Ignition Combustion Studies
with Hydrogen in a Two-Stroke DI-HCRI Diesel David A. Blank, Ph.D.
Engine”, SAE Paper 2007-01-0135, 2007. Fernworth House, Landour Cantt.
12. Biruduganti, M., Gupta, S., et. al., “Air Separation Mussoorie-248179 (Uttarakhand)
Membranes – an Alternative to EGR in large Bore India
Natural Gas Engines”, ICES2009-76054. dablank@hcriti.com
13. Blank, D. A., “Combustion Control 91-135-263-0119 / 91-931-944-0834
Chemical-Kinetics Studies with Natural Gas in
HCRI Enhanced Four-Stroke DI SI”, SAE Paper NOMENCLATURE, ABBREVIATIONS
2011-01-1895 and also JSAE Paper 20119330,
Symbols
2011.
M third body species
14. Blank, D.A., Pouring, A.A. and Lu, J., “Flame m mass
Quenching in the Micro-Chamber Passages of I.C. m& , Mdot mass-flow rate
Engines with Regular-Symmetric Sonex Piston P, Pj pressure, system j pressure
Geometry” SAE Paper 2001-28-0002, 2001. T, Tj temperature, system j gas temperature
15. Blank, D.A., "Radical Controlled Autoignition in a x mass ratio
HCRI Hydrogen DI Four-Stroke Diesel Engine φ fuel-air equivalence ratio
with Reduced Heat Rejection", SAE Paper
2007-01-0013, 2007. Subscripts
16. Broyden, C.G., Mathematics of Computation, vol. b burned mass
19, pp. 577-593, 1965. ex exhaust
17. Minkowycz, W.J., Sparrow, E.M., et. al., IMEP indicated mean effective pressure
in intake
Handbook of Numerical Heat Transfer, John m-c mini-chamber
Whiley & Sons, Inc., New York, 1988. v vent
18. Blank, D.A. and Pouring, A.A. “Frozen Equilibrium
and EGR Effects on Radical-Initiated H2 Abbreviations
Combustion Kinetics in Low-Compression D.I. ANL Argonne National Laboratory
Engines Using Pistons with Micro-Chambers”, ATAC Active-Thermo-Atmosphere Combustion
ATM, Atm Atmosphere
JSAE Paper 2003-0306, 2003 (and SAE Paper
ATDC After-Top-Dead-Center
2003-01-1788, 2003). BDC Bottom-Dead-Center
19. Semenov, N.N., Some Problems in Chemical BTDC Before-Top-Dead-Center
Kinetics and Reactivity, Princeton University CA Crank Angle
Press, 1958. COV Coefficient of Variation
20. Walker, R. W., “Free Radicals in Combustion CR(s) Compression Ratio(s)
deg degree
Chemistry”, Sci. Progress, Oxford, Part 2, Vol. 74,
DI Direct Injection
Num. 294, pp. 163-188, 1990. DISI DI Spark Ignition
21. Vanpee, M., “On the Cool Flames of Methane”, HCCI Homogeneous Charge Compression Ignition
Combust. Sci. and Tech., Vol. 93, pp.363-374, HCRI Homogeneous Combustion Radical Ignition
1993. ICE Internal Combustion Engine
22. Blank, D. A., "Hydrogen Lean-Combustion M-C, M-Cs Mini-chamber, Mini-chambers
Studies in a Four-Stroke DI Radical-Ignition NG Natural Gas
NOx Nitrogen Oxides (NO and NO2)
Diesel Engine with EGR", SAE Paper PCSI Premixed Charge Spark Ignition
2007-01-1887 and JSAE Paper 20077212, 2007. TI Thermal ignition
23. Blank, D. A. and Pouring, A. A. “Frozen TS Toyota-Stoken
Equilibrium and EGR Effects on Radical-Initiated TDC Top-Dead-Center
H2 Combustion Kinetics in Low-Compression D.I. RI Radical ignition
Engines Using Pistons with Micro-Chambers”, RIS(A) Radical ignition Species (Augmented)
RPM Revolutions Per Minute
JSAE Paper 2003-0306 and SAE Paper
SI Spark ignition
2003-01-1788, 2003. TDC Top-Dead-Center
24. Reyes, S. C., Iglesia, E. and Kelkar, C. P. SOC Start of Combustion
“Kinetic-Transport Models of Bimodal Reaction
Sequences -I. Homogeneous and
Heterogeneous Pathways in Oxidative Coupling

11
APPENDIX Nj
m& n , j
+ ∑n =1 2 m j cv , j
(Vn2, j ) (A2)
For this simulation the open-system conservation of species
equation for species k (within system j, where for the main In Equation (A2) Tj is the temperature of system j. The first
chambers j=1 [during main chamber combustion event j = term on the right hand side of Equation (A2) is due to the
b1, b2, b3 and u as per Figure 4, otherwise j = 1 for the work transfer across the control volume of each system (and
overall main chamber] and for the mini-chambers j=2 is thus zero for system j=2); the second term is due to heat
[during mini- chamber combustion event j = b1, b2, and u transfer between the chamber gas of each system and the
various surface areas (Al) of the chambers; the third term is
as per Figure 5, otherwise j=2 for the overall mini-chamber])
due to the heat released during the chemical reactions
is modeled as:
between the k different species contained in each chamber;
∂Yk m& sys Yk
≅− (A1) the fourth term is due to the energy transferred during mass
∂t mn flow across the Nj boundaries of each system; and the final
1 m ω& W term is due to net kinetic energy addition or subtraction from
+ n (∑ in m& inYk ,in − ∑ out m& out Yk ) + n k k each system due to the various velocities at the Nj
m m ρ
where, within a given system j, Yk is the mass fraction of boundaries. The variables Pj , c v , j , v j , m j and ρj refer
species k, m is the instantaneous value of the mass of the to the pressure, specific heat at constant volume, the
system and the superscript n denotes the value at the end of specific volume, the mass and the density of system j.
a time step ∆t . Wk represents the molecular weight of The term uk represents chemical specific internal energy of
species k and ρ is the density of the gases in the system. species k. & ),
Each system has mass flow rate terms ( m
The term m& sys is the rate of change of the mass of system surface area terms (Al) (at their respective interior surface
temperature (Tc), and with their overall combined
j. The terms m& in and m& out & ),
are mass flow rate terms ( m
radiative-conductive-convective heat transfer coefficients
into or out of system j, respectively. Finally, ω
& k , the (Uc)), velocity terms (Vj), sensible specific enthalpy and
formulation for the combustion rate term for species k is sensible specific internal energy terms ( h , u ), etc. The
given in [3]. The chemical-kinetics mechanism for various meaning of the specific enthalpy expression used in
CH4-C2H6-Air used in this combustion study is given in [5].
This specification includes 84 chemical reactions and 26 the fourth term of Equation (A2), h ′ , are given in [3] and
species. The present calculation has been updated to [15]. For our simulations Uc is simply equal to hc, the
include the inert gas argon and the presence of two species combined convective-radiative heat-transfer coefficient
of unmixed fuel (CH4 and C2H6) to account for the turbulent between the gases of a chamber and a chamber surface.
3
mixing. Also, the kinetics for 4 additional species to account The empirical formulations for computing this coefficient in
for NOx formation. Thus, 33 species are accounted for. The the different chambers is given in [3] and [15]. In the
additional kinetics reactions for NOx used in this study are computations, the wall temperatures are as specified in
25
given in Table A-1 below. This is based on the following Table 1. When the main chamber is divided into the burned
Arrhenius temperature dependence for the forward rate of and unburned mass regions during the combustion phase,
 − Ei  the area for the heat transfer at the inside surface of the
reaction: k f = Ai T β i exp  head is divided into two sub-areas. For these sub-areas the
 Rc T
i
 value of T j in the second right hand side term of Equation
Table A-1. Nitrogen Oxide Mechanism with values for k fi
(A2) is not set to T1 , but rather to Tu (see Equation
Reaction Ei βi Ai
(A4a)) (for the area outside the time varying radius rbit and
N+ NO = N2 + O 3.500E+13 0.0 330.0
N + O2 = NO + O 2.650E+12 0.0 6400.0 to Tb (see Equation (A4c)) (inside this radius). Further, the
N + OH = NO + H 7.333E+13 0.0 1120.0 equation of state for the pressure, density and temperature
N + CO2 = NO + CO 1.900E+11 0.0 3400.0
N2O + O = N2 + O2 1.400E+12 0.0 10810.0 ( Pj , ρj , Tj) inter-relationship is based on the
N2O + O = NO + NO 2.900E+13 0.0 23150.0 assumption that all the air, fuel and combustion products in
N2O + H = N2 + OH 4.400E+14 0.0 18880.0 each system of the engine behave like ideal gases, and is
N2O+ OH = N2 + HO2 2.000E+12 0.0 21060.0 thus provided by:
N2O + M = N2+ O + M 1.300E+11 0.0 59620.0 K
Yk
NO+ HO2= NO2 + OH 2.110E+12 0.0 -480.0 P = ∑ρ RT (A3)
NO2 + O = NO + O2 3.900E+12 0.0 -240.0 k =1 Wk
NO2+ H= NO + OH 1.320E+14 0.0 360.0 Also, the transient momentum equation for the mass
NO+O+M=NO2+M 1.060E+20 -1.41 000.0 exchange between the main and mini chambers is
given in references [3] and [4] and the port momentum
The open-system conservation of energy equation for both
equations for mass exchange between the manifold
the main chambers (j=1) and mini-chambers (j=2) is
and main chamber during intake and exhaust are
modeled as [3, 15]:
given in reference [11]. For properly quantifying the
Lj
dT j AU various sub-processes of the RIS(A)-PCSI cycle during
= - Pj dv j + ∑ l cl (Tc − T j ) individual periods of combustion, spatial variations of the
l =1 m j cv , j
l
dt cv , j dt temperature and species mass fraction values with flame
K N front expansion within both the mini-chamber and
1 m& n , j
∑
j
- (u k ω& k Wk ) j + ∑ (h j′, n − u j ) main-chamber is highly significant. To account for these
ρ j cv , j k =1 n =1 m j c v , j spatial variations, the solution regions of the mini-chamber
and the main-chamber are each divided into three solution
zones during their individual combustion periods: the

12
unburned mass zone, the flame front and the burned mass
zone (Figures 4-5). The M-C and main chamber burned
mass zones are further subdivided into a number of
additional sub-solution zones for various thermal and
chemical-kinetics calculations. At the same time, for each
chamber the conservation of the energy and mass must
also be rigidly maintained globally (i.e., globally for the entire
mini-chamber and globally for the entire main chamber). In
each chamber, from the beginning of a cycle (starting at the
beginning of intake) until the beginning of flame front
expansion, all of the gases are considered as a part of the
unburned mass zone (u). This is thus the starting condition
for each chamber calculation. Because of this, it is
convenient to start out using the earlier formulation of the
DI-HCRI code of [3, 15] (in its entirety) to follow the
chemistry of these unburned mass zones (in each chamber).
It is also convenient to continue to use this formulation [3,
15] (with only minor variations), to continue to track these
unburned mass zone solutions during the entire combustion
event of each individual chamber (main and mini). Finally,
after the individual completions of the flame front
propagation in each chamber, it is again convenient to use
the formulation [3, 15] in its entirety to track the fully
burned gases of each chamber the rest of the way through
the cycle till the end of the exhaust event. In order to
accomplish this solution strategy, during the periods when
combustion is occurring in each chamber, its burned mass
zone and its flame front mini-zone are computed separately
from the unburned mass zone solutions (that are being
accounted for via the formulation of [3, 15]). Then, at the
conclusion of the combustion event for each chamber, the
chemical make up of these burned mass sub-zones (with
their separately computed chemical constituencies
-computed via/within multiple sub-zones calculations) are
merged back into the unburned mass zone solution (via a
novel weighted averaging procedure). For each chamber,
this thus occurs at the conclusion of its combustion event.
Further, both for the global solution for each chamber
(needed for maintaining conservation of energy) and
unburned mass solution for each chamber, the development
provided in [3, 15] is fully descriptive of the formulation
needed to solve the conservation of mass, energy and
species in the mini-chamber, main chamber, vents and port
flow regions.
The only modification to this formulation during the
combustion phase portions is the means by which the
chamber unburned mass temperatures ( Tu ) are specified
while the temperature of the burned mass zones are also
being computed. Finally, except for temperature
calculations of the burned mass solution zone and its
sub-zones, the formulation of [3] is still fully descriptive of
the conservation of species solutions within the individual
burned mass sub-solution zones. The temperatures for
these individual sub-solution zones and also for the overall
burned mass solution zone must be found via a different
formulation. Thus, during the combustion phase in a
particular chamber, the formulation of [3] is used separately
for computing the chemistries of each of the large burned
mass sub-zones and the unburned mass zone – while both
global and local energy calculations are being coordinated
for two full zones (u and b) by making creative use of the
formulation of [3] and Equation (A2) to calculate the
effective temperatures of the chambers as a whole from the
global energy balance for each chamber. During the period
of combustion for the main chamber the conservation of
energy formulation (Equation (A2)) is used in the current
work instead to compute the global temperature of the main
chamber ( T1 ) (i.e., the effective temperature of the entire
main chamber) in order to rigorously maintain energy
conservation for the overall main chamber calculation. The
same is done during the period of combustion for the mini-
13
chamber to find its effective (entire chamber) temperature
( T2 ). From these meticulous energy balances, when the
chamber undergoing combustion is the main chamber, the
temperature of the unburned mass zone in the main
chamber ( Tu ) can then be found using the following perfect
gas relationship:
R1T1 x RT
Tu = + b b b (A4a)
(1 − xb ) Ru ( xb − 1) Ru
where, the individual gas constants Ru = R / Wu ,
Rb = R / Wb and R1 = (1 − xb ) Ru + xb Rb , and where
xb is the burned mass ratio for the chamber. The subscript
b indicates overall value of the burned mass solution zone
for the chamber and accordingly Tb is the overall
temperature of the burned mass zone of the main chamber
in this case. Again, here T1 is the global temperature of
the main chamber. To save in nomenclature, the same zone
and sub-zone formulations are used for both the main and
mini chamber descriptions (to avoid the use of extra
subscripts). Thus, for example, when the chamber in
question is instead the mini-chamber
R2T2 x RT
Tu = + b b b (A4b)
(1 − xb ) Ru ( xb − 1) Ru
where (1 − xb ) Ru + xb Rb = R2 instead. During the
individual mini and main chamber combustion periods, the
global temperature (effective conservation of energy
temperature) for the entire chamber under consideration,
(either T1 or T2 ) is computed for each time step from a strict
accounting of the heat of combustion, the sum of the mass,
chemical species and energy exchange with the other
chamber, the sum of the heat transfer exchanges via the
interior chamber surfaces, etc. (see [3,15] for details). Also
the energy generated by both the pre-ignition combustion
chemistry of the unburned mass zone and the chemical heat
release during flame front combustion is strictly accounted
for in the computation of each of these effective
conservation of energy temperature ( T1 and T2 ) during
the periods in which their chambers are in combustion. To
facilitate this strict accounting during an individual chamber
combustion phase, the third term on the right hand side of
Equation (A2) is replaced by an equivalent expression for
the combined internal heat release due to combustion for
both the unburned mass zone (first expression) and the
advancing flame front (second expression):
K
1
{
ρ j cv , j
∑ (u ω& W
k =1
k k k ) j } (1 − xb )
1 ∂m
+{ { b }t Y fuel ,u ( ∆h 0fuel − (1 + s )∆h products
0
)}
ρ j cv , j ∂t
The constituency of the natural gas fuel for this calculation is
considered to be a mixture of 95.0% methane and 5%
ethane by mass. The term s is the mass of oxygen in
grams needed to consume 1 gram of fuel. At the same time
an accurate computation of the temperature of the burned
mass behind the expanding flame front of a chamber, Tb ,
is accomplished by the tracking of the individual
temperatures of hundreds of thousands of extremely thin
hemi-spherically shaped shells of mass (mass slivers)
formed as the flame advances incrementally through a
myriad of computational time steps ( ∆t ). Initially, these
hemispherical mass slivers rapidly ascend from the
 instantaneous temperature of the unburned gas ( Tu ) (just expressions m1 , T1 and P1 dispersed throughout the
ahead of the flame front) to the terminal temperature Tbi above formulation must be replaced with m2 , T2 and P2 ,
just behind the flame front. This terminal temperature is respectively.
dependent on the local mass fraction of the fuel and this
instantaneous temperature of the unburned gas just prior to The burned mass fractions ( xb ), as a function of time (or
the flame front’s reaching this sliver. The “initial” terminal crank angle), are regressed from the burn rates for the RI
value of this mass sliver’s temperature Tbi is found from species enhanced flame front propagation. Because the
the solution of the following transcendental equation after it simulation study herein is designed to study the high load
has been integrated from the start to the end of the flame lean limit combustion event of one of the ANL test runs, the
front combustion process within this silver: rates in the simulation calculations are based on
experimentally determined ANL data regressed from the
pressure history for this experimental run using Heywood’s
∂
{ c v (Tbi ) Tbi − Y fuel ,u (∆h 0fuel − (1 + s)∆h products
0
) } =0 formulation.26 However, the time varying values of xbt can
∂t
not be taken directly from this experimental ANL data.
Using the evolving history of the temperatures of the During the simulation calculations this data must be
collection of these individual mass slivers, the overall translated. This is because the Kistler 6013B pressure
temperature of the burned mass zone, Tb , can be found transducer used during the ANL testing was measuring
global pressure changes that are resulting from the
from the numerical integration of: combination of both the combustion chemistry of flame front
and the RI induced pre-combustion chemistry in the
1 xb 1 xb . I 1 xb , I +1
Tb ≡
xb ∫
0
Tbit dt ≅
xb ∫0
Tbit dt +
xb ∫ xb . I
Tbit dt (A4c) “unburned” zone ( u ) ahead of the flame front (occurring at
lower temperatures). In conventional engines, the intensity
of pre-combustion chemistry is usually negligible. But as
1 xb . N 1 xb . i
∫ ∫
mentioned, this is not so for RIS(A) - SI. Thus, for the main
.... + Tbit dt + Tbit dt chamber calculations,
xb xb , N −1 xb xb , N
∂mb t
m mbt ≅ mbt − ∆t + { } ∆t (A6a)
where, the burned mass ratio xb ≡ b , is the fraction of26
∂t
m1
∂m ∂mb ,tot ,1 t ∂Y fuel ,u t t
the total main chamber mass ( m1 ) that has been consumed { b }t ≈ { } −{ } mu (A6b)
by the flame front to become a part of the burned mass ∂t ∂t ∂t
t t
( mb ) behind the flame front (the various burned gas where, mu ≡ m1 − mb , Y fuel ,u is the composite species
mass fraction of the NG fuel mixture in the “unburned” mass
sub-zones shown in Figure 4). The values of Tbi (the
zone, (as obtained via the solution of Equation (A1) for the
interval average values) stored (retained) are updated at the ∂mb ,tot ,1
end of each time step to account for four things: isentropic unburned zone), { }t is the overall burn rate of
expansion (and compression), global heat loss from the ∂t
burned mass zone to the cylinder head, global heat loss both zones b and u determined experimentally and m1 is
from the burned mass zone to the unburned mass zone (via
considered constant for this calculation.
internal gas conduction), and global heat mixing losses to
the unburned mass zone (during only the final 2% of the
In order to know the approximate surface areas of the flame
mass burn). This enables all of the stored values of Tbi to front for some of the main chamber heat transfer
accurately reflect what is happening globally in reference to calculations, the radius of the burned mass volume must be
the overall burned mass zone of the main chamber (in terms known. These areas are assumed to be the combination of
of overall pressure changes and heat transfer). Thus: an advancing hemi-spherically shaped area (the surface
( γ bi −1) / γ bi area of the flame front shown in Figure 4) and of the
t t − ∆t  P1t  expanding circular area (top area of the burned mass area)
T bi , s ≅T bi  t − ∆t  (A5a) adjacent to the head in (Figure 4). These areas together
 P1  represent the periphery of the burned mass volume shown
by the position of the flame front in Figure 4. Assuming the
Tbit ≅ Tbit , s − ∆Tirr (A5b) head to be flat (as in the case of the ANL NG test engine)

where ∆Tirr is due to the combination of heat transfer and the radius rbit needed for the burned mass heat loss
mixing losses, and the subscripts s and irr represent the calculations is found via the following formulation:26
terms “isentropic” and “irreversible”, respectively. The Vbt 2π ( rbit ) 3  ρ bt  1  
variable γ bi is the ratio of the specific heats at constant ybt ≡ ≅ ≅ 1 + t  t − 1 
V1 3V1  ρ u  xb  
pressure ( c p (Tbi ) ) and constant volume ( cv (Tbi ) ), where
where the individual densities ρb and ρu (for the
the individual specific heats are functions of the temperature unburned gas zone) are found via Equation (A3) with the
Tbi and the species mass fractions in the burned mass t
pressure of the main chamber ( P1 ) known to be the value
zone. The external heat transfer with the cylinder head is t
also accounted for accurately in the overall conservation of of the pressure for both large zones (b and u). P1 is
energy solution for T1 (via Equation (A2)). obtained from the combined solution of Equations (A1-A3)
for the entire main chamber (the global main chamber
Following this same convention, the Tb values for the solution) and from the main chamber global density

mini-chamber are also computed. For this the

14
 m1
ρ1 ≅ . Finally, V1 is the volume of the main
V1
chamber and is changing with time. As the flame front
progresses in the main chamber, a series of three of the
thousands of hemispherical burned mass slivers are chosen
and used to keep representative track of the detailed
chemistry occurring in their sub-zones throughout the rest of
the combustion event. The temperatures for these mass
slivers are already being maintained (along with many
thousands of other temperatures) via the calculations
required for Equations (A4c) and (A5). These mass slivers
are then considered to represent sub-zones 1, 2 and 3
(show in Figure 4) of the overall burned mass zone.
Initially, just before the flame front reaches a hemi-spherical
mass sliver, the temperature and species mass fractions of
this sliver are the same as the temperature species mass
fractions of the unburned zone. Thus each of these 3
representative mass sliver temperatures is also initially Figure A1: Methane-Oxygen Cool Flame Region
considered to be at Tu and the values of the 33 species (between curves) (2 CH4 + O2)21
mass fractions of the sliver are considered to be exactly the
same as those already being tracked via Equation (A1) for
the unburned mass zone. As the flame front reaches the
location of one of these slivers, a special calculation is made
using Equation (A1) (exactly as is presented) using heat
addition from the burned mass zone to raise the
temperature of this sliver up to the value of Tbi . Then,
during every time step afterward (until a little after the end of
the combustion event), the chemistry of this representative
sliver is updated using revised values of Tbi obtained from
the solution of Equation (A4c) for this sliver and the special
solution of Equation (A1) for the 33 species mass fractions
of this sliver. First sub-zone 1, then later both sub-zones 1
and 2 and finally all 3 sub-zones are updated in this manner
(for all of their mass-fractions and temperatures). The
calculations for these 3 representative main chamber
sub-zones continue until after the end of the combustion
event main chamber and to the point where the mass jets
coming out of the mini-chambers are facilitating extensive
turbulent mixing. At this point these 3 sub-zones are then
merging back into what is left of the unburned gas solution
(as it is finally burning) (via a special weighted averaging
procedure) shortly after the last 2 % of the unburned zone is
considered to have been ignited. This last 2% of the
unburned zone is assumed to ignite over time by its being
mixed (via high turbulence) with the adjacent fringe of the
burned region as all 3 burned mass sub-zones are merged
back into the unburned mass zone solution (via a novel
weighted averaging procedure) at the conclusion of the
main chamber combustion event. From this point on in the
main chamber calculations until the end of the cycle, the
development provided in reference [3] is again fully
descriptive.

For equivalent mini-chamber simulation calculations during

combustion the same general procedures are followed with
a few small differences. Because experimental data is not
available for the burn rates in the M-Cs, these rates are
instead estimated from flame speed data which must again
be transformed via Equation (A6), with the mass in question
being understood to be for M-C (and thus using the
subscript 2 in place of 1). The remainder of the formulation
given after equation (A5) must be similarly modified and
heat transfer calculations must account for geometry
differences. Also, because of the shorter time for the M-C SI
combustion event, only 2 representative M-C burned mass
sub-zones are used (Figure 5).

Finally, a sample methane cool flame region is given in

21
Figure A1.

15