Feed-Forward Control Design of Fuel Distribution on Advanced Dual-

Fuel Engines with Varying Intake Valve Closing Timings

Mateos B. Kassa, Carrie M. Hall, Andrew Ickes, and Thomas Wallner

Abstract— Cylinder-to-cylinder variations of the combustion reactivity controlled compression ignition (RCCI) were
process in multi-cylinder engines has been identified as one of developed as a strategy for addressing the limitations of
the main sources of efficiency losses in advanced internal earlier low temperature combustion strategies, which include
combustion engines. This study examines the dynamics of an difficulties in controlling combustion timing and the inability
engine operated with late intake valve closure (LIVC) timings in to achieve high loads [1]. Dual fuel combustion techniques
a dual-fuel combustion mode in which a high reactivity fuel is
typically use two fuels with different reactivities in order to
directly injected into the cylinders and a low reactivity fuel is
port injected into the cylinders. Despite the benefits of LIVC and adjust the overall fuel reactivity levels based on operating
dual fuel strategy, combining these two techniques resulted in conditions, thereby controlling the combustion timing.
efficiency losses due to the variability of the combustion process. Furthermore, by port-injecting the low reactivity fuel and
The difference in power production across cylinders ranges from direct injecting the high reactivity fuel, improved in-cylinder
9% at an IVC of 570 °ATDC to 38% at an IVC of 610 °ATDC reactivity stratification can be achieved resulting in more
and indicates an increasingly uneven fuel distribution as the optimal combustion durations and acceptable pressure rise
intake valve remains open longer in the compression stroke. This rates, a primary load-limiting factor for many HCCI-type
paper describes an approach for controlling the amount of fuel (homogenous charge compression ignition) combustion
injected into each cylinders’ port of an inline six-cylinder heavy-
strategies [2,3]. The benefits of the dual fuel engine in terms
duty dual fuel engine designed to minimize the variations in fuel
distribution across cylinder in late IVC operations. of performance and emission have been demonstrated on
Since measuring the actual fuel reaching each cylinder is not a single cylinder engines [2,3], as well as multi-cylinder
practical option on stock engines, this study aimed to produce a engines [4-6].
feedforward control technique that would not depend on Improvements in fuel efficiency and reductions in
cylinder-specific fuel flow measurements. Using a model emissions have also been achieved through the manipulation
developed on an engine simulation software, the optimal fuel of the intake valve timing. Specifically, late intake valve
injection amounts at each cylinder port across 125 operating closing (LIVC) has been shown to improve fuel efficiency by
conditions (consisting of sweeps of five engine speeds, five engine reducing pumping losses [7]. By closing the intake valves
loads, and five IVC timings) were determined by using PID
after conventional timing during the compression stroke,
controllers that adjust the amount of fuel injected at each port
based on the amount of fuel reaching the cylinders. The results reduced effective compression ratios are achieved and peak
from the PID controller were expressed as a function of the compression pressures and temperatures are reduced [1]. The
operating condition (engine speed, engine load, and IVC timing); impact of LIVC on emissions has also been well studied.
and consequently, could be reproduced without relying on the Nevin et al. [8] incorporated a hydraulically actuated variable
feedback of the amount of fuel reaching each cylinder but rather IVC system onto a single cylinder diesel engine and this study
on a feedforward model that solely relies on the operating revealed that late IVC resulted in a 90% NOx reduction at
parameters. The feedforward control strategy, unlike the PID high speed and intermediate load with constant airflow and no
controller, is easily implementable on a stock engine and EGR. With the help of late IVC, less EGR is required to
significantly reduces the variations in fuel distribution across
achieve the same NOx emissions, while CO and PM were
cylinders.
reduced by nearly 70%. He et al. [9] also experimentally
investigated the feasibility of using late IVC to improve the
I. INTRODUCTION performance and emissions of a diesel engine with a fully
flexible valve actuation system. They found that late IVC
Due to stringent emissions regulations and higher fuel reduces NOx emissions by approximately 25–50% and
efficiency demands, efforts to combine the benefits of spark- reduces PM by more than 95% at some operating conditions.
ignition and compression-ignition engines have led to the This study explores the combination of these two
development of advanced combustion engines featuring dual- techniques, dual-fuel combustion and LIVC, and the
fuel combustion. The dual-fuel engine is capable of delivering challenges associated with their combined usage. Despite the
similar and in some cases improved efficiency compared to benefits of LIVC for controlling trapped in-cylinder charge
conventional compression-ignition engine without the high mass and subsequently the peak compression pressure and
emissions [1-6]. Dual fuel combustion concepts including temperature and the benefits of dual fuel strategy to control
fuel reactivity and improve combustion efficiency, combining

M. Kassa and C. Hall are with the Department of Mechanical, Materials

and Aerospace Engineering, Illinois Institute of Technology, Chicago, IL
60616, USA. (e-mail: [email protected], [email protected])
A. Ickes and T. Wallner are with Argonne’s Center for Transportation
Research, Argonne National Laboratory, Lemont, IL, 60439, USA. (email:
[email protected], [email protected])
these two techniques resulted in efficiency losses due to the and their performance is described in the Results section.
variability of the combustion process across cylinders. Concluding remarks and a discussion of future work will
Combustion variability is a common problem in advanced follow at the end.
combustion modes. While such advanced combustion modes
II. EXPERIMENTAL SETUP
can provide high efficiencies, the utilization of exhaust gas
recirculation and two fuels of differing reactivities as well as The challenges associated with dual fuel engines utilizing
early fuel injection can introduce higher variability in the LIVC were observed on a dual-fuel heavy-duty engine. The
quantity and composition of the trapped mass consequently engine specifications are given in Table 1. The engine
leading to variations in the combustion process [10,11]. features an air system configured for high EGR delivery,
Cycle-to-cycle and cylinder-to-cylinder variations of the utilizing two-stage turbocharging (with interstage cooling)
combustion process are observed as differences in and a dual-pass EGR cooler with high and low temperature
combustion phasing, heat release rates, peak pressure, as well stages. The engine also features a fully flexible intake valve
as the indicated mean effective pressure (IMEP). These actuation system as well as separate direct injection and port
variations in the combustion process have been found to be injection fuel delivery systems. The direct injection system is
primarily caused by variation in gas motion in the cylinder used for the high reactivity fuel and a port fuel injection
during combustion; variation in the amounts of fuel, air, and system is used for the low reactivity fuel. The high reactivity
recycled exhaust gas trapped in the cylinder each cycle; and fuel used is ultra-low sulfur diesel and the low reactivity fuel
variation in the mixture of the composition within the cylinder investigated is compressed natural gas (CNG). For all tests,
each cycle [12]. Introducing LIVC along with a dual fuel the engine cooling system and air system restrictions were
combustion technique has been found to exacerbate these tuned to be representative of the conditions encountered in
combustion variations. production vehicles. Each cylinder was equipped with a
This study leverages a GT-Power simulation model of the Kistler 6125B pressure transducer which supplied pressure
experimental engine to better understand the distribution of measurements at a resolution of 0.1°CA, which allowed
the port injected fuel across cylinders under various operating precise calculations of the power production from each
conditions and identify the main factors affecting the gas cylinders.
exchange process and fuel distribution. The model showed
late intake valve closures allow a portion of the trapped air TABLE I. ENGINE SPECIFICATION
and port injected fuel to flow back out of the cylinders into Displacement Volume 12.4L
the intake manifold. The fuel that is pushed back in the intake
manifold is then unevenly redistributed across the cylinders Number of Cylinders 6
favoring the cylinders furthest away from the inlet of the Stroke 166 mm
intake manifold, largely due to the dominating direction of the
flow in the intake manifold. Thus, an effective fueling Bore 126 mm
strategy is necessary to achieve a uniform power production Connecting Rod 251 mm
across the cylinders and optimize the engine performance. Compression Ratio 17
In the simulation model, PID controllers were added at
each cylinder port in order to adjust the amount of fuel
injected at the port based on the amount of fuel reaching the III. GT-POWER MODEL DEVELOPMENT
cylinders. As the dominating direction of the flow in the
intake manifold transports most of the fuel that reaches the In order to address the cylinder-to-cylinder variations in
common plenum towards the cylinders furthest away from the power production observed from the experimental testing, a
inlet of the manifold, the PID controllers end up limiting the simulation model was also developed. The engine model
amount of fuel injected at these cylinder ports and injecting allowed the gas exchange process to be more thoroughly
more fuel at the cylinder ports closest to the inlet of the investigated and enabled the development of control
manifold. This control strategy works effectively at achieving strategies necessary to address the uneven distribution of the
uniform fuel distribution across the cylinders; however, this port injected fuel across cylinders in LIVC operations. The
approach relies on a feedback system requiring the model used for the simulation studies was built using GT-
measurements of the amount of fuel entering each cylinder, Power and is illustrated in Figure 1. The model features an
which is not practical from an experimental standpoint. The inline six-cylinder compression ignition engine fitted with a
goal of this paper is to reproduce the performance of the PID port fuel injection system. Figure 1 illustrates the cylinders
controllers without relying on the feedback mechanism and connected to the intake manifold through the intake ports and
instead use a control design that relies on a feedforward model intake runners, as well as the fuel injectors connected to the
that determines the optimal amount of fuel injected at each cylinders (direct injection) and at the intake ports (port-
port based on the operating condition. injection). Further details about GT-Power engine models and
In the following sections, the engine used for this study the other components of the model can be found in [13].
is described in Experimental Setup followed by the The engine geometric dimensions and piping dimensions
description of the GT-Power model in GT-Power Model were adjusted to match the experimental setup and the effect
Development. The different fueling strategies (conventional, of the turbocharger and EGR loop were simulated by dictating
PID based, and feedforward) are then separately discussed the inlet conditions (pressure and temperature) to agree with
experimental results. This model is used to study the gas conventional fueling strategy in LIVC operations results in
exchange process and investigate the cylinder-to-cylinder significant variations in the trapped fuel across cylinders.
variations in fuel distribution under the various fueling Figure 2 illustrates the consequential cylinder-to-cylinder
strategies described in the next sections. variations in the IMEP observed experimentally, and the trend
of increasing variations as the intake valve is maintained open
longer in the compression stroke. Each line in Figure 2
represents the IMEP of a specific cylinder at the varying IVC
timing. The bars are used to represent the cylinder-to-cylinder
variation (ratio of the standard deviation to the mean), and the
values for variation should be read using the secondary
vertical axis on the right. Similar results are also seen from
the GT-Power simulations in which the amount of fuel
reaching each cylinder is monitored at varying IVC timings
with the conventional fueling strategy (Figure 3).

22 14%

Cyl.-to-Cyl. Variation
12%
20
10%

IMEP [bar]
18 8%
16 6%
4%
14
2%
12 0%
570 580 590 600 610 620
IVC Timing (°ATDC)

Figure 2. Experimental results of cylinder-to-cylinder variations in IMEP in

LIVC operations.

230 20%
Trapped Fuel [mg/cycle]

Cyl.-to-Cyl. Variation
210
15%
190
170 10%
150
5%
130
110 0%
570 580 590 600 610
Figure 1. Diagram of GT-Power Model (Only intake side displayed). IVC Timing (ATDC)
IV. CONVENTIONAL FUELING STRATEGY
The conventional port fuel injection strategy consists of Figure 3. GT-Power simulation results of cylinder-to-cylinder variations in
injecting the same amount of fuel at each cylinder port. In fuel distribution in LIVC operations.
standard IVC timings (where the valves close near bottom
dead center), there is little or no backflow from the cylinders As discussed previously, the conventional fueling
and such an injection strategy effectively achieves uniform strategy results in uneven fuel distribution due to fuel being
fuel distribution across the cylinders. However, in LIVC pushed back into the intake manifold from the cylinders
operations, there is significant backflow from the cylinders during the compression stroke and getting redistributed
during the compression stroke in which the intake valves are unevenly across the cylinders due to the dominating direction
still (partially) open leading to an uneven re-distribution of of the flow in the manifold, which directs most of the fuel
the port-injected fuel across cylinders. As such, the reaching the common plenum towards the cylinders furthest
away from the inlet of the intake manifold (cylinders 4
through 6). One way to address the uneven distribution of the
port-injected fuel across the cylinders is to adjust the amount
of fuel injected at each port based on the amount of fuel
reaching the cylinders. Such a method is discussed in the
following section, in which a PID controller is added at each
fuel injector to adjust the fuel distribution across cylinders
over a wide range of operating conditions.

V. PID CONTROLLER BASED FUELING STRATEGY

A feedback mechanism to adjust the amount of fuel at
each port represents an effective strategy for achieving a (a) (b)
uniform fuel distribution, albeit not being practical from an Figure 5. Step response of fuel flow at the valves (blue: cylinder 1, green:
experimental standpoint or easily implementable on a stock cylinder 6, black: injector input, dashed: IVC at 570 °ATDC, solid: IVC at
engine. In this section, the development of the PID controllers 620 °ATDC) for (a) low speed low load and (b) high speed high load
in the GT-Power model is discussed and the result of the conditions.
controllers are presented as a prelude to the subsequent
section discussing the approach undertaken to reproduce the The PID controllers were used to successfully identify
results of the PID controller using a feedforward model—a the fueling amount at each port that yields uniform fuel
more practical approach at achieving a uniform fuel distribution across cylinders throughout 125 operating
distribution. conditions consisting of a sweep of five engine speed, five
The diagram illustrated in Figure 4 displays the engine loads, and five IVC timings (Table II). This approach
components added to the base model in order to implement in identifying the optimal fueling amount at each port
the PID controllers. The controllers, added for each cylinder, represents an idealistic approach as it relies on the
adjust the amount of fuel injected at each cylinder’s port based measurements of the fuel passing through each cylinder
on the amount of fuel passing through the corresponding valves, which are not readily available on typical stock
cylinder’s valves during an engine cycle. The port fuel engines. However, this case presents a best case scenario to
injectors use the desired air-to-fuel ratios as inputs and have which simpler approaches can be compared. In the following
limitations in the amount of fuel that can be injected per cycle. sections, an approach in which the results of the PID
As such, the intermediate components convert the output of controllers is reproduced without relying on a feedback
the PID controller (fuel flow rate) into air-fuel ratio (using the system is discussed.
air flow signal) and assure that the input to the injector is
TABLE II. OPERATING POINTS
within the prescribed constraints.
Variables Min Max Increments
Engine Speed [RPM] 400 2000 400
Engine Load- 110 540 86
Total Fueling [mg/cycle]
IVC [°ATDC] 570 610 10

VI. FEED FORWARD MODEL BASED FUELING STRATEGY

Figure 4. Port-fuel injection PID controller setup. The results from the PID controllers were used to develop
a feedforward model, in which the performance of the PID
controllers would be replicated without the use of a feedback
The gains of the PID controllers were adjusted manually system. In order to develop the feed forward model, the
by observing the response of the fuel flow to a step input at optimal fueling at each port was fitted as a function of the
each cylinder valve across different operating points. Figure operating conditions for each IVC timing. As illustrated in
5 illustrates the responses of the fuel flow at the valves of Figure 6, the optimal fueling at a cylinder port across a wide
cylinders 1 and 6 at two widely varying operating points. The operating condition range can be well-fitted as a function of
goal in setting the gains of the PID controllers was to assure the total fueling (analogous to engine load) and the engine
stability of the controllers. The characteristics of the transient speed. Although, Figure 6 illustrates a specific fitting
response were of little importance at this stage (as long as the corresponding to the fueling at the cylinder 3 port at 610
controller converges to the optimal fueling within the set
°ATDC IVC, similar fitting accuracies were achieved for each
number of simulation cycles). As such no detailed model of
cylinders across all 5 IVC timings as can be seen from the
the fuel flow was necessary, and appropriate gains can be well
coefficient of determination (R2 values) displayed in Table III.
estimated strictly from observing the step responses.
Table III also displays the fitting coefficient where the fitting
equation is as follows:
̇ = 𝑎 + 𝑏 ∗ 𝜔 + 𝑐 ∗ 𝐹𝑡𝑜𝑡
𝐹𝑜𝑝𝑡 ̇ 
 ̇ and 𝐹𝑡𝑜𝑡
where a, b, and c are the fitting coefficiens; 𝐹𝑜𝑝𝑡 ̇ are Fitting Coefficients
IVC Cyl. R2
the optimal and total fueling rate respectively in grams per a b c
second (g/s); and 𝜔 is the engine speed in revolutions per
5 -7.04E-03 -1.82E-04 1.74E-01 0.9939
min (RPM).
6 -3.84E-02 -2.27E-04 1.72E-01 0.9913

1 -2.22E-02 3.48E-04 1.41E-01 0.9936

2 -1.52E-02 4.03E-04 1.33E-01 0.9948

3 7.61E-02 -1.40E-05 1.72E-01 0.9916

590
4 1.39E-02 -2.48E-04 1.88E-01 0.9976

5 -2.60E-02 -2.10E-04 1.84E-01 0.9968

6 -2.63E-02 -2.79E-04 1.83E-01 0.9927

1 -1.70E-02 3.83E-04 1.35E-01 0.9946

2 2.91E-03 4.25E-04 1.29E-01 0.9947

3 2.23E-02 1.06E-04 1.59E-01 0.9972

580
4 -2.52E-03 -2.55E-04 1.89E-01 0.9975

5 -1.61E-02 -2.62E-04 1.90E-01 0.997

6 8.72E-03 -3.95E-04 1.98E-01 0.9932

1 -1.97E-02 4.34E-04 1.29E-01 0.9943

2 3.19E-02 4.66E-04 1.23E-01 0.9933

3 -2.79E-03 1.11E-04 1.60E-01 0.9967

570
4 -1.09E-02 -2.84E-04 1.93E-01 0.9974

5 -1.69E-02 -2.90E-04 1.93E-01 0.9968

6 2.08E-02 -4.40E-04 2.03E-01 0.9923

The representation of the optimal fueling at each port as a
linear function of engine load and engine speed demonstrates
that the optimal fueling can be well predicted from the
operating conditions without having to rely on a feedback
Figure 6. Optimal fueling fitting illustration and corresponding residual
plot.
mechanism. As the measurement amount of fuel entering each
cylinders is not generally available on an engine, a model that
only relies on the operating parameter provides a more
In addition to the coefficient of determination, the residual practical approach at reducing the cylinder-to-cylinder
plot in Figure 6 also attests to the accuracy of the fit across the variations observed in LIVC operations. In the next section,
operating conditions. The plot of the estimation errors the performance of the feedforward model based fueling
illustrates the relative magnitude of the fitting error margins strategy relative to the conventional and the optimal (PID
across the operating range considered. controller based) fueling strategy is discussed.

VII. RESULTS
Three fueling strategies have been discussed thus far: the
TABLE III. OPTIMAL FUELING FITTING COEFFICIENTS AND COEFFICIENT conventional fueling strategy, which consists of injecting the
OF DETERMINATION same amount of fuel at each cylinder port; the PID controller
Fitting Coefficients based fueling strategy, in which the amount of fuel injected at
IVC Cyl. R2 each cylinder port is adjusted based on the measured amount
a b c
of fuel that reaches the cylinders; and the feedforward model-
1 8.69E-02 1.75E-04 1.57E-01 0.9945
based fueling strategy, in which the amount of fuel injected at
2 6.78E-02 2.11E-04 1.51E-01 0.9943 each port is determined from the engine speed, load, and IVC
610
3 7.34E-02 8.95E-05 1.62E-01 0.9952 timings.
4 -1.57E-02 -1.80E-04 1.86E-01 0.9993 The performance of each of these fueling strategies in terms
of the variations of the fuel distributed across the cylinders
5 -7.49E-02 -1.52E-04 1.78E-01 0.9974
throughout the studied 125 operating points is illustrated in
6 -1.37E-01 -1.45E-04 1.66E-01 0.9887
Figure 7. The total fuel injection amount, the engine speed,
1 -2.53E-02 2.60E-04 1.59E-01 0.9899 and the IVC timing which serve to characterize the operating
600
2 -1.87E-02 3.09E-04 1.48E-01 0.9928 point, are displayed in the subplots of Figure 7. The resulting
3 5.73E-02 3.87E-05 1.70E-01 0.9951 cylinder-to-cylinder variations of the fuel reaching the
4 3.32E-02 -1.99E-04 1.77E-01 0.993
cylinders with each fuel injection strategy are also shown in
 the figure. VIII. CONCLUSION AND FUTURE WORK
Several factors such as uneven trapped mass composition
as well as uneven fuel distribution can lead to significant
variations in the combustion process across cylinders—one
jjj cylinder might exhibit very high peak pressures while another
cylinder misfires. Such cylinder-to-cylinder variations are
more frequently observed in engines leveraging advanced
combustion strategies and represent critical limiting factors to
attaining optimal performance.
In this study, in which an engine featuring a dual fuel
combustion strategy and LIVC conditions, the uneven
distribution of the port-injected fuel resulted in significant
cylinder-to-cylinder variations. A feedforward model-based
fueling strategy is developed in this work in order to reduce
these cylinder-to-cylinder variations of the combustion
process. The fueling strategy replicates the performance of a
feedback (PID) control system without relying on the
measurements of the fuel flow at the valves. The feedforward
model-based fueling strategy effectively reduces cylinder-to-
cylinder variations and represents a practical and easily
implementable approach for
Future works consist of implementing this approach on an
experimental engine and investigating the performance of the
feedforward model based fueling strategy in both steady-state
and transient conditions.

ACKNOWLEDGMENT
Experimental data utilized in this study was generated by a
Figure 7. Illustration of variations in fuel reaching each cylinders with project supported by the U.S. Department of Energy (DOE),
respect to the different fueling strategies across the studied operating points. the National Energy Technology (NETL) office, under
cooperative agreement “SuperTruck – Development and
As should be expected, the PID controller based fueling Demonstration of a Fuel-Efficient Class 8 Tractor & Trailer”
strategy basically eliminates the variations in fuel distribution DOE Contract: DE-EE0003303.
whereas significant cylinder-to-cylinder variations are The submitted manuscript has been partially created by
observed with the conventional fueling strategy. Although the UChicago Argonne, LLC, Operator of Argonne National
feedforward model-based fueling strategy does not achieve Laboratory (“Argonne”). Argonne, a U.S. Department of
the same uniformity as the PID controllers, this approach still Energy Office of Science laboratory, is operated under
represents a significant improvement over the conventional Contract No. DE-AC02-06CH11357. The U.S. Government
fueling strategy and significantly reduces the cylinder-to- retains for itself, and others acting on its behalf, a paid-up
cylinder variations to more tolerable levels. nonexclusive, irrevocable worldwide license in said article
The average cylinder-to-cylinder variation in fuel to reproduce, prepare derivative works, distribute copies to
distribution across the 125 operating conditions using the the public, and perform publicly and display publicly, by or
on behalf of the Government.
feedforward model based fueling strategy is 3.5%, whereas
the conventional fueling strategy leads to an average of 20.6%
cylinder-to-cylinder variation. The high cylinder-to-cylinder
variations (>6%) observed with the feedforward model based
REFERENCES
fueling strategy are generally associated with low engine
speeds. The highest eight cylinder-to-cylinder variations are [1] Ickes, A., Hanson, R., Wallner, T. 2015. Impact of Effective
Compression Ratio on Gasoline-Diesel Dual-Fuel Combustion in a
all at an engine speed of either 400 or 800 RPM, which Heavy-Duty Engine Using Variable Valve Actuation. SAE
suggests that some adjustments to the fitting parameters at [2] Kokjohn, S., Hanson, R., Splitter, D., and Reitz, R., "Experiments and
these engine speeds can potentially improve the performance Modeling of Dual-fuel HCCI and PCCI Combustion Using In-
of the fueling strategy. Cylinder Fuel Blending," SAE Int. J. Engines 2(2):24-39, 2010,
doi:10.4271/2009-01-2647.
[3] Hanson, R., Kokjohn, S., Splitter, D., and Reitz, R., "An Experimental
Investigation of Fuel Reactivity Controlled PCCI Combustion in a
Heavy-Duty Engine," SAE Int. J. Engines 3(1):700-716, 2010,
doi:10.4271/2010-01-0864.
[4] Daisho, Y., Yaeo, T., Koseki, T., Saito, T. et al., "Combustion and
Exhaust Emissions in a Direct-injection Diesel Engine Dual-Fueled
with Natural Gas," SAE Technical Paper 950465, 1995,
doi:10.4271/950465.
[5] Belaid-Saleh, H., Jay, S., Kashdan, J., Ternel, C. et al., "Numerical
and Experimental Investigation of Combustion Regimes in a Dual
Fuel Engine," SAE Technical Paper 2013-24-0015, 2013,
doi:10.4271/2013-24-0015.
[6] Zhang, Y., Sagalovich, I., De Ojeda, W., Ickes, A., Wallner, T. and
Wickman, D. 2013. Development of Dual-Fuel Low Temperature
Combustion Strategy in a Multi-Cylinder Heavy-Duty Compression
Ignition Engine Using Conventional and Alternative Fuels. SAE
International Journal of Engines, 6 (3), pp. 1481--1489.
[7] Helmantel A, Denbratt I. HCCI operation of a passenger car di diesel
engine with an adjustable valve train. SAE paper 2006-01-0029; 2006.
[8] Nevin RM, Sun Y, Gonzalez MA, Reitz RD. PCCI investigation using
variable intake valve closing in a heavy duty diesel engine. SAE paper
2007-01-0903; 2007.
[9] He X, Durrett R, Sun Z. Late intake valve closing as an emissions
control strategy at tier 2 bin 5 engine-out NOx level. SAE paper 2008-
01-0637; 2008.
[10] Bittle, J., Zheng, J., Xue, X. and Jacobs, T. 2012. Cylinder-to-cylinder
variation sources in diesel low temperature combustion and the
influence they have on emissions. Proceedings of the Central States
Section of the Combustion Institute, Park City, Utah.
[11] Selim, Mohamed Y.e. "Sensitivity of Dual Fuel Engine Combustion
and Knocking Limits to Gaseous Fuel Composition." Energy
Conversion and Management 45.3 (2004): 411-25, 2004.
[12] Heywood, John B. “Causes of Cycle-by-Cycle and Cylinder-to-
Cylinder Variations.” Internal Combustion Engine Fundamentals.
New York: McGraw-Hill, 1988. 419-426
[13] GT-Power User’s Manual, “Multiple Cylinder Turbocharged DI
Diesel Engine.” GT-Suite Engine Performance Tutorials Version 7.4,
Gamma Technologies, 2014. 69-92