Content uploaded by Maciej Mikulski
Author content
All content in this area was uploaded by Maciej Mikulski on Aug 30, 2024
Content may be subject to copyright.
Applied Thermal Engineering 246 (2024) 122940
Available online 13 March 2024
1359-4311/© 2024 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Research Paper
Variable valve actuation for efcient exhaust thermal management in an
off-road diesel engine
Jeyoung Kim
a
,
*
, Marko Vallinmaki
b
, Tino Tuominen
c
, Maciej Mikulski
a
a
University of Vaasa, School of Technology and Innovations, Efcient Powertrain Solutions (EPS) research group, Wolfntie 32, FI-65200 Vaasa, Finland
b
AGCO Power, Linnavuorentie 8-10, FI-37240 Linnavuori, Finland
c
VTT Technical Research Centre of Finland, PL 1000, 02044, Espoo, Finland
ARTICLE INFO
Keywords:
VVA
After-treatment thermal management
Predictive combustion model
Diesel combustion
Emissions
Off-road diesel engines
ABSTRACT
Exhaust thermal management (ETM) is crucial for effective emission mitigation in integrated exhaust after-
treatment systems of modern off-road diesel powertrains. However, conventional ETM strategies incur a sig-
nicant fuel efciency penalty. This study addresses the issue by investigating the application of variable valve
actuation (VVA) for efcient ETM. For the rst time, this investigation is conducted on a representative state-of-
the-art off-road powertrain platform. It explores four VVA strategies with unprecedent level of rigour, employing
a model-based approach that enables extended insights beyond stand-alone testing. Experiments with an EU
Stage-V off-road diesel engine provide the baseline for validating a one-dimensional model in GT-Suite. A
meticulously calibrated, predictive combustion model enables precise cross-evaluation of how VVA strategies
affect exhaust gas temperature (EGT), efciency, engine-out emissions and combustion characteristics, consid-
ering all trade-offs. VVA simulations are performed at three low-load operating points, where engine operation
borders catalyst light-off temperature (LOT). The ndings impartially conrm that cylinder deactivation (CDA)
and intake modulation are the most promising VVA strategies for off-road engines, with EGT increments sur-
passing +250 ◦C and +150 ◦C respectively, accompanied by minor fuel penalties (up to +3.5 %). CDA
demonstrated fuel savings of up to −2.5 % at certain points, due to reduced pumping and friction losses. Intake
modulation displayed large reduction in engine-out NO
x
(>90 %) and minimal penalties in carbon emissions
(HC, CO, and soot). The results underscore VVAs potential as an efcient ETM option to help the next generation
of off-road diesels to comply with upcoming EPA Tier 5 emission legislation.
1. Introduction
Compression ignition (CI) will remain dominant for heavy-duty (HD)
off-road applications for the foreseeable future. High energy density of
the fuel, combined with good thermal efciency and robustness, support
this status quo [1]. Although the on-road sector has made rapid progress
towards electrication, the low power density of Li-ion batteries means
this roadmap is not fully feasible for the off-road sector [2]. Hybrid-
isation offers additional advantages through a variety of energy- and
thermal-management strategies. Furthermore, low-carbon renewable
fuels provide an immediate route towards decarbonisation in energy-
demanding applications. Therefore, it is important to implement effec-
tive exhaust aftertreatment systems (EATS) and exhaust thermal man-
agement (ETM) to achieve net-zero emissions in CI engines.
Despite their merits, CI engines are notorious for their emissions of
harmful air pollutants, particularly nitrogen oxides (NO
x
) and particu-
late matter (PM) due to the presence of both locally fuel-rich zone and
high ame temperature [1,3]. High-pressure multiple injection [4,5]
and exhaust gas recirculation (EGR) [6] were developed to address these
issues. However, relying solely on in-cylinder emission control strategies
proves inadequate for simultaneously reducing both emissions due to
the inherent trade-off between NO
x
and PM in conventional CI engines.
Consequently, the adoption of EATS has become necessary to ensure
compliance with the increasingly stringent emission regulations in off-
road applications, such as EU Stage-V or US EPA Tier 4 nal limits.
Contemporary EATS is heavily integrated, as illustrated in Fig. 1. A
diesel oxidation catalyst (DOC) reduces hydrocarbon (HC), carbon
monoxide (CO), and soluble organic fraction through catalytic reactions
with oxygen [7,8]. Subsequently, a diesel particulate lter (DPF) cap-
tures PM via ltration. Additionally, selective catalytic reduction (SCR)
technology addresses NO
x
through reaction with urea. The urea is
* Corresponding author.
E-mail address: jeyoung.kim@uwasa. (J. Kim).
Contents lists available at ScienceDirect
Applied Thermal Engineering
journal homepage: www.elsevier.com/locate/apthermeng
https://doi.org/10.1016/j.applthermaleng.2024.122940
Received 10 January 2024; Received in revised form 6 March 2024; Accepted 12 March 2024
Applied Thermal Engineering 246 (2024) 122940
2
decomposed into NH
3
, through hydrolysis and thermolysis, where NH
3
serves as a reductant in the SCR [9]. Higher urea concentration in the
catalyst results in greater NO
x
reduction, yet it is crucial to recognize
that not all ammonia (NH
3
) is absorbed by the catalyst surface, with
excess being released into the exhaust stream [9,10]. Consequently,
precise control of urea injection is imperative [9]. Alternatively, an
ammonia oxidation catalyst (AOC) may be incorporated to manage
ammonia slips from the SCR. Although modern diesel engines primarily
emit NO
x
in the form of NO [11], this is converted to NO
2
in the DOC and
SCR [11]. The conversion of NO to NO
2
impacts both DPF passive
regeneration [12] and NO
x
reduction in the SCR [8]. Nevertheless, the
adoption of EATS introduces inherent challenges, including higher
vehicle costs for consumers, increased backpressure in the exhaust sys-
tem, additional maintenance requirements and the periodic need to
replenish the urea reservoir [13].
Regulation can inuence the conguration of integrated EATS. For
example, EPA Tier 4 nal regulations in North America do not restrict
the number of particulate matters (PN-particle number), so DPFs are less
common [14]. Instead, the EPA regulation focus on enhancing the
effectiveness of DOCs to address emissions concerns [14]. Conversely,
DPFs are becoming widespread in Europe to comply with new restriction
of particle number (PN) in EU Stage-V regulations.
Contemporary EATS demonstrate excellent performance. L´
opez et al.
[15] tested a state-of-the-art, vanadium-based SCR catalyst on a
US2007, 8.9-litre diesel engine and achieved NO
x
conversion efciency
of up to 92 % during non-road transient cycle (NRTC) tests. Higher
conversion efciency is achievable with dual-dosing, double SCR sys-
tems [16,17] or by adopting a pre-turbine layout [18,19]. The modern
DOC can remove up to 90 % of unburned HC and as much as 99 % of CO
at high exhaust gas temperature (EGT) [20]. Hu et al. [21] implemented
DOC, catalytic DPF and SCR for a non-road diesel engine, demonstrating
an average reduction of more than 90 % in emissions of CO, HC, NO
x
,
PN, and PM in non-road steady cycle test. However, it was reported that
the performance of EATS diminishes notably when EGT is relatively low.
Feng et al. [22] further examined the effect of exhaust gas thermal
conditions on emissions in a non-road diesel engine by analysing
experimental results of hot and cold NRTC tests. EATS consisted of DOC,
partial oxidation catalyst, SCR, and AOC. Low EGT in the cold cycle
attributed to lower conversion efciency. Specically, PM conversion
efciency decreased by 31 % compared to the hot cycle, exceeding
China-IV PM standards. Notably, it took approximately 400 ~ 500 s for
the SCR catalyst temperature to attain 200 ◦C in the NRTC cold mode
[14]. This time constitutes one-third of the entire mode duration, so it
exerted a substantial inuence on NO
x
emissions. Consequently, NO
x
,
HC, CO, and CO
2
exhibited markedly lower conversion efciencies
during the early stage (0–600 s) of the cold NRTC mode.
EGT is one of crucial factors, along with exhaust mass ow rate and
exhaust gas composition, to inuence the performance of DOC and SCR
Nomenclature
Abbreviations
1D one-dimensional
2EVO secondary exhaust valve opening
2IVO secondary intake valve opening
AFR air–fuel ratio
AOC ammonia oxidation catalyst
AVG average values
BDC bottom dead centre
BMEP brake mean effective pressure
BSFC brake specic fuel consumption
BSNO
x
brake specic nitrogen oxides (g/kWh)
CAC charge-air cooler
CAD crank angle degree
CCO cylinder cut-out
CDA cylinder deactivation
CHR cumulative heat release
CI compression ignition
CO carbon monoxide
DOC diesel oxidation catalyst
DPF diesel particulate lter
EATS exhaust aftertreatment systems
ECU engine control unit
EEVO early exhaust valve opening
EGR exhaust gas recirculation
EGT exhaust gas temperature
EIVC early intake valve closing
ETM exhaust thermal management
EVC exhaust valve closing
EVO exhaust valve opening
FS full scale
HC hydrocarbon
HD heavy-duty
HRR heat release rate
IAF intake air mass ow
IMEP indicated mean effective pressure
IMEP
g
gross indicated mean effective pressure
IMEP
n
net indicated mean effective pressure
IVC intake valve closing
IVO intake valve opening
LEVO late exhaust valve opening
LIVC late intake valve closing
LOT light-off temperature
MFB mass fraction burned
NO
x
nitrogen oxides
NRTC non-road transient cycle
NVO negative valve overlap
OP operating point
PM particulate matter
PMEP pumping mean effective pressure
PN particle number
RGF residual gas fraction
SCR selective catalytic reduction
SOI start of injection
TDC top dead centre
VGT variable geometry turbine
VVA variable valve actuation
Fig. 1. Modern integrated exhaust aftertreatment system for diesel engines.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
3
[8,22]. High EGT accelerates catalytic reactions, providing high EATS
conversion efciency. High EGT also is benecial for DPF regeneration
[23]. Conversely, low EGT slows catalytic reactions, resulting in poor
performance of the DOC and SCR system. The light-off temperature
(LOT) - the temperature at which conversion efciency reaches 50 % - is
normally used to indicate the effectiveness of the catalyst.
The LOT varies over each application by catalyst metal composition,
space velocity, exhaust gas composition and design of the aftertreatment
brick [20]. Guardiola et al. [8] noted that various emission species
exhibit distinct LOTs, with HC and CO having LOTs of 190 ◦C and
150 ◦C, respectively. Furthermore, DOC ageing increased the LOT of HC
and CO to 250 ◦C and 205 ◦C respectively, but catalytic monolith length
of the DOC showed no signicant effect on LOT. Guardiola et al. [24]
and Tharad [25] reported that the LOT of the modern DOC is 200 ◦C.
According to Villamaina et al. [26] and Ma et al. [27], the LOT of an SCR
system is 170 ◦C – 270 ◦C with different catalyst compositions. Addi-
tionally, the SCR system requires a temperature above 200 ◦C to
decompose urea and thus avoid solid deposits [28]. Despite some vari-
ations in these temperature ranges, Vos et al. [29] and Ko et al. [11]
claimed that EATS necessitates catalyst temperatures exceeding 250 ◦C
to 300 ◦C for effective emission mitigation.
Reaching this threshold can be challenging in real world operation,
particularly during the cold-start, warm-up phase and when idling. EGT
can be as low as 100 ◦C when an engine is idling [30,31]. Lauren et al.
[32] and Magee [33] consistently reported that EGT remains below
250 ◦C when engine load is lower than 30 % in contemporary off-road
and on-road diesel engines. Various ETM techniques can be used to
satisfy the increasing demand for high EGT over a wider range of
operating conditions. To this end, most production engines already
incorporate either intake or exhaust throttling. These slow-response,
airpath-based ETM measures can be combined with fast in-cylinder
thermal management via late post injection.
Bai et al. [30] demonstrated an EGT greater than 200 ◦C at low load
(3 % load) using intake throttling in a 7-litre, HD diesel engine. Lauren
et al. [32] evaluated intake and exhaust throttling in off-road diesel
engines, raising EGT by more than 200 ◦C. Both studies observed that
excessive throttling could raise EGT, but that this approach leads to poor
gas exchange process since the airpath is highly restricted. The two
studies reported high penalties of break specic fuel consumption
(BSFC), of up to 14.3 % and 20 % respectively. Wu et al. [34] employed a
late post injection strategy on a six-cylinder, HD diesel engine, showing
that EGT was increased by delaying post injection timing and increasing
post injection quantity. In the end, EGT was increased by about 60 ◦C,
but accompanied by a 3.4 % reduction in the brake thermal efciency.
Gosala et al. [1] implemented exhaust throttling in a HD diesel engine
via an over-closed variable geometry turbine (VGT) vane, coupled with
late post injection. Fuel efciency deteriorated by up to 58 % at idling
condition, while EGT increased by around +115 ◦C.
It is evident from the above reviews that there is a trade-off between
EGT increment and fuel efciency. Traditional ETM strategies demon-
strate a signicant fuel penalty that is deemed unacceptable in an off-
road sector which values high fuel efciency. Variable valve actuation
(VVA) offers a more promising solution by mimicking the throttling
effect of ETM while mitigating the trade-off with efciency.
Conventionally, VVA has been widely implemented in spark-ignition
engines to improve volumetric efciency and engine performance over a
wide operating envelope. However, VVA is subjected to more rigorous
mechanical constraints in diesel engines. Their high compression ratio
means the clearance between piston and cylinder head at the top dead
centre (TDC) is relatively small. This prohibits deep valve pockets on the
piston surface and allows only a small margin for advancing intake valve
opening (IVO) and retarding exhaust valve opening (EVO) [35,36].
Consequently, production diesel engines have shown only limited VVA
applications, such as Miller timing, addressed through a mechanical
cam-based partial or two-step VVA system.
However, an advanced, fully exible, cam-less VVA system could
provide more control beyond mechanical limitations. For example, it
could control all valves independently at each cylinder. This would
enable sophisticated VVA strategies such as asymmetric valve opening
or cylinder deactivation (CDA), as well as more precise cycle-to-cycle or
cylinder-to-cylinder control [37]. This could improve gas exchange,
combustion stability, emissions and thermal management. Nevertheless,
cam-less valvetrains face challenges, including precise and robust con-
trol, potentially increased parasitic losses and complexity of VVA cali-
bration [38]. Multiple engine performance benets need to be proven
before moving this technology from proof of concept to the production
stage.
Table 1 collates the most relevant, up-to-date research involving
VVA for HD diesel engine ETM. The discussion includes only a very brief
account of the individual contributions, but a more detailed analysis of
the subject can be found in a review article by Kim et al. [39]. Table 1
shows that various studies have demonstrated that VVA is capable of
increasing EGT in HD diesel engines. Intake valve modulation, such as
early intake valve closing (EIVC) and late intake valve closing (LIVC) is
by far the most researched technique. It is relatively easy to implement
by a mechanical, cam-driven partial VVA system, and has been realised
as Miller timing for effective in-cylinder NO
x
reduction. Gehrke et al.
[40] investigated advanced and retarded IVC timing. These raised EGT
by up to +58 ◦C and +116 ◦C respectively, with minor BSFC deterio-
ration (<2.6 %). Guan et al. [41] applied LIVC and raised EGT increment
by up to +52 ◦C, with a BSFC penalty of +5.3 %. Ojeda [42] demon-
strated EIVC and elevated EGT by up to +90 ◦C, while also improving
fuel efciency by 4.5 %.
Table 1 reveals large variations in fuel penalty and EGT increment
between individual studies, even in the same VVA category. For
instance, Gehrke et al. [40] and Guan et al. [41] reported a negative
impact on BSFC, while Ojeda [42] observed a positive impact on fuel
efciency in EIVC and LIVC. Vos et al. [29] indicated minor fuel benet
(−1.5 %) in CDA, but Magee [33] reported a BSFC improvement of up to
−10 %, with similar EGT increment. Zhang et al. [43] observed better
BSFC by −1.2 % in secondary intake valve opening (2IVO), while
Wickstr¨
om [44] reported a BSFC penalty of up to +9 %. Apart from EGT
and BSFC penalty, the majority of the studies included analysis on how
VVA strategies affected emissions. Once again, similar VVA strategies
produced inconsistent emission trends. In LIVC, Gehrke et al. [40]
indicated PM up by 234 % and constant NO
x
, while Guan et al. [41]
reported NO
x
decreased by 13.5 % and PM down by 81 %. In secondary
exhaust valve opening (2EVO), Joshi et al. [45] achieved NO
x
reduction
of 80 %, while Zhang et al. [43] reported a constant NO
x
level.
The above discussion highlights the conicting results between
works covering similar VVA strategies on different operating platforms
and at different operating points. The effectiveness of each VVA strategy
is strongly inuenced by the engine-specic valve proles, injection
parameters, emission control strategy, calibration, etc. Backpressure
level in the exhaust can vary between positive and negative, which will
impose opposite effects on in-cylinder backows. This huge span of
cross-dependencies makes it impossible to use these results to assess
whether VVA is a suitable next development step for off-road engines.
Furthermore, most of the relevant works that provide reliable
experimental results on VVA effects are technical papers, and so provide
limited insight into the mechanisms underpinning the demonstrated
performance/emission results. Most of the VVA strategies considered in
Table 1, aside from their direct impact on ETM and pumping work,
involve a feedback loop to combustion and emission formation, through
changes to in-cylinder trapped mass/ratio of residuals. Sole multi-
cylinder engine experiments are not enough to thoroughly character-
ise the phenomena related to VVA-ETM, implying the signicance of
predictive models as supporting tools.
The present study aims to address the above knowledge gaps by
utilizing a cutting-edge, EU Stage-V compliant off-road diesel engine as
the baseline for a comprehensive simulation of all available VVA stra-
tegies. A well-validated predictive combustion model enables accurate
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
4
estimation of the trade-offs between EGT, efciency, and emissions.
These methodological intricacies are tailored to the objective of iden-
tifying the most promising VVA strategy for efcient aftertreatment
ETM. The study not only presents demonstrated performance results but
also conducts a detailed analysis of the underlying mechanisms
responsible for the observed trade-offs. Consequently, an efcient and
rapid warm-up of EATS by VVA will facilitate compliance with up-
coming EPA Tier 5 emission legislation including a new idle NO
x
stan-
dard and low-load cycle test [46]. This initiative will enhance air quality
and suppress adverse environmental impacts such as global warming
and rapid climate change.
2. Methodology
2.1. Experimental platform
The baseline for the research is a four-stroke, four-cylinder, off-road
diesel engine. It was developed by AGCO Power for typical use in agri-
culture tractors and combined harvesters. Table 2 summarises the en-
gines technical data. The experimental work to obtain reference data
shown in Fig. 4 was performed at the VTT Technical Research Centre in
Finland for model calibration and validation.
With a four-valve, “Mexican-hat” combustion chamber design, the
engines displacement is about ve litres. Diesel fuel was distributed
using a common-rail direct injection system with solenoid-type in-
jectors, capable of multi-pulse operation. Injection and other engine
control parameters are managed by a fully-open engine controller from
MoTeC. The engine calibration was adapted from the original engine
control unit (ECU) by running reference tests and gathering information
of the actuator settings in different load points. Fuel consumption was
measured by an AVL 733 s fuel scale system. The change in fuel mass was
recorded over a 60-second measurement window and then fuel mass
ow was calculated based on the known mass and length of the mea-
surement window.
The engine was equipped with single-stage BorgWarner turbo-
charger. An electronic wastegate controlled boost pressure. The hot
pressurised air from the compressor was cooled by a water-cooled
charge-air cooler (CAC). A hot-lm type air ow meter from ABB
measured total air mass ow at the entry of air path. Various tempera-
ture and pressure readings were collected at several locations in the
intake and exhaust line, as well as in the cooling and lubrication circuits
for monitoring and validation. Fig. 2 is a schematic diagram of the en-
gine test bench and instrumentation. The concentrations of gaseous
components in the exhaust, such as CO, HC and NO
x
, were measured
with a rack of different emission measurement devices, listed with other
instrumentation in Table 3. Particulate concentrations were measured
with an AVL MSS analyser. All emission measurements were sampled
from the raw gases taken after the turbocharger. The exhaust system did
not include a lambda sensor, so the excess air ratio (λ) was calculated,
based on the air mass ow and fuel ow measurement.
In-cylinder pressure prole was measured at only one selected cyl-
inder using a piezoelectric pressure transducer (AVL GU22C) with a
charge amplier, while sensing dynamic pressure change. Zero-level
correction for the cylinder pressure was performed by calculations
within AVL IndiCom measurement software, using pressure difference of
two points along the compression line and pre-dened polytrophic co-
efcient. The calculation assumes compression is adiabatic between the
two points. Measurements of high-frequency pressure and engine rota-
tional speed were triggered via an optical encoder with 0.1 CAD reso-
lution. Engine rotational torque was sensed by an HBM T10F torque
Table 1
Summary of the most relevant research works concerning the use of VVA for efcient ETM.
Ref Type of Engine VVA strategy Operating conditions Results
Speed / load ΔEGT ΔBSFC ΔNOx ΔPM or Soot
[RPM] / [bar] [℃] [%] [%] [%]
Gehrke et al. (2013)
[40]
HD diesel engine,
1.6L, single-cylinder
EIVC 1250 / 5.4 (BMEP) +58 +1.2 0 +104
LIVC +116 +2.6 0 +234
Guan et al. (2019)
[41]
HD diesel engine, 2.0L, single-cylinder, EGR LIVC 1150 / 2.2 (BMEP) +52 +5.3 −13.5 −81
Ojeda (2010)
[42]
HD diesel engine, 6.4L, eight-cylinder
HP*-EGR
EIVC 2050 / 4.3 (BMEP) +90 −4.5 0 −66
Joshi et al. (2022)
[45]
HD diesel engine,
six-cylinder,
HP*-EGR, VGT
2EVO 800 / 1.3 (BMEP) +52 +3 −80 +300
Wickstr¨
om (2012)
[44]
HD diesel engine, 2.0L, single-cylinder 2IVO 1200 / 25 % load +144 +9 −81 >2.5 FSN*
Zhang et al. (2016)
[43]
HD diesel engine,
1L, single-cylinder
2EVO – / 1.3 (IMEP) +78 −1.6 0 >0.1 FSN*
2IVO +55 −1.2 −39 <0.04 FSN*
Vos et al. (2021)
[29]
HD diesel engine,
six-cylinder,
HP*-EGR, VGT
CDA 2200 / 3.9 (BMEP) +117 −1.5 +4 +19
Magee (2014)
[33]
HD diesel engine,
6.7L, six-cylinder,
HP*-EGR, VGT
CDA 1200 / 13 % load +100 −10 NA NA
ΔEGT (þ: increased temperature; –: decreased temperature) ΔBSFC (þ: deteriorated or increased fuel consumption; –: improved or reduced fuel consumption).
ΔNO
x
/ ΔPM or Soot (þ: deteriorated or increased emissions; –: improved or reduced emissions).
FSN*: Filter smoke number HP*: High-pressure.
Table 2
Technical data of the base engine.
Type Off-road diesel engine
Conguration Four-stroke, four-cylinder
Combustion mode Conventional diesel combustion
Rated power / speed 149 kW / 1900 RPM
Max. engine load 23.8 bar (BMEP)
Boosting system Single-stage turbocharger with electronic wastegate
(Max. boost 3.5 bar. abs*)
Injection system Common-rail direct injection with multi- pulse injection
Max. fuel injection
pressure
2000 bar
Valvetrain Fixed valvetrain by mechanical camshaft
No. of valves per cylinder Four
IVO/IVC (at 1 mm lift) 375.6 CAD / 541.2 CAD (TDCF
**
0 CAD)
EVO/EVC (at 1 mm lift) 160.9 CAD/ 345.7 CAD (TDCF
**
0 CAD)
Exhaust aftertreatment
system
DOC, DPF, SCR
Emission class EU Stage-V
abs* - absolute, TDCF
**
- top dead centre when ring.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
5
transducer.
The engine was directly coupled with a 570 kW eddy current dyna-
mometer (Froude AC570F) via a driveshaft with a slight horizontal
offset angle. At the engine side, the driveshaft is mounted to a rubber
damper to dampen the torsional vibrations. Coolant and oil tempera-
tures were controlled by the engine’s own thermostats, and so may have
varied to some extent, depending on engine load and speed. The dyna-
mometer was controlled by its own automation system, based on a
programmable logic controller and in-house software. The automation
system governed all thermal parameters and acquired all the low-
frequency signals.
The engine was run for a pre-dened time at each operating point
before taking a measurement, ensuring that the engine was thermally
stabilised. All the measurements were taken during the last 60 s of a 500-
second measurement stage. All the low-frequency data collection was
performed with an in-house data acquisition system, consisting of
modules from National Instruments. The high-frequency in-cylinder
pressure trace was recorded for 100 consecutive engine cycles with an
AVL IndiModul 622. The measured in-cylinder pressure data were post-
processed using AVL Concerto to analyse combustion characteristics and
compute general performance. A digital low-pass lter was applied to
lter out combustion-induced high-frequency noise in the measured raw
pressure signals. The clean pressure signals were used to compute gross
and net indicated mean effective pressure (IMEP), and also the pumping
mean effective pressure (PMEP), based on following equations, Eqs. (1)–
(3).
IMEPg=1
Vs∮pdv =Area of high pressure loop
Vs
(1)
IMEPn=1
Vs∮pdv
=(Area of high pressure loop) − (Area of pumping loop)
Vs
=IMEPg−PMEP (2)
PMEP =1
Vs∮pdv =Area of low pressure loop
Vs=IMEPg−IMEPn(3)
where, p is instantaneous in-cylinder pressure, v is instantaneous in-
cylinder volume, Vs is swept volume of a cylinder.
For combustion analysis, the gross heat release rate (HRR) and the
gross cumulative heat release (CHR) were calculated, based on the rst
law of thermodynamics (energy balance) and a single-zone model, while
considering in-cylinder heat loss (dQloss
dθ) as follows Eqs. (4)–(7):
HRR =γ
γ−1pdV
dθ +1
γ−1Vdp
dθ +dQloss
dθ (4)
dQloss
dθ =A×hc× (T−Tw)+
εσ
(T4−Tw4)(5)
hc=130 ×v−0.06 ×p0.8×T−0.4× (u+1.4)0.8(6)
CHR =∫EOC
SOC
HRRdθ (7)
where, v is instantaneous cylinder volume, p is instantaneous cylinder
Fig. 2. Schematic diagram of engine test bench with instrumentation.
Table 3
Instrument devices and accuracy.
Type Transducer Measurement
range
Accuracy
(FS =Full Scale)
Air mass ow ABB Sensymaster 0–4000 kg/h 0.6 % (10–100 % of
meas. range)
Fuel
consumption
AVL Fuel balance
733 s
0–1800 g 0.12 % of FS
Torque HBM T10F 0-5000Nm Accuracy class 0.1
(DIN EN ISO 376)
Pressure
- Cylinder AVL GU22C 0–250 bar 0.2 %
- Int./Exh.
manifold
Danfoss MBS 33 0–4 bar 0.8 % of FS
- Intake pipe Vaisala PTB100A 500–1100 mbar 0.15 mbar
- Exhaust pipe Mûller MHDS 400
mbar
0–400 mbar
(rel*)
0.1 % of FS
Temperature K type
thermocouple
0–1200 ◦C 1.5 ◦C
Exhaust composition
- CO Maihak AG NDIR.
UNOR
1–1000 ppm 2 %
- HC J.U.M Engineering
HFID
1–100 ppm 2.5 % of FS
- NO
x
ECO Physics CLD
822
1–5000 ppm 1 %
- PM
concentration
AVL MSS 483 0–50 mg/m
3
Max. 3 % at DR* of
2–10
Max. 10 % at DR* of
10–20
DR*: Dilution rate rel*: Relative.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
6
pressure, T is instantaneous cylinder bulk temperature, Tw is cylinder
wall temperature, A is heat transfer area, hc is convective heat transfer
coefcient, u is mean piston speed, γ is ratio of specic heat, SOC is crank
angle at start of combustion and EOC is crank angle at the end of com-
bustion.
The change of cylinder volume (V) and pressure (p) was calculated
based on 0.1 CAD. The ratio of specic heat (γ) was calculated, based on
bulk cylinder temperature (T)computed by ideal gas law at each crank
angle step. The heat loss model considered convective and radiation heat
transfer which is signicant in diesel engines. The convective heat
transfer coefcient (hc) was estimated by Hohenberg model [47] for
diesel engines with swirl. The heat loss model was further corrected in
the simulation process. The 5 %, 50 % and 90 % mass fraction burned
(MFB) combustion phasing indicators were derived directly from the
CHR curve. The model validation uses the average over 100 cycles for all
cylinder pressure, temperature, IMEP, PMEP and combustion-relevant
parameters.
2.2. 1D engine model set-up
A one-dimensional (1D) engine model was used to perform a model-
based investigation. The research engine model was originally devel-
oped by AGCO Powers engine performance development team, using the
1D commercial code, GT-Suite, as depicted in Fig. 3. EAT was not
necessary because this studys main aim was to investigate EGT incre-
ment accurately, along with combustion, engine-out emissions and fuel
efciency with various VVA strategies. Accordingly, EAT was not
included in the engine model, although its effect on ow restriction was
considered by adding an orice component after the turbochargers
turbine. This is a common practice to simplify the model without losing
the physical behaviour.
The virtual 1D engine model represents the entire engine system,
such as intake and exhaust airpath line, turbocharger, cylinders, ECU,
injectors and even cranktrain system. The turbocharger is located near
the inlet and outlet boundary and was encrypted. Turbocharger per-
formance was computed, based on its compressor and turbine perfor-
mance map. Boost pressure was controlled by a boost controller with
target boost level. A semi-predictive intercooler model was used for
Fig. 3. 1D research engine model.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
7
realistic cooling of compressed charges. The cranktrain system (piston,
connecting rod, crankshaft, bearing, etc.) was modelled for reliable
estimation of mechanical friction over different in-cylinder conditions of
VVA.
The study used the predictive combustion model, DIPulse for reliable
predictions of EGT, emissions and efciency. DIPulse predicts burning
rate (combustion) and associated emissions for direct-injection diesel
engines with single and multi-pulse events. VVA has a direct impact on
in-cylinder mixture formation and combustion, so use of the DIPulse
model captures how VVA inuences combustion. The model estimates
burn rate prole from given injection prole and in-cylinder conditions.
High-quality injection data is required to obtain an accurate combustion
prole from DIPulse: this study used an extensive injection map and
detailed geometry of the injectors as input data.
The DIPulse model works closely with the injector and ECU models.
The ECU model has a brake mean effective pressure (BMEP) controller to
maintain a constant engine load for the present study. It triggers an
increase or decrease in injection rate if the current BMEP varies from the
target value. Based on imposed injection prole, DIPulse estimates the
burn rate prole. This feedback loop is executed until the current BMEP
level is within a certain tolerance of the target BMEP.
The injector model supports multiple injection strategies, such as
pilots, main and post injection. The controller adjusts only main injec-
tion quantity by varying its prole (duration) without changing injec-
tion timing and pressure. Pilot and post injection events remain
constant.
A nite element solver was used to calculate heat losses in the cyl-
inders, wall surface temperature of cylinder liner, piston, and heads,
including valves and valve guides. The results were imposed on the
Hohenberg model to calculate in-cylinder heat loss [47]. The Colburn
heat transfer correlation was adopted for intake and exhaust pipes, while
wall temperature of exhaust pipes was numerically estimated [48].
Turning to emission characteristics, an extended Zeldovich mechanism
was used to model in-cylinder NO
x
formation, and the Hiroyasu model
was applied for soot emission [49].
2.3. Predictive combustion and emission model calibration
The predictive combustion model plays a key role in the present
study to properly capture the associated effects of VVA strategies on
EGT, efciency and emissions. The baseline engine model was thor-
oughly parametrized and calibrated by the engines manufacturer, AGCO
Power. However, at low-load conditions that involve post injection, the
baseline model calibration was not predicting the heat release and
emissions within the accuracy targets. These were set at below 1 % root
mean square error for the reproduced in-cylinder pressure and 5 %
maximum error for NO
x
emissions respectively. Relevant parameters of
the DIPulse model and a reaction rate multiplier of the extended
Zelodvich mechanism in the NO
x
sub-model were recalibrated in order
to achieve these accuracy targets.
Basically, the DIPulse model discretizes the cylinder into three
thermodynamic zones: main unburned zone; spray unburned zone; and
spray burned zone [50]. The models working principle is to trace fuel
from injection stage to burning stage. Four sub-models are used to
capture four phenomenological diesel combustion behaviours –
entrainment, ignition, premixed combustion and diffusion combustion.
Each behaviour is mathematically described, as shown in Eqs. (8)–(11)
respectively.
dminj
dt = − CENT
minjuinj
u2
du
dt (8)
τ
IGN =CIGN
ρ
−1.5exp(3500
Tp)[O2]−0.5(9)
dmPM
dt =CPM mPM k(t−tIGN )2f([O2]) (10)
dmDF
dt =CDF mDF
k
√
Vs
3
√f([O2]) (11)
where, minj is injection mass ow rate; uinj is velocity at injector
nozzle; u is velocity at spray tip;
τ
IGN is ignition delay;
ρ
is pulse gas
density; Tp is pulse gas temperature; [O2]is oxygen concentration; mPM is
premixed mass; k is turbulence kinetic energy; tIGN is time at ignition;
mDF is diffusion combustion mass (remaining unmixed fuel and
entrained gas mass); Vs is cylinder swept volume; CENT is entrainment
rate multiplier; CIGN is ignition delay multiplier; CPM is premixed com-
bustion rate multiplier; and CDF is diffusion combustion rate multiplier.
Entrainment rate in Eq. (8) describes spray penetration and fuel
mixing with surrounding gases when fuel is injected at each pulse. It was
derived from conservation of momentum based on empirical spray
penetration law [50]. The entrainment rate or air mass is inversely
proportional to injection velocity (u) at spray tip. The entrainment rate
is further modied by the entrainment rate multiplier (CENT). Ignition
delay in Eq. (9) represents the time delay between start of injection and
start of combustion, which has a signicant effect on the amount of
premixed charge and premixed combustion. The ignition delay is
computed at each pulse, based on oxygen content ([O2]), pulse temper-
ature (T) and pulse gas density (
ρ
), based on the Arrhenius expression. It
can be modied by the ignition delay multiplier (CIGN ).
Premixed combustion in Eq. (10) is the early stage of diesel com-
bustion. This is rather rapid combustion of premixed charge accumu-
lated during the ignition delay period. The premixed mass (mPM ),
ignition delay (t−tIGN) and its mixing by turbulence (k) inuence the
combustion directly. However, the rate of combustion is kinetically
limited [50] and can be modied by a premixed combustion rate
multiplier (CPM ). After premixed combustion, the remaining unburned
fuel and entrained gas start to mix and burn in a diffusion-limited phase.
This is called diffusion combustion, as described in Eq. (11). It can be
reduced by engine load or long injection duration due to spray-wall and
spray-spray interactions [50]. The diffusion combustion rate multiplier
(CDF) can adjust the combustion rate.
The four multipliers (CENT,CIGN,CPM ,CDF) described above need to be
calibrated with experimental data in order to accurately estimate com-
bustion behaviour. This allows adjustment to suit the intensity of each
phenomenon. Calibration was performed over 10 selected operating
points (OPs) with the engine manufacturers own xed valve settings.
These calibration OPs are denoted by black markers in Fig. 4, encom-
passing engine speeds and load regions critical for the ETM analysis. The
multipliers were tted using GT-Suites multi-parameter optimiser.
Fig. 4. Operating points for model calibration and validation.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
8
Tuning was global level, not case-dependent. Ultimately, one set of four
multipliers was obtained, valid for the entire operating envelope.
Following calibration, both the predictive combustion and engine model
underwent validation at three low load OPs, identied as Mode1,
Mode2, and Mode3 in red in Fig. 4. The validation will be discussed
further in Chapter 3.1. These three OPs, employing the xed valve set-
tings provided by the engine manufacturer, establish the baseline con-
ditions for VVA simulations.
The NO
x
model was calibrated with experimental data in the same
manner. A NO
x
calibration multiplier was tuned to adjust the net rate of
NO
x
formation. However, a multiplier for soot emissions was not cali-
brated in the present study because it is extremely challenging to predict
it exactly [51]. In addition, HC and CO emissions are typically an order
of magnitude lower in CI engines compared to NO
x
and soot, and typi-
cally not an issue when considering EPA Tier 4 nal emission limits.
Thus, these sub-models were not explicitly calibrated. To this end, the
model predictions of soot, HC and CO are to be treated as informative in
terms of trends but not absolute values. Absolute level accuracy was
validated from the NO
x
emission results in this model-based study, but
only for VVA cases close to the baseline valvetrain setting. For this
reason, the absolute NO
x
level prediction is not guaranteed for signi-
cantly early or late and unconventional valve timings.
2.4. VVA strategies and simulation conditions
VVA simulation was performed at three low-load OPs, identied as
Mode1, Mode2, and Mode3 in Fig. 4. Table 4 shows simulation condi-
tions. These three OPs, using the engine manufacturer’s xed valve
settings, serve as the baseline for investigating and analysing the effect
of VVA. The EATS was suffering from particularly low conversion ef-
ciency at these points. All are below 30 % load but range from low to
high engine-speeds. Based on the literature review described in the
introduction, four VVA strategies were identied as interesting for diesel
engine ETM. Each strategy involves two distinctive techniques of
achieving the ETM target. The four strategies selected for VVA simula-
tion are:
•Intake modulation including EIVC and LIVC
•Exhaust modulation including early exhaust valve opening (EEVO)
and late exhaust valve opening (LEVO)
•Internal EGR modulation including negative valve overlap (NVO)
and 2EVO
•Cylinder deactivation (CDA) and cylinder cut-out (CCO)
Three to four different valve proles were reproduced at each tech-
nique, allowing the sensitivity of each strategy on the EGT increment to
be studied. Fig. 5 shows the characteristics of each valve technique.
Fig. 5(a)–(d) illustrate valve proles of the intake and the exhaust
modulation strategies. IVC timing and EVO timing were selected as a
main parameter in the intake and exhaust modulation strategy respec-
tively, since excessive IVO and EVC modulation are constrained by low
valve clearance distance due to the high compression ratio in diesel
engines. For instance, advancing IVO and delaying EVC are ruled out in
order to avoid collision with the piston. In addition, constant valve
overlap was maintained throughout the intake and exhaust modulation,
because change of valve overlap could inuence in-cylinder charge
mixture and thermal condition at IVC, which strongly affects EGT. This
makes EGT analysis difcult since many factors are involved. For this
reason, only one parameter at each technique was modulated up to 80
CAD, with an interval of 20 CAD, in both the intake and exhaust
modulation.
Fig. 5(e)–(h) depict the corresponding valve proles for the internal
EGR and cylinder deactivation/cut-off strategies. In the internal EGR
modulation, internal EGR fraction (or residual fraction) was changed by
either NVO or 2EVO. In the NVO strategy, the high internal EGR fraction
was retained by advancing EVC for early termination of the exhaust
process. Although retarding IVO did not affect the amount of internal
EGR, it provided promising potential to reduce pumping loss. Hence,
IVO was retarded, together with advancing EVC, with an interval of 20
CAD simultaneously up to 60 CAD for better efciency, as seen in Fig. 5
(e).
In the 2EVO strategy, the internal EGR fraction was controlled by
reopening exhaust valves during the intake stroke. During the second
opening, high exhaust pressure pushed burned gases into the cylinder,
increasing the internal EGR effect. The period of the 2EVO was main-
tained while the peak valve lift was changed in three steps (2 mm, 3.5
mm and 5 mm) to vary the internal EGR fraction, as shown in Fig. 5(f).
Higher peak valve lift during 2EVO enlarged the valve opening area,
thus increasing the residual fraction. In fact, 2IVO was considered to be a
promising strategy during the initial literature stage. However, it was
not included in the nal simulation study because it was identied that
2EVO outperformed 2IVO by introducing hot gases directly into the
combustion process [43].
In the CDA and CCO strategy, two of the four cylinders were deac-
tivated. Fuel injection was suspended for the deactivated cylinders. CDA
also shut off both intake and exhaust valves in the deactivated cylinders,
as seen in Fig. 5(g), while CCO maintained all valve operation as normal,
regardless of cylinder deactivation, as seen in Fig. 5(h).
All the simulation was conducted under constant load and boost
pressure conditions at each operating point as the baseline (Table 4)
regardless of VVA. The simulation varied only valve timing and prole.
The rest of the system remained unchanged, exactly as the baseline.
Thus, any changes in behaviour are strongly attributable to VVA. Since
all strategies were performed at exactly the same engine speed and
torque level, this enables extensive back-to-back comparisons such as
burn rate, combustion phase, gas exchange, mass transfer, emissions and
fuel consumption, as well as EGT. All are comprehensively discussed in
the next chapter. Note that in the present study, EGT refers to turbine
outlet temperature before entering the EATS.
3. Results and discussion
3.1. Model validation – standard engine operation without VVA
Before investigating VVA, the calibrated model was validated over
three low-load OPs (Mode1, Mode2, Mode3), as shown in Fig. 4. Fig. 6
compares the simulation results with the experimental results without
VVA at these points, showing absolute and percentage error for selected
key parameters. The key parameters can be divided into two groups. The
rst three (IMEP
g
, P
max
and MFB50) are good indicators to validate the
predictive combustion model because they are directly inuenced by the
predicted in-cylinder and combustion prole. The error limits illustrated
by blue dashed lines were adopted from the recommendation by GT-
Suite [50]. The target limits are not that rigorous, to cover uncertainty
of the predictive model over the wide operating ranges. The remaining
six parameters can be used to validate the general engine performance.
Their 5 % error limits are typical of those adopted in numerical engine
studies to provide sufcient accuracy in representing general engine
behaviours.
Overall, the DIPulse model predicted combustion behaviour well in
Mode1 and Mode2. Both P
max
and IMEP
g
were well within the accuracy
Table 4
Reference operating points with parameters relevant for ETM (Note that infor-
mation on valve actuation and injection is further encapsulated in Fig. 5 and
Fig. 7, respectively).
Simulation conditions Mode1 Mode2 Mode3
Engine speed [rpm] 1000 1425 1900
Engine load [%] 28.8 34.3 29.8
Boost pressure ratio [–] 1.06 1.18 1.46
EGT [◦C] 280 329 293
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
9
Fig. 5. The scope of simulated valve proles: (a) EIVC; (b) LIVC; (c) EEVO; (d) LEVO; (e) NVO; (f) 2EVO; (g) CDA (deactivated cylinder); (h) CCO (all cylinders) and
CDA (activated cylinders).
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
10
target of 5 % and 5 bar. In addition, the error of MFB50 did not exceed
0.5 CAD. But the models predictions were less accurate in the mentioned
parameters in Mode3. Here, although the P
max
was well-captured, the
IMEP
g
was underpredicted, with 8 % error compared to the measure-
ment results. This correlated with a large error (2.5 CAD) in estimating
MFB50.
Fig. 7 compares simulation and experimental results for in-cylinder
pressure and gross heat release rate (HRR) at Mode3. They help
explain the discrepancies in the IMEP
g
and MFB50 results. It is apparent
that the model predicts the prole of HRR very well until approximately
30 CAD after TDC. The simulations inaccuracy evidently stems from the
under-predicted HRR from the late post injection (SOI4). The model
predicts here a distinctively pre-mixed combustion, while the experi-
mental HRR shows an elongated afterburning phase. Furthermore, the
total amount of heat released in this phase is underpredicted by the
model, which results in lower simulated in-cylinder pressure during
expansion. This explains the earlier noted discrepancy in IMEP
g
for
Mode3 (Fig. 6).
There are two possible reasons for the models inaccuracy around late
post-injection prediction. The rst pertains to the accuracy of model
input. The DIPulse model uses injector mass-ow rate to calculate HRR.
This input cannot be directly measured during engine operation and
instead is obtained from injector characterisation, assuming mean-value
rail pressure during all injection events. However, in real engine oper-
ation the common-rail pressure drops momentarily after the main post-
injection, causing reduced fuel rates and slower, more mixing-controlled
combustion for the following post-injection dose. Secondly, while this
inadequacy of the input data could be mitigated by model tuning, large
post injection is seldomly represented in the calibration data. It is noted
that other validation points – Mode1 and Mode2 - also incorporate post
injections, but the order of magnitude of these is lower than in Mode3.
In other words, the models predictivity at Mode3 is far outside its
calibration matrix, forming the most challenging case for simulation.
Bearing this in mind, the model results are considered acceptable,
especially as the combustion reproduction inaccuracies diminish when
considering engine-level parameters relevant to the present study. This
is clearly seen in the BSFC and exhaust thermal state parameters in
Fig. 6. The models inaccuracy in Mode3 is obviously larger than in other
validation cases, but well in line with the 5 % target for EGT and BSFC.
The same applies to NO
x
emissions (BSNO
x
), which were predicted with
relative error below 3.5 % in all cases.
In conclusion, the model is predictive in terms of VVAs effects on
combustion, and its accuracy in terms of thermal state, performance and
emission parameters substantiates its use for the current research.
Nevertheless, in extreme valve timings, the model predictions should be
regarded as indicative in terms of trends rather than absolute values,
especially where combustion-related effects are considered.
3.2. Effect of intake modulation (EIVC and LIVC)
Fig. 8 presents the results of simulating EIVC and LIVC, showing
changes to the governing engine parameters with respect to the baseline
at the validated points of Mode1-3. It is evident that one can elevate EGT
by as much as +150 ◦C by both advancing and retarding the IVC timing
by the theoretically allowable interval. The increase in the EGT is due
primarily to reducing the intake air ow (IAF), as shown in Fig. 8(b).
Intake ow reduction potential, and hence the EGT increment, are
stronger with EIVC and are a direct result of shortened intake duration
and reduced valve lift (see Fig. 5a for reference). LIVCs effect on IAF and
ΔEGT is qualitatively the same as for EIVC, but is due to different phe-
nomenology. Note that here the maximum valve lift is unchanged
(Fig. 5b) and the restriction of intake ow comes from increased back-
ow to the intake manifold when intake valves are still open during the
compression.
In both cases, the reduced IAF inuence in-cylinder heat capacity.
Fig. 9 illustrates in-cylinder heat capacity of excess air at Mode2,
showing that intake modulation reduces in-cylinder heat capacity
Fig. 6. Absolute or percentage errors of combustion and general engine performance parameters: gross indicated mean effective pressure (IMEP
g
); peak in-cylinder
pressure (P
max
); crank angle of 50% fuel mass fraction burned (MFB50); intake air mass ow (IAF); brake specic fuel consumption (BSFC); intake air temperature
(IAT); exhaust gas temperature at turbine outlet (EGT); exhaust pressure at turbine outlet (P
ex
); brake specic NO
x
(BSNO
x
) at Mode1, Mode2 and Mode3.
Fig. 7. Comparison of normalised cylinder pressure (P
cyl
) and normalised gross
heat release rate (HRR) at Mode3 (the least accurate).
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
11
compared to the baseline. The reduction of mass ow by EIVC/LIVC
incurred low heat capacity, resulting in less heat being absorbed during
combustion. Consequently, the lower heat capacity elevates cylinder
temperature, increasing the EGT (Fig. 8a).
Fig. 10(a) shows combustion temperature in the burned zone for
EIVC and LIVC at Mode2. With retarded IVC timing, the burned zone
temperature beyond 30 CAD after TDC increased due to the low heat
capacity effect. Thus, when the exhaust valve opened, hotter combustion
gases were discharged, raising the EGT. Even though EIVC and LIVC
appear to have similar combustion temperatures in the gure, EIVCs
actually is slightly higher where EIVC had slightly more reduction in air
ow than LIVC at the same degree of IVC shifting, as seen in Fig. 8(b).
This reduced the heat capacity (Fig. 9) and then raised EGT. It is
apparent that in-cylinder heat capacity is a main parameter to control in-
cylinder thermal status and EGT when using EIVC and LIVC.
Both EIVC and LIVC had greatest effect on EGT at their most extreme
IVC timings. The EGT increment is +171 ◦C (IVC-80) and +140 ◦C
(IVC+80) respectively at Mode2, corresponding with 36 % (IVC-80) and
33 % (IVC+80) of ow reduction. EIVC produced +31 ◦C higher EGT
than LIVC because its ow restriction was three percentage points
Fig. 8. Simulation results of EIVC and LIVC: (a) ΔEGT; (b) ΔIAF; (c) ΔBSFC; (d) ΔPMEP; (e) Δignition delay; and (f) in-cylinder air–fuel ratio (AFR).
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
12
greater, lowering in-cylinder heat capacity. LIVC performed better at
Mode1, while EIVC was better at Mode3. These results reect the
maximum reduction of air mass ow at each operating point. Therefore,
the effect of EIVC and LIVC on EGT is dependent on the reduction of total
air mass ow, which determines in-cylinder heat capacity.
Fig. 8(c) shows an inconsistent BSFC trend. BSFC in Mode1 and 2 was
similar to the baseline up to ±60 CAD but then increased by more than 2
% at IVC-80 and IVC+80. However, fuel consumption in Mode3 was 2 %
lower than the baseline. In general, BSFC is closely related to pumping
work, but in this study the BSFC trend does not match the PMEP trend,
depicted in Fig. 8(d). PMEP was reduced in most cases of intake mod-
ulation. The similar trend was observed in literature [52,53]. It is noted
that lower trapped mass with EIVC and LIVC attributes to reduced
resistance during compression, leading to a decrease in pumping losses.
Thomas [53] highlighted that a higher degree of IVC modulation pro-
vides higher PMEP reduction potential. However, reduction of PMEP
does not contribute to decreasing BSFC in the present study, except in
Mode3. This indicates that there must be another factor, other than
pumping losses, inuencing BSFC. For this reason, combustion behav-
iour was examined.
In fact, EIVC and LIVC exert a strong impact on combustion. Shifting
IVC timing reduced effective compression ratio so that cylinder pressure
and temperature were lower than the baseline at the end of compression
stage. Fig. 11 shows the bulk in-cylinder temperature during
compression stroke at Mode2 for different LIVC cases. Cylinder tem-
peratures at the point of pilot and main injections with IVC+80 were 60
to 70 ◦C lower than the baseline. Consequently, peak cylinder pressure
and peak cylinder temperature gradually reduced with shifting IVC
timing, as observed in Fig. 10(a) and (b). The same trend was observed
with EIVC. Low cylinder temperature with EIVC and LIVC means the
air–fuel mixture takes longer to reach self-ignition temperature,
increasing the ignition delay time. Larger shifts in IVC timing lead to
even lower cylinder temperatures and hence greater ignition delays.
Fig. 8(e) shows ignition delay was prolonged by up to 10 CAD at IVC-80
and IVC+80. This delay helped to enhance premixed combustion, as
shown in Fig. 12, because the additional air–fuel mixing time increased
the amount of premixed mixture.
Ultimately, the long ignition delays as shown in Fig. 8(e) shifted the
whole combustion process. This is most noticeable at extreme IVC tim-
ings, IVC−80 and IVC+80, where longer ignition delay slowed com-
bustion (long combustion duration) and delayed MFB50. This indicated
that more fuel is burned when the piston is moving downward, not near
TDC. This caused less effective work transfer and so increased BSFC.
Therefore, BSFC was affected by the combined effects of PMEP and
combustion.
Mode1 and 2 show stable BSFC with moderate shifts of IVC timing up
to ±60 CAD, where reduction of PMEP is offset by delayed combustion.
However, it seems that the reduction of PMEP is stronger than the
delayed combustion, so BSFC was reduced at Mode3. At the most
Fig. 9. Trace of specic heat capacity (c
p
) of excess air at Mode2 dur-
ing combustion.
Fig. 10. Simulation results of EIVC and LIVC at Mode2: (a) burned zone temperature and (b) normalised in-cylinder pressure (P
cyl
).
Fig. 11. Simulation results of LIVC at Mode2: bulk cylinder temperature.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
13
extreme IVC conditions (IVC−80 and IVC+80), BSFC increased sharply
because combustion is slowed too much by excessive phase delay, so
milder intake modulation is suggested in order to avoid increasing fuel
consumption. Another measure entails advancing the combustion phase
to mitigate the delay. Two methods are available for advancing the
combustion phase: adjusting injection timing and internal or external
EGR. De Ojeda [42] demonstrated maintenance of combustion phase
(MBF50) by advancing injection timing with EIVC, improving BSFC by
4.5 % and raise EGT about 95 ◦C. Guan et al. [41] employed internal
EGR with LIVC to reduce ignition delay by introducing hot residual
gases. This resulted in reducing the BSFC penalty by 0.7 % while
increasing EGT by 62 ◦C.
Fig. 13 shows how the change of combustion characteristics inu-
enced engine-out emissions. LIVC and EIVC reduced peak combustion
temperature due to low effective compression ratio and delayed com-
bustion phase, as shown in Fig. 10(a). The lower peak combustion
temperature gradually reduced NO
x
emissions (Miller effect), as seen in
Fig. 13(a). The maximum NO
x
reduction of more than 97 % was
observed at IVC+80 and IVC-80. Some points are even below EU Stage-V
NO
x
limit (0.4 g/kWh). However, Fig. 13(b) shows that extreme IVC
modulations drastically increased soot emission because the air mass
ow was substantially decreased. This enriched the fuel mixture, as
shown in Fig. 8(f), promoting soot formation. Furthermore, the reduc-
tion in excess air constrained soot oxidation, despite increased com-
bustion temperature.
Fig. 13(a) and (b) depict a typical diesel engine trade-off between
soot and NO
x
up to ±60 CAD. However, this trade-off disappears at
IVC−80 and IVC+80, where both NO
x
and soot were mitigated simul-
taneously. This is typical when combustion moves from mixed-control to
premixed mode and can be clearly seen from Fig. 12
′
s HRR plot. Towards
particularly early/late valve timings, elongated ignition delay enables
more time for fuel to premix, and increased local oxidizer availability
Fig. 12. Simulation results of EIVC and LIVC at Mode2: normalised gross HRR.
Fig. 13. Simulation results of engine-out emissions from EIVC and LIVC: (a) ΔNO
x
; (b) ΔSoot; (c) ΔHC; and (d) ΔCO.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
14
suppresses soot formation. Note however, that the results of the pre-
dictions of the combustion/emission sub-model at extreme valve timings
have not been validated, and so should be treated with caution.
Fig. 13(c) and (d) show that modulating IVC timing increased HC and
CO emissions. These emissions are unburned and partially burned fuels,
representing incomplete combustion. Both EIVC and LIVC reduced the
fresh charge ow and trapped air mass, so there was insufcient oxygen
for the injected fuel. Even though increased ignition delay gave more
mixing time, locally rich mixture resulted in incomplete combustion.
Guan et al. [41] and Garcia et al. [55] highlighted that low combustion
temperature also contributed to incomplete combustion, thereby
increasing HC and CO emissions. Moreover, the oxidation process of HC
and CO in the combustion chamber is suppressed by the reduced excess
air. Nevertheless, the DOC’s improved conversion efciency due to
elevated EGT will effectively mitigate HC and CO, leading to low tailpipe
HC and CO emissions.
In short, intake modulation could elevate EGT more than 150 ◦C by
controlling in-cylinder thermal status with air ow. Extreme IVC
modulation raises EGT signicantly, but produces considerable changes
in combustion characteristics, such as long ignition delay, so entails a
minor fuel penalty due to delayed combustion. The fuel penalty can be
minimised by adjusting fuel injection timing or introducing EGR. Intake
modulation also gave signicant reductions of NO
x
via the Miller effect,
but soot, HC and CO emissions deteriorated due to the rich mixture.
Increasing the air–fuel ratio can reduce these emissions but will also
reduce the EGT increment.
3.3. Effect of exhaust modulation (EEVO and LEVO)
Fig. 14 plots simulation results of exhaust modulation. Fig. 14(a)
shows an increasing EGT trend as EVO timings move further away from
the baseline. Both extreme EEVO and LEVO achieved the highest EGT
increments. EVO−80 raised EGT by +101 ◦C; EVO+80 raised EGT by
+68 ◦C. But these excessive EVO timings also carried signicant BSFC
penalties of +16.6 % (EVO−80) and +13.3 % (EVO+80), as seen in
Fig. 14(c). Compared to intake modulation, the EGT increment is rather
Fig. 14. Simulation results of EEVO and LEVO: (a) ΔEGT; (b) ΔIAF; (c) ΔBSFC; (d) ΔPMEP; and (e) AFR.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
15
low and the BSFC penalty is high. Unlike intake modulation, the EGT
increment is not related to air ow, since air ow (IAF) remains con-
stant, as shown in Fig. 14(b). Constant boost pressure and unchanged
intake valve prole gave constant intake ow.
Instead, exhaust modulations ability to raise EGT is associated with
how EVO timing affects the point at which combustion ends. In the
power stroke, EVO timing determines when combustion will be halted.
Advancing EVO timing discharges more hotter combustion gases
because combustion ceases earlier. This moves the temperature at EVO
towards peak temperature regions, as shown in Fig. 15(a). The high in-
cylinder thermal status was achieved by reduced expansion work (pre-
vent temperature drop) and short residual time (less heat losses).
However, advancing EVO timing gradually reduced the gross work
because combustion was terminated earlier. This degraded both torque
and power. To maintain the same engine performance, fuel injection
needs to be increased to compensate performance drop, causing a high
BSFC penalty with EEVO.
Delaying EVO timing postpones the end of combustion even after
bottom dead centre (BDC). After BDC, piston is moving upward but
exhaust valves are still closed, so the in-cylinder charge is compressed.
In our case, cylinder pressure rose to 3 bar where usually exhaust
pressure is below 2 bar at low load, even though it depends on turbo-
charger operation. This promoted the high cylinder temperature seen in
Fig. 15(b) and thus higher EGT. The later the EVO timing, the higher the
EGT, due to re-compression. However, the re-compression induced high
pumping work during the gas exchange process. Once again, more fuel
injection is inevitable to maintain the same performance, resulting in a
high BSFC penalty, even though expansion work is maximised in LEVO
cases.
EEVO demonstrated better EGT performance than LEVO for the same
degree of EVO modulation. The main reason is that EEVO directly
released hot gases, while LEVO was involved with long expansion and
re-compression which increased heat losses. Both exhaust modulations
caused rather high BSFC penalties of up to +13.3 % (EVO+80) and
+16.6 % (EVO−80). The BSFC increment stems from pumping work
(LEVO and EEVO) and reduction of gross work (EEVO). It is noted that
EVO modulation strongly affects the gas exchange process and power
cycle. However, neither EEVO nor LEVO has a major effect on com-
bustion behaviour because intake ow and in-cylinder charge condition
(at IVC) before the combustion remained similar to the baseline. As a
result, combustion and the cylinder pressure proles were fairly
consistent for all cases. This combustion stability makes implementation
of exhaust modulation easier because combustion recalibration is not
needed.
Fig. 16 provides engine-out emissions for all EEVO and LEVO cases.
Variation of NO
x
emission was rather small (less than 20 %) because
combustion was not inuenced by EVO modulation. This is related to the
relatively stable peak combustion temperatures, seen in Fig. 15(a),
compared to other strategies (see Fig. 10a for reference). Fig. 16(b) and
(c) shows greater EVO modulation increased both soot and HC. Soot
formation was promoted by rich mixture with changing EVO, as seen in
Fig. 14(e). Unlike intake modulation, the rich mixture was created by
increased fuel injection. Overall, EEVO indicated slightly higher soot
than LEVO. This is because EEVO had less residual time, which inter-
rupted the oxidation process.
EEVO also showed higher HC than LEVO. The reduced combustion
duration with EEVO curtailed complete combustion, which increased
HC. This effect was strongest at most advanced EVO case (EVO−80)
where post-injected fuel had less time to be burned: LEVO gave longer
combustion time, enabling a more complete burn. LEVOs longer com-
bustion also gave more oxidation time. This combination of long com-
bustion and oxidation time contributed to slightly lower HC with LEVO.
Fig. 16(d) s CO emission trend is noticeably different from the others.
Two factors caused LEVOs high CO emissions. One is increased incom-
plete combustion of the rich mixture; the other is increased partial
oxidation of HC from the long oxidation process. In contrast, CO with
EEVO was lower than the baseline. This is probably attributable to
reduced combustion efciency. Note that CO has very low activation
energy and so is generally easier to oxidise than HC. CO emission is
preliminary attributed to local oxygen deciencies, so less HC under-
going combustion leaves more oxygen surplus for complete oxidation of
CO.
Exhaust modulation increased EGT by up to +100 ◦C, but with a high
BSFC penalty of more than +10 %. EGT increment is controlled by
changing combustion duration, but it has negative effect on fuel ef-
ciency due to loss of expansion work (EEVO) and high pumping work
(EEVO and LEVO). Furthermore, different combustion duration pro-
duced different emission trends. Despite changing combustion duration,
combustion characteristics were not inuenced, so NO
x
emissions were
not heavily affected by EVO modulation. Lastly, excessive EVO modu-
lation should be avoided because it carries a high fuel penalty.
3.4. Effect of internal EGR (NVO and re-introduction of exhaust)
Fig. 17 presents the simulation results of i-EGR. First, increasing NVO
and increasing valve lift of 2EVO both promoted more dilution effect.
This is indicated by residual gas fraction (RGF) in Fig. 17(a). Fig. 17(b)
shows that both NVO and 2EVO elevated EGT and that the increase is
proportional to the EGR effect. The EGT increment was +205 ◦C
(NVO60) and +120 ◦C (2EVO with 5 mm lift) respectively. The EGT
Fig. 15. Simulation results at Mode1: (a) burned zone temperature in EEVO; and (b) bulk cylinder temperature in LEVO.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
16
increment mechanism is rather similar to intake modulation. A higher i-
EGR suppressed fresh charge induction, which reduced in-cylinder heat
capacity, as seen in Fig. 17(c). This raised combustion temperature and
then EGT. Fig. 18(a) displays increasing combustion temperature due to
the effect of low heat capacity with increasing dilution effect in NVO and
2EVO at Mode2.
However, the reduction mechanism of air ow is different. Intake
modulation directly restricted intake ow by shifting IVC timing. On the
other hand, i-EGR utilised RGF to constrain intake ow. Increasing the
NVO period by advancing EVC trapped more RGF because the exhaust
cycle was halted earlier. High RGF in the cylinder prevents induction of
fresh charges in the next cycle, cutting incoming air mass ow.
2EVO allowed backow during intake stroke because exhaust pres-
sure is usually higher than intake pressure at low-load operation.
Increasing the maximum valve lift increased effective valve opening
area, thus raising RGF. However, air ow reduction occurred at the same
cycle, as shown in Fig. 18(b), which illustrates instantaneous mass ow
via intake and exhaust valves. With both intake and exhaust valves open,
the higher degree of backow (high RGF) countered the incoming ow,
creating resistance for the incoming charge. Fig. 18(b) clearly shows the
backows suppression of the incoming charge.
Fig. 17(b) and (c) show that increasing RGF with either NVO and
2EVO restricted incoming charge ow, which improved EGT. Although
direct comparison of NVO and 2EVO is difcult, NVO produced the
higher EGT. This is mainly because NVO achieved more EGR in the
current study. NVO60 raised RGF to 29 %, which cut intake air ow by
40 %. 2EVO with 5 mm valve lift raised RGF to 24 %, reducing air ow
by 31 %. Thus, NVO60 showed greater EGT increment. However, the
RGF of NVO20 was lower than 2EVO with 2 mm lift, so 2EVO raised EGT
further. It is evident that RGF is the main parameter to control EGT in i-
EGR. Results with 2EVO in particular reveal the strong relationship
between RGF and EGT. In 2EVO, RGF increment is more noticeable at 2
mm lift, so the EGT increment also is large. But with higher lifts, the
incremental RGF diminishes when compared with the NVO cases, so the
increase in EGT also begins to atten.
Although a large amount of EGR is effective in raising EGT, it harms
fuel efciency. Both NVO and 2EVO exhibit an increasing BSFC trend in
Fig. 17(d). NVOs BSFC penalty was up to +15.7 % (NVO60), while
2EVOs was less severe, at up to +6.5 % (5 mm lift). NVOs higher BSFC
penalty resulted from increased pumping work, as shown in Fig. 17(e).
This is mainly due to re-compression by EEVC and re-expansion by LIVO
during NVO, which is clearly visible in Fig. 19. Cylinder pressure was
raised by more than 10 bar during the gas exchange process, which has a
negative effect. This behaviour became more severe with a long NVO
period at Mode1 and 2. However, PMEP was reduced at Mode3. In fact,
the baseline pumping loop is rather large due to substantial pressure
gradient between intake and exhaust. NVO reduced Mode3 pumping
losses by narrowing the pumping loop via re-compression and re-
expansion. Nevertheless, PMEP is not the main cause of increased
BSFC, because 2EVO and NVO (Mode3) still carry a BSFC penalty,
despite their pumping work being lower than the baseline. Further
analysis of the combustion behaviour was required to investigate the
BSFC penalty in i-EGR.
Fig. 20 presents the predictive combustion models HRR at Mode2.
Fig. 16. Simulation results of engine-out emissions from EEVO and LEVO: (a) ΔNO
x
; (b) ΔSoot; (c) ΔHC; and (d) ΔCO.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
17
Both NVO and 2EVO show reduced premixed combustion behaviour
with increasing EGR rate, which is opposite to intake modulation. This
behaviour was observed at other operating points. Since less fuel is
burned at the early stage of the combustion, more fuel is burned at the
later stage, which is diffusion combustion mode. Therefore, combustion
duration was prolonged with increasing EGR fraction. It is widely known
that EGR slows down combustion due to reduced oxygen content, which
corresponds to our results. This phenomenon is clearly visible in Fig. 20.
NVO60s combustion duration in particular is much longer than that of
2EVO with 5 mm lift, reecting the higher EGR effect of NVO60 in
Mode2.
Ultimately, the slow and delayed combustion caused ineffective
work transfer and reduced performance because more fuel was burned
far from the sweet spot near TDC, and more slowly. Injection of more
fuel is essential to maintain the same performance, contributing to
deteriorating fuel efciency in both cases. Fig. 20 shows that NVO60 in
particular underwent more changes in combustion phase and duration
than 2EVO with 5 mm lift, and thus recorded a larger fuel penalty.
Fig. 17. Simulation results of NVO and 2EVO: (a) RGF; (b) ΔEGT; (c) ΔIAF; (d) ΔBSFC; (e) ΔPMEP; and (f) AFR.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
18
The reduced intensity of premixed combustion in both NVO and
2EVO with increasing EGR resulted from less excess oxygen. Introducing
hot exhaust gases when injection was initiated does not reduce in-
cylinder mixture temperature, so ignition delay time or mixing time
was not affected negatively, unlike intake modulation. However, less
oxygen concentration with EGR yielded poor air–fuel mixing and less
premixed charge. This reduced premixed combustion. The lower HRR in
premixed combustion reduced the peak cylinder pressure and lowered
peak combustion temperature, as shown in Fig. 18(a). Although both
intake modulation and i-EGR exhibited similar cylinder-pressure prole
trends, their differing ignition delays caused hugely different combus-
tion behaviour.
NVO and 2EVO show similar emission trends in Fig. 21. These are
typical of diesel engines with EGR. First, Fig. 21(a) shows huge NO
x
reduction (up to ~ −99 %) with increasing EGR, due to the reduction in
peak combustion temperature, as seen in Fig. 18(a). However, other
emissions rose. Fig. 21(b) shows soot was increased by up to +145 %.
The air–fuel mixture became rich due to reduced excess air, as seen in
Fig. 17(f), promoting soot formation. Also, the lower ame temperature
and less excess oxygen reduced the rate of soot oxidation/re-burning
[56]. HC and CO emissions increased by up to +239 % and +107 %
respectively, as shown in Fig. 21(c) and (d). These are the results of
increased incomplete combustion, stemming from poor air–fuel mixing
with high EGR rates. Additionally, the low oxygen concentration sup-
pressed oxidation, contributing to increasing carbon emissions. NVO60
had the highest EGR effect and also recorded most of the highest soot,
HC and CO emissions.
Internal EGRs dilution effect has demonstrated the capability to raise
EGT by more than +200 ◦C. Increasing the EGR fraction raises EGT by
restricting in-cylinder air mass, thus lowering heat capacity. However,
this huge improvement in EGT is accompanied by a considerable BSFC
penalty, because EGR harms combustion. This penalty can be mitigated
to some extent by adjusting injection timing. EGR suppresses NO
x
by up
to 99 %, but the rich mixture increases other emissions.
3.5. Effect of cylinder deactivation (CDA) and cylinder cut-out (CCO)
Fig. 22 illustrates how CDA and CCO affect EGT and IAF over
different boost pressure levels. The model gradually varied boost pres-
sure, as shown in Fig. 22(a). The same boost pressure was imposed in
both CDA and CCO operation at each boost level (v1, v2, and v3) except
Mode2 (boost v3), where the turbocharger could not deliver the target
boost. Fig. 22(b) shows CDA enhanced EGT up to +264 ◦C (boost v1)
from its baseline, whereas CCO raised EGT by only +101 ◦C (boost v1).
Despite this difference in EGT increment, both strategies have high
Fig. 18. Simulation results of NVO and 2EVO at Mode2: (a) burned zone temperature; and (b) instantaneous mass ow rate via intake and exhaust valves.
Fig. 19. Simulation results of log P – log V diagram from NVO and 2EVO
at Mode2.
Fig. 20. Simulation results of NVO and 2EVO at Mode2: normalised gross HRR.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
19
overall combustion temperature compared to baseline at Mode1, as seen
in Fig. 23(a). The same trend was observed at other operating points.
This is explained later.
The primary effect of high combustion temperature is increased
HRR, as shown in Fig. 23(b). Since the two deactivated cylinders did not
produce any work, the activated cylinders need to work almost twice as
hard to maintain the same performance, as seen in Fig. 24(c). Hence,
more fuel was injected, raising combustion temperature compared to the
baseline, as shown in Fig. 23(a).
The secondary effect is boost pressure. Increasing boost pressure
with both CDA and COO reduced EGT. This is associated with the heat
capacity effect, as discussed before. Increasing boost pressure delivered
more air into the activated cylinders, as observed in Fig. 22(c). This
increased the heat capacity, which suppressed high temperature incre-
ment during combustion. For this reason, the high boost pressure
showed lower combustion temperature after 30 CAD from TDC in Fig. 23
(a), leading to lower EGT. However, it was observed that the primary
effect is much stronger than the secondary effect in the present study.
It was noted above that CDA and CCO produced broadly similar
combustion temperatures, but that CDA obtained an EGT that was more
than +100 ◦C higher than CCO. This is mainly due to the different gas
exchange process arising from different valve motions in the deactivated
cylinders. Fig. 25 illustrates contour plots of temperature along the
exhaust pipe lines at Mode1 (boost v3). CCO allowed the deactivated
cylinders to breathe as usual, but because combustion did not occur,
they discharged rather cold charges, as shown in Fig. 25(a). The mixing
of the cold charges with hot gases from activated cylinders meant a large
amount of thermal energy was lost, restricting the EGT increment. In
fact, the impact of CCOs cold gases could even reduce EGT below the
baseline level at Mode2, as seen in Fig. 22(b).
In contrast, all valves were shut down in CDAs deactivated cylinders,
so there was no mixing, as shown in Fig. 25(b). Instead, some hot gases
owed back to the exhaust line of deactivated cylinders, heating up the
exhaust pipe. Even though there are some losses during backow, it is
not that signicant compared to CCO. This contributed to maintaining a
high EGT in CDA. This demonstrates that the valve operation of deac-
tivated cylinders strongly affects EGT. Since the charges were not
equally distributed in CDA, the total air mass ow in Fig. 22(d) could not
provide correct in-cylinder air mass distribution. Instead, net charge
ow per activated cylinder is a more useful parameter to estimate heat
capacity effect.
Furthermore, differences in the gas exchange process in the deacti-
vated cylinders have different impact on fuel efciency, as seen in
Fig. 24(a). CCO showed a high BSFC penalty of up to 15 % against the
baseline (boost v1 at Mode3), whereas CDAs penalty was no greater
than +2.9 % (boost v1 at Mode1). CDA even displayed better BSFC than
the baseline at some points with high boost pressure. Fig. 24(b) indicates
that CCOs larger fuel penalty stemmed mainly from its higher pumping
work. To investigate the factors that induce a large pumping work,
Fig. 26 compares PMEP in both activated and deactivated cylinders
Fig. 21. Simulation results of engine-out emissions from NVO and 2EVO: (a) ΔNO
x
; (b) ΔSoot; (c) ΔHC; and (d) ΔCO.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
20
(CYL
a
/CYL
d
) with average values (AVG) at boost v2, showing that high
pumping work in both CCO’s activated and deactivated cylinders
mostly.
The high PMEP in the activated cylinder is likely from high back
pressure due to over-closed wastegate because CCO has rather low
exhaust enthalpy at turbine inlet suffering from low EGT (Fig. 22b).
However, the opening of the waste-gate cannot be traced back due to the
encryption of the turbocharger model in the present study. This effect
was rather small at Mode1 owing to rather low boost requirement.
However, it was more severe at Mode2/3 when boost control was
Fig. 22. Simulation results of CDA and CCO: (a) boost pressure; (b) ΔEGT; (c) ΔIAF per activated cylinder; and (d) Δtotal IAF.
Fig. 23. Simulation results of CDA and CCO at Mode1: (a) burned zone temperature; and (b) normalised gross HRR.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
21
Fig. 24. Simulation results of CDA and CCO: (a) ΔBSFC; (b) ΔPMEP; (c) Δnet IMEP; and (d) AFR in activated cylinder.
Fig. 25. Contour plot of exhaust temperature at Mode1 (boost v3): (a) CCO; and (b) CDA.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
22
actively involved with increased boost pressure. Even though CDA
increased averaged PMEP up to 40 % at Mode1 (Fig. 24b), this incre-
ment is very small as seen in Fig. 26.
The high pumping work in the deactivated cylinders is attributed to
normal valve operation. Fig. 27 presents P-V diagram of a deactivated
cylinder in log scale for both cases at Mode3. It clearly shows that CCO
has large pumping loop due to normal valve operation. However, CDA
had no pumping loop because valve operation was not involved. In a
similar manner, it became severe at Mode2/3 in Fig. 26 due to high back
pressure with wastegate control while CCO’s PMEP at Mode1, was less
affected due to low boost demand. It is noteworthy that boost control
could inuence the gas exchange process and pumping loop to some
extent in CDA/CCO. Consequently, CCO incurred a higher fuel penalty
because it needs to produce more work to compensate for pumping
losses in both activated and deactivated cylinders except Mode1. This
explains why CCOs IMEP is slightly higher than CDAs in Fig. 24(c).
Another interesting trend is that increasing boost pressure could
reduce the BSFC penalty. This relates to combustion behaviour. Fig. 23
(b) shows ignition delay was short due to the high in-cylinder temper-
ature derived from increased engine load. Premixed combustion was
enhanced in both CDA and CCO with increasing boost pressure, because
this delivered more air which improved air–fuel mixing and the amount
of premixed charge. Even though ignition delay was slightly shorter,
premixed combustion was enhanced by the abundant excess air from
increased boost pressure. The extra premixed combustion at high boost
pressure released more heat, raising peak combustion temperature, and
advanced the combustion phase as shown in Fig. 23(b). Since more fuel
burned earlier near TDC, this promoted more effective work transfer.
Consequently, BSFC at high boost pressure (v3) was reduced by up to
−3.2 % (CDA) and −1.7 % (CCO) compared to low boost pressure (v1).
BSFC could be improved even further by optimising injection timing to
achieve the same MFB50 as the baseline.
Fig. 23(b) shows that combustion was longer and slower with CDA
and CCO compared with the baseline. This is mainly due to a longer
main fuel injection duration. Injection pressure and start of injection
(SOI) were xed, so injection duration needs to be extended to deliver
more fuel. This lengthened the whole combustion process. In practice,
injection pressure is supposed to be increased with increased engine load
in CDA and CCO, but this was not considered in the present study.
Raising the injection pressure would reduce the long combustion dura-
tion to some extent. CCOs combustion duration was slightly longer than
CDAs at all three boost levels, since CCO needed increased fuelling and
hence had a longer main fuel injection duration. Combustion duration
was shorter at high boost pressure because increased premixed com-
bustion burned more fuel at the early stage of the combustion.
Fig. 28 depicts engine-out emissions, and shows that both CDA and
CCO display similar trends with increasing boost pressure. Fig. 28(a)
shows that raising boost pressure also increases NO
x
emission. This is a
consequence of increased peak combustion temperature due to
enhanced premixed combustion releasing more heat at the beginning of
combustion. However, note that NO
x
emission was reduced by 50 % at
low boost pressure (v1) for both CDA and CCO because their peak
combustion temperature was lower than the baselines, as seen in Fig. 23
(a). This mainly is due to maintained injection pressure, despite
increased load. In real-world operation, higher peak combustion tem-
perature is expected with increasing injection pressure with CDA and
CCO. But even though this means engine-out NO
x
emission would be
higher, the increased EGT will enhance EATS efciency, so tailpipe NO
x
emissions will be lower.
Soot, HC and CO emissions were increased by cutting half of the
cylinders (boost v1), as seen in Fig. 28(b), (c), and (d). The increased
main fuel injection quantity created locally rich mixture, as seen in
Fig. 24(d), causing high soot formation. Fuel injection was more than
doubled in the activated cylinders; there was insufcient excess air to be
mixed completely, resulting in incomplete combustion. Higher boost
pressure provided more excess air, promoting better air–fuel mixing and
alleviating locally rich mixture. This suppressed formation of soot, HC
and CO but also lowered EGT. Additionally, increased oxygen concen-
tration encouraged more oxidation. But these emissions never fell below
their baseline levels, since formation was still greater than oxidation,
except in the case of CO emission at CDA operation. However, several
experimental studies have demonstrated that soot, HC and CO emissions
Fig. 26. Comparison of PMEP in activated and deactivated cylinders for CDA and CCO at Mode1, Mode2 and Mode3 (boost v2).
Fig. 27. Simulation results of CDA and CCO: log P – log V diagram of deacti-
vated cylinder at Mode3.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
23
can be cut by more than 50 % in CDA and CCO by increased oxidation,
high air fuel ratio (AFR) and increased injection pressure [57,58].
Both CDA and CCO improved EGT, mainly by increased heat release
in activated cylinders. CDA achieved extremely high EGT increment of
more than +250 ◦C, with only a minor BSFC penalty (<2.5 %). CCO gave
an EGT increment of up to +101 ◦C, but with a BSFC penalty of up to 15
%. Both generated similar in-cylinder thermal conditions under same
boost pressure, but the differing valve prole in deactivated cylinders
caused different outcomes for EGT, BSFC and emissions, due to the
mixing effect. Raising boost pressure could improve BSFC and emissions
(soot, HC and CO) to some extent, but reduces the EGT benets. How-
ever, the maximum achievable boost level is limited by the character-
istics of turbocharger. Despite of noise, vibration and harshness arising
from unbalanced cylinder ring in CDA and CCO, it can be effectively
Fig. 28. Simulation results of engine-out emissions from CDA and CCO: (a) ΔNO
x
; (b) ΔSoot; c) ΔHC; and (d) ΔCO.
Table 5
Summary of VVA impact on ΔEGT; ΔIAF; ΔBSFC; ΔPMEP; combustion; and engine-out emissions.
Category VVA ΔEGT ΔIAF ΔBSFC ΔPMEP Combustion effect EGT control mechanism ΔNO
x
ΔSoot ΔHC ΔCO
Intake modulation EIVC ++ – +– More premixed
combustion
Air ow reduction
(low heat capacity)
— + + ++
LIVC ++ – -+– – + + +
Exhaust modulation EEVO +-+ +++ + No major effect Combustion duration + ++ ++++ –
LEVO +-+ +++ +++ + + + +
Internal EGR NVO ++ – +++ ++ Less premixed
combustion
Air ow reduction
(low heat capacity)
— ++ +++ ++
2EVO +– +– — ++ +++ +
Cylinder deactivation CDA +++ – -+– Increased heat
release
– +++ +++ +
CCO +– ++ + – ++++ ++++ +
ΔEGT (þ: increased temperature; –: decreased temperature) ΔIAF (þ: increased mass ow; –: decreased mass ow).
ΔBSFC (þ: deteriorated or increased fuel consumption; –: improved or reduced fuel consumption) ΔPMEP (+: increased pumping work; -: decreased pumping work).
ΔNO
x
/ ΔSoot / ΔHC / ΔCO (þ: deteriorated or increased emissions; –: improved or reduced emissions).
ΔEGT: +<100 ◦C, 100 ◦C <++<150 ◦C, 200 ◦C <+++, ΔIAF: –<-30 %, −30 %<-<-1%, −1%<-+<+1%.
ΔBSFC: −1%<-+<+1%, +1%<+<+5%, +5%<++<+10 %, +10 %<+++, ΔPMEP: −50 %<–<-10 %, −10 %<-<0%, 0 %<+<+50 %, +50 %<++<+150 % +150
%<+++.
ΔEmissions: —<-90 %, −90 %<–<-50 %, −50 %<-<0%, 0 %<+<+50 %, +50 %<++<100 %, +100 %<+++<+400 %, +400 %<++++.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
24
mitigated in diesel engines [59,60]. Equally, it has been demonstrated
that smooth transient performance can be achieved without severe
torque variations during the switching phase [61,62].
3.6. Comparison of all VVA strategies
All VVA strategies have been evaluated at the same operating points
under constant load conditions, enabling direct compassion. Table 5
summarises the case-averaged effects of VVA on EGT, IAF, BSFC, PMEP,
combustion and emissions. Additionally, Fig. 29 illustrates EGT-BSFC
performance. These graphs highlight common characteristics. First, all
VVA strategies indicate potential to increase EGT above the baseline,
with the exception of CCO at Mode2, where EGT declined. Second,
increasing EGT is always accompanied by increasing fuel penalty,
regardless of engine operating point, but again with one exception:
intake modulation at Mode3 showed better BSFC than baseline. In
general, the fuel penalty is mainly associated with either high pumping
work or degraded combustion, or a combination of both. Optimising
combustion parameters can minimise the fuel penalty stemming from
deterioration of combustion. In contrast, it is rather difcult to reduce
pumping work, since it is determined by valve timing and proles as
well as turbocharger control. Nevertheless, combustion proved to be the
dominant effect on BSFC in the present study, so optimising the com-
bustion phase remains a viable option for minimising the BSFC penalty
to some extent.
Fig. 29 shows the general trend of a trade-off between EGT and BSFC,
with a few exceptions. Real-world implementation demands that the
most fuel-efcient options are considered. Fig. 29s plots can be divided
into two sub-regions to evaluate VVA strategy in terms of fuel efciency.
The ideal strategy will be located near the top left corner, indicating
simultaneous high EGT increment and low BSFC penalty. This points to
cylinder deactivation as our studys most fuel-efcient strategy, since it
elevated EGT by more than +250 ◦C, even with lower BSFC than the
baseline. It is followed by intake modulation, with high potential of EGT
increment above +150 ◦C and a minor fuel penalty of up to +3.5 %. The
low BSFC penalty in both cases stems from low pumping work. Table 5
shows that intake modulation also has the benet of a relatively low
emissions penalty compared to others. However, both strategies have
further potential to reduce their BSFC penalty by optimisation, which
was outside the scope of this study. Therefore, cylinder deactivation and
intake modulation emerge as promising solutions for efcient exhaust
thermal management.
Poorer strategies are near the bottom right corner of Fig. 29, where
EGT increment is rather low, but fuel penalty is excessive. Cylinder cut-
out and exhaust modulation proved to be extremely inefcient in this
study, with fuel efciency deterioration of more than +15 % for an EGT
elevation of no more than +100 ◦C above baseline. Hence, they should
be avoided. Their poor fuel efciency is due mainly to high pumping
losses, indicating that the gas exchange process affected by VVA plays an
important role in fuel efciency. Suppressing their high fuel penalty
would entail comprising their EGT increment by avoiding extreme valve
modulation. This is applicable to all strategies. However, the latest
research has demonstrated signicant reduction of pumping loss in CCO
is attainable by partial breathing of non-red cylinders: this leads to
similar EGT-BSFC performance as CDA [63]. The present study applied
only one VVA strategy at a time, but others have demonstrated a com-
bination of more than two VVA strategies could minimise the BSFC
penalty and maximise EGT [29,33,64,65].
The most promising VVA strategies revealed by this study are to be
subjected to experimental verication in follow-up work. The simulation
activities continue towards model-based co-optimisation of different
strategies, and extending into advanced combustion concepts such as
low temperature combustion.
4. Conclusions
This study investigated all promising VVA strategies (intake modu-
lation, exhaust modulation, internal EGR, cylinder deactivation and
cylinder cut-out) to evaluate their potential for efcient exhaust thermal
management at low-load conditions. Notably, such a comprehensive
analysis has not been conducted for a state-of-the-art off-road engine
platform in any other published study to date. The combination of
cutting-edge model-based investigations and meticulously conducted
validation experiments provides signicantly greater insight than
afforded by standalone engine testing in prior works. The key ndings
improve the fundamental understanding of VVA’s mechanisms and their
interaction. The results support the assertion that it is feasible for the
next generation of engines for non-road mobile applications to meet the
substantially tightened emission limits anticipated by the EPA Tier 5
legislation. These ndings can be summarised as follows:
•All strategies supported by a fully-exible VVA system have a posi-
tive effect on increasing EGT, but fuel efciency and emissions are
mostly negatively affected (trade-off).
•The most efcient VVA strategies are CDA and intake modulation.
Both could elevate EGT by more than +150 ◦C, facilitating efcient
catalytic conversion, with minor negative or positive effect on fuel
consumption.
•CDA has the best EGT/efciency ratio. Under challenging conditions
of high engine-speeds and low loads, where achieving high conver-
sion rates is particularly difcult, one can simultaneously improve
the fuel economy by 2.5 % while achieving EGT increment of 166 ◦C.
Alternatively, adjusting waste-gate settings, gives as much as 245 ◦C
elevation in EGT with statistically same engine efciency.
Fig. 29. EGT-BSFC performance: (a) Mode1; (b) Mode2; and (c) Mode3.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
25
•The most inefcient strategies are exhaust modulation and cylinder
cut-out. A high fuel penalty of up to +16.5 % was observed but EGT
increment was below 100 ◦C and even can be lower than baseline.
•EGT increment is achieved mainly by three mechanisms: reduction of
in-cylinder trapped air mass (intake modulation and internal EGR);
increased heat release rate (CDA and CCO); and change of combus-
tion duration (exhaust modulation).
•BSFC penalty is attributed to inefcient gas exchange process (high
pumping work) and ineffective work transfer by shifted combustion
phase.
•VVA affects combustion characteristics such as ignition delay, pre-
mixed combustion and combustion phase by changing in-cylinder
mixture conditions. However, combustion can be corrected or opti-
mised by altering fuel injection timing or by other methods, such as
external EGR. This could minimise BSFC penalty.
•VVA generally produces more engine-out emissions such as soot, HC,
and CO with increasing EGT due to a rich mixture. However, the
higher EGT is expected to accelerate catalytic reaction and so reduce
tailpipe emissions. Moreover, the formation of engine-out emissions
can be suppressed by increasing air–fuel ratio and injection pressure.
•Intake modulation excels in this regard, demonstrating signicant
engine-out NO
x
reduction (>90 %) and minimal adverse effects on
carbon emissions (HC, CO, and soot), all while maintaining excellent
thermal and performance results.
CRediT authorship contribution statement
Jeyoung Kim: Writing – review & editing, Writing – original draft,
Visualization, Validation, Software, Methodology, Investigation, Formal
analysis, Data curation, Conceptualization. Marko Vallinmaki: Soft-
ware, Resources, Methodology. Tino Tuominen: Validation, Resources.
Maciej Mikulski: Writing – review & editing, Supervision, Funding
acquisition.
Declaration of competing interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
Data availability
The data that has been used is condential.
Acknowledgements
This research was conducted in the Clean Propulsion Technologies
project, funded by Business Finland, Finland (ref. 38485/31/2020). The
authors would like to thank Tino Tuominen, Mårten Westerholm, Sami
Nyyss¨
onen, Christer S¨
oderstr¨
om, Ari-Pekka Pellikka and Petri S¨
oderena
from VTT Technical Research Centre of Finland for performing engine
tests and sharing experimental data. Thanks go also to Marko Vallinmaki
and Pekka Nousiainen from AGCO Power for sharing the engine model
and technical support on model modication and engine simulation.
Lastly, thanks to David Wilcox for proofreading the manuscript.
References
[1] D.B. Gosala, A.K. Ramesh, C.M. Allen, M.C. Joshi, A.H. Taylor, M.V. Voorhis, G.
M. Shaver, L. Farrell, E. Koeberlein, J. McCarthy Jr., D. Stretch, Diesel engine
aftertreatment warm-up through early exhaust valve opening and internal exhaust
gas recirculation during idle operation, Int. J. Engine Res. 19 (7) (2018) 758–773,
https://doi.org/10.1177/1468087417730240.
[2] H. Fang, Challenges with the ultimate energy density with li-ion batteries, IOP
Conf. Ser.: Earth Environ. Sci. 781 (2021), https://doi.org/10.1088/1755-1315/
781/4/042023.
[3] D.B. Gosala, C.M. Allen, A.K. Ramesh, G.M. Shaver, J. McCarthy Jr., D. Stretch,
E. Koeberlein, L. Farrell, Cylinder deactivation during dynamic diesel engine
operation, Int. J. Engine Res. 18 (10) (2017) 991–1004, https://doi.org/10.1177/
1468087417694000.
[4] X. Li, H. Gao, L. Zhao, Z. Zhang, X. He, F. Liu, Combustion and emission
performance of a split injection diesel engine in a double swirl combustion system,
Energy 114 (Nov. 2016) 1135–1146, https://doi.org/10.1016/j.
energy.2016.08.092.
[5] B. Wang, Autoignition of light naphtha and its surrogates in a rapid compression
machine, Energy Sci. Eng. 7 (1) (Feb. 2019) 207–216, https://doi.org/10.1002/
ese3.270.
[6] B.G. Jeong, J.H. Won, K.C. Oh, H.S. Heo, S. Bae, H.J. Seo, H. Kim, Characteristics of
integrated air control and low-pressure exhaust gas recirculation valve for diesel
engines, Int. J. Automot. Technol. 21 (1) (2020) 239–247, https://doi.org/
10.1007/s12239-020-0023-x.
[7] Y. Liang, X. Ding, J. Dai, M. Zhao, L. Zhong, J. Wang, Y. Chen, Active oxygen-
promoted NO catalytic on monolithic Pt-based diesel oxidation catalyst modied
with Ce, Catal. Today 327 (2019), https://doi.org/10.1016/j.cattod.2018.06.008.
[8] C. Guardiola, B. Pla, J. Mora P´
erez, D. Lefebvre, Experimental determination and
modelling of the diesel oxidation catalysts ageing effects, Proceedings of the
Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering
233 (12) (Oct. 2019) 3016–3029, https://doi.org/10.1177/0954407018814305.
[9] B. Pla, P. Piqueras, P. Bares, A. Aronis, Simultaneous NOx and NH3 slip prediction
in a SCR catalyst under real driving conditions including potential urea injection
failures, Int. J. Engine Res. 23 (7) (Jul. 2022) 1213–1225, https://doi.org/
10.1177/14680874211007646.
[10] H. Zhang, J. Wang, Ammonia coverage ratio and input simultaneous estimation in
ground vehicle selective catalytic reduction (SCR) systems, J. Franklin Inst. 352 (2)
(Feb. 2015) 708–723, https://doi.org/10.1016/j.jfranklin.2014.06.009.
[11] A. Ko, Y. Woo, J. Jang, Y. Jung, Y. Pyo, H. Jo, O. Lim, Y.J. Lee, Complementary
effects between NO oxidation of DPF and NO2 decomposition of SCR in light-duty
diesel engine, J. Ind. Eng. Chem. 80 (Dec. 2019) 160–170, https://doi.org/
10.1016/j.jiec.2019.07.045.
[12] P. Jiao, Z. Li, B. Shen, W. Zhang, X. Kong, R. Jiang, Research of DPF regeneration
with NOx-PM coupled chemical reaction, Appl. Therm. Eng. 110 (Jan. 2017)
737–745, https://doi.org/10.1016/j.applthermaleng.2016.08.184.
[13] A. García, J. Monsalve-Serrano, D. Villalta, R. Lago Sari, Performance of a
conventional diesel aftertreatment system used in a medium-duty multi-cylinder
dual-mode dual-fuel engine, Energy Conv. Manag. 184 (Mar. 2019) 327–337,
https://doi.org/10.1016/j.enconman.2019.01.069.
[14] K. Lee, J. Lee, S. Lee, K. Oh, S. Jang, Fuel Consumption and Emission Reduction for
Non-Road Diesel Engines with Electrically Heated Catalysts, Catal. 13 (6) (2023),
https://doi.org/10.3390/catal13060950.
[15] Y.M. L´
opez - De Jesús, P.I. Chigada, T.C. Watling, K. Arulraj, A. Thor´
en,
N. Greenham, P. Markatou, NOx and PM Reduction from Diesel Exhaust Using
Vanadia SCRF®, SAE Int. J. Engines 9 (2) (2016) https://doi.org/10.4271/2016-
01-0914.
[16] Volkswagen, Clean and cultivated: the 2.0 TDI engine with new Euro 6d emission
standard (2020). https://www.volkswagen-newsroom.com/en/press-releases/cle
an-and-cultivated-the-20-tdi-engine-with-new-euro-6d-emission-standard-6721.
(Accessed 11 Sep 2023).
[17] BOSCH, Exhaust-gas treatment with double-injection (2019). https://www.bosch
-mobility.com/en/solutions/exhaust-gas-treatment/exhaust-gas-treatment-with
-double-injection-technology. (Accessed 11 Sep 2023).
[18] Y. Zhu, W. Zhou, C. Xia, Q. Hou, Application and Development of Selective
Catalytic Reduction Technology for Marine Low-Speed Diesel Engine: Trade-Off
among High Sulfur Fuel, High Thermal Efciency, and Low Pollution Emission,
Atmosphere 13 (5) (2022), https://doi.org/10.3390/atmos13050731.
[19] Y. Zhang, C. Xia, D. Liu, Y. Zhu, Y. Feng, Experimental investigation of the high-
pressure SCR reactor impact on a marine two-stroke diesel engine, Fuel 335 (2023)
127064, https://doi.org/10.1016/j.fuel.2022.127064.
[20] P. Tayal, Light off temperature based approach to determine diesel oxidation
catalyst effectiveness level and the corresponding outlet NO and NO2
characteristics, (2014). http://docs.lib.purdue.edu/open_access_theses/694.
[21] J. Hu, J. Liao, Y. Hu, J. Lei, M. Zhang, J. Zhong, F. Yan, Z. Cai, Experimental
investigation on emission characteristics of non-road diesel engine equipped with
integrated DOC +CDPF +SCR aftertreatment, Fuel 305 (Dec. 2021), https://doi.
org/10.1016/j.fuel.2021.121586.
[22] R. Feng, Z. Sun, G. Li, X. Hu, B. Deng, S. Xiong, The impact of thermal status on
emissions of a non-road diesel engine equipped with aftertreatment system under
transient operation, Case Studies in Thermal Engineering 36 (Aug. 2022), https://
doi.org/10.1016/j.csite.2022.102244.
[23] Z. Meng, C. Chen, J. Li, J. Fang, J. Tan, Y. Qin, Y. Jiang, Z. Qin, W. Bai, K. Liang,
Particle emission characteristics of DPF regeneration from DPF regeneration bench
and diesel engine bench measurements, Fuel 262 (Feb. 2020), https://doi.org/
10.1016/j.fuel.2019.116589.
[24] C. Guardiola, B. Pla, P. Bares, J. Mora, An on-board method to estimate the light-off
temperature of diesel oxidation catalysts, Int. J. Engine Res. 21 (8) (Oct. 2020)
1480–1492, https://doi.org/10.1177/1468087418817965.
[25] V. Tharad, DIESEL OXIDATION CATALYST - DIESEL PARTICULATE FILTER /
CATALYTIC DIESEL FILTERS (2021). https://www.linkedin.com/pulse/diesel-oxi
dation-catalyst-particulate-lter-catalytic-vijay-tharad. (Accessed 20 Feb 2024).
[26] R. Villamaina, I. Nova, E. Tronconi, T. Maunula, M. Keenan, Effect of the NH4NO3
addition on the low-T NH3-SCR performances of individual and combined fe- and
cu-zeolite catalysts, Emiss. Control Sci. Technol. 5 (4) (2019) 290–296, https://doi.
org/10.1007/s40825-019-00140-3.
J. Kim et al.
Applied Thermal Engineering 246 (2024) 122940
26
[27] S. Ma, W. Gao, Z. Yang, R. Lin, X. Wang, X. Zhu, Y. Jiang, Superior Ce–Nb–Ti oxide
catalysts for selective catalytic reduction of NO with NH3, J. Energy Inst. 94 (Feb.
2021) 73–84, https://doi.org/10.1016/j.joei.2020.11.001.
[28] D.W. Stanton, Systematic development of highly efcient and clean engines to
meet future Commercial vehicle greenhouse gas regulations, SAE Int. J. Engines 6
(3) (Sep. 2013) 1395–1480, https://doi.org/10.4271/2013-01-2421.
[29] K.R. Vos, G.M. Shaver, M.C. Joshi, A.K. Ramesh, J. McCarthy Jr., Strategies for
using valvetrain exibility instead of exhaust manifold pressure modulation for
diesel engine gas exchange and thermal management control, Int. J. Engine Res. 22
(3) (2021) 755–776, https://doi.org/10.1177/1468087419880634.
[30] S. Bai, J. Han, M. Liu, S. Qin, G. Wang, G. Li, Experimental investigation of exhaust
thermal management on NOx emissions of heavy-duty diesel engine under the
world Harmonized transient cycle (WHTC), Appl. Therm. Eng. 142 (July. 2018)
421–432, https://doi.org/10.1016/j.applthermaleng.2018.07.042.
[31] K. Boriboonsomsin, T. Durbin, G. Scora, K. Johnson, D. Sandez, A. Vu, Y. Jiang,
A. Burnette, S. Yoon, J. Collins, Z. Dai, C. Fulper, S. Kishan, M. Sabisch, D. Jackson,
Real-world exhaust temperature proles of on-road heavy-duty diesel vehicles
equipped with selective catalytic reduction, Sci. Total Environ. 634 (2018)
909–921, https://doi.org/10.1016/j.scitotenv.2018.03.362.
[32] M. Laur´
en, T. Karhu, M. Laivola, J. Ekman, S. Niemi, K. Spoof-Tuomi, Different
methods to improve the exhaust gas temperature in modern stage V 0ff-road diesel
engine over transient emission cycles, SAE Technical Papers (2020), https://doi.
org/10.4271/2020-01-0903.
[33] M.E. Magee, Exhaust Thermal Management Using Cylinder Deactivation and Late
Intake Valve Closing, (2014). https://docs.lib.purdue.edu/open_access_theses
/214/.
[34] B. Wu, Z. Jia, Z. guo Li, G. yi Liu, X. lin Zhong, Different exhaust temperature
management technologies for heavy-duty diesel engines with regard to thermal
efciency, Appl. Therm. Eng. 186 (2021) 116495, https://doi.org/10.1016/j.
applthermaleng.2020.116495.
[35] T. Tomoda, T. Ogawa, H. Ohki, T. Kogo, K. Nakatani, E. Hashimoto, Improvement
of diesel engine performance by variable valve train system, Int. J. Engine Res. 11
(5) (2010) 331–344, https://doi.org/10.1243/14680874JER586.
[36] F.J. Arnau, J. Martín, B. Pla, ´
A. Au˜
n´
on, Diesel engine optimization and exhaust
thermal management by means of variable valve train strategies, Int. J. Engine Res.
22 (4) (2021) 1196–1213, https://doi.org/10.1177/1468087419894804.
[37] Z. Lou, G. Zhu, Review of advancement in Variable valve actuation of internal
combustion engines, Appl. Sci. 10 (4) (2020), https://doi.org/10.3390/
app10041216.
[38] A. di Gaeta, C.I. Hoyos Velasco, V. Giglio, Modelling of an electro-hydraulic
variable valve actuator for camless engines aimed at controlling valve lift
parameters, Int. J. Control 94 (10) (2021) 2857–2873, https://doi.org/10.1080/
00207179.2020.1737333.
[39] J. Kim, A. Soleimani, P. Nousiainen, M. Axelsson, M. Mikulski, Variable valve
actuation for next generation marine and off-road engines - a comprehensive
review oriented on meeting future emission legislation. https://ssrn.com/abstract
=4760136.
[40] S. Gehrke, D. Kov´
acs, P. Eilts, A. Rempel, P. Eckert, Investigation of VVA-based
exhaust Management strategies by means of a HD single cylinder Research engine
and rapid prototyping systems, SAE Int. J. Commer. Veh. 6 (1) (Apr. 2013) 47–61,
https://doi.org/10.4271/2013-01-0587.
[41] W. Guan, H. Zhao, Z. Ban, T. Lin, Exploring alternative combustion control
strategies for low-load exhaust gas temperature management of a heavy-duty diesel
engine, Int. J. Engine Res. 20 (4) (2019) 381–392, https://doi.org/10.1177/
1468087418755586.
[42] W. De Ojeda, Effect of Variable valve timing on diesel combustion Characteristics,
SAE Technical Papers (Apr. 2010), https://doi.org/10.4271/2010-01-1124.
[43] X. Zhang, H. Wang, Z. Zheng, R.D. Reitz, M. Yao, Effects of late intake valve closing
(LIVC) and rebreathing valve strategies on diesel engine performance and
emissions at low loads, Appl. Therm. Eng. 98 (2016) 310–319, https://doi.org/
10.1016/j.applthermaleng.2015.12.045.
[44] A. Wickstr¨
om, Variable Valve Actuation Strategies for Exhaust Thermal
Management on a HD Diesel Engine (2012). https://www.diva-portal.org/smash
/record.jsf?pid=diva2%3A541747&dswid=-9603.
[45] M.C. Joshi, G.M. Shaver, K. Vos, J. McCarthy Jr., L. Farrell, Internal exhaust gas
recirculation via reinduction and negative valve overlap for fuel-efcient
aftertreatment thermal management at curb idle in a diesel engine, Int. J. Engine
Res. 23 (3) (2022) 369–379, https://doi.org/10.1177/1468087420984590.
[46] CARB, Tier 5 Rulemaking Workshop II - Proposed Emission Standards, 2023. https
://ww2.arb.ca.gov/sites/default/les/2023-10/%232%20Proposed%20Emission
%20Standards-ADA-10232023.pdf. (Accessed 25 Feb 2024).
[47] G.F. Hohenberg, Advanced approaches for heat transfer calculations, SAE
Technical Papers (Feb. 1979), https://doi.org/10.4271/790825.
[48] A.P. Colburn, A method of correlating forced convection heat-transfer data and a
comparison with uid friction, Int. J. Heat Mass Transf. 7 (12) (Dec. 1964)
1359–1384, https://doi.org/10.1016/0017-9310(64)90125-5.
[49] H. Hiroyasu, T. Kadota, M. Arai, Development and use of a spray combustion
modeling to predict diesel engine efciency and pollutant emissions : Part 1
combustion modeling, Bulletin of JSME 26 (214) (1983) 569–575, https://doi.org/
10.1299/JSME1958.26.569.
[50] G. Technology, “GT-SUITE Engine Performance Application Manual ver 7.1,” 2022.
[51] Gamma Technologies, “Engine Performance Application Manual (Version 2022),”
2022.
[52] P. Wang, Z. Hu, L. Shi, X. Tang, Y. Liu, K. Deng, Experimental investigation of the
effects of miller timing on performance, energy and exergy characteristics of two-
stage turbocharged marine diesel engine, Fuel 292 (Jan. 2021), https://doi.org/
10.1016/j.fuel.2021.120252.
[53] D.M. Thomas, Experimental setup and testing of a Variable valve actuation enabled
cam-less natural gas engine. https://doi.org/10.25394/PGS.21685760.v1.
[55] E. Garcia, V. Triantopoulos, A. Boehman, M. Taylor, J. Li, Impact of Miller Cycle
Strategies on Combustion Characteristics, Emissions and Efciency in Heavy-Duty
Diesel Engines, SAE Technical Papers, (April. 2020), https://doi.org/10.4271/202
0-01-1127.
[56] D. Agarwal, S. Kumar, A. Kumar, Effect of exhaust gas recirculation (EGR) on
performance, emissions, deposits and durability of a constant speed compression
ignition engine, Appl. Energy 88 (8) (2011) 2900–2907, https://doi.org/10.1016/
j.apenergy.2011.01.066.
[57] Y. Liu, A. Kuznetsov, B. Sa, Simulation and analysis of the impact of cylinder
deactivation on fuel saving and emissions of a medium-speed high-power diesel
engine, Appl. Sci. 11 (16) (2021), https://doi.org/10.3390/app11167603.
[58] K.R. Vos, G.M. Shaver, A.K. Ramesh, J. McCarthy Jr., Impact of Cylinder
Deactivation and Cylinder Cutout via Flexible Valve Actuation on Fuel Efcient
Aftertreatment Thermal Management at Curb Idle, Front. Mech. Eng. 5 (Aug.
2019), https://doi.org/10.3389/fmech.2019.00052.
[59] T. Reinhart, A. Matheaus, C. Sharp, B. Peters, M. Pieczko, J. McCarthy Jr.,
Vibration and emissions quantication over key drive cycles using cylinder
deactivation, Int. J. Powertrains. 9 (4) (2020) 315–344, https://doi.org/10.1504/
IJPT.2020.111245.
[60] M. Pieczko, J. McCarthy, J. Hamler, Mitigating vibration for a heavy-duty diesel
cylinder deactivation truck, SAE Technical Papers (2021) 1–8, https://doi.org/
10.4271/2021-01-0661.
[61] D.B. Gosala, C.M. Allen, G.M. Shaver, L. Farrell, E. Koeberlein, B. Franke,
D. Stretch, J. McCarthy Jr., Dynamic cylinder activation in diesel engines, Int. J.
Engine Res. 20 (8–9) (2019) 849–861, https://doi.org/10.1177/
1468087418779937.
[62] C.M. Allen, M.C. Joshi, D.B. Gosala, G.M. Shaver, L. Farrell, J. McCarthy Jr.,
Experimental assessment of diesel engine cylinder deactivation performance
during low-load transient operations, Int. J. Engine Res. 22 (2) (2021) 606–615,
https://doi.org/10.1177/1468087419857597.
[63] D.B. Gosala, G.M. Shaver, J. McCarthy Jr., T.P. Lutz, Fuel-efcient thermal
management in diesel engines via valvetrain-enabled cylinder ventilation
strategies, Int. J. Engine Res. 22 (2) (2021) 430–442, https://doi.org/10.1177/
1468087419867247.
[64] K.R. Vos, G.M. Shaver, M.C. Joshi, J. McCarthy Jr., Implementing variable valve
actuation on a diesel engine at high-speed idle operation for improved
aftertreatment warm-up, Int. J. Engine Res. 21 (7) (2020) 1134–1146. https://doi.
org/10.1177/1468087419880639.
[65] M.C. Joshi, Opportunities to improve aftertreatment thermal management and
simplify the air handling architectures of highly efcient diesel engines
incorporating valvetrain exibility. https://doi.org/10.25394/PGS.11527854.v1.
J. Kim et al.
