Experimental and computational assessment of inlet swirl effects on a gasoline compression ignition (GCI) light duty diesel engine

Interestingly, early heat release was shown to be insensitive to changes in swirl ratio Rs during early mixing controlled combustion; however, later during combustion, the heat release r

The light-medium load operating regime (4-8 bar net IMEP)

presents many challenges for advanced low temperature

combustion strategies (e.g HCCI, PPC) in light-duty, high

speed engines In this operating regime, lean global

equivalence ratios (Φ<0.4) present challenges with respect to

autoignition of gasoline-like fuels Considering this intake

temperature sensitivity, the objective of this work was to

investigate, both experimentally and computationally, gasoline

compression ignition (GCI) combustion operating sensitivity to

inlet swirl ratio (Rs) variations when using a single fuel

(87-octane gasoline) in a 0.475-liter single-cylinder engine

based on a production GM 1.9-liter high speed diesel engine

For the first part of this investigation, an experimental matrix

was developed to determine how changing inlet swirl affected

GCI operation at various fixed load and engine speed

operating conditions (4 and 8 bar net IMEP; 1300 and 2000

RPM) Here, experimental results showed significant changes

in CA50 due to changes in inlet swirl ratio For example, at the

4 bar net IMEP operating condition at 1300 RPM, a reduction

in swirl ratio (from 2.2 to 1.5) caused a 6 CAD advancement of

CA50, while increasing swirl ratio (from 2.2 to 3.5) resulted in a

2 CAD retard of CA50 This advancement in CA50 at the 1.5

swirl ratio operating point was accompanied with significant

increases in NOx emissions (from 0.2 to 1.6 g/kg-fuel) Minor

adjustments in injection strategy could be made to maintain

NOx emissions less than 1 g/kg-fuel

In subsequent experiments at 4 bar net IMEP, first equivalence

ratio, then CA50 were matched in an effort to further isolate the

effects of changing swirl ratio In these later cases conditions

allowed for a 25C reduction in the required inlet temperature at

the lower swirl condition (from 77C to 52C when reducing swirl

from 2.2 to 1.5) Experimental measurements were numerically

simulated to help analyze the combustion behavior and

emissions characteristics using a 3D-CFD code coupled with

detailed chemistry This numerical investigation quantified the

thermal and mixing effects of swirl ratio variation on mixture conditions before ignition and subsequent influence on ignition timing, in-cylinder pressure profile, and emissions

Introduction

Global energy consumption forecasts continue to predict increasing demand for liquid hydrocarbon fuels for the foreseeable future For example, the EIA projects liquid fuel consumption in the transportation sector to increase 46% by

2035, relative to 2008 levels [1] As a result, concerns related

to excessive urban air pollution as well as consumption of finite petroleum resources has prompted governmental agencies to develop increasingly stringent vehicle emissions and fuel consumption standards As a result, the two primary IC engine combustion strategies, gasoline spark-ignition and diesel compression ignition, have necessarily evolved For example, and in general, engine manufacturers have attempted to improve the fuel consumption of spark-ignited engines by a combination of reduced displacement, intake turbocharging and/or implementation of direct injection fuel systems In contrast, with inherent advantages related to thermal efficiency (but challenges related to PM and NOx emissions), diesel engine development has focused on increased fuel injection pressures (to enhance air-fuel mixing prior to ignition) and the use of exhaust gas recirculation (EGR) to reduce peak combustion temperatures (and higher NOx formation rates) and the implementation of exhaust gas aftertreatment

In parallel, advanced combustion research over the last 30 years has focused on the development of low temperature combustion (LTC) strategies The primary objective in an LTC strategy is to develop an air-fuel charge prior to autoignition devoid of locally rich mixture concentrations that can lead to either excessive PM (due to insufficient mixing) or NOx emissions (where locally near-stoichiometric mixture concentrations can result in high peak combustion temperatures) By avoiding excessive PM and NOx formation

Experimental and Computational Assessment of Inlet

Swirl Effects on a Gasoline Compression Ignition (GCI)

Light-Duty Diesel Engine

2014-01-1299 Published 04/01/2014

Paul Loeper, Youngchul Ra, David Foster, and Jaal Ghandhi

Univ of Wisconsin

CITATION: Loeper, P., Ra, Y., Foster, D., and Ghandhi, J., "Experimental and Computational Assessment of Inlet Swirl

Effects on a Gasoline Compression Ignition (GCI) Light-Duty Diesel Engine," SAE Technical Paper 2014-01-1299, 2014,

doi:10.4271/2014-01-1299

Copyright © 2014 SAE International

regions, simultaneous reductions in both can be achieved

HCCI served as an early example of LTC, and as Najt, Foster,

Onishi et al [2, 3] demonstrated, short (nearly-volumetric)

combustion durations of a highly premixed charge lead to

thermal efficiencies exceeding 40% (in addition to significant

PM and NOx emission reductions)

However, the combination of kinetically controlled combustion

phasing and short combustion durations presented challenges

with respect to controllability and engine load limitations

Christensen and Johansson [4, 5] demonstrated the ability to

use EGR and/or intake turbocharging to expand high load

operation In addition, as demonstrated by Noda, Sjoberg,

Aroonsrisopon et al [6, 7, 8, 9], HCCI combustion phasing and

duration was shown to be sensitive to temperature and mixture

concentration gradients (i.e stratification) As a result, varying

levels of mixture and/or temperature stratification were

observed to affect combustion durations leading to

more-acceptable pressure rise rates at a given speed/load operating

condition The ability to use mixture and temperature

stratification to enable better control over combustion phasing

and duration has since led to the development of multiple

variations of HCCI (e.g PPC, PPCI, GCI, RCCI, etc.) Varying

the inlet air swirl ratio is one method that can be used to vary

in-cylinder temperature stratification

Traditionally, increasing inlet air swirl has been used as a

strategy in high speed diesel engines to enhance air-fuel

mixing prior to autoignition And when using a highly reactive

fuel such as diesel, enhanced mixing (or increased

homogeneity) has been observed to reduce specific fuel

consumption due to shorter combustion durations and more

optimal combustion phasing; as well as reductions in PM, CO,

and UHC emissions In an optical diesel engine, Miles [10]

confirmed combustion processes were significantly affected by

increasing Rs from 1.5 to 3.5 with reductions in ignition delay,

peak heat release rates, and pressure rise rates Interestingly,

early heat release was shown to be insensitive to changes in

swirl ratio (Rs) during early mixing controlled combustion;

however, later during combustion, the heat release rate then

increased with increasing Rs Here, combustion luminosity

imaging showed increased activity in the squish and bowl

regions of the combustion chamber during these periods

Whereas swirl effects have been extensively studied in high

speed diesel applications as discussed previously, a review of

the literature reveals less cohesion with respect to swirl effects

in LTC applications using low reactivity fuels (e.g gasoline)

For example, in a model developed by Aceves et al [11]

simulating propane HCCI, reduced inlet swirl (from 4.3 to 0.43)

was examined as a strategy to reduce UHC and CO emissions

Due to the enhancement of heat transfer resulting from high

inlet swirl (and corresponding increase in boundary layer

thickness), the authors reasoned increasing inlet swirl results

in a overall cooler in-cylinder charge, thus slowing (or

potentially quenching) CO and UHC oxidation kinetics At the

lower swirl ratio, their model demonstrated negligible effects

with respect to temperature distribution and CO/UHC emission

levels Experimental results from Christensen and Johansson

[12] (using both PRF50 and pure iso-octane fuels) compared

combustion phasing and emissions performance between a flat-top and square bowl piston at two different Rs (2.0 vs 2.8)

in a 1.6-liter engine at 1200 RPM In results obtained using

pure iso-octane at =0.4, combustion phasing advanced

(between 1 and 2 CAD) for a fixed inlet temperature when swirl was increased to 2.8 Additionally, the high swirl case resulted

in improvements in combustion and thermal efficiencies, as well as reductions in HC emissions

Following, Sjoberg et al [9] investigated using varying inlet swirl ratios to shape HCCI heat release rates in a 0.98-liter single cylinder engine (based on a Cummins B-series diesel 6-cylinder engine) at 1200 RPM The authors hypothesized combustion durations could be extended by increasing inlet air swirl This would result in increased heat transfer, which would then create a larger temperature distribution (or thicker temperature boundary layer) At a fixed CA50 (7.2 ATDC) and equivalence ratio (Φ=0.381), increasing Rs from 0.9 to 3.6 increased CA10-90 from 4.64 CAD to 5.94 CAD (a 28% increase) In these cases, fixed CA50 was achieved by increasing IVC temperature in the higher swirl case (in excess

of 15C higher than the low swirl case) As a result of higher IVC temperatures (i.e less dense), gross IMEP was reduced in the higher swirl case Therefore, when fueling was increased to maintain fixed gross IMEP (4.43 bar), the combustion duration only increased 11.6% in the higher swirl scenario (5.18 vs 4.64 CAD) and was accompanied by a 3.7% increases in ISFC In contrast, by maintaining a fixed Rs of 0.9, the authors observed similar increases in CA10-90 by reducing coolant temperatures from 100C to 50C, but with fewer penalties in fuel consumption (0.5%) As a result, the authors questioned the merits of using swirl enhancement to extend combustion duration i.e

increased heat transfer (and reduced efficiency) rates negated most benefits

Objectives

Considering the previously discussed results and conclusions related to inlet swirl effects in LTC strategies, the objectives of the work to be presented can be divided into three parts:

1 Experimental assessment of inlet swirl effects in a GCI strategy over a larger load and speed range, than previously investigated Results concerning attempts to isolate swirl effects by first maintaining fixed equivalence ratio, and then fixed CA50 as well, will be presented

2 Utilize CFD simulation to provide insight into how the in-cylinder physics change due to variations in inlet swirl ratio (e.g air-fuel mixing vs temperature effects)

3 Given the results of experiments and CFD simulation, discuss the merits of using variable inlet swirl as a control strategy in GCI operation

Experiment

Engine Setup

The engine used in this study is based on a GM 1.9-liter EuroIV light-duty 4-cylinder diesel engine The production cylinder head is mounted to a single cylinder Labeco CLR

crankcase A re-entrant piston bowl (16.6:1 compression ratio)

was developed specifically for diesel LTC experiments, and

remains in place for these GCI experiments Turbocharging

conditions can be simulated through PID control of inlet and

exhaust charge tank pressures Cooled exhaust gas

recirculation (EGR) can be driven by maintaining a differential

pressure between the surge tanks (shown in Figure 1)

Additional engine specifications are listed in Table 1

Figure 1 Engine test cell experimental setup at the University of

Wisconsin-Madison / GM Collaborative Research Laboratory

Table 1 Single-cylinder engine specifications

Similar to the cylinder head and piston, fueling system

hardware closely resembles the production engine

configuration as well An off-engine fuel cart utilizing a Bosch

CP3.3 pump delivers fuel to a Bosch common rail and

CRIP2-MI injector Rail pressure (maintained by inlet metering and

high pressure bypass valve), injection timing, and duration

parameters are managed through flash commands to a

rewritable ECU via ETAS INCA calibration software The

experimental injector tip used in this study has 7-holes and a

155° included cone angle Fuel injection system specifications

are listed in Table 2

Table 2 Bosch fuel injection system specifications

A Kistler 6125B piezoelectric transducer is used for high resolution cylinder pressure measurements A BEI encoder provides 10 pressure measurements per crank angle degree; high speed data are then averaged over 200 cycles

A primary objective of this research is to use readily available 87-octane gasoline without addition of ethanol Fuel properties are shown in Table 3

Table 3 Specifications of gasoline used in experiments

Horiba emission analyzers were used to monitor both exhaust and intake gas compositions Five analyzers, including CO2 (Horiba model FMA-220), O2 (AIA-220), CO, NOx (CLA-220), and HC (FIA-236-1), monitored exhaust gas concentrations while 2 additional analyzers (CO2 and O2) monitored intake gas composition Heated sample lines maintained gas samples at 191°C prior to entering the emissions bench where after the samples were cooled to condense water (except for the HC sample) Particulate measurements were monitored using an AVL Smoke Meter (415S); further discussion of PM

measurements will be omitted since PM measurements never exceeded 0.01 g/kg-fuel injected

More specific to the work presented here, inlet swirl ratio adjustments were made through the adjustment of butterfly valves housed within a swirl plate (located between the intake runner and engine cylinder head), as shown in Figure 2

Figure 2 The swirl plate situated between the intake runner and cylinder head features both helical and tangential intake ports Intake swirl can be adjusted by opening and/or closing butterfly valves (19 positions) located in the ports

As seen in Figure 2, butterfly valve adjustments in the helical

and tangential ports can provide a range of inlet swirl ratios, as

shown in Figure 3

Figure 3 Inlet swirl ratios as a function of butterfly valve position in

either helical or tangential ports Note Rs=2.2 represents the baseline

inlet swirl condition in which both helical and tangential valves are fully

open

The inlet swirl ratio has been constant, at Rs=2.2, for the

majority of GCI work within the UW ERC-GM CRL Therefore,

for each of the load and speed operating conditions

investigated here, Rs=2.2 serves as a baseline operating point

Experimental Test Conditions

In order to assess the effects of variable swirl ratio in a GCI

operating strategy, three load/speed conditions were selected

encompassing the light-medium load operating regime (see

Figure 4) at inlet swirl ratio values of 1.5, 2.2, and 3.5

At each load-speed operating condition, three experimental

cases were selected in an effort to isolate inlet swirl effects, as

shown in Table 4

Figure 4 Experimental test matrix consisting of swirl investigations at three operating conditions

Table 4 For each operating condition, three separate cases were developed to try and isolate inlet swirl effects in GCI operation

In case 1, both inlet temperature and IMAP were held constant, which demonstrated engine response due to varying levels of intake throttling Due to these inherent throttling effects, in case

2 the IMAP was adjusted to maintain fixed equivalent ratios for all three swirl ratios investigated Lastly, the inlet temperature was adjusted in case 3 to maintain fixed combustion phasing (CA50) and allow for further isolation of inlet swirl effects in GCI In all three cases, fuel flow was adjusted as necessary to maintain fixed net IMEP (either 4 bar or 8 bar)

Numerical Approach

Physical Models

For simulating the spray processes and the subsequent mixing and combustion of fuel/air mixtures in the combustion chamber, various physical sub-models were employed in the present KIVA-ERC-CHEMKIN code The code is based on KIVA3V Release 2 [13] coupled with the CHEMKIN II library [14] The added sub-models include models related to drop breakup, collision and coalescence, drop deformation, drop evaporation, wall impingement and vaporization, etc

A hybrid primary spray break-up model that is computationally efficient as well as comprehensive enough to account for the effects of aerodynamics, liquid properties and nozzle flows was employed [15] In this model, the injected fuel “blobs” are

tracked by a Lagrangian method while the break-up of each

blob is calculated from considerations of jet stability using

Kelvin-Helmholtz (KH) instability theory For secondary and

further break-up processes, a Kelvin Helmholtz (KH) - Rayleigh

Taylor (RT) hybrid model was used In the present study, the

model constants were used as suggested by Beale and Reitz

[15] since, due to high volatility of gasoline, the fuel distribution

in the cylinder is not as sensitive to spray model constants as it

is in diesel-fueled spray cases

A droplet collision model based on the stochastic particle

method [13], in which the collision frequency is used to

calculate the probability that a drop in one parcel will undergo a

collision with a drop in another parcel, was used assuming all

drops in each parcel behave in the same manner The

probability of coalescence is determined by considering the

Weber number that includes the effects of density and surface

tension of the liquid droplets

Droplet deformation in terms of its distortion from sphericity is

modeled using a forced, damped harmonic oscillator model,

where the surface tension and viscosity of the droplet are the

major properties used in the restoring force and damping

terms, respectively [16] Distortions of the droplets affect the

momentum exchange between the droplets and the ambient

gas; and subsequently, drop velocities (or relative velocity

between the drop and the gas) that are governing parameters

in the breakup and evaporation processes as well

The droplet vaporization model [17] considers the evaporation

of spray droplets using the Discrete Multi-Component (DMC)

approach under temperatures ranging from flash-boiling

conditions to normal evaporation The improved model

accounts for variable internal droplet temperatures, and

considers an unsteady internal heat flux with internal

circulation, and a model for the determination of the droplet

surface temperature The model uses an effective heat transfer

coefficient model for the heat flux from the surrounding gas to

the droplet surface Also, the variable density of the gasoline

surrogate fuel as a function of temperature is considered in the

governing equations and the relevant sub-models The

effective heat transfer coefficient calculated in the model is also

used to determine the amount of fuel to be treated as vapor

when the drop surface temperature reaches the critical

temperature while the drop interior is still in the sub-critical

condition The model has been well tested for evaporation of

both sprays and single drops at various pressure and

temperature conditions, including flash-boiling

Effects associated with spray/wall interactions, including

droplet splash, film spreading due to impingement forces, and

motion due to film inertia were considered in a wall film

sub-model, in addition to calculations of film transport on

complex surfaces with heating and vaporization of the film, and

separation and re-entrainment of films at sharp corners [18]

For the turbulence calculation, the RNG k-ε model [19] was

used

In the two-phase transport equations, droplets are treated as point sources and the wall film fuel flow is not resolved on the computational grid Therefore, it is assumed that in a

computational cell where droplets or wall film parcels exist, the liquid vaporizes under the prevailing mixture conditions and the vapor mixes completely with the gaseous mixture within the cell Thus, stratification of gaseous species within a single cell

is not resolved

The physical models employed in the present study have been extensively validated for diesel spray injections Although the models were not extensively validated with gasoline sprays in the present study The performance of the models in gasoline application was tested by the authors in the previous study [20] Lastly, with respect to simulating inlet swirl, the initial velocity profiles were developed using the Bessel function such that the overall swirl ratio matched experimental flow-bench data The authors believe these resultant velocity profiles fairly represent swirl motion in the experimental engine setup

Combustion Models

The ignition/combustion characteristics of automotive fuels are often represented using blends of two hydrocarbons, typically the two primary reference fuels, i.e., iso-octane (iC8H18) and n-heptane (C7H16) It is widely accepted that the oxidation processes of n-heptane and iso-octane well represent the ignition and combustion characteristics of diesel and gasoline fuels, respectively In the present study, a skeletal reaction mechanism for primary reference fuel oxidation with 49 species and 149 reactions [20, 21] was used to calculate the detailed chemical kinetics of combustion The mechanism has been well validated using data from HCCI [22] and direct injection engine experiments [23], as well as with the ignition delay time data obtained in shock tube tests [24] for various temperatures, pressures and fuel compositions In the present study, gasoline was modeled as PRF 87, i.e., 87 % iso-octane and 13 % n-heptane for physical properties and chemistry calculation Assuming a well-stirred reactor in each cell, changes of species concentration were obtained from the chemical reaction calculations, which are directly integrated with the transport calculations in the CFD code For the calculation of

NOx formation, a 4 species (N, NO, N2O and NO2) and 12 reaction NOx mechanism was used that has been reduced from the GRI NOx mechanism (available online) and added to the PRF reaction mechanism A phenomenological soot model, modified from the Hiroyasu soot model [25] was employed to predict soot emissions For oxidation of soot, the Nagle-Strickland-Constable (NSC) model was employed in the soot model

Computational Conditions

The gasoline listed in Table 3 was considered for the computations The operating conditions in Table 5 with injection pressure of 500 bar were used with injection timing as a

parametric variable A small-bore light-duty diesel engine with

the injection system described above (see Tables 1 and 2) was

used for the simulations Double injections through a 7-hole

injector with an included spray angle of 155° were modeled In

the computations, the first injection pulse was assumed to be

made such that the injected fuel vaporizes completely and

mixes uniformly with the air before IVC timing This assumption

corresponds to operation of the engine where the first pulse

was injected during valve overlap leaving the remainder of the

intake stroke to vaporize and mix before intake valve closure

The injection timings of the second pulses were fixed at −31°

ATDC except for the case of Rs 1.5 in Case 3, the timing of

which was −35° ATDC The initial and boundary conditions for

a baseline engine operation were first obtained from the

measured data of engine operation Then, the initial and

boundary conditions were adjusted to match measured

pressure profiles of motoring operation

The simulated cylinder gas pressures at intake valve closure

(IVC) were obtained from the experimental data and the gas

temperatures at IVC were estimated considering the

corresponding mass and pressure in the cylinder It is notable

that, due to heat transfer from the cylinder walls, mixing of

fresh air with hot internal residual gases, and a slight

compression of the gas mixture during the period between

BDC and IVC, the gas temperatures at IVC are normally higher

than the intake port gas temperatures No EGR was

considered, but internal residual exhaust gases were taken into

account to estimate the initial composition at IVC The injection

pressure was 500 bar, which was significantly lowered from

those of high load operations using the same engine [20]

Detailed computed operating conditions are listed in Table 5

Table 5 Simulated operating conditions Numbers in parenthesis are

for the case with swirl ratio of 1.5 in Case-3

For double injection cases considered, the proportion of fuel

injected during the first pulse (hereafter, denoted as 1st split

ratio, S1), which was assumed to be completely vaporized and

uniformly mixed with air before IVC, was fixed at 56%, except

for the 1.5 swirl ratio case of Case-3 (67%)

In Figure 5, the injection rate shape profile of a 2nd pulse for the

baseline double injection is shown The start of injection

commands (SOIC) of the 1st pulse (not shown in the figure) is

−350° ATDC, and the 2nd pulse timing is −31° ATDC In the

figure, the rates are normalized by the maximum value during the injection When the injection amount for each pulse was changed as the engine operating conditions varied, the duration of each pulse was changed accordingly based on the measured injection rate profiles The pressure wave

interactions between the 1st and 2nd pulses can be neglected since the pulse dwell is long enough for such wave interactions

to damp out

Figure 5 Injection rate shape of a 2nd pulse for the baseline double injection case (swirl ratio=2.2 of Case-1) used in simulations Injection pressure was 500 bar and pulse duration was 562 µs at engine speed

of 1300 rev/min Rates are normalized by the maximum value during the injection

Three dimensional computational grids with the crevice volume resolved as an elongated top land region were employed To save computation load, a 1/7th sector of the full 360° mesh with periodic boundaries (corresponding to one plume from the seven-hole injector nozzle) was used The average cell dimensions were 1.2 to 1.8 mm and 0.6 to 4.1 mm in the radial and vertical directions, respectively, with twenty cells

azimuthally (see Figure 6) To resolve the crevice region, i.e., the gap between the piston and cylinder wall above the top

ring, two radial cells were used with three vertical cells This

grid resolution was found to be sufficient to ensure grid insensitivity of the spray sub-models and the combustion model during this study

Figure 6 Vertical cross-section view of the computational grid with crevice volume resolved as a thin annulus Azimuthal angle of the sector span was 51.4 degrees

Experiments - Case 1

During the first set of experiments (case 1), inlet swirl effects

were assessed at two engine speeds, 1300 and 2000 RPM

This engine speed range is representative of low-to

medium-load operation in a light duty vehicle Additionally, if inlet swirl is

to be used as part of a comprehensive GCI control strategy,

case 1 parameters capture engine response absent of any

changes in intake boost pressure or temperature As will be

presented, the inlet swirl range investigated resulted in

substantial changes in combustion phasing (CA50) As a result,

these variances necessitated the selection of intake operating

parameters (IMAP and Tin) such that excessive combustion

instability (>3% COV of IMEP) or pressure rise rates (>10 bar/

deg) were avoided Table 6 shows baseline engine operating

parameters for case 1 experiments (fixed IMAP, Tin, and IMEP)

Table 6 In case 1 experiments, inlet temperature and IMAP remained

fixed at 4 bar net IMEP Depending on engine speed, inlet temperature

and IMAP differed in order to capture the full range of combustion

phasing for the inlet swirl ratios investigated

Note at 2000 RPM, due to reductions in both mixing time and

the progression of autoignition chemistry, higher intake

temperature and pressures were required to ensure gasoline

autoignition For both engine speeds investigated in case 1

experiments, the fuel injection strategy remained fixed, and

consisted of two injections at an injection pressure of 500 bar

Additionally, 70% of fuel mass was injected early during the

intake stroke (−350° ATDC; also referred to as % premix) with

the remaining 30% of fuel mass injected at −31° ATDC [26, 27]

As shown in Figure 7, at 1300 RPM, the effects on combustion

phasing (and pressure rise rates) due to varying inlet swirl

levels are significant, and indicates in-cylinder temperature

distribution as the dominant process For example, reducing

the inlet swirl ratio from 2.2 to 1.5 results in the advancement

of CA50 by over 6 CAD (from 6.9° ATDC to 0.6° ATDC) This

advanced, short combustion duration event, results in high

pressure rise rates (10.8 bar/deg) and high NOx emissions (9.1

g/kg-FI), and indicates a hotter, more-thermally homogeneous

mixture distribution prior to ignition, as described by Aceves et

al [9, 11]

In contrast, increasing inlet swirl from 2.2 to 3.5 was observed

to retard combustion phasing (CA50 retards from 6.9° ATDC to

10.2° ATDC) Pressure rise rates are reduced as well (from 4.3

to 3.2 bar/deg), while combustion duration (in this case,

CA10-75) increases from 8.1 to 10.3 CAD (an increase of

27%) In the case of increasing inlet swirl, these results agree

with previous assessments of increased swirl ratios in LTC

strategies (specifically HCCI) As inlet swirl is increased, heat

transfer from in-cylinder gases to the wall is enhanced resulting

in an overall cooler charge temperature This reduction in temperature causes combustion phasing to retard [6, 9]

Figure 7 Experimental cylinder pressure and heat release rates for 4 bar net IMEP operation at 1300 RPM given changes in inlet swirl With Tin and IMAP fixed at 65C and 130 kPa, respectively, CA50 varies substantially (∼10 CAD over the inlet swirl range), and advances as inlet swirl is reduced

The results of a similar experimental set at 2000 RPM is shown

in Figure 8

Figure 8 Experimental cylinder pressure and heat release rates for 4 bar net IMEP operation at 2000 RPM given changes in inlet swirl With Tin and IMAP fixed at 80C and 150 kPa, respectively Volumetric efficiency effects are more apparent at the higher engine speed At Rs=1.5, cylinder pressure during compression is significantly lower causing CA50 to retard, relative to Rs=2.2 and 3.5

Given fixed IMAP and Tin at 2000 RPM, the combustion phasing variations due to changing inlet swirl are in the opposite direction as observed at 1300 RPM (see Figure 7 and Figure 8) Using throttle plates in the intake runner to adjust inlet swirl levels effectively reduces the volumetric efficiency of the engine, and these effects figure more prominently at 2000 RPM For example, at 1300 RPM, reducing swirl from 2.2 to 1.5 results in a reduction of volumetric efficiency from 94.8% to 86.9% In contrast, at 2000 RPM, the same inlet swirl variation reduces volumetric efficiency from 94.8% to 74.4% This reduction results in a decrease in cylinder pressure (TDC pressure is reduced from 63.2 bar to 50.8 bar, a 19.6% reduction) causing CA50 at Rs=1.5 to retard 1.3 CAD (from 8°

to 9.3° ATDC) Increasing Rs from 2.2 to 3.5 causes

combustion phasing to advance 1 CAD (from 8° to 7° ATDC)

and results in higher pressure rise rates as well (3.9 to 5.2 bar/

deg)

Figure 9 In a comparison of case 1 experimental results, CA50

exhibits more variation at 1300 RPM than at 2000 RPM Further, at

1300 RPM, CA50 advances as swirl ratio is reduced; in contrast, at

2000 RPM, CA50 advances (although less) with increased inlet swirl

Intake throttling resulting from required swirl plate adjustments at

Rs=1.5 and 3.5 reduces volumetric efficiency and effectively creates a

globally-richer mixture concentration

Although variations in combustion phasing were observed at both engine speeds, the results at 2000 RPM were less significant While results to be discussed for case 3 (matched CA50 and Φ) will provide a better understanding of swirl effects

at 2000 RPM, the effects of increasing inlet swirl at 2000 RPM given fixed IMAP and Tin follow results observed in

conventional diesel combustion; that is, increased swirl enhances air-fuel mixing and shortens ignition delay

Figure 9 compares CA50, Φ, and volumetric efficiency as a function of inlet swirl between the two engine speeds investigated at 4 bar net IMEP (recall, fueling rate was adjusted

as necessary to maintain constant load)

Interestingly, at 2000 RPM, the variation in CA50 over the swirl range considered is less than that observed at 1300 RPM Specifically, at 1300 RPM, CA50 varies almost 10 CAD (over the inlet swirl range investigated), as opposed to 2 CAD at

2000 RPM Regardless of these opposing trends in CA50 at

1300 and 2000 RPM, NOx emission trends are similar, and appear to be dominated by (and show sensitivity to) local mixture concentrations, which influences ignition location within the combustion chamber, and subsequently, peak combustion temperatures (as will be seen in CFD results, shown Figure 14b and c) For example, although Rs=1.5 causes combustion phasing to retard at 2000 RPM, NOx emissions remain highest (as shown in Figure 10); similar to results at 1300 RPM Similar NOx trends throughout the swirl range investigated are

observed, i.e., increasing NOx with reduced inlet swirl UHC

and CO emissions both increase as inlet swirl is increased for both engine speeds Specific to UHC trends, increasing inlet air turbulence appears to result in overly lean regions and cooler temperatures within the combustion chamber, causing oxidation kinetics to quench (and corroborated by CFD results) The CO trends between engine speeds are similar as well; however, at 2000 RPM, a substantial reduction in CO was observed (more so than at 1300 RPM) when inlet swirl was increased from Rs=2.2 to 3.5 (277 to 131 g/kg-FI) CO emissions for this piston bowl geometry have been shown to

be affected by the ability to promote CO oxidation in the squish region [28, 29]; this reduction may indicate sufficient local mixture enrichment in this region

In order to analyze the in-cylinder combustion behavior with swirl ratio variation, numerical simulations were performed for the engine operation at 1300 rev/min

Figure 11 compares predicted pressure and heat release rate profiles with measured data In the figure, predicted (or calculated) and experimental results are presented with solid

and dashed lines, respectively It is seen that the change of

pressures during the compression and expansion strokes with

swirl ratio variation is well captured by the prediction The

predicted ignition timings are in good agreement with

experiments for all three swirl ratios, while pressure rise is

slightly over-predicted for the cases with swirl ratios of 2.2 and

3.5 While experimental heat release is derived from cylinder

pressure data, numerical calculations consist of chemical heat

release only (and absence of wall heat transfer)

Figure 10 Experimental comparison of emission levels at 4 bar net

IMEP given case 1 operating conditions NOx trends are similar,

regardless of engine speed, over the inlet swirl range investigated This

behavior indicates NOx emission rates are dominated by local mixture

concentrations, as opposed to phasing effects (which could increase or

decrease mixing time) UHC emissions increase with increasing inlet

swirl, which could indicate more crevice volume entrapment

Interestingly, CO emissions for both engine speeds peak at Rs=2.2; at

Rs=1.5, reduced heat transfer (and overall hotter charge) facilitates

more-complete CO oxidation while at Rs=3.5, CO levels are reduced

due to an enrichment of squish region mixture concentration (as will be

shown in CFD results)

Figure 11 Comparison of predicted and measured pressure and heat release rate profiles for engine operations at 1300 rev/min in Case-1

Figure 12 Comparison of predicted and measured IMEP and emissions for engine operations at 1300 rev/min in Case 1 (a) IMEP and NOx emissions, (b) UHC and CO emissions

The predicted IMEP matched the experimental values of ∼4bar, and the NOx emissions are in good agreement with measured data both in trend and quantitatively, as shown in Figure 12a UHC emissions are slightly over-predicted, while CO emissions are significantly over-predicted, as seen in Figure 12b The over-prediction of CO emissions is attributed to the underprediction of CO oxidation during the expansion stroke after the main ignition (CA >10° ATDC) The first explanation for this is that while numerical calculations showed higher peak heat release rates (Figure 11), cumulative heat release was predicted lower, thus resulting in lower CO oxidation during piston expansion Secondly, the underpredicted mixing of high temperature burned gases with unburned charge in the squish

region could be another reason for the numerical results

leading to higher CO emissions Except for the discrepancy in

CO emissions, the numerical simulations predict the engine

performance and emissions trends quite well

For the baseline operating conditions (Rs=2.2 in Case 1),

distributions of spray droplet, gas temperature and fuel

equivalence ratio in the cylinder are shown for various crank

angles in Figure 13 The snapshot plots provide a means for

characterizing combustion behavior through temperature and

equivalence ratio distributions along a characteristic plane; in

this case, the spray axis It is seen in Figure 13a that a small

fraction of spray droplets enter the squish region, and impinge

on the piston-top surface resulting in a thin film of fuel on the

wall It is also seen that the wall film fuel layer moves in the

direction of in-cylinder swirl during compression Figure 13b

shows the temperature distribution on the spray axis plane

The (black) isothermal contour shown in the figure indicates

T=1400 K locations Ignition is predicted to occur at around +4°

ATDC in the middle of the re-entrant bowl region High

temperature burned gases are mainly seen in the bowl bottom,

re-entrant and squish regions and move towards the cylinder

liner being convected by the reverse squish flow as the piston

descends Local maximum gas temperature peaks at slightly

above 2500 K around 10° ATDC (Figure 13d)

Targeting the bowl lip region in the cylinder, the fuel spray is

split between the bowl and squish regions Due to the

interaction between the squish flow (in the direction of the bowl

center) and the spray-induced flow, the fuel vapor is directed to

the re-entrant region and flows along the bowl surface while

being affected by swirl flow This helps form charge mixtures

favorable for ignition in the bowl re-entrant region On the

contrary, the localized wall film fuel forms locally rich regions

near the piston-top surface in the squish region (clearly seen at

TDC in Figure 13c), which burns after the first ignition occurs

(+4° ATDC) Typically, upon ignition, locally rich mixtures serve

to enhance combustion heat release; however, in the squish

region where rich mixtures appear, cooling by evaporation and

wall heat transfer serves to suppress reactions and increase

ignition delay

It is notable that the maximum equivalence ratio in the cylinder

decreases gradually as more mixing of fuel and air evolves

until the ignition timing (see ϕmax values in Figure 13c) Since

the equivalence ratio is calculated using the reactants only, the

equivalence ratio of a local lean/rich mixture approaches zero/

infinite, once ignition occurs This is why the local maximum

equivalence ratio is seen to increase to ϕmax=2.46 at the timing

of ignition (+4° ATDC) in Figure 13c Further mixing of burned

gases and unburned mixtures (likely to be lean) increases the

uniformity of the in-cylinder mixtures and maximum

equivalence ratios fall on the lean side (+10° ATDC)

Figure 13 In-cylinder distributions of spray droplets, gas temperature and equivalence ratio for the simulation baseline operating conditions (a) Spray drop distribution at various cranks angles before ignition Spray axis planes are plotted for reference (b) Gas temperature distributions in the spray axis plane Iso-contour lines are for 1400 K (c) Equivalence ratio distributions in the spray axis plane Iso-contour lines are for f=0.5 and local maximum equivalence ratio at each crank angle is indicated, as well (d) Profiles of average and local maximum gas temperatures in the cylinder