a computational investigation of thea computational ... · equivalent cetane number 47 ca =...

A computational investigation of theA computational investigation of theA computational investigation of the A computational investigation of the effects of swirl ratio and injection effects of swirl ratio and injection pressure on mixture preparationpressure on mixture preparationpressure on mixture preparation pressure on mixture preparation

in a lightin a light--duty diesel engineduty diesel engine

F. Perinia, A. Dempseya, R. Reitza

D. Sahoob, B. Petersenc, P. MilesbD. Sahoo , B. Petersen , P. MilesaUniversity of Wisconsin Engine Research Center

bSandia National LaboratoriescFord Motor CompanycFord Motor Company

ERC Student Seminar – 01/22/2013

University of Wisconsin -- Engine Research CenterERC Seminar – 01/22/2013

slide 1

OutlineOutlineOutlineOutline Experimental vs. Numerical study details

Motivation and problem setup

Model improvementp Compressible connecting rod assembly

Local mixture preparation study Local mixture preparation study Non-reacting conditions

Model accuracy impact on fired operation Wall heat transfer sensitivity to swirl ratio and injection pressure


slide 2

OpticalOptical engineengine experimentalexperimental setupsetupf ASME ICES2012 81234 Engine specifications

Bore x stroke [mm] 82.0 x 90.4 Unit displacement [cm3] 477.2 Compression ratio 16 4 : 1

cf. ASME ICES2012-81234

P1 = Half of squish regionP1 = Half of squish regionCompression ratio 16.4 : 1Squish height at TDC [mm] 0.88

Bosch CRIP 2.2 Injector

P1 Half of squish region

P2 = Piston bowl rim

P3 = Inner bowl region

P1 Half of squish region

P2 = Piston bowl rim

P3 = Inner bowl regionBosch CRIP 2.2 InjectorSac volume [mm3] 0.23 Number of holes 7 Included angle [deg] 149

rede

ck [c

m]

CA = -15 deg ATDC

-0.5

0.0

0.5

P1P2

rede

ck [c

m]

CA = -15 deg ATDC

-0.5

0.0

0.5

P1P2

Hole diameter [mm] 0.14 Hole protrusion [mm] 0.3 Fuel properties r axis [cm]

dist

ance

from

fir

0 1 2 3 4

-2.0

-1.5

-1.0

0.5

P3

r axis [cm]

dist

ance

from

fir

0 1 2 3 4

-2.0

-1.5

-1.0

0.5

P3

p pComposition [mole fractions] 75% nC7H16 25% iC8H18 Fluorescent tracer [mass fraction] 0.5% C7H8

266 nm UV horizontal laser sheet Images at (degrees ATDC):

r axis [cm]

0 0.5 1 1.5 2 2.5 3 3.5 4

r axis [cm]

0 0.5 1 1.5 2 2.5 3 3.5 4


slide 3

Equivalent Cetane Number 47

CA = [-17.5:2.5:-5.0]

OpticalOptical engineengine experimentalexperimental setupsetup IIIIi l N ti R ti 7 Diesel PPCI cases

(ASME ICES2012-81234)

Low-load

Non-reacting mixture

Reacting mixture

Intake charge composition [mole 100% N2

10% O2 81% N2 Low load

Highly dilute Slightly boosted

R d

fractions] 9% CO2 Intake pressure [bar] 1.5 Intake temperature [K] 300 372

Rs and pinj sweeps

Very low PM and NOx

Si ifi t UHC d CO

Engine speed [rpm] 1500IMEP [bar] --- 3.0 Global equiv. ratio [-] --- 0.3 I j t d f l [ ] 0 0088 0 0088 Significant UHC and CO

incomplete oxidationof bulk gas mixtures

Injected fuel mass [g] 0.0088 0.0088

Start of Injection [deg] -23.0 0.1, -23.3 0.1 Parameter sweeps: *Rs = 1.55,

*Rs = 2.20,*pinj = 860 bar *pinj = 860 barswirl ratio R [ ]g

comb. efficiency Crucial role of mixing

and chemical kinetics

Rs 2.20, *Rs = 3.50, *Rs = 4.50, *Rs = 2.20, *R = 2 20

pinj 860 bar*pinj = 860 bar *pinj = 860 bar *pinj = 500 bar * 1220 b

- swirl ratio, Rs [-] - injection pressure, - pinj [bar]

*baseline case


slide 4

and chemical kinetics *Rs = 2.20, *pinj = 1220 barbaseline case

ComputationalComputational modelmodel setupsetup The ERC version of KIVA3v-R2 is employed Improvements to:

Phenomenon Submodel

Spray breakup KH-RT instability, Beale and Reitz

Near-nozzle flow Gas-jet theory, Abani et al. Spray j y

Droplet collision O’Rourke model with ROI (radius-of-influence)

Wall film O’Rourke and Amsden

p y

Evaporation Discrete multi-component fuel, Ra and Reitz

Turbulence RNG k- ε, Han and Reitz

C b ti Detailed chemical kinetics with sparse

Fuel

G id f l ti t d (SAE 2012 01 0143)

Combustion panalytical Jacobian, Perini et al.

Reaction kinetics Reduced PRF mechanism, Ra and Reitz

Chemistry


slide 5

Grid from resolution study (SAE 2012-01-0143)

CompressibleCompressible connectingconnecting rodrod modelmodel IIMotivation Motoring pressure trace match requires

model calibration

over-predictionaround TDCmodel calibration

Incomplete piston/head geometry modeling Charge blow-by to the crankcase Compressibility in silica piston assemblies

around TDC

under-predictionis not negligible (Aronsson et al., SAE2012-01-1604)

Using Geometrical CR leads to significantdi ti i th d l

under predictionduring expansion

pressure overprediction in the model

Engine grid’s CR typically reducedb tifi i ll i i i lby artificially increasing crevice volume Affects trapped mass, wall heat transfer, pollutants Still difficult to match pressure trace shape


slide 6

CompressibleCompressible connectingconnecting rodrod modelmodel IIII*figure not in scale

Target squish = 0.88 mm

C ld l 1 04 Th l i 0 31

Measured thermalexpansion

Cold clearance = 1.04 mm

Compression 0.15 mm (TDC)

Thermal expansion 0.31 mm

TARGET POS.

SETTINGSETTINGMeasured extended

piston assemblycompliance

Experimental setup accounts for deviations from rigid slider-crank does not consider bearing and crankshaft clearances

compliance

does not consider bearing and crankshaft clearances Thermal expansion less affected by engine operation

Ring friction is dominant scarcely reached by hot gases Dynamic squish height can improve modelling pollutants


slide 7

Dynamic squish height can improve modelling pollutants

CompressibleCompressible connectingconnecting rodrod modelmodel IIIIII

MODEL DETAILS Static intertial forces are

Ap

dtd

2sr c

b

||F

F

neglected Global compliance is modeled

as an effective connecting rod length, c

F

z

tan1i dcsdzP

g g ,

Extended piston + bearings + crankshaft

MODEL CALIBRATION

.costan

1sin2

dtdtP

6 x 106 motored trace match

Need to reduce CR:Wall heat transfer

modeling inaccuracy?MODEL CALIBRATION “Rigid” grid: CR= 16.7,

squish = 0.73 mm (cold height+thermo) very low compliance (k = 4.5e4 N/mm)

5

6

[Pa]

y p ( / )(5 times lower than measured)

“Calibrated” grid: CR=16.3, sq = 0.73 mm k = 1.0e5 N/mm (half than measured) 3

4

pres

sure

CR = 15.7, rigidCR = 16.7, k = 4.5e07 N/m


slide 8 -30 -20 -10 0 10 20 302

crank angle [degrees ATDC]

CR = 16.3, k = 1.0e08 N/mexperiment

LocalLocal equivalenceequivalence ratioratiopredictionprediction


slide 9

SwirlSwirl ratioratio effectseffects I I –– CA = CA = --17.5 17.5 degdeg

KIVA, -17.5 CA

m]

Experiment, -17.5 CA

KIVA, -17.5 CA

m]

Experiment, -17.5 CA

2

Rs = 1.55, pinj = 860 bar Rs = 4.5, pinj = 860 bar

P1, y

axi

s [cm

-2

0

2

-2

0

2

1

2

P1, y

axi

s [cm

-2

0

2

-2

0

2

1

2

P

-2 0 2

s [cm

]

2

-2 0 2

2

0

2

3

P

-2 0 2

s [cm

]

2

-2 0 2

2

0

2

3

Jet deflection very well predicted at high Rs

P2, y

axi

s

-2 0 2

-2

0

-2 0 2

-2

0

0

1

P2, y

axi

s

-2 0 2

-2

0

-2 0 2

-2

0

0

1

2Jet deflection very well predicted at high RsExcessive distortion at Rs = 1.55 lack of

penetration into the squish region2 0 2

axi

s [cm

]0

2

2 0 2

0

2

2

4

y ax

is [c

m]

0

2

0

2

2

4

6penetration into the squish region


slide 10

x axis [cm]

P3, y

-2 0 2

-2

x axis [cm]

-2 0 2

-2

0x axis [cm]

P3, y

-2 0 2

-2

x axis [cm]

-2 0 2

-2

0

2

SwirlSwirl ratioratio effectseffects II II –– CA = CA = --5.0 5.0 degdegRs = 1.55, pinj = 860 bar Rs = 4.5, pinj = 860 bar

KIVA, -5.0 CA Experiment, -5.0 CA 1.5

KIVA, -5.0 CA Experiment, -5.0 CA1L k f i i l l i t f i f

1, y

axi

s [cm

]

-2

0

2

-2

0

2

0.5

1

1, y

axi

s [cm

]

2

0

2

2

0

2

0.5

1Lack of mixing no overly lean mixture forming fromthe jets and remaining near the center

P

-2 0 2

[cm

]

2

-2 0 2

2

0

1

P1

-2 0 2

-2

[cm

]

2

-2 0 2

-2

2

0

1

Spray penetration very well captured

P2, y

axi

s

2 0 2

-2

0

2 0 2

-2

0

0

0.5

P2, y

axi

s [

2 0 2

-2

0

2 0 2

-2

0

0

0.5Spray penetration very well captured

Squish region and central inner bowl plane

-2 0 2

axis

[cm

]0

2

-2 0 2

0

2

0.5

1-2 0 2

axis

[cm

]

0

2

-2 0 2

0

21

1.5


slide 11

x axis [cm]

P3, y

-2 0 2

-2

x axis [cm]

-2 0 2

-2

0

x axis [cm]

P3, y

-2 0 2

-2

x axis [cm]

-2 0 2

-2

0

0.5

SwirlingSwirling flow flow structurestructureC i i h PIV b P (SAE 2011 01 1285)

25

Rs = 2.20Rs = 3 50 12

14

Comparison with PIV measurements by Petersen (SAE 2011-01-1285)

tangential velocities [m/s], CA = -50.013

tangential velocities [m/s], CA = -35.0 tangential velocities [m/s], CA = -25.0

15

20

eloc

ity [m

/s]

Rs 3.50 Rs = 4.50

8

10

12

eloc

ity [m

/s]

[cm

]

11

12

13

5

10

tang

entia

l v

( k ) i f [2 ] 2

4

6

tang

entia

l v

CA = -50.0 CA = -35.0 CA = -25.0

( k ) i f [25]CA = -35° ATDC

z ax

is

9

10

0 5 10

0 1 2 3 4 50

radius [cm] from cylinder axis

(marks) experiments from [25](lines) KIVA, Bessel fit = 2.20

0 1 2 3 4 50

2

radius [cm] from cylinder axis

(marks) experiments from [25](lines) KIVA, Bessel fit = 2.20

r axis [cm]

0 1 2 3 4 5

8

r axis [cm]

0 1 2 3 4 5

r axis [cm]

0 1 2 3 4 5

Predicted velocity profile accuracy deteriorates when approaching TDC high angular momentum from the squish volume forced inward OK with the model’s geometry but not seen in the experiments


slide 12

g y p- impact of valve recesses in the head and cut-outs on the piston

InjectionInjection pressurepressure effectseffects I I –– CA=CA=--5.05.0

KIVA, -5.0 CA Experiment, -5.0 CA 1

KIVA, -5.0 CA Experiment, -5.0 CA 1.5

Rs = 2.2, pinj = 500 bar Rs = 2.2, pinj = 1220 bar

Fuel does not reach back

P1, y

axi

s [cm

]

-2

0

2

-2

0

2

0.5

P1, y

axi

s [cm

]

-2

0

2

-2

0

2

0.5

1into plane P2

P

-2 0 2

[cm

]

2

-2 0 2

2

0

1

P

-2 0 2

[cm

]

2

-2 0 2

2

0

1pinj far from typical valuesfor validation of diesel

P2, y

axi

s

2 0 2

-2

0

2 0 2

-2

0

0

0.5

P2, y

axi

s

2 0 2

-2

0

2 0 2

-2

0

0

0.5spray models

I ffi i t t ti-2 0 2

axis

[cm

]

0

2

-2 0 2

0

21

1.5-2 0 2

axis

[cm

]

0

2

-2 0 2

0

2

0.5

1Insufficient penetrationBetter mixing in P1


slide 13

x axis [cm]

P3, y

-2 0 2

-2

x axis [cm]

-2 0 2

-2

0

0.5

x axis [cm]

P3, y

-2 0 2

-2

x axis [cm]

-2 0 2

-2

0

InjectionInjection pressurepressure, , mixturemixture stratificationstratification Cumulative azimuthal equivalence ratio distributions

100 %KIVA, Rs = 2.2, pinj = 500 bar

100 %Experiment, Rs = 2.2, pinj = 500 bar

Histograms at fixed distances from cylinder axisRs = 2.2, pinj = 500 bar Rs = 2.2, pinj = 1220 bar

100 %

]

j 100 %

0 %

20 %

40 %

60 %

80 %

dist

ribu

tion

of

[-]

P10 %

20 %

40 %

60 %

80 % P1

0 %

20 %

40 %

60 %

80 %

dist

ribu

tion

of

[-]

P10 %

20 %

40 %

60 %

80 %

P1

Most deviations are in the central region impact on ignition and pollutants!

0.1 0.9 1.6 2.4 3.1 0 %

radius [cm]

60 %

80 %

100 %

ion

of

[-]

j

60 %

80 %

100 %

0.1 0.9 1.6 2.4 3.1 0 %

radius [cm]

0.1 0.9 1.6 2.4 3.1 0 %

radius [cm]

60 %

80 %

100 %

tion

of

[-]

j

60 %

80 %

100 %j

0.1 0.9 1.6 2.4 3.1 0 %

radius [cm]

Close match at the squish plane OK for emissions

0.0 0.5 0.9 1.4 1.9 0 %

20 %

40 %

radius [cm]

dist

ribu

t

P2

100 %j

100 %j

0.0 0.5 0.9 1.4 1.9 0 %

20 %

40 %

radius [cm]

P2 0.0 0.5 0.9 1.4 1.9

0 %

20 %

40 %

radius [cm]

dist

ribu

t

P2

100 %j

100 %

0.0 0.5 0.9 1.4 1.9 0 %

20 %

40 %

radius [cm]

P2

20 %

40 %

60 %

80 %

dist

ribu

tion

of

[-]

P3 20 %

40 %

60 %

80 %

P3 20 %

40 %

60 %

80 %

dist

ribu

tion

of

[-]

P320 %

40 %

60 %

80 % P3


slide 14

0.9 1.2 1.6 1.9 2.2 0 %

radius [cm]

0.9 1.2 1.6 1.9 2.2

0 %

radius [cm]

0.1 0.25 0.25 0.5 0.5 0.75 0.75 1.01.0 1.25 1.25 1.5 1.5

0.9 1.2 1.6 1.9 2.2 0 %

radius [cm]

0.9 1.2 1.6 1.9 2.2

0 %

radius [cm]

FiredFired engineengine operationoperation I I –– RRss sweepsweep

40AHRR, pinj = 860 bar, Rs sweep

/deg

]

KIVA, Rs = 1.55Exp, Rs = 1.55

Experiment: ignitionadvances at Rs = 1.55

i h i t i th

20

30

t rel

ease

[J/

KIVA, Rs = 2.20Exp, Rs = 2.20KIVA, Rs = 3.50Exp, Rs = 3.50

richer mixtures in the squish region

At higher Rs richer

10

pare

nt h

eat At higher Rs, richer

mixtures are due to fuelmore strongly confined to

the bowl

-15 -10 -5 0 5 10 15 200


app

the bowl

KIVA: - similar ignition timings good average match- inverse ignition behavior at Rs under investigation:

1) Wall heat transfer


slide 15

1) Wall heat transfer. 2) over-predicted jet deflection more homogeneous and

leaner mixtures longer ignition delay

FiredFired engineengine operationoperation II II –– ppinjinj sweepsweep

40AHRR, Rs = 2.20, pinj sweep

deg]

KIVA, pinj = 500 bar

exp, pinj = 500 bar

LT HR timing well captured at the experimental pressures

20

30

rele

ase

[J/d KIVA, pinj = 860 bar

exp, pinj = 860 bar

KIVA, pinj = 1220 bar

exp, pinj = 1220 barHT HRR peaks well captured Wall heat transfer

10

20

aren

t hea

t

Lower and delayed AHRR peak at pinj=500 bar in

-15 -10 -5 0 5 10 15 200


appa

injKIVA lack of penetration in

the central region6.5

7 x 106pressure trace, Rs = 2.20, pinj sweep

pinj = 500 bar

pinj = 860 bar

p = 1220 barg [ g ]

Computational Investigation at higherinjection pressures Misfiring conditions at pinj 1500 bar

5.5

6

pres

sure

[Pa]

pinj = 1220 bar

pinj = 1500 bar

pinj = 1750 bar

pinj = 2000 bar


slide 16

Misfiring conditions at pinj 1500 bar Wall heat transfer

-20 -10 0 10 204.5

5


ConcludingConcluding RemarksRemarks II

Aim: assess+improve the accuracy of KIVA modelling of anoptical light duty diesel engine operated in LTC (PPC) mode, withrespect to:p quantitative equivalence ratio distributions provided by the experiments at three in-cylinder planes understanding and exploring the role of mixing and wall understanding and exploring the role of mixing and wallheat transfer on combustion development

Non-reacting operation and equivalence ratio distribution Elastic extended piston – connecting rod assembly model

Significantly improved motored pressure curve match Need to lower geometrical CR wall heat transfer?


slide 17

ConcludingConcluding RemarksRemarks IIII

Non-reacting operation and equivalence ratio distribution Mixture dynamics and equivalence ratio stratification before ignition

Very accurate penetration is predicted with a refined grid Very accurate penetration is predicted with a refined grid Over-predicted swirl when approaching TDC jet deflection and under-predicted penetration at pinj

Under predicted turbulent mixing crucial to emissions Under-predicted turbulent mixing, crucial to emissions

Fired engine operation At increasing Rs: predicted HTHR timing delays

measured HTHR timing advances

The model responds to wall heat transfer and over-predicted spray deflection

the experiments instead show that mixing rules over ignition timing


slide 18

the experiments instead show that mixing rules over ignition timing

ConcludingConcluding RemarksRemarks IIIIIIFi d i iFired engine operation Injection pressure plays a major role on combustion development

Increased impact area + greater spray jet momentum Delayed ignition timing can lead to misfire at very high pressures

F W kF W kCFD model impact on mixing and ignition

Future WorkFuture Work

Turbulent transport generalized RNG k- model

Fluid flow solver accuracy solution tolerances and numerics Fluid flow solver accuracy solution tolerances and numerics

Wall heat transfer Impact of wall temperatures C j t h t t f


slide 19

Conjugate heat-transfer

Future WorkFuture Work

UHC and CO emissions Investigate impact of the reaction mechanism on predicted emissions

160

180

200

220

[g/k

g f]

UHC, CA=35.0

100

120

140

160

CO

@ E

VO

125k cells grid48k cells gridexperiment

1 2 3 4 580

100

swirl ratio, Rs [-]

CO, CA=35.0

Identif and alidate

Simulation

Identify and validatein-cylinder emissionsources


slide 20

Simulation

ThanksThanks forfor youryour attentionattention!!QuestionsQuestions??


slide 21

BackBack--upup slidesslidesBackBack--up up slidesslides


slide 22

Laser Laser sheetsheet imagingimaging –– missingmissing zoneszones

Laser sheet

distortion whenpassing throughcomplex surfaces

(266nm, UV)

injector tip “shade”

LLens

Mirror CameraMirror


slide 23

KIVA KIVA planeplane slicesslices reconstructionreconstruction

Generate a plane representation:

kjikjiPPP vvvuuuzyxplane

O

z

P v

x

y u

Compute intersectingpoints of KIVA mesh 3

4

points of KIVA mesh with plane

1

2


slide 24-2 -1 0 1 2

0

KIVA KIVA planeplane slicesslices reconstructionreconstruction

Reconstruct data at those values using Delaunay Triangulation (left)

Pursue cubic spline interpolation at a more refined grid, p p g ,using point positions from the laser sheet images (right)

4

4.5

4

2.5

3

3.5

cm]

3

4

1

1.5

2

y ax

is [c

1

2

-2 -1 0 1 2-0.5

0

0.5

x axis [cm]-2 -1 0 1 2

0


slide 25

x axis [cm]

a computational investigation of thea computational ... · equivalent cetane number 47 ca =...

Documents