1Zheng

An Enabling Study of Diesel Low-Temperature Combustion via Adaptive Control

Ming Zheng, Graham T Reader, Usman

Asad,Yuyu

Tan, Xiaoye

Han, Kelvin Xie, and Meiping

Wang

Clean Diesel Engine Research LaboratoryUniversity of Windsor

Technical Session 1: Advanced Combustion Technologies, Part 14:00–4:20pm Monday August 4 2008, Grand Foyer

U.S. Department of Energy 14th

Diesel Engine-Efficiency and Emissions Research (DEER) ConferenceHyatt Regency Dearborn Hotel, August 4-7, 2008

Dearborn, Michigan

Contact: mzheng@uwindsor.ca

2Zheng

LTC Enabling Prospects1.

Optimize control by separating the time scales of fuel air mixing and ignition

2.

Stabilizing LTC operations –

on cliff operation of ultra low NOx emissions and acceptable fuel efficiency

3.

Guide transient combustion control within LTC mode when major engine operating parameters such as boost, EGR, and engine speed varies

4.

Raise engine Load level in LTC5.

Mode shifts between conventional and LTC

6.

Multi-cylinder EGR, fuel, and air distributions7.

Biodiesel Impact –

Cetane, oxygen content, volatility,

viscosity, biodegradation, high pressure compressed solid

3Zheng

Diesel LTC Challenges1.

The fuel efficiency of the LTC cycles is commonly mired by the high levels of hydrocarbon (HC) and carbon monoxide (CO) emissions. The fuel-efficiency of HCCI engines is often compromised by the high levels of HC and CO emissions that may drain substantial amount of fuel energy (5~15% in low-load cases) from the engine cycle.

2.

Moreover, the combustion process becomes less robust and enters into narrower operating ranges and with higher instabilities compared to conventional high temperature combustion (HTC) operations –

LTC is closer to the flame-out limits than HTC.

3.

The scheduling of early fuel delivery in HCCI engines has lesser

leverage on the exact timing of auto-ignition that may even occur before the compression stroke completes when a high compression ratio of conventional diesel cycles is applied, which may cause excessive efficiency reduction and combustion roughness.

4.

The high HC and CO emissions are attributed to the relatively low volatility of diesel fuels, the lowered combustion efficiency of

the lean and/or EGR weakened cylinder charge, the non-homogeneity of the cylinder charge, and the fuel condensation and flame quenching on the surfaces of the combustion chamber.

4Zheng

5Zheng

Research Platform –

non compromised for control performance

The research platform consists of an advanced common- rail diesel engine modified for the intensified single

cylinder research and a set of embedded real-time (RT) controllers, field programmable gate array (FPGA) devices, and a synchronized personal computer (PC) control and measurement system. Up to 12 fuel injection pulses per cylinder per cycle have been applied to modulate the homogeneity history of the cylinder charge in mixed mode combustion in order to improve the phasing and completeness of combustion under independently controlled exhaust gas recirculation (EGR), intake boost, and exhaust backpressure.

6Zheng

Experimental Setup -

Capable of multiple parallel 1st

priority control tasks

Engine Type 4 Cylinder, Ford “Puma” Displacement [cm3] 1998 Bore x Stroke [mm] 86 x 86 Compression Ratio 18.2:1 Combustion System Direct Injection Injection System Common-rail; PRail ≤ 160 MPa

7Zheng

Adaptive Combustion Control Platform

-2 0

1 1 5

2 5 0

2 7 0 .5 2 8 3 .3 2 9 6 .1 3 0 8 .9 3 2 1 .7 3 3 4 .5 3 4 7 .3

2

High speed network

Fast Communication

-1

0

1

2

3

4

280 290 300 310 320 330

1

Power DriveReconfigurable

-8

0

8

278 288 298 308 318 328 338

3

Signal Conditioning

Engine Control Parameters

Prompt Communication

Cylinder Pressure EncoderZ & A Signals, 0.1°CA Resolution

Common Rail Diesel Engine

Injector

Data Flow Legend: ControlFeedback

Host & RT-FPGA Controller

RT-FPGA Controller

Cycle-by-CycleCA50 Estimation

Common Rail HP Pump Pressure Control

Combustion Pattern Recognition and Characterization

RT Controller

t ~ after each combustion cycle

Decision Making for FPGA

t ~ after a number of combustion cycles

Programming & Control Platform

Host PCs/Data Storage

Local Decision Making

Reconfigurable I/O

t ~ within one combustion cycle

FPGA VI, 96 sync DI/O8 16bit A/I, 8 16bit A/O3 million gates, 40MHz

ADAPTIVE FUELING CONTROL SYSTEM INTEGRATION

COMBUSTION CHARACTERIZATION

Power DrivesEV Driving Cycle Reconfigurable

Output upto 20A 200VMin. Dwell 10 s b/w EVs PCs: Boost & Exhaust

Back-pressure Control

CLEAN DIESEL ENGINE LAB: COMBUSTION CONTROL PLATFORM III

IMEP Estimation for within Cycle Control

Smart NOx & dual λ Sensors

EGR Estimation and Application StrategiesNOx conformation etc.

Prompt Communication

8Zheng

Fuel Injection Scheduling1.

Up to 12 fuel injection pulses per cylinder per cycle have been applied to modulate the homogeneity history of the HCCI operations in order to better phasing and completing the combustion process.

2.

Empirical studies have been conducted under independently controlled exhaust gas recirculation (EGR), intake boost, and exhaust backpressure.

9Zheng

Challenges in Digital Combustion Control

1.

Adaptation step relaxation versus prompt performance modulation

2.

Cylinder pressure noise filtration versus signal sharpness

3.

Simplex feedback control versus model based forward control

4.

Fuel injection pulse numbers versus total injection time window of the least condensation

10Zheng

Experimental Case Outline

1.

Single shot with heavy EGR to separate the time domains between injection and combustion

2.

Multiple early shots with moderate EGR to improve homogeneity

3.

Multiple early plus main to gain power output4.

Speed and boost transients

11Zheng

DIESEL LTC CHALLENGES•

Prolonged Ignition Delay to enable LTC

0

0.05

0.1

0.15

0.2

0.25

30 40 50 60 70 80EGR Ratio [%]

Indi

cate

d So

ot [g

/kW

-h]

0

1

2

3

4

IME

P [b

ar] &

Ind

NO

x,TH

C

SootNOxIMEPTHC

IncreasedCycle-to-Cycle

Variation toUnstable Operation

Decreased Thermal Efficiency

LTCSlope 2

HTCSlope 1

Free NOx Reductionwithout Soot Penalty

Single Shot Experiments (EGR Sweep)Engine Speed: 1400 RPM

Pintake: 1 bar (abs)Tintake: 30°C

[g/k

W-h

]

Ignition delayprolonged by >50%

12Zheng

New LTC Emission Trade-off

10

100

1000

10000

0 1 2 3 4 5 6Smoke [FSN]

log

(TH

CC

1 )

60MPa & 45kPa90MPa & 45kPa90MPa & 75kPa120MPa & 45kPa150MPa & 45kPa150MPa & 75kPa

HTC

LTC

Single Shot Experiments (EGR Sweep)CA50% HR: ~365°CASpeed: 1500 RPM IMEP: ~8 barEGR Rate: 0 ~ 66%

Increasing EGR Increasinginjectionpressure

0

1

2

3

4

5

6

0 50 100 150 200 250 300NOx [ppm]

Smok

e [F

SN] 60MPa & 45kPa

90MPa & 45kPa90MPa & 75kPa150MPa & 45kPa150MPa & 75kPa

High injection pressure effectiveto reduce smoke in HTC

Increasing EGR

Single Shot Experiments (EGR Sweep)CA50% HR: ~365°CASpeed: 1500 RPM IMEP: ~8 barEGR Rate: 0 ~ 66%

LTC Regime

13Zheng

Diesel LTC vs. Engine Load

Relatively high CO & HC penalty due to large injection quantity/shot; Use of heavy EGR to improve combustion

phasing

0

20

40

60

80

100

120

320 340 360 380 400Crank Angle [°CA]

Cyl

inde

r Pre

ssur

e [b

ar]

0

0.2

0.4

0.6

0.8

Nor

mal

ized

HR

R [1

/°C

A]

Multi-pulse HCCI - 3 Injections/cycleSpeed: 1200 RPM Injection Pressure: 120 MPaFuelling: 45 mg/cycle Pintake: 1.5 bar absTintake: ~45°C

IMEP: 7 barEGR: 45 %CA50%: 350.1°CAO2exh: 4.7%Indicated (g/kWh)NOx: 1.02Soot: 0.1CO: 12.9THCC1: 7.2

IMEP: 7.5 barEGR: 61 %CA50%: 361.3°CAO2exh: 0.65%Indicated (g/kWh)NOx: 0.07Soot: 0.01 CO: 14.4THCC1: 3.9

Injection Scheduling#1: 450 μs@260°CA#2: 600 μs@280°CA#3: 200 μs@335°CA

Smok

eles

sR

ich

Com

bust

ion

0

40

80

120

160

200

320 340 360 380 400Crank Angle [°CA]

Cyl

inde

r Pre

ssur

e [b

ar]

0

0.2

0.4

0.6

0.8

Nor

mal

ized

HR

R [1

/°C

A]

Multi-pulse HCCI - 8 Injections/cycleSpeed: 1200 RPM Injection Pressure: 120 MPaFuelling: 53.7 mg/cycle Pintake: 2.2 bar absTintake: ~45°C

IMEP: 9.1 barEGR: 52 %CA50%: 348.9°CAO2exh: 2.8%Indicated (g/kWh)NOx: 0.1Soot: 0.002CO: 10.3THCC1: 3.2

IMEP: 9.7 barEGR: 58 %CA50%: 353.7°CAO2exh: 0.3%Indicated (g/kWh)NOx: 0.02Soot: 0.002CO: 10.7THCC1: 3.7

Injection Scheduling# 1,3,5,7: 290 μs@260°,274°,288°,302°CA# 2,4,6: 280 μs@267°,281°,295°CA#8: 200 μs@335°CA

Smok

eles

sR

ich

Com

bust

ion

Reducing CO & HC penalty with 8 Injections/ cycle; Higher boost and heavy

EGR to improve LTC

14Zheng

Combustion Cycle Efficiency Calculations•

Exhaust hydrocarbon energy with respect to fuel input at different engine loads

0

2

4

6

8

10

0 1000 2000 3000 4000 5000Exhaust THCC1 [ppm]

Exh

aust

TH

C w

rt Fu

el [%

]

2

4

2012

86

Lines ofConstant IMEP

AssumptionsPintake: 1 bar absLHV: 42.9 MJ/kgη ind : 45%EGR: 0 %

15Zheng

Engine Cycle Simulation•

Heat release rates used as input for simulations

320 330 340 350 360 370 380 390 400CA50 [°CA]

Hea

t Rel

ease

Rat

e [1

/°C

A]

Duration variation

Phase variation

(TDC)

Shape variation

Splitting

16Zheng

Engine Cycle Simulation ResultsEffect of CA50 and combustion duration on ηind

, pmax

& (dp/dθ)max

(1500 RPM, IMEP: 6 bar, Pint

: 1.15 bar abs)

Conventional DieselTypical HCCI Conventional DieselTypical HCCI Conventional DieselConventional DieselTypical HCCI

17Zheng

Engine Cycle Simulation Results•

ηind

comparison between –

Upper: different heat release shapes; Lower: single hump and split combustion

CA50 [°CA]

Single hump Double hump Long tail

Split combustion

Hea

t Rel

ease

Dur

atio

n [°

CA]

Split combustion340 350 360 370 380 390 400 410 420

20

40

60

8

020

4

0

60

80

Single hump Double hump Long tail

CA50 [°CA]

Single hump Double hump Long tail

Split combustion

Hea

t Rel

ease

Dur

atio

n [°

CA]

Split combustionSplit combustion340 350 360 370 380 390 400 410 420

20

40

60

8

020

4

0

60

80

Single hump Double hump Long tailSingle hump Double hump Long tail

18Zheng

Equivalent Combustion Cycle Deficiency Calculations

•

Equivalent THC penalty with CA50 off-phasing

-20

-15

-10

-5

0

5

10

15

0 1000 2000 3000 4000 5000Equivalent Exhaust THCC1 [ppm]

Off-

phas

ing

from

CA

ηm

ax [°

CA

]

2 4

2086

Lines ofConstant IMEP

AssumptionsPintake: 1 bar absEGR: 0 %LHV: 42.9 MJ/kgCAη max : ~365°CA

12

Per

cent

age

Dec

reas

e in

ηm

ax [%

]

27

0

5

15

1

5

12

20

RetardedTiming

AdvancedTiming

19Zheng

Cycle Based Adaptive Control to Improve LTC Transients

1.

SOI synchronized at optimized HR phasing via dPmax

timing modulation

2.

dPmax

ceiling limitation via pilot quantity modulation

3.

IMEP compensation via pilot and main modulations

4.

IMEP top-up control via post quantity modulation

20Zheng

Transient Peak Pressure Rise Timing Modulation after Entering LTC with High EGR (single shot)

Injection Timing Control

348

352

356

360

364

368

372

0 1000 2000 3000 4000Engine Cycle # [-]

SOI,

CA

(dp/

d θ) m

ax [°

CA]

0

5

10

15

20

25

IMEP

& (d

p/d θ

) max

[bar

]

Adaptive Control Tests - EGR, Speed & Boost Transients1000 rpm/50kPa → 1500/100 →1000/50

CA (dPmax) ModulationPinj: 90MPa

High EGR ON

1000 rpm50 kPa

1500 rpm100 kPa

1000 rpm50 kPa

CA dPmax

IMEPStart of Combustion

dPmax

21Zheng

Emission Result when Entering LTC with High EGR under Peak Pressure Rise Timing Modulation

Injection Timing Control

0

100

200

300

400

8580 9080 9580Time [sec]

NO

x &

TH

C [p

pm],

EG

R [%

]

0

1000

2000

3000

CO

[ppm

], S

peed

[RPM

]

Adaptive Control Tests - EGR, Speed & Boost Transients1000 rpm/50kPa → 1500/100 →1000/50

CA (dPmax) ModulationPinj: 90MPa

NOx < 20 ppm

Sharp EGR Increase

1000 rpm50 kPa

1500 rpm100 kPa

1000 rpm50 kPaCO

THCNOx

EGR

Speed

22Zheng

Peak Pressure Rise Timing and Rate Modulation

0

2

4

6

8

10

12

14

16

0 500 1000 1500Engine Cycle #

(dp/

d θ) m

ax [b

ar]

320

330

340

350

360

370

380

SO

I pilo

t, SO

I mai

n, C

A d

p max

[°C

A]1000 rpm

50 kPa

1500 rpm100 kPa

1000 rpm50 kPa

Adaptive Control Tests - Speed & Boost Transients1000 rpm/50kPa → 1500/100 →1000/50

dPmax & CA (dPmax) ModulationPinj: 90MPa

CA dPmax

SOCpilot

dPmax

SOCmain

SOC: Start of CombustionPilot and Main Injections:Timing and Quantity Control

23Zheng

Emission Results under Peak Pressure Rise Timing and Rate Modulation

Pilot&Main: timing & quantity control

0

100

200

300

400

500

4800 4900 5000 5100Time [sec]

NO

x &

THC

[ppm

], EG

R [%

]

0

1000

2000

3000

4000

5000

6000

CO

[ppm

], S

pped

[RP

M],

Boo

st [k

Pa]

Adaptive Control Tests - Speed & Boost Transients1000 rpm/50kPa → 1500/100 →1000/50

dPmax & CA (dPmax) ModulationPinj: 90MPa

1000

rpm

50 k

Pa 1500 rpm

100 kPa1000 rpm50 kPa

CO THC

NOxEGR

SpeedBoost

24Zheng

Peak Pressure Rise Timing and Rate Modulation under High EGR (NOx ~30ppm) Transients

330

340

350

360

370

380

0 1000 2000 3000Engine Cycle # [-]

SOI,

CA

(dp/

d θ) m

ax [°

CA

]

0

3

6

9

12

15

18

21

24

(dp/

d θ) m

ax, I

MEP

[bar

]

Adaptive Control Tests - Speed & Boost Transients1000 rpm/50 kPa → 1500/100 →1000/50

Pinj: 120 MPa

IMEP dPmax

1000 rpm

1500 rpm

SOCmain

dPmax & CA (dPmax) Modulation

SOCpilot

CA (dPmax)

high speed&boost high low d&b t

low speed&boost

SOC: Start of Combustion

25Zheng

Transient Low NOx Confirmation with NOx Sensor

0

20

40

60

80

100

120

3200 3400 3600 3800 4000Time [sec]

NO

x [p

pm],

Boo

st [k

Pa]

, EG

R [%

]

0

500

1000

1500

2000

Spe

ed [R

PM

]

Adaptive Control Tests - Speed & Boost Transients1000 rpm/50kPa ® 1500/100 ®1000/50

Pinj: 120MPa

Smart Nox Sensor NOx Analyzer

EGR

SpeedBoost

dPmax & CA (dPmax) Modulation

26Zheng

LTC ADAPTIVE CONTROL STRATEGIES

•

Structure of the CDEL Adaptive Control System

27Zheng

SYSTEMATIC AND ADAPTIVE CONTROL RESULTS

NOx as function of intake oxygen

0

400

800

1200

1600

2000

8 10 12 14 16 18 20 22Intake Oxygen [%]

NO

x [p

pm] 60MPa & 45kPa

90MPa & 45kPa90MPa & 75kPa120MPa & 45kPa150MPa & 45kPa150MPa & 75kPa100MPa & 50kPa130MPa & 50kPa

Single Shot LTC ExperimentsCA50% HR: ~365°CASpeed: 1500 RPM EGR Rate: 0 ~ 66%

IMEP ~8 bar

~6.5 bar~8.5 bar

NOx decreasesmonotonically

with EGR

-0.02

0.03

0.08

0.13

0.18

0.23

300 310 320 330 340 350 360 370 380Crank Angle [°CA]

Hea

t-Rel

ease

Rat

e [1

/ °C

A]

Diesel LTC Tests-3 Injections/CycleEngine Speed: 1800 RPMIMEP: 5.9~6.6 barTintake: 50~60°C

EGR: 41%IMEP: 5.9 barSmoke: 1.1FSN

EGR: 47%IMEP: 6 barSmoke: 0.96FSN

EGR: 56%IMEP: 6.6 barSmoke: 0.42FSN

IncreasingEGR

Last Injection

Heat-release rate for soot

reduction

28Zheng

Adaptive Control ON/OFF Comparison (High EGR, NOx ~20pm)

325

335

345

355

365

375

0 2500 5000 7500Engine Cycle # [-]

SO

I, C

A (d

p/d θ

) max

[°C

A]

0

4

8

12

16

20

IME

P &

(dp/

d θ) m

ax [b

ar]

Adaptive Control Tests - Speed & Boost Transients1000 rpm/50 kPa → 1500/100 →1000/50

Pinj: 90 MPa

SOCmain

dPmax

1000 rpm1500 rpm

dPmax & CA (dPmax) Modulation

SP:13 bar/°CA

SOCpilot

IMEP

Adaptive Control OFF Pilot OFF

11 bar/°CA 9 bar/°CA

CA dPmax

SOC: Start of Combustion

29Zheng

Pressure Rise Curves under Peak Rise Timing and Rate Control

-4

-2

0

2

4

6

8

10

12

14

355 360 365 370 375

Crank Angle [°CA]

(dP

/dθ)

max

[bar

/°C

A]

SP: 13 bar

SP: 11 bar

SP: 9 bar

OFF with pilot

OFF

Adaptive Control Tests - Speed & Boost TransientsSpeed: 1000 rpm Boost: 50 kPaIMEP: 6.7~7.2 bar Pinj: 90 MPa

IMEP ~6 bar withAdaptive OFF

All Cases: High EGR NOx ~20ppm

30Zheng

IMEP Compensation (high EGR, NOx ~30ppm)

100

150

200

250

300

350

400

450

0 1000 2000 3000 4000Engine Cycle # [-]

Ada

pted

Pos

t Inj

Dur

atio

n [ μ

s]

5

6

7

8

9

10

11

12

13

IME

P [b

ar]

Adaptive Control Tests - Speed & Boost Transients1000 rpm/50 kPa → 1500/100 →1000/50 Pinj: 150 MPa Baseline IMEP: 8.5 bar

IMEP Postduration

1000 rpm1500 rpm

Post-TDC IMEPModulation within each

Combustion Cycle

ΔIMEP

Control OFF

Main: 420μs@359°CAPost: 180~350μs@372°CA

Main: 380μs@359°CAPost: 250μs@372°CA

31Zheng

Sample cylinder pressure and heat release characteristics with

adaptive control

0

10

20

30

40

50

60

70

80

90

100

300 330 360 390 420Crank Angle [°CA]

Cyl

inde

r Pre

ssur

e [b

ar]

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

Nor

mal

ized

Hea

t Rel

ease

Rat

e [1

/°C

A]

1000 RPM, Pint=1.5 bar abs(dp/dθ)max

=8.6 bar/°CA

1000 RPMPinj=1200 barCommanded fuel injection:230µs@330°CA450µs@353°CAAdaptive control:(dp/dθ)max=10 bar/°CACA of (dp/dθ)max=363°CAAcutal fuel injection:130µs@335°CA510µs@355°CA 1500 RPM

Pint=2 bar abs(dp/dθ)max

=6.3 bar/°CA

TDC

2-shots

mpilot ↓mmain

↑

-4

-2

0

2

4

6

8

10

300 330 360 390 420Crank Angle [°CA]

Cyl

inde

r Pre

ssur

e [b

ar]

TDC

dPmax Ceiling <10bar/°CA

Pilot turned ONto lower dPmax

Pilot turned OFFbecause of no needfor lowering dPmax

32Zheng

Progresses in LTC Control -

Reduce reliance on de-NOx after-treatment

1.

The simulations and empirical results indicate that the combustion phasing dominates the maximum attainable fuel efficiency of the engine. However, the phasing domination cedes to high HC when the fuel efficiency is severely deteriorated such as by excessive EGR.

2.

The energy deficiency of typical LTC heat release patterns has been further quantified by comparing with HC and CO emissions with combustion phasing deficiency across the engine load spectra.

3.

Adaptive control strategies based on cylinder pressure and heat release characteristics are implemented to stabilize and enable the low-temperature combustion from mid to high loads especially when high boost and EGR are applied.

4.

Further, oxygen and NOx sensors at the intake and exhaust of the

engine are devised to comprehend the transient impacts of EGR, boost, and load variations.

5.

The multi-pulse scheduling is effective to prevent premature ignition and elevated NOx and soot.

33Zheng

Prospective LTC Load Control Improvements

1.

The mode of EGR enabled LTC

is suitable for low load operations, in which a single shot of fuel is delivered close to the top dead center (TDC). The heat release phasing is fully controllable via injection timing control and thus high energy efficiency in attainable.

2.

The mode of early injection HCCI

is suitable for mid load operations, in which the fuel is delivered in multiple events and by milliseconds prior to TDC and thus the heat release phasing is not directly controllable. EGR is commonly applied suppress premature ignition and combustion noise.

3.

The mode of split burning LTC

is suitable for high load operations, in which a partial amount of fuel is delivered to produce HCCI combustion and the remaining for post TDC late combustion. The latter may be benefited from the virtual EGR produced by the prior HCCI burning and timed to best eliminate combustibles and raise power output.

34Zheng

ACKNOWLEDGEMENTS

The research is sponsored by the Canada Research Chair

program. NSERC, CFI, OIT,

AUTO21, Ford Motor Company, University of Windsor and other non-disclosed OEMs have supported the research programs at Clean Diesel Engine Lab.