CONTENTS

Introduction

Engine setup

Test gasses

ECC sensor

ECC sensor results

Overview of dual-fuel combustion processes

Combustion measurements and data processing

Dual-fuel combustion results

Conclusions

2

TNO ENGINE TESTSDUAL-FUEL COMBUSTION ON 6 CYLINDER TRUCK ENGINE

3

Motivation

For light-duty engines the spark ignition (SI) concept is common,

whereas for heavy-duty engines also dual-fuel engines are increasingly

considered as a viable alternative. This is because the dual-fuel

technology promises higher efficiency, higher load performance, and

the fall back option to 100% diesel operation.

Objectives

• Validation of the gas composition sensor developed in WP3 on an

environment that is typical for present and future heavy-duty gas

engines:

• Validation of MN – knock relation for conventional dual-fuel (CDF)

using natural gas and diesel;

• Outlook towards the meaning and value of MN for advanced dual-

fuel combustion concept RCCI.

• Potential assessment of sensor signals for control purposes in dual-

fuel applications.

ENGINE SETUP

4

Engine 6 cylinder in-line, 8 liter, 250 kW

Compression

ratio18:1

Bore/stroke

ratio0.8

Fuel injection

equipment

Common rail system, maximum

injection pressure 2000 bar, multiple

injection capability

Air pathWaste-gated turbocharger, intake

throttle

EGR pathHigh-pressure, cooled EGR system with

EGR valve

Engine test

fuels

• 4 different natural gas compositions:

• G20

• MZ70

• TUBS “Mix9”

• Dutch Natural Gas (DNG)

• Diesel fuel type: EN590;

gas sensor location

ENGINE TEST FUELSvol.% (rel.

uncertainty)

G20 MZ70 “Mix9”

definition

“Mix9”

in the bottles

Dutch NG (DNG)

CH4 100 80.11 89.997 89.991 80.85

C2H6 0 19.88 5.717 5.730 (±2% rel.) 3.74

C3H8 0 0.01 1.797 1.786 (±2% rel.) 0.61

i-C4H10 0 0 0 0 0.09

n-C4H10 0 0 2 2.026 (±2% rel.) 0.10

i-C5H12 0 0 0 0 0.03

n-C5H12 0 0 0 0 0.02

N2 0 0 0.490 0.467 (±2% rel.) 13.02

...

MN (method) 100 70.5 (NPL) 69 (NPL) 68.9 (NPL) 84.2 (NPL)

5

ECC SENSOR INSTALLATION

ECC sensor developed and validated in WP3 is installed in fuel line of dual fuel engine

ECC sensor measures gas composition and Methane Number

3 weeks of experiments

Points of attention:

Large enough vessel to reduce flow

No direct flow on the sensor chips

Thermal isolation from engine

Particle and droplet filter

6

ECC Sensor

ECC Sensor in vessel Location of vesselChip contamination

during first test

ECC SENSOR RESULTSDuring operation of engine:

Pressure fluctuations between 3-4 bara

Temperature fluctuations between 20-40 °C

Flow between 0-40 kg/h (~0-10 m/s)

Chip response depends heavily on temperature

Under no-flow conditions T-sensor ≠ T-chips

→ major error in T-compensation

Only use flow regime for data processing

Start-up time < 10 minutes

7

ECC SENSOR RESULTS

8

G20 G20DNG MZ70 MIX9

Testing week with all four gasses was used for calibration (=week3)

Only methane numbers are assessed

Full compositions introduce too many errors

The difference between measured and set-point

Is high when starting flow

T-stabilization

Is high when switching gas

Short no-flow period

Shows large scatter

Response time can be fast under ideal conditions

< 5 min for G20→MZ70 transition

ECC SENSOR RESULTSOther two weeks of testing used for validation

Week2 follows set-point MN’s well

Week1 shows large deviations due to fast switching

and some maintenance, that introduced (humid) air

into the fuel lines

Standard deviations of all flows (including transients)

is larger tan for the steady state flows

Typical response times are 10-20 minutes after gas

changes, although when compositions change more

gradually, faster responses can be expected.

9

G20 DNG

DNG MZ70MZ70 DNGTest Type Max

deviationStandard

deviation for all flows

Standard deviation during steady state flow

Week 1 Validation* 15 3.1 2.1

Week 2 Validation 10 2.6 1.8

Week 3 Calibration 14 4.1 2.1

*for week 1, the first day was omitted, because of too many uncertainties in the test.

DUAL-FUEL (CDF AND RCCI) COMPARED TO

SPARK-IGNITED MONO-FUEL COMBUSTION

10

TDC

Spark Ignition

deg aTDC

Natural gas

Combustion

TDC

RCCI – Reactivity Controlled Compression Ignition

deg aTDC

Diesel

Combustion

Natural gas

TDC

CDF – Conventional Dual-Fuel

deg aTDC

Diesel

Combustion

Natural gas

injection window

What are the End-gas

properties for Dual-Fuel?

How are these affected by

the gas Methane Number?

How are these affected by

engine control parameters?

SI (end-gas=1) CDF (end-gas > 1)

diesel

NG/air

End-gas

End-gas

• Combustion starts in center through local

ignition with spark plug

• Flame propagates outward, compressing

the unburned mixture ahead of the flame

• When auto-ignition conditions are

reached in end-gas, knock can occur

• Injection of diesel creates a mixture

of diesel, with entrained air/NG

mixture

• Combustion starts on edge of diesel

jets, under rich conditions

• Combustion propagates outward;

once diesel is consumed end-gas of

NG/air remains

• When auto ignition conditions are

reached, knock can occur in end-gas

RCCI (end-gas >> 1)

End-gas?

• Diesel is injected early such that it

largely premixes with the NG/air

mixture

• Auto ignition occurs in multiple

locations

• Combustion propagates through

diesel/NG/air mixture without sharply

defined flames

• No clear definition of end-gas, only

gradients are present

SIMILARITY OF KNOCK IN DUAL-FUEL

COMPARED TO SPARK-IGNITED ENGINE

Schematic cylinder cross-sections

MEASUREMENT OVERVIEW

Conventional Dual-Fuel (CDF) and

Reactivity Controlled Compression Ignition (RCCI)

Measured engine operating points: see figure

Most extensive dataset at main operating point indicated with the

filled marker.

Selection based on different air/lambda requirement for CDF and

RCCI using the same engine hardware.

Engine operating conditions: see table

Main differences between CDF and RCCI is the lambda

12

SOI [CA aTDC] Lambda [-] BR [%]

CDF -5 to -20 1.2, 1.3, 1.4 70 to 90

RCCI -40 to -100 1.4, 1.9 70 to 90

Diesel injection

timing

Air-fuel eq. ratio

Blend ratio

KNOCK DETECTION AND INDICATORS

13

Knock: uncontrolled end-gas auto-ignition, resulting in excessive pressure gradients

Detection through

Engine block vibrations (acceleration or ‘knock’ sensor)

Cylinder pressure measurement and analysis

Different methods are in use:

PPR - maximum pressure rise rate

MAPO - maximum amplitude of pressure oscillations

used for CDF

IMPO - integral of pressure oscillations

Knock ratio - ratio of pressure oscillation integral before/after reference point

(location of peak pressure)

used for RCCI

High-frequent

oscillations

14

RCCI, MAPO vs Knock ratio (KBKR)

Sharp pressure oscillations seem to be less for RCCI (more ‘volume’ reaction compared to CDF),

therefore, the Knock ratio seems to be a better indicator.

Test

gas

MN

(NPL)

G20 100.0

MZ70 70.5

Mix9 68.9

DNG 84.2

KNOCK DETECTION AND INDICATORS

SELECTED RESULTS CDF, EFFECT OF MN / BR / LAMBDA

15

Same MN, MZ70 is

more sensitive, even

at higher lambda Higher MN

Sensitivity for lambda

(lower global lambda,

so lower lambda in

end-gas for same BR)

Test

gas

MN

(NPL)

G20 100.0

MZ70 70.5

Mix9 68.9

DNG 84.2

SELECTED RESULTS RCCI, EFFECT OF BR / MN

16

• For later diesel injection, less premixing takes place and zones with conditions that promote

knocking (lower local lambda and more diesel) are more likely

• For RCCI combustion, knocking behaviour increases for lower BR (higher end-gas lambda). This

response is opposite to CDF combustion

• It seems that for RCCI combustion, end-gas conditions are more determined by the amount of

diesel and not so much by the end-gas lambda

• For later diesel injection, more concentration differences/gradients are expected as diesel has less

time to mix. As knock indicators show a higher response for later SOI, it seems that diesel rich

areas promote knocking

• Much higher sensitivity of MIX9 compared to other gases not explained yet

Test

gas

MN

(NPL)

G20 100.0

MZ70 70.5

Mix9 68.9

DNG 84.2

CONCLUSIONSECC sensor validation on the engine

The accuracy of the MN measured using the gas quality sensor is 2-3 MN units

MN-knock relation in dual-fuel engine combustion

CDF needs adjusted SOI to achieve efficient heat release timing, similar to spark timing for SI engines

SOI advance is knock limited. Earlier phasing is possible for gases with higher MN

RCCI needs later SOI to achieve the same result in contrast to CDF and SI

Blend ratio (diesel amount) should be used to prevent knock, MN has less influence

Dual-fuel concept provides more than one control parameter to avoid knock: SOI, BR, Lambda

Lambda setpoint may be chosen based on gas quality to run the engine at optimum efficiency

MN sensor application for engine control

Application of a gas quality sensor for CDF enables new optimization strategies. Although sudden variations of gas

quality require a response time much shorter than that of the current sensor to prevent knock, information on gas quality

with a longer response time can be used to optimize the blend ratio and lambda to improve efficiency

17