OIL & GAS An accurate ‘octane number’ for LNG as a ... · - octane number, cetane number,...

Preview:

Citation preview

DNV GL © 2017

Ungraded

Thursday, April 6, 2017 SAFER, SMARTER, GREENERDNV GL © 2017

Howard Levinsky

Thursday, April 6, 2017

Ungraded

OIL & GAS

An accurate ‘octane number’ for LNG as a transportation fuel

1

Sander Gersen, Martijn van Essen, Gerco van Dijk and Howard Levinsky

DNV GL Oil&Gas and University of Groningen, the Netherlands

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Acknowledgement

We thank the TKI program of the Dutch government for financial support. We gratefully

acknowledge the participants in this project for their contributions: Shell and ENGIE for financial

support and their input regarding LNG in developing the algorithms, and Wärtsilä and FPT

Industrial for providing data, engine platforms and invaluable advice regarding engine operation.

2

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

LNG as a transportation fuel; Why is composition important?

3

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

LNG as a transportation fuel

LNG growing as transportation fuel

Lower emissions (no sulfur, low NOx, no soot) attractive marine fuel

+ Very low noise emissions: also attractive as truck fuel

Relatively low CO2 emissions compared to other fossil fuels

Abundant supply internationally

4

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

LNG has significant quality variations

5

LNG composition varies around the world.

Composition also changes with time due to ‘boil-off’ of lighter components

(N2, CH4)

But the variation in composition is less important than the resulting

variations in properties.

Market for traditional liquid fuels

(gasoline, diesel, HFO, etc.) has

already decided how to characterize

the fitness for purpose for the end

user:

- octane number, cetane number, etc.

Does not yet exist for LNG

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

The quality of commercially available LNG varies significantly across the globe

6

35.0 37.0 39.0 41.0 43.0 45.0

Australia – NWS

Australia – Darwin

Algeria – Skikda

Algeria – Bethioua

Algeria – Arzew

Brunei

Egypt – Idku

Egypt – Damietta

Equatorial Guinea

Indonesia - Arun

Indonesia – Badak

Indonesia –…

Libya

Malaysia

Nigeria

Norway

Oman

Peru

Qatar

Russia – Sakhalin

Trinidad

USA - Alaska

Yemen

Lower calorific value, MJ/m3(n)

17% variation32 point variation in

Methane number

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

LNG and its compositional variations

7

Variations bring differences in things like heating value, but also in knock resistance of

fuel…knock can (seriously) compromise engine performance

Essential for optimum engine performance to match fuel with engines

Need to characterized the knock resistance accurately

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Benefits for industry

8

Accurate knock prediction benefits:

– allows engine manufacturers to size and tune engines for a given range of LNG compositions

more precisely

– lowers CAPEX and OPEX for engine owner, and improves engine availability and safety

– Fuel suppliers can manage LNG quality more precisely to meet market demands, by

assuring that no LNGs are excluded from the market or overtreated without cause.

Better prediction will thus help the adoption of LNG as a cost-effective, clean and reliable

transportation fuel.

Dedicated knock algorithm for a specific engine type allows incorporation in a control system

to maximize knock-free engine performance for a wide range of fuel compositions.

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

TKI LNG project

1. Characterize the knock resistance of LNG fuels for three engines, a spark-ignited lean-burn

CHP engine, a dual-fuel engine used in the maritime sector and a stoichiometric truck engine.

2. Quantify possible differences in response to variations in LNG composition.

3. Use results in international discussions regarding standardization, but also for benefit of

individual parties to assess the risk of engine knock for their own engine/fuel combinations.

9

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Engine types selected for this study

10

CHP

E2876LE302

MAN

Marine

W34DF

Wärtsilä

Truck

FPT Cursor 9

Combustion system -mono-gas

-open chamber

-lean burn

-spark ignited

-dual fuel

-pilot injection

-ultra lean burn

- mono-gas

- open chamber

- stoichiometric burn

- spark ignited

Rated power 200kW

(6 cylinders)

3MW

(6 cylinders)

280 kW

(6 cylinders, in line)

CR 11 12 12

rpm 1500 750 600-2400

Bore*Stroke 128*166 340*400 117*135

Air factor 1.55 2 1

Intake manifold

pressure

2.07 3.5 -

BMEP 13 bar 22 bar 24 bar

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Recalling engine knock

Normal combustion vs. knocking combustion

Normal combustion end gas is consumed by propagating

flame front

t combustion < t autoignition

End gas

Unburned mixture

Burned mixture

Spark plug

Far end of combustion chamber

autoignition reactions in end gas

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Recalling engine knock

Knock is autoignition of end gas

competition between propagating flame front and autoignition reactions in end gas

Impacted by changes in fuel composition

autoignition chemistry, burn rate and heat capacity

Normal combustion vs. knocking combustion

Knocking combustion end gas spontaneously ignites

t autoignition < t combustion

End gas

Unburned mixture

Spark plug

Far end of combustion chamber

autoignition reactions in end gas

autoignition

Burned mixture

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Engine Knock model

DNV GL approach: understanding and describing

(changes) in end-gas autoignition process with varying fuel

gas composition

– based on combustion properties rather than purely on

empirical methods

– calculate end-gas autoignition during burn cycle with

varying conditions rank fuels for knock

– Incorporates changes in:

– autoignition chemistry (>2000 reactions/300 species)

– Burn rate (computed SL), and

– thermophysical properties (heat capacity)

13

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Ignition chemistry: verify/alter using RCM autoignition studies

Rapid Compression Machine

- test and optimize chemical mechanism used in knock model

- revealing ‘rules’ on autoignition behavior

0

5

10

15

20

25

30

35

40

0 10 20 30 40 50

Pre

ssure

(b

ar)

Time (ms)

Measured and computed RCM pressure traces

Measurement

Simulation

Mechanical compression

Autoignition delay time

Compressionend pressure

Autoignition event

Pure fuelsCH4, C2H6, n-C4H10, i-C4H10 and H2

Binary mixtures of CH4 with...

...C2H6, C3H8, n-C4H10, i-C4H10, n-C5H12, i-C5H12, neo-C5H12, H2, CO, N2 and CO2

Ternary mixtures of CH4

with......H2 and CO

Pipeline fuels Dutch natural gas

The building blocks of the model are supported by a stringent testing program, verified with experimental data

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Verify model predictions: phasing and knock measurements (KLST) in CHP engine

Make - type MAN - E2876LE302

Rated powerRated speed

208 kW1500 rpm

Bore x strokeC.R.

128 x 166 mm11 : 1

Configuration

6 cyl. in-line, turbocharged-intercooled

Combustion spark ignition, open chamber, lean-burn

Gas composition

- on-stream adjustment- flow-independent- verification w. gas chromatography

Source gas streams

- max. 6- calibrated for Dutch natural gas/CH4, C3H8, H2, CO, CO2, N2, but other

fuels possible

- flow range 150 to 400 m3s/h

p/stream

engine blending station

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

How well can the traditional tools predict engine knock? (SAE Int. J. Fuels Lubr. 9(1):2016)

16

For same MN, cannot distinguish KLST of ±2°CA, constant KLST spread ~14-15 points

Spread is far outside the experimental uncertainty

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Sharper scalpel: DNV GL method (SAE Int. J. Fuels Lubr. 9(1):2016)

Simulate burn cycle, under conditions w/wo knock, verify

using engine measurements.

Use methane/propane scale to rank gases for knock

(Propane Knock Index, PKI) and convert to 0-100 MN scale.

Easy-to-use computational algorithm (digital product)

Compare with measurements in engine

17

Set to 90% burn by increasing initial temperature in simulations for test fuel and compute propane equivalent under same conditions.

0

20

40

60

80

100

120

5 7 9 11 13 15 17 19

Pre

ssu

re, b

ar

Crank angle timing, ms

Measured

Simulation (non-knocking)

Simulation (knocking)

Autoignitiondelay time

Fixed referencetiming

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Results: KLST vs. PKI Methane Number (SAE Int. J. Fuels Lubr. 9(1):2016)

18

Results are within experimental uncertainty, much better

characterization of impact of fuel composition on knock.

Not only for alkanes, but also for renewable components,

H2, CO and CO2

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Other engine platforms?

The analysis of the physics and chemistry

related to engine knock shows that the

different platforms can be grouped based on

combustion behavior

19

Dual fuel engine

Spark ignited engine

Autoignition calculations (P=120bar, =2)

Crossing lines indicates that ranking of fuels can depend on engine conditions

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Modeling results Dual-Fuel engine(Marine engine)

20

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Properties Dual-Fuel engine

21

CHP

E2876LE302

MAN

Marine

W34DF

Wärtsilä

Truck

FPT Cursor 9

Combustion system -mono-gas

-open chamber

-lean burn

-spark ignited

-dual fuel

-pilot injection

-ultra lean burn

- mono-gas

- open chamber

- stoichiometric burn

- spark ignited

Rated power 200kW

(6 cylinders)

3MW

(6 cylinders)

280 kW

(6 cylinders, in line)

CR 11 12 12

rpm 1500 750 600-2400

Bore*Stroke 128*166 340*400 117*135

Air factor 1.55 2 1

Intake manifold

pressure

2.07 3.5 -

BMEP 13 bar 22 bar 24 bar

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Modeling burn rates: variable ignition timing for diesel pilot

22

Determination of ignition timing:

Autoignition Diesel at 167 CA

for basis LNG (derived from

measured heat release profile)

Temperature and pressure at 167

CA is 788K and 58bar

Simulation of LNG + 10%

propane at 167 CA ignition

Peak pressure too high and too

early

Simulation of LNG + 10%

propane at 788 K

Heat capacity of LNG/air

mixture impacts ignition

timing and phasing

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Differences in knock behavior between engine types

23

PKI MN is roughly 6 points higher for the dual-fuel engine at 14.5% ethane in methane than for the spark-ignited engine

Maximum difference in methane number found between the spark-ignited and diesel-pilot engine for these LNG composition is 4.5 MN points

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Can we neglect pentanes?

Simulate compositions with varying fractions of pentane

isomers in spark-ignited engine.

1% n-pentane decreases MN by 20 points

0.15% pentane (as seen in some GIIGNL data) gives

difference of 2.5 MN points…too large to neglect.

Therefore, included in the algorithm.

24

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Outlook 2017

Significant differences in ranked knock resistance between spark-ignited and diesel-pilot engines;

both lean burn. What about different stoichiometry? Examine stoichiometric truck engine.

Truck engine (FPT Cursor 9) :

– Truck engine tests with a different fuel compositions (phasing and knock experiments)

– Modeling Truck engine

– Development dedicated PKI MN algorithm truck engine

Compare the results for the different engine platforms for a wide variety of LNG compositions, to

quantify the possible differences in ranking of fuel compositions.

25

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

Conclusions

More accurate characterization of knock resistance: based on changes in physics and chemistry.

within the experimental uncertainty, including the effects of renewable fuels.

Dual-fuel engine gives a different ranking from the spark-ignited engine.

caused by impact different regimes of temperature and pressure on physics and chemistry.

Expect different stoichiometry (truck engine) will also respond differently.

‘simply’ using one method to characterize knock may not be enough.

Results can serve as input for international discussions on standardization, but can also be used to

assess the risk of engine knock for individual engine/fuel combinations.

Propane-based scale allows testing (new) engines using methane-propane mixtures, rather than

complex fuel compositions.

The algorithm(s) can be used for feed-forward fuel-adaptive engine control to optimize engine

performance for a wide range of fuel compositions.

Together with Shell demonstrated 6% improved fuel efficiency in spark-ignited engine

26

DNV GL © 2017

Ungraded

Thursday, April 6, 2017

SAFER, SMARTER, GREENER

www.dnvgl.com

A ‘correct’ octane number for LNG

Technical background of the TKI project

27

Howard Levinsky

Howard.Levinsky@dnvgl.com

+31-50-7009739

Recommended