Combustion Engines Finland

1. FromFuel Properties to
Engine Performance
Michal Wojcieszyk
21 March 2019

2. 2
Agenda
1. ADVANCEFUEL project in brief
2. Alternative fuels and their properties vs light-duty engine
performance
3. Approach and data selection
4. Modeling methodology
5. Example of results
6. Fuel Blend Property Calculator
7. Key observations

3.  Part of EU Horizon 2020
 Coordination and Support Action of EU Commission
 Facilitating market roll-out of advanced liquid biofuels in
transportation sector between 2020 and 2030 and beyond
3
Partners: Stakeholders:

4. 4
Importance of the research
• Increased market acceptance and end-use of renewable fuels
• Support for decision makers and fuel producers
• Assessment of future potential of alternative fuels
• Providing market stakeholders with state-of-art knowledge and
sophisticated, user-friendly tools with integrated calculators,
standards, and recommendations.

5. Structure of the problem
5

6. 6
Analyzed SI fuels:
• Gasoline
• Methanol (CH3OH)
• Ethanol (C2H5OH)
• Butanol (C4H9OH)
Analyzed CI fuels:
• Diesel
• HVO
• FAME (biodiesel)
• BTL/GTL
• DME
Selected SI input properties:
• Octane Number
• Heat of Vaporization
• Net Calorific Value
• Auto Ignition Temperature
• Carbon content
Selected CI input properties:
• Net Calorific Value
• Cetane Number
• Density
• Viscosity
• Oxygen content
• Carbon content
Alternative fuels and properties

7. Driving conditions vs fuel properties – CI
engine

8. Engine data selection
NEDC

9. 9
• Character of the data: multi input, single output
• Approach: data-driven black-box modeling
• Mathematical methodology: multilinear regression
• Validation: residual analysis and cross-validation
• Input and output parameters represented as relative (%) changes in
reference to standard diesel/gasoline.
Modeling methodology
= + + + ( )
- fuel consumption [%]
X – alternative fuel concentration [%]
A(X)...D(X) – fuel property [%]
a...d – model coefficients.

10. How to select input properties ?
• Properties that are measured
in the literature sources.
• Statistical significance
analysis (t-test, p-value for t-
test). All input properties
have to be mathematically
significant and justified.

11. Model proposal – CI fuel consumption (FC)
2
– Fuel Consumption
– Density
– Viscosity
– Cetane number
– Lower heating value
2 – Oxygen content
[Units – % changes]
Note: Model under
continuous development!

12. 12
Model proposal – SI fuel consumption (FC)
Note: Model under continuous development!

13. Fuel Blend Property Calculator - FBPC

14. 14
Key observations
• Engine performance is influenced both by fuel properties and driving
conditions.
• Driving cycle (WLTC, NEDC) approach for light-duty engines seems to be the
most suitable one from the end-user point of view.
• While studying an effect of fuel properties on engine performance, the black-
box modeling was applied and multilinear regression executed.
• Developed models both for SI and CI case represent the impact of fuel
properties on engine performance with high accuracy (FC and CO2). Models
are under continuous development.
• Fuel consumption of alternative fuel or its blends can be predicted based on
known set of fuel properties. It is highly dependent on NCV, density and CN
in CI case. NCV, density, oxygen content and RON matter for SI engine.

15. Thank you for your attention !
Michal Wojcieszyk
michal.wojcieszyk@aalto.fi
Yuri Kroyan
yuri.kroyan@aalto.fi
Martti Larmi
martti.larmi@aalto.fi
• Energy Conversion Research Group https://vimeo.com/321946937