Available online at www.sciencedirect.

com

Energy Procedia 29 (2012) 138 – 146

World Hydrogen Energy Conference 2012

Heat loss comparison between hydrogen, methane, gasoline

and methanol in a spark-ignition internal combustion engine
Joachim Demuynck*, Michel De Paepe, Ivan Verhaert, Sebastian Verhelst
Ghent University, Sint-Pietersnieuwstraat 41, 9000 Gent, Belgium

Abstract

Understanding the heat transfer mechanism from the combustion gases to the walls inside an internal combustion
engine is key in the search for higher efficiencies, higher power outputs and lower emissions. Current heat transfer
modeling concepts have been reported to be inaccurate for hydrogen engines. To investigate the heat transfer
mechanisms in a hydrogen engine, we have measured the instantaneous heat loss inside a spark-ignition engine at
three locations. To determine the effect of the throttle position, compression ratio, ignition timing and air-to-fuel ratio
in the entire parameter space systematically, techniques of Design of Experiments are applied. The experiment has
been repeated for methane, gasoline and methanol to compare hydrogen with other fuels and to build a database for
the development of a fuel independent heat transfer model. The paper shows that the effect of the engine factors is
similar for all the fuels. However, the heat loss to the cylinder walls of hydrogen is only at the same level of that of
the other fuels for very lean mixtures. The engine efficiency drastically reduces for rich mixtures as a consequence,
indicating that a lean mixture in combination with boosting and external gas recirculation should be used to obtain
high power outputs with high efficiencies for port fuel injected engines.

©
© 2012
2012Published
Publishedby byElsevier
ElsevierLtd.
Ltd.Selection and/or peer-review under responsibility of Canadian Hydrogen and
Fuel Cell Association Open access under CC BY-NC-ND license.
Selection and/or peer-review under responsibility of Canadian Hydrogen and Fuel Cell Association

Keywords: internal combustion engine; heat loss; hydrogen; methanol; Design of Experiments

1. Introduction

It is well known that hydrogen is an interesting energy carrier or buffer because it can be produced in
several renewable ways and used in a wide variety of applications. For transportation, research mainly
focuses on fuel cells. Much less attention has been devoted to internal combustion engines (ICEs) on

* Corresponding author. Tel.: +32-9-2643302; fax: +32-9-2643590.

E-mail address: [email protected].

1876-6102 © 2012 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of Canadian Hydrogen and Fuel Cell Association.
Open access under CC BY-NC-ND license. doi:10.1016/j.egypro.2012.09.018
 Joachim Demuynck et al. / Energy Procedia 29 (2012) 138 – 146 139

Nomenclature

ANOVA analysis of variance

CCD central composite design
CFR Cooperative Fuel Research
CR compression ratio
DoE design of experiments
FCT fully closed throttle
IGN ignition timing
λ air-to-fuel equivalence ratio
RSM response surface method
RTD resistance temperature detector
TP throttle position
WOT wide open throttle

hydrogen, since the ICE is often readily dismissed as a future prime mover, for its low efficiency
(particularly at part load) and pollutant emissions. However, the ICE in itself is a scalable and sustainable
technology, because it is made out of abundantly available and recyclable materials [1]. Furthermore, its
efficiency is still being improved, with reported lab peak efficiencies up to 57% [2]. Consequently, it is
still a possible future prime mover if sustainable fuels are used, hydrogen being an option. The unique
properties of hydrogen indeed enable zero greenhouse gas and near-zero noxious emissions and a high
efficiency throughout the load range [3]. Compared to a fuel cell, a hydrogen-fuelled ICE has the
advantage of a reduced cost, both for the engine and fuel (lower purity needed). Furthermore, the same
ICE can be run on several fuels, enabling a transition from the current fuels to hydrogen.
To enhance the ICE development, computer tools are being developed to simulate the engine cycle. An
important part in these models is the prediction of the heat transfer inside the engine, since it has an effect
on all three improvement targets: the power output, the efficiency and the emissions. The research on the
heat loss in hydrogen engines indicates that the heat loss mechanism differs from other fuels [4, 5] and
that existing models are not capable to capture that difference [6, 7]. To investigate the heat loss
mechanism of hydrogen in comparison with that of other fuels, measurements have been carried out in a
spark ignition engine on hydrogen, methane, gasoline and methanol. This paper investigates the effect of
four engine factors on the heat flux in addition to that of the fuel: the throttle position (TP), the
compression ratio (CR), the ignition timing (IGN) and the air-to-fuel ratio (λ). A systematic variation of
the engine factors is needed [8], so techniques of Design of Experiments (DoE) have been applied.
140 Joachim Demuynck et al. / Energy Procedia 29 (2012) 138 – 146

2. Measurement method

2.1. Measurement equipment

The engine used in this research is a four-stroke single-cylinder spark ignition engine based on a CFR
(Cooperative Fuel Research) engine operated at a constant speed of 600 rpm. A cross section is given in
Fig. 1, showing the possible sensor positions in the cylinder wall. It is equipped with PFI (port fuel
injection) and has a variable compression ratio. Two types of injectors are available in the intake
manifold, one for gaseous fuels and one for liquid fuels. The injection and ignition is controlled by a
MoTeC M4Pro electronic control unit. The compression ratio has to be kept below 10, because the
moving piston would otherwise damage the heat flux sensor. The details of the engine are given in Table
1. It has recently been revised, resulting in a larger bore diameter and different valve timings compared to
older publications.

Fig. 1. Cross section of the CFR engine, P1: spark plug position, P2-P4: possible sensor positions, IV: intake valve, EV: exhaust
valve

Table 1. CFR engine properties

Bore 83.06 mm
Stroke 114.2 mm
Connecting rod length 254 mm
Swept Volume 618.8 cm³
IVO 10 °CA ATDC
IVC 29 °CA ABDC
EVO 39 °CA BBDC
EVC 12°CA ATDC

The heat flux and wall temperature are measured at positions P2, P3 and P4 with a Vatell HFM-7
sensor which consists of a thermopile (heat flux signal, HFS) and an RTD (resistance temperature
detector). Vatell claims that the sensor has a response time of 17μs. The Vatell AMP-6 amplifier was
used as a current source for the RTD and as an amplifier for both output signals. The measurement
positions are at the same height in the cylinder wall and are equally distributed around the circumference
of the cylinder. The spark plug was mounted in position P1.
In-cylinder pressure was measured with a water-cooled Kistler 701A piezoelectric sensor (mounted in
P4 or P2). Inlet and outlet pressure were measured with two Kistler 4075A10 piezoresistive pressure
sensors. The inlet pressure was used to reference the in-cylinder pressure. Gas flows were measured with
Bronkhorst Hi-Tec F-201AC (gas) and F-106BZ (air) flow sensors. Liquid fuel mass flow rate is
 Joachim Demuynck et al. / Energy Procedia 29 (2012) 138 – 146 141

measured gravimetrically. Finally, type K thermocouples were used to measure coolant, oil and inlet and
exhaust gas temperatures.
All the signals were acquired with a National Instruments PXI system. Crank angle resolved signals
(HFM and pressure signals) were acquired synchronously with a PXI-6143 S-series card every 0.5 °CA
(sample rate of 7.2 kHz) during 100 consecutive cycles. The other signals were averaged over time and
acquired with a PXI-6224 M-series card at a sampling rate of 1 Hz.

2.2. Experimental design

The experiment was designed according to DoE methods (Design of Experiments) described in ref.
[9]. The purpose of the DoE is to vary the several factors in a systematic way and to define the minimum
required number of combinations to investigate the heat flux in the entire parameter space, because it was
not possible to run all the combinations.
The peak heat flux to the cylinder walls was the investigated dependent variable. A Response Surface
Method (RSM), more specifically a Central Composite Design (CCD), will be used, so each factor will be
tested at 5 levels (coded as -2, -1, 0, 1 and 2). Since the fuel and measurement position are categorical
factors, they will not be included in the DoE design and the RSM will be repeated for the several levels of
these factors (three times for the different measurement positions and four times for the different fuels).
The levels of IGN and λ are fuel dependent to cover the optimal and widest possible range for each fuel.
A combined wide range of IGN and λ is not possible for hydrogen, due to the occurrence of abnormal
combustion at the extreme combinations (e.g. an early ignition for rich mixtures). Consequently, λ has
been varied at the lean side for hydrogen to cover the widest possible range for IGN. Stoichiometric
measurements will later be conducted to confirm the findings beyond the lean mixtures. An overview of
the extreme levels of the continuous factors is given in Table 2.

Table 2. Overview of the factors’ extreme levels

factor fuel level level code

fully closed throttle (FCT) -2
TP all
wide open throttle (WOT) 2
8 -2
CR all
10 2
38°ca BTDC -2
methane
10°ca BTDC 2
5°ca BTDC -2
hydrogen
-15°ca BTDC 2
IGN
25°ca BTDC -2
gasoline
5°ca BTDC 2
25°ca BTDC -2
methanol
5°ca BTDC 2
1.2 -2
methane
0.8 2
2.2 -2
hydrogen
1.4 2
λ
1.2 -2
gasoline
0.8 2
1.3 -2
methanol
0.7 2
142 Joachim Demuynck et al. / Energy Procedia 29 (2012) 138 – 146

A full factorial CCD, containing all the possible combinations, consists out of three parts. First, all the
possible combinations of the factors at the levels -1 and 1, the so called cubical points, being used to test
the main effects and interactions. Second, the extreme values of a certain factor (-2 or 2) at the center
level of the other factors (0) to test non-linear effects of that certain factor. Third, the replication of the
center point (all factors at level 0) to test the experimental error. Running the full factorial CCD for all the
measurement positions for a certain fuel was not possible on one day, so a fractional factorial design with
a reduction in the number of cubical points needed to be designed.
No information was available in literature on which interactions could be neglected, so the full
factorial experiment was run in one measurement position (P2) on methane to provide this information.
The Analysis of Variance (ANOVA) for the maximum heat flux of that experiment showed that all the
main effects and the quadratic effects of IGN, TP and λ were significant. None of the interactions turned
out to be significant. However, including the variation of the heat flux within the engine cycle (adding the
degree crank angle as a factor), revealed that the first order interactions of IGN could be significant.
Consequently, they were considered in the final design of the fractional factorial design. The degree crank
angle was not kept as a factor in that final design, since the experimental model of the instantaneous heat
flux does not accurately fit the measurements points in the entire parameter space, in contrast to that of
the peak heat flux. To have a clean estimation of all the main effects and the interactions of IGN with CR,
TP and λ, a fractional factorial design of half of all the possible cubical points could be run (8 instead of
16). The full factorial experiment on methane also showed that the effect of CR was linear, so the two
most extreme values were not run for that factor in the fractional factorial design. Consequently, the final
CCD design for each measurement position and fuel consists out of 16 combinations (2 center points, 6
extreme values and 8 cubical points), which means a total of 192 cases (=16x3x4).

3. Results and discussion

The statistical analysis of each CCD experiment results in an experimental surface of the maximum
heat flux over the entire parameter space. The results are very similar for the three measurement
positions, so only the results at one position (P3) will be presented here. None of the interactions between
IGN and the other factors turned out to be significant, so the effect of a certain factor is independent of
the level of the other factors. Consequently, the effect of a certain factor can be plotted at the center level
of the other factors. In each graph plotted in the next sections, the mean value of the plotted line would
change in case the level of the other factors would vary, but not the shape of the line.
In addition to the experimental surface of the maximum heat flux, surfaces can also be obtained for the
indicated work. Fig. 2 plots the effect of the throttle position on the indicated work output for the different
fuels. The left side of the graph (63°) corresponds with the WOT value and the ride side (87°) with the
FCT value. The throttle has not been varied between 0° and 63° because this does not reduce the air flow
and, hence, the work output (because of the low engine speed). Fig. 2 shows that the level of the work
output differs between the fuels over the parameter space, having an influence on the peak heat flux
value. Especially the work output on hydrogen (lean mixtures) can be significantly lower (up to 50%)
than that of the other fuels. For hydrogen, the throttle position only affects the indicated work starting
from 70°, because less air can flow into the engine due to the low density of the fuel. Therefore, to be able
to compare the absolute values of the heat fluxes over the parameter space in addition to the factor effects,
the heat flux has been corrected for the work output in the following graphs.
 Joachim Demuynck et al. / Energy Procedia 29 (2012) 138 – 146 143

Fig. 2 The indicated work differs between the fuels over the parameter space

The effect of the throttle position and compression ratio on the peak heat flux is plotted in Fig. 3. The
left side of the graph shows that the peak heat flux is reduced when the throttle opening is decreased,
because the ingoing mass and, hence, released heat is reduced. The effect for the three fuels is very
similar, with the average heat flux level going up from gasoline over methane and methanol to hydrogen.
For the first three fuels, the heat flux slightly increases before dropping because the turbulence generated
by the throttle counteracts the effect of the reducing ingoing mass, which has also been reported before
[4]. For hydrogen, this effect is not visible in the heat flux trace. The throttle position must have a lower

Fig. 3 The effect of the throttle position and the compression ratio
144 Joachim Demuynck et al. / Energy Procedia 29 (2012) 138 – 146

effect on the turbulence level of the ingoing flow, because of the long fuel injection duration. It is not
plotted here, but the importance of the heat flux significantly increases when the intake flow is throttled.
For methane for example, the total amount of heat losses to the walls, expressed as a percentage of the
energy content in the fuel, increases from 30 to 45% going from WOT to FCT.
The right hand side of Fig. 3 plots the effect of the compression ratio. Increasing the compression ratio
increases the heat flux because the cylinder pressure and, consequently, gas temperature rise. For
gasoline, there does not seem to be an effect of the compression ratio, but this must be caused by a
measurement error. Again, the average heat flux level increases from gasoline over methane and
methanol to hydrogen.
The effects of the ignition timing and air-to-fuel equivalence ratio are plotted in Fig. 4. The left hand
side of the graph shows that of the ignition timing, early ignition timings being at the far left side. The
effect for all the fuels is again very similar (except for gasoline, which must be caused by a measurement
error). The heat flux decreases linearly when the ignition timing is retarded, because the peak gas
temperature is reduced. Hydrogen, again results in the highest heat flux levels.

Fig. 4 The effect of the ignition timing and the air-to-fuel equivalence ratio

The effect of the mixture richness is plotted at the right side of Fig. 4. Gasoline, methane and methanol
have been varied around the stoichiometric air-to-fuel ratio, resulting in a maximum in the peak heat flux
at the rich side. The air-to-fuel equivalence ratio for which the heat flux peaks differs for each fuel. The
maximum in the peak heat flux has not been reached for hydrogen, since only lean mixtures have been
used as explained above. Extra measurements in the center level of the other factors will be run to find
that maximum for hydrogen. It is expected that peak heat fluxes could be 3 times higher than that of the
other fuels. However, for the first time in this paper, Fig. 4 shows that the heat flux level of hydrogen can
be at the same level as that of the other fuels if lean mixtures are used. Earlier results already
demonstrated the negative effect of the soaring heat flux on the engine efficiency for stoichiometric
hydrogen mixtures in the CFR engine [10]. These results are repeated in Fig. 5 to demonstrate that the
indicated efficiency drops from 29 to 23% because the amount of heat loss increases from 23 to 37%
when the air-to-fuel ratio is changed from 2 to 1. The indicated efficiency does not drop as strongly as the
 Joachim Demuynck et al. / Energy Procedia 29 (2012) 138 – 146 145

increase in the heat loss because the combustion efficiency improves. Similar results have been shown on
a production type engine [11], so lean hydrogen mixtures should be used to obtain very high engine
efficiencies for port fuel injected engines. To reach high power outputs while maintaining those very high
efficiencies, supercharging in combination with EGR can be used [12]. This does not apply for direct
injection engines, where stratification could be used to have lean mixtures near the wall.

Fig. 5. The increasing heat loss for stoichiometric operation on hydrogen significantly reduces the indicated efficiency

4. Conclusions

This paper has presented instantaneous heat flux measurements in a spark-ignition engine. The heat
loss of hydrogen was compared with that of gasoline, methanol and methane. The effect of the throttle
position, compression ratio, ignition timing and air-to-fuel equivalence ratio have been investigated for all
the fuels. Design of Experiments was used to determine the minimum required amount of factor
combinations that needed to be run.
The paper showed that the effect of the engine factors were very similar for all the fuels. Only for very
lean mixtures, the heat loss to the cylinder walls of hydrogen turned out to be at the same level as that of
the other fuels. The results also indicated the negative effect of the higher heat losses for richer mixtures
on the engine efficiency. Consequently, to maintain a high engine efficiency throughout the entire load
range, boosted and lean hydrogen mixtures should be used in combination with EGR for port fuel injected
engines.
The presented dataset can now be used, together with that obtained on motored operation [13], to
investigate the possibility of a fuel independent heat transfer model. In contrast to the maximum heat flux
discussed here, the instantaneous heat flux will be investigated for that purpose to include the variation
within the engine cycle.

Acknowledgements

The authors like to acknowledge the suggestions and technical assistance of Koen Chielens and Patrick
De Pue. The research is carried out in the framework of a Ph.D. which is funded by a grant (SB-81139) of
the Institute for the Promotion of Innovation through Science and Technology in Flanders (IWT-
Vlaanderen). The experimental equipment is funded by a Research Grant (1.5.147.10N) of the Research
Foundation - Flanders (FWO). These financial supports are gratefully acknowledged.
146 Joachim Demuynck et al. / Energy Procedia 29 (2012) 138 – 146

References

[1] Abbott, D., Keeping the Energy Debate Clean: How Do We Supply the World's Energy Needs? Proc. IEEE, 2010. 98: p. 42-
66.
[2] Reitz R.D., Hanson R.M., Kokjohn S.L. and D.A., S., Fuel Reactivity Controlled Compression Ignition (RCCI) - A Practical
Path to High-Efficiency, Ultra-Low Emission Internal Combustion Engines, in SAE 2011 High Efficiency IC Engines Symposium.
2011: Detroit, USA.
[3] Verhelst, S. and Wallner, T., Hydrogen-fueled internal combustion engines. Progress in Energy and Combustion Science,
2009. 35(6): p. 490-527.
[4] Demuynck, J., Raes, N., Zuliani, M., De Paepe, M., Sierens, R. and Verhelst, S., Local heat flux measurements in a hydrogen
and methane spark ignition engine with a thermopile sensor. International Journal of Hydrogen Energy, 2009. 34(24): p. 9857-
9868.
[5] Shudo, T. and Nabetani, S., Analysis of Degree of Constant Volume and cooling Loss in a Hydrogen Fuelled SI Engine.
2001. SAE paper 2001-01-3561.
[6] Demuynck, J., De Paepe, M., Huisseune, H., Sierens, R., Vancoillie, J. and Verhelst, S., On the applicability of empirical
heat transfer models for hydrogen combustion engines. International Journal of Hydrogen Energy, 2011. 36(1): p. 975-984.
[7] Shudo, T. and Suzuki, H., Applicability of heat transfer equations to hydrogen combustion. JSAE Review, 2002. 23(3): p.
303-308.
[8] Demuynck, J., De Paepe, M., Sierens, R., Vancoillie, J. and Verhelst, S. Literature review on the convective heat transfer
measurements in spark ignition engines. in 13th European Automotive Congress (EAEC 2011), paper EAEC2011_A46. 2011.
Valencia, Spain.
[9] Berger, P.D., Experimental Design with Applications in Management, Engineering, and the Sciences, ISBN: 0-534-35822-5.
2002, Belmont, USA: Duxbury.
[10] Demuynck, J., De Paepe, M., Sierens, R. and Verhelst, S. Heat transfer comparison between methane and hydrogen in a
spark ignited engine. in World Hydrogen Energy Conference. 2010. Essen, Germany.
[11] Vancoillie, J., Demuynck, J., Sileghem, L., Van De Ginste, M. and Verhelst, S., Comparison of the renewable
transportation fuels, hydrogen and methanol formed from hydrogen, with gasoline - Engine efficiency study. International Journal
of Hydrogen Energy, 2012. 37(12): p. 9914-9924.
[12] Verhelst, S., Maesschalck, P., Rombaut, N. and Sierens, R., Increasing the power output of hydrogen internal combustion
engines by means of supercharging and exhaust gas recirculation. International Journal of Hydrogen Energy, 2009. 34(10): p. 4406-
4412.
[13] Demuynck, J., De Paepe, M., Vancoillie, J., Sileghem, L. and Verhelst, S., Applying Design of Experiments to Determine
the Effect of Gas Properties on In-Cylinder Heat Flux in a Motored SI Engine. SAE Int J Engines, 2012. 5(3).