You are on page 1of 10

Energy 47 (2012) 411e420

Contents lists available at SciVerse ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Combustion characteristics and optimal factors determination with Taguchi


method for diesel engines port-injecting hydrogen
Horng-Wen Wu*, Zhan-Yi Wu
Department of System and Naval Mechatronic Engineering, National Cheng Kung University, Tainan 701, Taiwan, ROC

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 19 April 2012
Received in revised form
8 September 2012
Accepted 10 September 2012
Available online 6 October 2012

This study applies the L9 orthogonal array of the Taguchi method to nd out the best hydrogen injection
timing, hydrogen-energy-share ratio, and the percentage of exhaust gas circulation (EGR) in a single DI
diesel engine. The injection timing is controlled by an electronic control unit (ECU) and the quantity of
hydrogen is controlled by hydrogen ow controller. For various engine loads, the authors determine the
optimal operating factors for low BSFC (brake specic fuel consumption), NOX, and smoke. Moreover, net
heat-release rate involving variable specic heat ratio is computed from the experimental in-cylinder
pressure. In-cylinder pressure, net heat-release rate, A/F ratios, COV (coefcient of variations) of IMEP
(indicated mean effective pressure), NOX, and smoke using the optimum condition factors are compared
with those by original baseline diesel engine. The predictions made using Taguchis parameter design
technique agreed with the conrmation results on 95% condence interval. At 45% and 60% loads the
optimum factor combination compared with the original baseline diesel engine reduces 14.52% for BSFC,
60.5% for NOX and for 42.28% smoke and improves combustion performance such as peak in-cylinder
pressure and net heat-release rate. Adding hydrogen and EGR would not generate unstable combustion due to lower COV of IMEP.
2012 Elsevier Ltd. All rights reserved.

Keywords:
Port-injecting hydrogen
A diesel engine
Taguchi method
Injection timing
EGR percentage
Combustion performance

1. Introduction
A diesel engine is one of the most effective energy conversion
systems. It is widely applied to many power plants, such as a power
source for land vehicles, commercial marine vessels, stationary
power plants, and so on [1]. However, because of the ever stricter
emissions regulation, it is necessary to nd new ways of decreasing
exhaust emissions from engines and on alternative fuels for solving
these crises. Hydrogen is expected to be an ideal alternative for
fossil fuel systems [2] and also considered to meet energy, environment and sustainable development needs [3]. The following
descriptions are the characteristics of hydrogen, such as it gains
much higher ame speed and larger diffusion speed [4] so it can
benet the energy efciency and emissions. In addition, it is a long
term renewable, recyclable and non-polluting fuel since it is
without carbon [5]. The ammability limits of hydrogen vary from
an equivalence ratio of 0.1e7.1 [6], and the engine is hence operated
with a wide range of air/fuel ratio. The lower heating value of
hydrogen is 120 MJ/kg which is higher than that of diesel fuel
(42 MJ/kg). The auto-ignition temperature of hydrogen (858 K) is

* Corresponding author. Tel.: 886 6 2747018x223; fax: 886 6 2747019.


E-mail address: z7708033@email.ncku.edu.tw (H.-W. Wu).
0360-5442/$ e see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.energy.2012.09.027

also higher than that of diesel (453 K) [6]. Hydrogen replacement of


air in real conditions is based on volume. As a result, there will be
good amount of replacement of air [7e9].
Moreover, Rakopoulos et al. [10] indicated that one of the most
signicant advantages that hydrogen might have as a fuel was its
potential for increased second-law efciency, due to fundamental
differences in the mechanism of entropy generation during
combustion with respect to the usual hydrocarbon-based fuels.
Recent years, because of good combustion characteristics of
hydrogen, some scholars used hydrogen to upgrade engine
performance tested on various systems, such as spark-ignited
gasoline engine [11,12] and continuous hydrino thermal power
system [13]. Rakopoulos et al. [10] and Bysveen [14] mixed
compressed natural gas with hydrogen to improve the engine
efciency. For diesel engine, since the auto-ignition temperature
of hydrogen is too high, it is difcultly compressed by direct
injecting into the cylinder. Thus, researchers used hydrogen with
port conducting or port injecting method. Saravanan et al. [15]
inducted hydrogen into a single-cylinder direct injection diesel
engine at the intake port. The results demonstrated that with 30%
hydrogen enrichment, the brake thermal efciency increased from
22.78% to 27.9% while smoke and NOX (only under lean burn
condition) emissions decreased; however, too high hydrogen
enrichment (over 50%) would cause the knock. Moreover, they also

412

H.-W. Wu, Z.-Y. Wu / Energy 47 (2012) 411e420

using the EGR (exhaust gas circulation) (0e25%) with 7.5 lpm
hydrogen ow rate increased the brake thermal efciency, and
signicantly reduced NOX, but the exhaust gas emission
increases by 40% with 20% EGR compared with 0% EGR at full
load [16]. In addition, NOX increased without EGR with 7.5 lpm
hydrogen ow rate. Moreover, James et al. [17] evaluated a series
of commercially available natural gas fuel injectors, originally
designed for heavy-duty diesel application, for use with hydrogen
fuel in an electronic fuel-injected internal combustion engine.
They concluded that performance of the injectors evaluated was
within the necessary tolerances for hydrogen applications with
internal combustion engines. Saravanan et al. [18,19] studied on
port injection of hydrogen and varied injection timing, injection
duration and ow rates of hydrogen for various loads. Their results
showed that using port-injected hydrogen increased engine efciency with a greater reduction in emissions, especially the start of
injection at 5 ATDC. However, NOX increased while the hydrogen
ow rate increased at higher loads. Adam et al. [20] discussed the
performance and emission analysis of hydrogen fueled compression ignition engine with variable water injection timing. The
hydrogen was injected at the intake port at xed injection timing
from 0 CA to 40 CA and constant ow rate of 5 lpm with water
injected from 20 BTDC to 20 ATDC within injection duration of
20 CA and 40 CA. The results indicated that water injection timing
of 20 ATDC and duration of 20 CA had showed better engine
performance due to increased gross indicated work and indicated
thermal efciency.
The experimental design using Taguchi method has been
applied in the industry for nding factors that are necessary for
achieving objectives. Lu et al. [21] used the Taguchi method to
determine the optimum design of natural-circulation solar-waterheater. Wu and Ku [22] used this method for experiments to
acquire the primary optimum setting of the operating parameters
in proton exchange membrane fuel cells (PEMFCs). As Nataraj et al.
[23,24] proved the Taguchi method to be a useful technique for the
simultaneous optimization of several engine parameters and also
for predicting the effect of various design parameters on diesel
emissions.
Furthermore, Saravanan et al. [25] investigated the combined
effect of fuel injection timing, EGR ratio, and fuel injection pressure
in controlling the NOX emission of a stationary diesel engine with
diesel by Taguchis L9 orthogonal arrays. Ganapathy et al. [26]
expressed the view that Taguchis approach based on thermodynamic model application is extended to study the suitability of
different types of biodiesel under research for optimization of
engine performance. Applying orthogonal arrays and reliable
quality Taguchi method reduces the number of experimental runs.
The main advantage of the Taguchi method for optimization is its
use of orthogonal arrays for design simplifying the task of planning
experiments greatly.
From the above papers cited, the use of hydrogen as a fuel in
a diesel engine has been studied; however, the authors found that
most of the experiments with adding hydrogen into the engine
discussed how the ow rate of hydrogen affected engine performance and emissions at a xed EGR ow rate by full-factorial
experiments. In addition, there are barely previous investigations
to study combustion effect and emission of a hydrogen/diesel dual
fuel engine by Taguchi method. Therefore, the authors use the
Energy-Share method [27] (it means the same total fuel energy
input) to investigate the engine emissions on a single-cylinder
diesel engine. Based on this method, brake specic energy
consumption (BSEC) is almost the same at a xed load and engine
speed. As a result, brake specic fuel consumption (BSFC) is discussed instead of BSEC to know how the hydrogen fuel affects the
BSFC of the diesel fuel. As far as the diesel engine, the NOX and

smoke are more important. From the cited literature, adding


hydrogen reduced smoke but adding EGR increased it; in some case
adding hydrogen increased NOX but it decreased with increasing
EGR. Thus, this study then applies the Taguchis orthogonal array L9
(34) to determine the optimum operating condition of hydrogen
injection timings, hydrogen-energy-share ratio and percentage of
EGR for low BSFC, NOX and smoke emission. Moreover, using the
Taguchi method predicts uses 67% less experiment run number.
This research using Taguchi method takes 27 (3 factors  3
levels  3 loads) times, but using full factors experimental method
runs 81 times. The results of this study may be of interest to
engineers attempting to develop optimal hydrogen/diesel dual fuel
engine performance.
2. Experimental setup
The experiments were conducted under various hydrogen
injection timings, hydrogen-energy-share ratios, and percentage of
EGR for various engine loads and at a rated speed of 1800 rpm. Fig. 1
shows a block diagram of the experimental setup used in this study.
The engine used in this study was Kubota RK-125 of a 624 cm3
single-cylinder, water cooled, and four-stroke diesel engine. The
main specications of the engine are listed in Table 1. The test setup
consists of hydrogen injection system, hydrogen cylinder, ame
arrestor, EGR system, and emissions (CO, CO2, HC, NOX, and smoke)
measuring equipment.
The injection system includes a hydrogen injector, an electronic
control unit (ECU) and hydrogen ow controller. Hydrogen is
provided by a hydrogen cylinder through a hydrogen ow
controller in the pipe to a hydrogen injector. Hydrogen ow rate
was measured by using a thermal mass ow controller, GFC-37AVAD. Its precision is 1.5% of full scale. Hydrogen injector used
SYM Ltd. 06416-H5B-000 HE15W which injection pressure was
25 kPa, and voltage range was 8e18 V. The injection timing of the
hydrogen injector is controlled by an ECU, DELPHI 25362590JB,
CPU 8bit/8 MHz.
In the EGR system, the exhaust had to be cooled using the heat
exchanger before it was inducted into the cylinder. The exhaust gas
temperature was kept around 28  1  C by varying the ow rate of
cooling water. Gas ow rate was then monitored by the EGR ow
meter while EGR valve was adjusted to obtain the appropriate EGR
ratio. EGR ow rate was measured by the Dwyer Series DS-300
Standard valves, which was rated at 13.7 bar and 93.3  C. The
cooling gas was separated from the moisture by a separator before
the cooling gas was inducted into the surge tank. A Kistler 6001
pressure transducer was used in conjunction with a charge
amplier, and Kistler 5011B to a data obtainment board was used to
record the cylinder pressure. The crank angle degree signals
required were obtained from a shaft encoder providing a resolution
of 1 crank angle and an accuracy of 0.05 crank angle. The
cylinder gas pressure and the crank angle degree signals were
acquired and processed quickly by analog/digital signal converter
NI PCI-6259 connected with a computer. BOSCH EAM3.011 was
used to measure smoke emission, and RI-803(type T) was used to
measure CO/HC/CO2 emissions. ACHO Physics CLD-60 was for
measuring NOX emission. The measurement uncertainties were
determined and shown in Table 2 using the analysis proposed by
Abernethy et al. [28].
3. Methodology descriptions
The hydrogen-energy-share ratio is a concept of equal input
energy. The hydrogen-energy-share ratio is dened as the ratio of
energy of hydrogen fuel to total energy, which is obtained from the
following equation:

H.-W. Wu, Z.-Y. Wu / Energy 47 (2012) 411e420

413

Fig. 1. Experimental apparatus.

Table 1
The specications of the engine.
Bore  stroke
Displacement volume
Compression ratio
Rated output
Rated speed
Injection pressure of diesel fuel injector
Start of injection
Injection pressure of hydrogen injector

94  90 mm
624 cm3
18
9.2 kW
2400 rpm
21.57e22.56 MPa
21.5 w 23.5 BTDC
250 kPa

Table 2
Measurement uncertainties.
Item

Uncertainty (%)

Pressure
Hydrogen ow rate
NOX
Smoke

0.5
1.5
1.0
2.0

414

H.-W. Wu, Z.-Y. Wu / Energy 47 (2012) 411e420

Hydrogen  energy  share

EH2
EH2 Ediesel

(1)

where EH2 is the lower heating value (MJ/kg) of hydrogen multiplied by the mass rate of hydrogen (kg/min) and Ediesel the lower
heating value of diesel multiplied by the mass rate of diesel.
(EH2 Ediesel) is always constant at each load.
EGR is a NOX reduction technique used in most gasoline and
diesel engines. EGR works by recirculating a portion of an engines
exhaust gas back to the engine cylinders. The denition of EGR ratio
[29] in this study is given as:

EGR%

VEGR

V V V

_ f g=h
m
BPkW

(3)

_ f is the fuel mass ow rate


where BP is the brake power (kW), and m
(g/h) which is based on both diesel and hydrogen.
Net heat-release rate can read the combustion stages and
events, but its detail formation model is complicated. This study
expresses the heat-release model [30] as



dQ
1
dV
dp
pV dk
kp


V
J=CA
k1
dq
dq
dq
k  12 dq

(4)

This net heat-release rate is on the basis of the mass and


energy conversion of the law of thermodynamics for open system,
where dQ/dq is the net heat-release rate (J/CA), q the crank angle
in degrees, and V the volume of the cylinder (m3). k is the specic
heat ratio calculated from the composition and temperature of the
in-cylinder gas [31], treated as a variable, and dk/dq is thus
obtained.
In engine testing, both the air mass ow rate and the fuel mass
ow rate are normally measured. The ratio of these ow rates is
useful in dening engine operation conditions [1]:

Air=Fuel ratioA=F

_a
m
_f
m

(5)

_ a is the air mass ow rate (g/h), and m


_ f is the fuel mass
where m
ow rate (g/h) which is based on both diesel and hydrogen.
The normal operating range for diesel engine is 18 & A/F & 70.
Because of many factors affecting combustion characteristics,
even if engine operates under steady conditions, combustion
characteristics are not same completely from cycle to cycle. If
variation is larger from cycle to cycle, then the engine will operate
unstably and will burn incompletely to affect combustion characteristics of engine. The formula of cyclic variation written by its
coefcient of variations (COV) [32] is as follows:

COV

1
S=N ratio 10log
n

n
X
1
y2
i1 i

where yi is the value of the quality characteristic measured from


the trial, the unit of S/N is dB, and n is the number of the test in
a trial. Using the design of experiment based on orthogonal array,
the parametric levels having the highest S/N ratio decides the
optimum combination of setting. Fig. 2 shows a ow chart of
Taguchi method [23,24] implemented in this study. The Taguchi
process begins dening the goal to identify and quantifying the
parameters. After selecting the design parameters and selecting
the appropriate orthogonal array, experiment is conducted to
obtain data. Executing a statistical analysis using the data of the
experiment nds out the optimum factor combination by
analyzing the response curve and the contributions of the factors.
Predicting the objective responses at the optimum factor combination checks the reproducibility of the results from this experiment. The nal step is to run conrmation experiments to verify
the engine parameter settings and to evaluate the predictive
capability.

4. Results and discussion


This study uses the L9 orthogonal array of the Taguchi method to
nd out the best (A) hydrogen injection angle, (B) hydrogenenergy-share ratio and (C) the percentage of exhaust gas circulation (EGR) in a single DI diesel engine. For various engine loads, this
paper determines the optimal operating factors for low BSFC, NOX,
and smoke. The authors then compare emissions and combustion
performance of the optimum condition factors with those of original baseline diesel engine. Table 3 shows the L9 orthogonal array
for the experiment work, and Table 4 lists the three factors and
their levels. The test results are discussed in the following sections.

s
y
q
P
2
N  1$ N
i 1 yi  y

(7)

(2)

EGR
airthe volume

f
where VEGR
is
of engines exhaust gas per cycle (m3/

cycle), Vair the volume of the intake air per cycle (m3/cycle), and Vf
the volume of the fuel per cycle (m3/cycle).
BSFC [1] is the ratio of fuel mass ow rate of an engine to its
output power given by the following equation.

BSFCg=kW$h

simplied the statistical design efforts by using orthogonal array


and statistical analysis to evaluate experimental data. Taguchi
dened a signal-to-noise (S/N) ratio which is an evaluation of the
stability of performance of an output characteristic. The S/N ratio
takes both average and variation into account. In order to obtain the
lowest emissions, the author used the smaller-the-better expression for the S/N ratio given by [33]

(6)

where COV is the coefcient of variation, s standard deviation, y the


mean value, and N cycle numbers of sample; N is equal to 40 in this
study.
Taguchi methods are statistical methods developed by Taguchi
[33] to improve the quality of manufactured goods. Taguchi

Fig. 2. Flow chart of the Taguchi method.

H.-W. Wu, Z.-Y. Wu / Energy 47 (2012) 411e420


Table 3
Taguchis L9 (34) orthogonal arrays.

Table 5
Contribution of factors for emissions at various loads.

No.

1
2
3
4
5
6
7
8
9

1
1
1
2
2
2
3
3
3

1
2
3
1
2
3
1
2
3

1
2
3
2
3
1
3
1
2

1
2
3
3
1
2
2
3
1

4.1. Statistical analysis


The average emission responses and the S/N ratios for each
factor are computed in the Taguchi process. An analysis of variance
(ANOVA) was performed to identify the most signicant factors and
to quantify their effects on BSFC and emissions such as NOX and
smoke. Table 5 summarizes the contribution of ANOVA results for
emission response at various loads while the negative contribution
is not been listed. Table 6 is the summary for optimal conditions at
various loads, which helps us nd out the optimal factors.
4.1.1. S/N response curves analysis
The S/N response curves are graphical representations of change
in performance characteristics with the variation in process
parameter level [23]. From Figs. 3e5, they show considerable
effects for BSFC and emissions at various loads.
For BSFC, at 30% load, the highest S/N ratio is observed at 10
BTDC (A1), 20% hydrogen (B3) and 40% EGR ratio (C3), at 45% load
A1B3C3, and A3B3C3 at 60% load. It is obvious that the S/N ratio of
factor A vary slightly because the hydrogen ratios are the same with
various hydrogen injection timing. As a result, choosing each level
for factor A causes little variation of BSFC. However, the S/N curves
of factor B vary signicantly at 20% hydrogen (B3). The reason is that
higher hydrogen ratio will replace a large amount of diesel, especially at higher loads. For factor C, EGR will reduce the oxygen
content of air so higher EGR can decrease the brake specic fuel
consumption.
For NOX, at 30%, load, the optimum factors are found to be at 10
ATDC (A3), 5% hydrogen (B1) and 40% EGR ratio (C3), at 45% load
A3B2C3, and A3B1C3 at 60% load. Although factor A and B have been
chosen for the optimum factor; however, from Figs. 3e5, there are
barely slope changes of the S/N curves. On the contrary, the slopes
of the S/N curves are more and steeper with the increase in the EGR
ratio. This is because the formation of NOX is noticeably affected by
EGR technology.
For Smoke emission, at 30% load, it is lowest at TDC (A2), 20%
hydrogen (B3) and 40% EGR (C3), at 45% load A1B3C1, and A3B3C1 at
60% load. There is absence carbon of hydrogen; therefore, the 20%
hydrogen is the optimal selection.
Figs. 3e5 also provide the information for S/N curves of CO, HC
and CO2. At 60% load for CO, it is lowest at A3B3C1. Since hydrogen
does not contain any carbon, using hydrogen produces lower CO.
Furthermore, with an increase in EGR, the concentration of oxygen
will reduce; as a result, lower oxygen concentration leads to

Table 4
Factors with levels.

A
B
C

415

Factors

Level 1

Level 2

Level 3

Hydrogen injections timing


Hydrogen-energy-share ratio
Percentage of EGR

10 BTDC
5%
20%

TDC
10%
30%

10 ATDC
20%
40%

Contribution (%)
Factors
30% load
A
B
C
45% load
A
B
C
60% load
A
B
C

BSFC

NOX

Smoke

e
50.5
49.2

0.64
0.52
97.65

33.68
19.41
34.84

e
84.16
8

0.14
0.62
99.12

e
66.6
26.2

e
89.96
9.71

0.67
e
98.2

e
90.42
8.46

incomplete combustion, so CO increases. For HC, the S/N curve of


factor B is found to increase with decreasing hydrogen ratio. As for
CO2, more hydrogen and less EGR will reduce CO2.
The optimum setting for various characteristics is determined
according to the above statements.
4.1.2. Optimal factors
The contribution of each factor for the quality objects will be
considered to decide the optimized combination. In Table 5, factor
A seems to be small contribution; however, for smoke at 30% load, it
gives almost 30% contributions. Using EGR system is known to
reduce NOX emission due to incoming air with the re-circulated
exhaust gas which dilutes the mixture with inert gas, lowers the
peak gas temperature, and reduces the amount of excess oxygen. As
a result, the factor C has more than 97% contributions listed in
Table 5. Furthermore, at 30% load, the contribution of factor B is

Table 6
Summary table for optimal conditions at various loads.
Factors

Levels

(a) 30% load


A
1
2
3
B
1
2
3
C
1
2
3
(b) 45% load
A
1
2
3
B
1
2
3
C
1
2
3
(c) 60% load
A
1
2
3
B
1
2
3
C
1
2
3

S/N ratio
for BSFC

49.51
49.35
48.76
49.59
49.21
48.82

S/N ratio
for NOX

S/N ratio for


smoke

Overall
optimum
A2

41.98
39.57
35.44

14.83
13.51
13.91
14.60
14.10
13.56
14.81
13.98
13.45

40.18
39.98
39.79
48.11
47.84
47.14

B3

C3

42.91
42.65
42.05

46.94
43.43
37.23

C3

A3
18.75
17.25
15.49

44.41
40.95
34.58

47.57
47.11
46.14
47.18
46.94
46.70

B3

A3

25.47
23.94
20.90
22.72
23.43
24.16

B3

C3

416

H.-W. Wu, Z.-Y. Wu / Energy 47 (2012) 411e420


-32

-48.4

NOx at 30% load

BSFC at 30% load


S/N (dB)

S/N (dB)

-48.8
-49.2
-49.6
-50

-36
-40

-44

A1

A2

A3

B1

B2

B3

C1

C2

C3

A1

-12

A2

B2

B3

C1

C2

C3

B1

B2

B3

C1

C2

C3

B1

B2

B3

C1

24
S/N (dB)

S/N (dB)

B1

CO at 30% load

Smoke at 30% load


-13
-14

22
20

-15

18

-16
A1

A2

A3

B1

B2

B3

C1

C2

A1

C3

A2

A3

-11.6

-34

CO2 at 30% load

HC at 30% load
-12
S/N (dB)

-35
S/N (dB)

A3

26

-36
-37

-12.4
-12.8

-38

-13.2

A1

A2

A3

B1

B2

B3

C1

C2

C3

A1

A2

A3

C2

C3

Fig. 3. S/N response curves for 30% load.

only 15.4%; however, at 45% and 60% loads the contribution of


factor rise to 64.94% and 90.28%, and it means that factor B affects
smoke emission obviously as the load increases.
The optimized combination is obtained through the analysis of
the S/N response curves and the contributions. As a result, Table 6
summarizes the optimum factors setting determined for each
response at various loads. In Table 6, the optimized combination is
determined by choosing the level with the highest S/N ratio and
much more contribution. For example, at 30% load, control factor A

is more signicant on smoke than on other objects; moreover, the


highest S/N ratio for smoke is A3. Factor B is nearly no contribution
to NOX. The S/N ratios for BSFC and for smoke are thus considered,
and the highest S/N ratio is B3. As for factor C, the contribution to
NOX is up to 97.13%, so C3 is conrmed to be the optimal factor.
According to the statements of above, the best combination is
A2B3C3 at 30% load, A3B3C3 at 45% and 60% load. The experiments of
engine emissions were carried out by each optimum factor and
compared with the original baseline diesel engine.

-32

-46.8

NOx at 45% load

BSFC at 45% load


-36
S/N (dB)

S/N (dB)

-47.2
-47.6
-48

-44

-48.4

-48

A1

A2

A3

B1

B2

B3

C1

C2

C3

A1

-14

A2

A3

B1

B2

B3

C1

C2

B2

B3

C1

C2

B2

B3

C1

C2

C3

24
Smoke at 45% load

CO at 45% load

22

-16

S/N (dB)

S/N (dB)

-40

-18

20
18
16
14

-20
A1

A2

A3

B1

B2

B3

C1

C2

A1

C3

-35

A3

B1

C3

-13
HC at 45% load

CO2 at 45% load


S/N (dB)

-36
S/N (dB)

A2

-37
-38

-14

-15

-39

-16

-40
A1

A2

A3

B1

B2

B3

C1

C2

C3

A1

A2

Fig. 4. S/N response curves for 45% load.

A3

B1

C3

H.-W. Wu, Z.-Y. Wu / Energy 47 (2012) 411e420


-34

-45.6

NOx at 60% load

BSFC at 60% load


-38
S/N (dB)

S/N (dB)

-46.2
-46.8
-47.4

-42
-46

-48

-50

A1

A2

A3

B1

B2

B3

C1

C2

C3

A1

-20

A3

B1

B2

B3

C1

C2

C3

B2

B3

C1

C2

C3

CO at 60% load
20

-22

S/N (dB)

S/N (dB)

A2

25
Smoke at 60% load

-24

15
10
5

-26
A1

A2

A3

B1

B2

B3

C1

C2

A1

C3

-35

A2

A3

B1

-14.5

HC at 60% load

-15
S/N (dB)

-37
S/N (dB)

417

-39
-41

CO2 at 60% load

-15.5
-16
-16.5

-43

-17

A1

A2

A3

B1

B2

B3

C1

C2

C3

A1

A2

A3

B1

B2

B3

C1

C2

C3

Fig. 5. S/N response curves for 60% load.

4.2. Conrmation experiments


An essential step in Taguchis optimization technique is to
conduct conrmation experiments for validating the predicted
results. A 95% condence interval (CI) for the predicted mean of
optimum quality characteristic on a conrmation test is thus estimated using the following two equations [34].

v
#
"
u
u
1
1
CI tFa;1;fe Ve

Neff Re

(8)

and

Neff

N
1 TDOF

freedom for the estimate of the mean optimum. After


substituting these values, the condence interval is CI 2.8.
The 95% condence interval of the predicted optimum NOX at
60% load is hence obtained as 36.52  2.8 dB. To test the predicted result, this study conducted a conrmation experiment
with the optimum settings of the process parameters determined
from the analysis. The S/N ratio of the mean NOX emission at 60%
load from the conrmation experiments is 36.12 dB, which falls
within the predicted 95% condence interval. Other optimum
emissions fall within the predicted 95% condence interval, too.
Therefore, the predictions made using Taguchis parameter
design technique agreed adequately with the conrmation
results.

(9)

where Fa;1;fe is the F-ratio required for 100 (1a) percent


condence interval, fe the degree of freedom (DOF) for error, Ve
the error variance, Re the number of replications for the conrmation experiment, and Neff the effective number of replications.
N is the total number of experiments, and TF is the total degree of

BSFC (g/kWh)

400
300
200
100

baseline diesel engine


Optimized diesel engine

0
30%

45%

60%

Load(%)
Fig. 6. Comparison of BSFC between original baseline diesel engine and optimized
engine.

Fig. 7. Comparison of in-cylinder pressure between original baseline diesel engine and
optimized engine.

418

H.-W. Wu, Z.-Y. Wu / Energy 47 (2012) 411e420

5%

70

Heat Release Rate (J/degree)

60
50

COV of IMEP(%)

baseline diesel engine at 30% load


Optimized diesel engine at 30% load
baseline diesel engine at 45% load
Optimized diesel engine at 45% load
baseline diesel engine at 60% load
Optimized diesel engine at 60% load

40

baseline diesel engine


Optimized diesel engine

4%
3%
2%
1%
0%

30

30%

45%

60%

Load(%)

20
Fig. 10. Comparison of COV of IMEP between original baseline diesel engine and
optimized engine.

10
300

0
360

380

400

420

440

baseline diesel engine

460

-10
Crank angle (degree)
Fig. 8. Comparison of net heat-release rate between original baseline diesel engine
and optimized engine.

250
NOx(ppm)

340

Optimized diesel engine

200
150
100
50

4.3. Comparison between baseline and optimized engine results


Comparisons of different combustion performance and emissions between original baseline diesel engine and optimized engine
are discussed in this section, which are shown in Figs. 6e10. Fig. 6
shows the change of BSFC between original baseline diesel engine
and optimized engine. The BSFC is clearly lower with the optimized
engine. For various loads, the BSFC with the optimized engine
reduces at least 14% fuel consumption rates per kilowatt. This
reduction is because inducting the hydrogen to the cylinder
replaces some diesel fuel and the engine is then at leaner equivalence ratio.
Fig. 7 is the variations of in-cylinder pressure between baseline
diesel engine and optimized engine with load. Compared with the
baseline engine, peak pressures in cylinder raise 13%e17% at the
optimized engine. Fig. 8 displays the comparisons of net heatrelease rate between original baseline diesel engine and optimized engine. In this gure, there is higher net heat-release rate for
the optimized engine while there is lower net heat-release rate for
the baseline diesel engine at various loads. The higher peak
in-cylinder pressure and net heat-release rate is due to the faster
combustion in hydrogen than in diesel with EGR. The comparison
of A/F ratios between baseline diesel engine and optimized engine
with load is shown in Fig. 9. Adding hydrogen reduces the

0
30%

60%

Load(%)
Fig. 11. Comparison of NOX emission between original baseline diesel engine and
optimized engine.

consumption of diesel fuel more due to a higher heating value of


hydrogen and the same energy input, so the A/F ratios of the
optimized engine are higher. The lean burn situations become more
efcient for the engine.
Fig. 10 shows the comparisons of COV of indicated mean effective pressure (IMEP) between baseline diesel engine and optimized
engine with load. Problems often occur when COV of IMEP exceeds
about 10% [1]. The COV of IMEP values for all conditions are lower
than 10% (from 0.9% to 2.5%). From Fig. 9 COV of IMEP with EGR and
hydrogen are higher; this is because EGR causes poor combustion
to increase post combustion and partial burning possibly, and thus
leads to more unstable engine operation. Furthermore, hydrogen
causes fast combustion. However, these are still small cyclic variations. As a result, adding hydrogen and EGR by using energy-share
ratio method would not generate unstable combustion.
In Fig. 11, NOX emission is effectively inhibited with the optimized engine. At 60% load, the reduction rates are up to nearly 66%.

60

25
baseline diesel engine

baseline diesel engine


Optimized diesel engine

20
smoke(%)

50
A/F ratio

45%

40
30

Optimized diesel engine

15
10
5
0

20
30%

45%

60%

Load(%)
Fig. 9. Comparison of A/F ratios between original baseline diesel engine and optimized
engine.

30%

45%

60%

Load(%)
Fig. 12. Comparison of smoke emission between original baseline diesel engine and
optimized engine.

H.-W. Wu, Z.-Y. Wu / Energy 47 (2012) 411e420

Similarly, in Fig. 12, the smoke emission with the optimized engine
reduces a lot compared with baseline diesel engine. At 45% load, the
reduction rate reaches to about 53%. The results indicate that for the
optimized engine there is signicant reduction in both NOX and
smoke emissions.
5. Conclusions
After obtaining the optimal operating factors for low BSFC, NOX,
and smoke of diesel/hydrogen mixtures with Taguchi method, and
comparing BSFC, NOX, smoke, and combustion performance between
the optimized engine and baseline diesel engine, the authors
conclude on the basis of the results and discussion as follows.
The Taguchi method was found to be a good method to nd
optimum combination for low BSFC, NOX, and smoke in a hydrogen/
diesel dual fuel engine. The predictions using Taguchis parameter
design technique agreed adequately with the conrmation results
on 95% condence interval. Using Taguchi method saves 67%
experiment run number in this research.
According to the analysis of S/N curves and ANOVA, hydrogenenergy-ratio affects BSFC and smoke more than other factors;
EGR affects NOX signicantly; hydrogen injections timing does not
have important effects on the objectives.
Compared with original baseline diesel engine, for BSFC, the
higher hydrogen ratio replaces much more amount of the diesel
fuel, and higher EGR reduces more oxygen content of air, so the best
combination can reduce BSFC at least 14.52% at various loads. For
NOX, because of the EGR technology, the best combination can
reduce more than 60.5% at various loads. For smoke, since there is
absence carbon of hydrogen in spite of EGR causing more smoke, the
best combination can still reduce at least 42.28% at various loads.
The best combination also improves combustion performance
as a result of the faster combustion in hydrogen than in diesel with
EGR. For in-cylinder pressure, it raises 13%e17% peak in-cylinder
pressure in the optimized engine. Net heat-release rate is higher
at the optimized engine than that at the baseline diesel engine at
various loads. For COV of IMEP, the best combination does not cause
unstable combustion.
Acknowledgment
The authors gratefully acknowledge the support of the National
Science Council of Taiwan, ROC, under grant NSC 98-2221-E-006259-MY3. Also, the authors would like to thank Dr. M Nataraj who
provided valuable papers.
Appendix
Nomenclature

A/F
ANOVA
BP
CA
CI
COV
DOF
E
EGR
F
IMEP
k
_
m

air/fuel ratios
analysis of variance
brake power, kW
crank angle, degrees
condence interval
coefcient of variance, %
degree of freedom
energy, J
exhaust gas recirculation, %
F-ratio which is a test statistic for multiple independent
variables
indicated mean effective pressure, bar
specic heat capacity, J kg1 K1
mass ow rate, g h1

N
n
NOX
p
ppm
Q
R
Re
rpm
S/N
T
TDOF

419

q
s

total number of experiments


number of the test in a trial
nitric oxides, ppm
pressure, bar
part per million, mg L1
heat source, J
universal gas constant, J mol1 K1
number of replications for the conrmation experiment
revolutions per minute
signal-to-noise ratio, dB
temperature, K
the total degrees of freedom for the estimate of mean
optimum
volume per cycle, m3/cycle
volume, m3
error variance
value of the measured quality characteristic
mean value
probability of committing a possible error in the
estimation
crank angle, degrees
standard deviation

Subscript
air
e
eff
f
H2
i

air
error
effective
fuel
hydrogen
species

V0
V
Ve
y
Y

References
[1] Heywood B. Internal combustion engine fundamentals. New York, U.S.A:
McGraw-Hill Book Company; 1988.
[2] Soid SN, Zainal ZA. Spray and combustion characterization for internal
combustion engines using optical measuring techniques e a review. Energy
2011;36:724e41.
[3] Li YL, Chen HS, Zhang XJ, Tan CQ, Ding YL. Renewable energy carriers:
hydrogen or liquid air/nitrogen? Appl Thermal Eng 2010;30:1985e90.
[4] Gomes Antunes JM, Mikalsen R, Roskilly AP. An investigation of hydrogenfuelled HCCI engine performance and operation. Int J Hydrogen Energy
2008;33:5826e8.
[5] Kawahara ETN, Piao Z, Fujita S. Hydrogen combustion and exhaust emissions
ignited with diesel oil in a dual fuel engine. SAE TECH paper 2001-01-3503;
2001. USA.
[6] White CM, Steeper RR, Lutz AE. The hydrogen-fueled internal combustion
engine: a technical review. Int J Hydrogen Energy 2006;31:1292e305.
[7] Hari Ganesh R, Subramanian V, Balasubramanian V, Mallikarjuna JM, Ramesh A,
Sharma RP. Hydrogen fueled spark ignition engine with electronic controlled
manifold injection e an experimental study. Renew Energy 2008;33:1324e33.
[8] Subramanian V, Mallikarjuna JM, Ramesh A. Effect of water injection and
spark timing on the NO emissions and combustion parameters of a hydrogen
fuelled spark ignition engine. Int J Hydrogen Energy 2007;32:1159e73.
[9] Subramanian V, Mallikarjuna JM, Ramesh A. Intake charge dilution effects on
control of NO emission in a hydrogen fueled SI engine. Int J Hydrogen Energy
2007;32:2043e56.
[10] Rakopoulosb CD, Scotta MA, Kyritsisa DC, Giakoumis EG. Availability analysis
of hydrogen/natural gas blends combustion in internal combustion engines.
Energy 2008;33:248e55.
[11] Wang S, Ji C, Zhang J, Zhang B. Comparison of the performance of a sparkignited gasoline engine blended with hydrogen and hydrogen-oxygen
mixtures. Energy 2011;36:5832e7.
[12] Yuksel F, Ceviz MA. Thermal balance of a four stroke SI engine operating on
hydrogen as a supplementary fuel. Energy 2003;28:1069e80.
[13] Mills Randell L, Zhao Guibing, Good William. Continuous hydrino thermal
power system. Appl Energy 2011;88:789e98.
[14] Bysveen M. Engine characteristics of emissions and performance using
mixtures of natural gas and hydrogen. Energy 2007;32:482e9.
[15] Saravanan N, Nagarajan G. An experimental investigation of hydrogenenriched air induction in a diesel engine system. Int J Hydrogen Energy
2008;33:1769e75.
[16] Saravanan N, Nagarajan G. Experimental investigation on a Di dual fuel engine
with hydrogen injection. Int J Energy Res 2009;33:295e308.

420

H.-W. Wu, Z.-Y. Wu / Energy 47 (2012) 411e420

[17] Heffel James W, McClanahan Michael N, Norbeck Joseph M. Electronic fuel


injection for hydrogen fueled internal combustion engines, SAE TECH paper,
vol. 981924; 1998.
[18] Saravanan N, Nagarajan G, Dhanasekaran C, Kalaiselvan KM. Experimental
investigation of hydrogen port fuel injection in Di diesel engine. Int J
Hydrogen Energy 2007;32:4071e80.
[19] Saravanan N, Nagarajan G. Performance and emission studies on port injection of hydrogen with varied ow rates with diesel as an ignition source. Appl
Energy 2010;87:2218e29.
[20] Adnan R, Masjuki HH, Mahlia TMI. Performance and emission analysis of
hydrogen fueled compression ignition engine with variable water injection
timing. Energy 2012;43:416e26.
[21] Lu SM, Li YCM, Tang JC. Optimum design of natural-circulation solar-waterheater by the Taguchi method. Energy 2003;28:741e50.
[22] Wu HW, Ku HW. The optimal parameters estimation for rectangular cylinders
installed transversely in the ow channel of PEMFC from a three-dimensional
PEMFC model and the Taguchi method. Appl Energy 2011;88:4879e90.
[23] Nataraj M, Arunachalam VP, Dhandapani N. Optimizing diesel engine
parameters for emission reduction using Taguchi method: variation risk
analysis approachepart I. Indian J Eng Mater S 2005;12:169e81.
[24] Nataraj M, Arunachalam VP, Dhandapani N. Optimizing diesel engine
parameters for emission reduction using Taguchi method: variation risk
analysis approachepart II. Indian J Eng Mater S 2005;12:505e14.

[25] Saravanan N, Nagarajan G, Sampath S. Multi response optimization of NOX


emission of a stationary diesel engine. Fuel 2010;89:3235e40.
[26] Ganapathy T, Murugesan K, Gakkhar RP. Performance optimization of Jatropha biodiesel engine model using Taguchi approach. Appl Energy 2009;86:
2476e86.
[27] Geo VE, Nagarajan G, Nagalingam B. Studies on dual fuel operation of rubber
seed oil and its bio-diesel with hydrogen as the inducted fuel. Int J Hydrogen
Energy 2008;33:6357e67.
[28] Abernethy RB, Benedict RP, Dowdell RB. Asme measurement uncertainty.
Trans ASME J Fluids Eng 1983;107:161e4.
[29] Abd-Alla GH. Using exhaust gas recirculation in internal combustion engines:
a review. Energy Convers Manage 2002;43:1027e42.
[30] Shudo T, Suzuki H. Applicability of heat transfer equations to hydrogen
combustion. JSAE Rev 2002;23:303e8.
[31] Ferguson CR. Internal combustion engine fundamentals: applied thermoscience. New York, U.S.A: John Wiley & Sons, Inc; 1986.
[32] Karim GA, Al-Alousi YH. Some considerations of cyclic variations in spark
ignition engines fuelled with gaseous fuels, SAE TECH paper, vol. 840232;
1984.
[33] Genichi Tauchi. Introduction to quality engineering, whiter plains. New York,
U.S.A: Kraus International Publications; 1986.
[34] Ross PJ. Taguchi techniques for quality engineering. New York, U.S.A:
McGraw-Hill; 1996.

You might also like