150

CHAPTER 7

OPTIMIZATION OF ENGINE PARAMETERS USING

DESIGN OF EXPERIMENTS

7.1 INTRODUCTION

The performance characteristics can be improved with biodiesel by

re-designing the compression ratio, injection system (pressure and timing) and
determining the optimum biodiesel blend. To develop a good injection
system, a parameter search to determine the influence of design parameters on
both the performance and exhaust emissions should be performed. However,
when this parameter search is executed experimentally, it involves a huge
expenditure of money and time. For this reason, the optimization of
parameters by simulations based on limited experimental data followed by
optimization on a computer are sparingly useful. In the past, Banapurmath et
al. (2008) Alonso et al. (2007) and Pandian et al. (2009) used the most
common optimization techniques for engine analysis such as response surface
method, grey relational analysis, non linear regression and genetic algorithm.
Maheswari et al. (2011) used non linear regression and found out optimum
value to be 13% biodiesel –diesel blend with an injection timing of 24°bTDC.
Win et al. (2005) used response surface methodology to optimize the
parameter such as load, speed and static injection timing of a diesel fueled CI
engine to reduce noise, fuel consumption and exhaust emissions. Ganapathy
et al. (2009) developed a thermo dynamic model of biodiesel engine in
combination with Taguchi optimization approach to determine the optimum
engine design and operating parameters and found out the performance of
 151

diesel engine using jatropha biodiesel as fuel. Jinlin Xue et al. (2011)
predicted the performance and exhaust emissions of blended fuels using
artificial neural network (ANN). Celik et al. (2005) observed that many
research about optimization and modification on engine and methodology for
measurements should be performed when petroleum diesel is substituted
completely by biodiesel. Many researchers indicated the need of research in
the areas of engine modifications so as to suit to higher blends without severe
drop in performance so that the renewability advantages along with emission
reduction can be harnessed to a greater extent.

From the above discussions, it is observed that lot of research work

carried out to study engine operating characteristics as a variable and
parameters to find out optimum for emissions and efficiency. The use of
modern computing technique to study the performance and emissions are
limited as Taguchi method, Response surface methodology and Regression
analysis and the predicting optimum performance of bio fuel is needed in a
complex situation. The present study focus on ways to eliminate the lacuna on
from the literature review to find out a simplified approach to get engine
operating characteristics, performance and emissions.

The first part of the chapter deals with the optimal selection of
these parameters based on sixteen test runs conducted using S/N ratio of
taguchi method. Next part of this study using multi object optimization, the
optimal combination was found out to minimize the BSFC, emissions and to
maximize the BTE. The second part deals with optimization of various
parameters using RSM. Third part of chapter deals with multiple regressions
to find out the equation to predict the emission and performance.From the
empirical model developed to find out the responses a C++ PROGRAM was
written to find an optimal solution
 152

Finally Validation is done between optimized results obtained and

actual confirmation results.

7.2 OBJECTIVE

i) To carry out multi-objective optimization on thermal

performance and engine exhaust emissions using RSM and
Taguchi technique and to find the optimum operating
conditions that gives maximum efficiency and minimum
emissions.

ii) To develop a model equation using multiple regression.

These mathematical equations are able to predict the engine
performance and emissions without doing experiments.

iii) Validation of the simulated values with experimental results

7.3 MATERIALS AND METHODS

A single cylinder 5.2 kW water cooled direct injection diesel engine

has a bore of 87.5 mm with an eddy current type dynamometer for loading is
selected for experimental study.

7.3.1 Taguchi Method

7.3.1.1 Taguchi procedure

The Taguchi method provides simple and effective solutions for

investigating the effect of parameters on the performance as well as in the
experimental planning. In this method, the signal –to-noise (S/N) ratio is used
to represent a performance characteristic and the largest value of the S/N ratio
is required. There are three types of S/N ratios-the lower-the-better, the –
higher-the better and the-more-nominal –the -better
 153

The criteria for optimization of the response parameters was based

on the smaller the better S/N ratio.

1 r 2 
S / N = −10 log ∑ y i  (7.1)
 r i=1 

yi represents the measured value of the response variable- i.

The S/N ratio with a higher –the-better characteristic can be expressed as

1 r 
S / N = −10 log ∑1 / y i2  (7.2)
 r i=1 

yi represents the measured value of the response variable.

The negative sign is used to ensure that the largest value gives an
optimum value for the response variable and therefore robust design.

7.3.1.2 Setting optimum conditions and prediction of response

variables

The next step in DOE analysis is determining optimal conditions of

the control parameters to give the optimum responses. In this work the
response variables to be optimized were BTE, has to be maximized and
B.S.F.C. and Emissions to be reduced as much as possible. Hence the
optimum parameter settings will be those that give maximum values of the
BTE and minimum values of BSFC, HC, and NOx. The optimum settings of
the parameters were achieved from the S/N tables of the control parameters.

The optimum value of response variable can be predicted using the

additivity law.
 154

x
OPT = T + ∑
i =1
(X i − T ) (7.3)

Where T is the overall mean value of the output response variable for the test
runs conducted. Xi is the design and control parameter value for the I level of
the parameter X.

7.3.2 Response Surface Methodology

Response surface methodology was employed in the present study

for modeling and analysis of response parameters in order to obtain the
characteristics of the engine. The design and analysis of experiment involved
the following steps:

The first step was the selection of the parameters that influence the
performance and emission characteristics. In this study, the compression ratio,
injection pressure, injection timing, fuel blends and power were considered as
the input parameters.

The compression ratio, CR was varied at four levels in steps of 0.2

from 17.5 to 18.1, the injection pressure was varied from (190bar - 230bar),
injection timing (23o-29o) Btdc. The karanja fuel blends; B was varied from
10% to 50%. The power, P was varied from 3.64 kW to 5.2 kW.

The advantage of using Design of Experiments is to evaluate the

performance of the engine over the entire range of variation of compression
ratio and other parameters with minimum number of experiments. The design
matrix was selected based on the 3 level factor design of response surface
methodology generated from the software “Design Expert” version 8.0.7.1,
contained 64 experimental runs as shown in Table 7.1.
 155

As per the run order, the experiments were conducted on the engine
and the responses were fed on the responses column.

Finally, the optimal values of the compression ratio, injection pressure,

injection timing fuel blends and power parameters were obtained by using the
desirability approach of the response surface methodology

7.4 RESULTS AND DISCUSSION

7.4.1 Optimization of Engine Parameters using Taguchi

Sixteen experiments, following the plan shown in Table 5.6

(chapter five), were performed on the engine .The corresponding S/N ratios
calculated are shown in Table 7.1 and the graph are presented in are shown in
Figures 7.1-7.5.

Table 7.1 Calculated S/N Ratio For single objective

BTE BSFC
Factors CO HC NOx
Run Number S/N ratio
1 30.63 12 -27.95 -29.54 -63.90
2 30.37 11.37 -12.04 -35.56 -62.27
3 28.94 10.17 -6.19 -37.38 -60.48
4 28.63 9.37 -4.4 -46.02 -56.77
5 27.60 8.17 -1.93 -33.06 -64.11
6 29.25 10.17 -5.68 -33.25 -62.27
7 30.63 11.70 -9.89 -39.27 -60.99
8 30.37 11.37 -33.97 -37.26 -55.95
9 29.54 10.17 -15.91 -32.04 -61.99
10 28.30 8.63 -27.95 -33.97 -60.83
11 30.63 11.37 -2.85 -41.80 -61.08
12 30.88 11.70 -17.72 -42.54 -56.41
13 29.54 10.45 -19.17 -30.37 -64.71
14 29.83 10.45 -4.4 -40.00 -62.27
15 29.25 9.89 -10.45 -32.25 -58.06
16 28.63 9.11 -4.58 -48.30 -55.91
 156

S/N ratio for BTH

31.5
31
30.5
30
S/N ratio

29.5
29
28.5
28
27.5
27
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Experimental runs

Figure 7.1 S/N ratios for Brake thermal efficiency

S/N ratio of BSFC

13
12
11
10
S/N ratio

9
8
7
6
5
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Experimental runs

Figure 7.2 S/N ratio for Brake specific fuel consumption

 157

Experimental runs
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
0

-5

-10

-15
S/N ratio

-20

-25

-30

-35 S/N ratio for CO

-40

Figure 7.3 S/N ratio for CO emissions

Experimental runs

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
-25

-30

-35
S/N ratio

-40

-45

S/N ratio of HC
-50

Figure 7.4 S/N ratio for HC emissions

 158

Ex perimental runs
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
-55
-56
-57
-58
-59
S/N ratio

-60
-61
-62
-63
-64
-65
S/N ratio of NOx
-66

Figure 7.5 S/N ratio for NOx emissions

7.4.2 Prediction of Output Response Variables

7.4.2.1 Prediction of Brake thermal efficiency (BTE)

For the engine performance, the response variable BTE was higher-
the –better. The criteria for optimization of the response parameters was based
on the higher the better S/N ratio.

The experimental results were substituted in equation (7.1) to

calculate the S/N ratios for all response variables and it is shown in
Figure 7.6. From the value the optimization of the engine parameters were
obtained .CR-17.9, IP-210 bar, IT-23° bTDC, B –B10, P - 4.16 kW is the
input parameter that predicts the output response BTE maximum.

From equation (7.4) the value of BTE is calculated from the

response table shown in Table 7.2

OPT BTE = Y + (CR 3 - Y ) + (IP2 - Y ) + (IT1 - Y ) + (B1 - Y ) + (P2 - Y ) (7.4)

 159

where Y bar is the average value of BTE from the experiments and CR, IP, IT,
B and P are the parameters chosen for the experiment.

The value of BTE predicted is 39 %.The output is very high, so only

multiobjective optimization is chosen.

S/N Ratio for BTHE

A B C
30.5

30.0

29.5
Mean of SN ratios

29.0

28.5

17.5 17.7 17.9 18.1 190 210 220 230 23 25 27 29

D E
30.5

30.0

29.5

29.0

28.5

10 20 30 50 3.64 4.16 4.68 5.20

Signal-to-noise: Larger is better

Figure 7.6 The value of S/N ratio for BTE

 160

Table 7.2 The Response table for BTE in % from the experimental
results

Static Biodiesel
Average Compression Injection
Injection fuel Power
S.No. Value ratio Timing
fraction
Pressure (P)
Y bar (CR) (IP) (IT)
(B)
1 30.5 32 31 33.5 30.5
2 30 33 30.25 32.5 33
30.25
3 31.25 31 30.5 26.25 28.5
4 29.25 30.5 29.25 28.25 29

7.4.2.2 Prediction of Brake Specific Fuel Consumption (BSFC)

The experimental results were substituted in equation (7.2). to

calculate the S/N ratios for all response variables and it is shown in Figure
7.7. From the value of optimization The engine parameters obtained were,CR-
18.1, IP-220 bar IT-29°bTDC, Blend –B30, Power -5.2 kW. That predicts the
minimum specific fuel consumption.

The response value of BSFC is calculated and it is shown in Table 7.3

BSFC = Y + (CR 4 - Y ) + (IP3 - Y ) + (IT4 - Y ) + (B3 - Y ) + (P4 - Y ) (7.5)

The value of BSFC is 0.4074 kg/kW hr.

 161

S/N Ratio for BSFC

A B C
11.4

10.8

10.2
Mean of SN ratios

9.6

9.0
17.5 17.7 17.9 18.1 190 210 220 230 23 25 27 29
D E
11.4

10.8

10.2

9.6

9.0
10 20 30 50 3.64 4.16 4.68 5.20

Signal-to-noise: Smaller is better

Figure 7.7 The value of S/N ratio for BSFC

Table 7.3 The Response table for BSFC in kg/kWh from the
experimental results

Biodiesel
Compression Static Injection
Average Injection Timing fuel Power
S.No. ratio
Y bar Pressure fraction (P)
(CR) (IP) (IT)
(B)
1 0.295 0.3137 0.29 0.267 0.3025
2 0.31 0.315 0.31 0.278 0.276
0.3053
3 0.29 0.29 0.29 0.355 0.317
4 0.3175 0.30 0.317 0.32 0.325

7.4.2.3 Emissions HC The experimental results were substituted in

equation (7.2) to calculate the S/N ratios for all response variables shown in
 162

Figure 7.8. From the value the optimization of engine parameters obtained
were,CR-17.9, IP-190 bar, IT-23° Btdc, Blend –B30, Power 5.2 kW that
predicts minimum HC.

The value of HC is calculated from the response table shown in Table 7.4

HC = Y + (CR 4 - Y ) + (IP4 - Y ) + (IT1 - Y ) + (B3 - Y ) + (P4 - Y ) (7.6)

The value of HC is 122 PPM.

S/N Ratio for HC Emissions

A B C
-30

-35
Mean of SN ratios

-40

-45
17.5 17.7 17.9 18.1 190 210 220 230 23 25 27 29
D E
-30

-35

-40

-45
10 20 30 50 3.64 4.16 4.68 5.20

Signal-to-noise: Smaller is better

Figure 7.8 The value of S/N ratio for HC

 163

Table 7.4 The Response table for HC in PPM from the experimental
results

Biodiesel
Compression Static Injection
Average Injection Timing fuel Power
S.No. ratio
Y bar Pressure fraction (P)
(CR) (IP) (IT)
(B)
1 91 37 114 89 49
2 64 64 70 73 81.5
123
3 87 83 72 107 103
4 109 167 94 87 117

7.4.2.4 Emissions NOx

The experimental results were substituted in equation (7.2). to

calculate the S/N ratios for all response variables shown in Figure 7.9.From
the value of optimization the engine parameters were obtained . CR-17.5,
IP-230 bar, IT-29° Btdc, Blend –B20, Power 4.16 kW.

The value of NOx is calculated from the response table shown in Table 7.5

NO X = Y + (C R 1 - Y ) + (IP4 - Y ) + (I T4 - Y ) + (B 2 - Y ) + (P2 - Y ) (7.7)

The value of NOx Is 318 PPM.

 164

S/N Ratio for NOx Emissions

C1 C2 C3
-56

-58

-60
Mean of SN ratios

-62

-64
17.5 17.7 17.9 18.1 190 210 220 230 23 25 27 29
C4 C5
-56

-58

-60

-62

-64
10 20 30 50 3.64 4.16 4.68 5.20

Signal-to-noise: Smaller is better

Figure 7.9 The values of S/N ratio of NOx

Table 7.5 The Response table for NOX in PPM from the experimental
results

Biodiesel
Compression Static Injection
Average Injection Timing fuel Power
S.No ratio
Y bar Pressure fraction (P)
(CR) (IP) (IT)
(B)
1 1154 1539 1157 1164 1024
2 1163 1250 1092 1195 1085
1117
3 1038 1190 1060 1097 1175
4 1111 651 1158 1012 1182

7.4.2.5 CO Emissions

The experimental results were substituted in equation (7.2) to

calculate the S/N ratios for all response variables. From the value the
 165

optimization of engine parameters were obtained. CR-17.9, IP-230 bar,

IT-29° Btdc, Blend –B10, Power 3.64 kW.

CO = Y + (CR 3 - Y ) + (IP1 - Y ) + (IT4 - Y ) + (B1 - Y ) + (P1 - Y ) (7.8)

The value of CO is 1.075 %.

Table 7.6 The Response table for CO in % from the experimental results

S.No. Biodiesel Power

Compression Static Injection
Average Injection Timing fuel (P)
ratio
Y bar Pressure fraction
(CR) (IP) (IT)
(B)
1 0.345 0.282 0.467 0.2725 0.1
2 0.415 0.352 0.37 0.275 0.33
0.356
3 0.262 0.457 0.317 0.48 0.312
4 0.4 0.335 0.2675 0.395 0.68

7.4.2.6 Optimization results

The single objective optimization gives different results and shown in

Table 7.7 and the results obtained are very high in case of efficiency and fuel
consumption, emissions. The input parameters are different for each response,
hence multi objective optimization is chosen.
 166

Table 7.7 Optimum parameter for single objective optimization

Controlled factors BTE BSFC CO HC NOx

A: compression ratio 17.9 18.1 17.9 17.7 17.9
B: Static Injection Pressure (bar) 210 220 230 230 190
C: Injection Timing (BTDC) 23 29 29 23 27
D.Biodiesel fuel fraction (%) 10 30 10 20 50
E. Power (kW) 4.16 5.2 3.64 3.64 3.64

7.4.3 Multi-Optimization Techniques

To obtain an optimal combination of engine parameters considering

performance and emissions multi-optimization techniques is used.

In multi-objective optimization we have to optimize the input

parameters which give maximum efficiency and minimum fuel consumption
and emissions.

Objective function I: To maximize efficiency

The objective function here is to maximize efficiency regardless of the

emissions.

Objective function η BTE= (7.9)

Multi-objective function II: To minimize emissions and fuel consumption

Objective function=( + + + ) (7.10)

The objective function is to minimize the emissions and fuel consumption

 167

Multi-objective function III: To minimize emissions, fuel

consumption and maximize efficiency.

In order to account for the diesel engine emissions and fuel

economy, a multi-objective function is formulated. The multi-objective
function is given in Equation. (7.11).

Objective function = 0.2 ( + + + + )

(7.11)

It can be seen from equation (7.11) that the objective function

assigns a equal weightage of 20% to the brake thermal efficiency, 20 % brake
fuel consumption and 20% to the emissions. Both the thermal performance
and the emissions (NOx, HC and CO) are made dimensionless by dividing
them by their respective maxima (for emissions) and taking the reciprocal of
η BTE=η max for the thermal performance. In view of the above, the final multi-
objective function is a dimensionless quantity that needs to be minimized and
can be considered to be a cost function objective function.

Table 7.8 Weighing factor of each response variable

Response Weighing
S.No.
variable Factor
1 BTE 0.2
2 BSFC 0.2
3 NOx 0.2
4 HC 0.2
5. CO 0.2
 168

The sum of the weighing factors for all the response variables
should be unity and the weighting factor assigned to each particular response
variable is determined on the basis of its relative importance. In this present
experimental work the weighting factor for each response variable assumed as
given in Table 7.8 .The weighting factor plays a very important role in this
type of analysis. The experimental results were used to Calculate S/N ratios
for all the response variables.

To obtain an optimal combination of engine parameters considering

performance and emissions multiple objective optimizations is used. The S/N
ratios is calculated using the following equation

S / N = W 1SN 1 + W 2 SN 2 + W 3 SN 3 + W 4 SN 4 + W 5 SN 5 (7.12)

Where, W1, W2, W3, W4 and W5 are the weighing factor of the response and
SN1,SN2, SN3, SN4 and SN5 are the signal to noise ratio of the response
variable. Using equation (7.12) S/N ratio for the combined objective is found
out and it is given in Table 7.9 and shown in Figure 7.10.

Table 7.9 S/N ratio for combined objective

Test runs S/N ratio

1 -15.752
2 -13.626
3 -12.988
4 -13.838
5 -12.666
6 -12.356
7 -13.564
8 -17.088
9 -14.046
10 -17.164
11 -12.746
12 -14.818
13 -14.852
14 -13.278
15 -12.324
16 -14.21
 169

S/N ratio for combined objective

-10
-20
-30
-40
-50
-60
-70
CR IP IT B P
Level 1 -56 -57.3 -55 -57 -62
Level 2 -55.7 -56.4 -53.4 -58 -55.4
Level 4 -58.8 -51.6 -58 -57 -55
Series4 -54.7 -59.9 -59.4 -52 -52.5
Design parameters

Figure 7.10 S/N ratio for combined objective

For the combined objective with equal weightage for all the
responses, the optimum setting is found out, CR-17.9, IP-200 bar, IT-23°
bTDC, B20. P -3.64kW is the optimal combination to achieve multiple–
performance characteristics of the engine.

The Taguchi method of DOE analysis is showing good

reproducibility and the values of the BTE, BSFC, HC, CO and NOx were
found to be 34.75%, 0.2393 kg/kWh, 22 ppm, 0.015% and 598 PPM
respectively.

Table 7.10 Values of response variable

Response
S.No. Values
variable
1 BTE 34.75 %
2 BSFC 0.2393 kg/kW h
3 NOx 598 PPM
4 HC 22 PPM
5 CO 0.015 %
 170

7.4.4 Optimization of engine parameters using RSM

In this study, RSM was applied to optimize the input parameter that
gives maximum efficiency, minimum fuel consumption and emissions.

RSM addresses optimization by:

1) Providing an understanding of how the test variables affect

the selected process response;

2) Determining the interrelationships between the test

variables;

3) And characterizes the combined effect that all the influential

test variables may have on process response.

This work investigates the influence of compression ratio, injection

pressure, injection timing and power on the performance and emissions of a
DI diesel engine using karanja biodiesel (10%, 20%, 30%, and 50%) blended-
diesel fuel. Tests were carried out using four different CR,IP,IT, blend and
power The experiments were designed using a statistical tool known as
Design of Experiments (DOE) based on response surface methodology
(RSM). The resultant models of the response surface methodology were
helpful to predict the response parameters such as Brake Specific Fuel
Consumption (bsfc), Brake Thermal Efficiency (BTE), Carbon monoxide
(CO), Hydrocarbon (HC) and Nitrogen Oxides (NOx). Optimization of
parameters was performed using the desirability approach of the response
surface methodology for better performance and lower emission.
 171

Table 7.11 The Experimental design matrix

Fuel
Run Compression Pressure Injection Blends Power BTE BSFC CO HC NOx
Order Ratio ο
bar Btdc % kW % kg/kWh % ppm ppm
1 17.5 190 23 10 3.64 34.84 0.24 0.22 69 1258
2 17.5 210 25 10 4.16 35.42 0.23 0.42 69 1319
3 17.5 220 27 10 4.68 34.94 0.24 1.11 70 1335
4 17.5 230 29 10 5.2 33.41 0.26 2.28 71 1306
5 17.5 190 23 20 3.64 31.31 0.28 0.14 69 1239
6 17.5 210 25 20 4.16 31.63 0.27 0.3 69 1282
7 17.5 220 27 20 4.68 30.89 0.28 0.84 74 1279
8 17.5 230 29 20 5.2 29.10 0.31 1.76 84 1232
9 17.5 190 23 30 3.64 29.45 0.30 0.11 69 1206
10 17.5 210 25 30 4.16 29.50 0.29 0.25 67 1230
11 17.5 220 27 30 4.68 28.50 0.31 0.72 81 1209
12 17.5 230 29 30 5.2 26.45 0.34 1.53 112 1143
13 17.5 190 23 50 3.64 30.68 0.27 0.17 67 1097
14 17.5 210 25 50 4.16 30.22 0.28 0.34 61 1084
15 17.5 220 27 50 4.68 28.69 0.30 0.93 110 1027
16 17.5 230 29 50 5.2 26.12 0.34 1.95 214 924
17 17.7 190 23 10 3.64 34.38 0.26 0.12 69 1061
18 17.7 210 25 10 4.16 35.28 0.24 0.26 70 1173
19 17.7 220 27 10 4.68 35.12 0.25 0.74 69 1241
20 17.7 230 29 10 5.2 33.91 0.27 1.58 68 1264
21 17.7 190 23 20 3.64 30.97 0.30 0.06 70 1075
22 17.7 210 25 20 4.16 31.60 0.28 0.17 70 1169
23 17.7 220 27 20 4.68 31.19 0.29 0.55 67 1218
24 17.7 230 29 20 5.2 29.72 0.32 1.20 62 1223
25 17.7 190 23 30 3.64 29.21 0.31 0.04 70 1074
26 17.7 210 25 30 4.16 29.59 0.31 0.13 71 1150
27 17.7 220 27 30 4.68 28.91 0.32 0.47 63 1181
28 17.7 230 29 30 5.2 27.18 0.35 1.04 47 1167
29 17.7 190 23 50 3.64 30.67 0.29 0.08 70 1031
30 17.7 210 25 50 4.16 30.53 0.29 0.20 74 1070
31 17.7 220 27 50 4.68 29.32 0.31 0.62 48 1064
32 17.7 230 29 50 5.2 27.07 0.34 1.34 73 1013
33 17.9 190 23 10 3.64 33.17 0.27 0.10 69 883
34 17.9 210 25 10 4.16 34.39 0.26 0.23 69 1048
35 17.9 220 27 10 4.68 34.56 0.25 0.68 72 1167
36 17.9 230 29 10 5.2 33.67 0.27 1.45 78 1242
37 17.9 190 23 20 3.64 29.87 0.31 0.05 69 929
38 17.9 210 25 20 4.16 30.83 0.30 0.15 66 1076
39 17.9 220 27 20 4.68 30.74 0.30 0.05 87 1177
40 17.9 230 29 20 5.2 29.59 0.32 1.10 133 1234
41 17.9 190 23 30 3.64 28.23 0.33 0.02 67 962
 172

Table 7.11 (Continued)

Fuel
Run Compression Pressure Injection Blends Power BTE BSFC CO HC NOx
Order Ratio
bar Btdc % kW % kg/kWh % ppm ppm
42 17.9 210 25 30 4.16 28.93 0.35 0.11 59 1090
43 17.9 220 27 30 4.68 28.57 0.31 0.42 122 1173
44 17.9 230 29 30 5.2 27.16 0.30 0.95 256 1211

45 17.9 190 23 50 3.64 29.91 0.31 0.06 61 984

46 17.9 210 25 50 4.16 30.09 0.30 0.18 33 1075
47 17.9 220 27 50 4.68 29.21 0.32 0.56 248 1121

48 17.9 230 29 50 5.2 27.28 0.35 1.23 70 1122

49 18.1 190 23 10 3.64 31.22 0.29 0.16 69 726
50 18.1 210 25 10 4.16 32.77 0.27 0.33 68 942
51 18.1 220 27 10 4.68 33.25 0.26 0.91 78 1114
52 18.1 230 29 10 5.2 32.69 0.27 1.90 98 1240
53 18.1 190 23 20 3.64 28.30 0.33 0.10 66 805

54 18.1 210 25 20 4.16 29.32 0.31 0.23 56 1003

55 18.1 220 27 20 4.68 29.54 0.31 0.68 133 1156
56 18.1 230 29 20 5.2 28.72 0.32 1.45 298 1264

57 18.1 190 23 30 3.64 26.50 0.35 0.07 60 870

58 18.1 210 25 30 4.16 27.52 0.33 0.18 31 1050
59 18.1 220 27 30 4.68 27.49 0.33 0.58 258 1184

60 18.1 230 29 30 5.2 26.40 0.35 1.26 74 1274

61 18.1 190 23 50 3.64 28.41 0.32 0.12 39 957
62 18.1 210 25 50 4.16 28.91 0.31 0.27 60 1100

63 18.1 220 27 50 4.68 28.35 0.32 0.76 71 1198

64 18.1 230 29 50 5.2 26.74 0.35 1.61 235 1251

7.4.4.1 Analysis of the model

The principal model analysis was based on the analysis of variance

(ANOVA) as it provides numerical information for the p value. The models
 173

found to be significant as the values of p were less than 0.05. The different
models for the responses were developed in terms of actual factors and are
given below as equations (7.13) to (7.17) and shown in table

Model equation for brake thermal efficiency: It can be seen that

the regression coefficients of the linear terms for compression ratio, injection
pressure, injection timing and blend. The quadratic terms in CR2 and B2, and
the interaction terms in CR*IP, CR*IT, CR*B, IP*IT and IT*B had
significant effects on the yield (P value <0.05).

The predicted model for BTE is

ηBTE = 29.618 - 0.734 CR - 1.58 IP + 0.806 IT -2.531B+0.19

P-0.0582 CR* IP + 0.7676 CR *B +.330CR*B- 1.575IP*IT+0.025 IP*IT-
0.803 IT*B-0.828 CR2 + 3.307B2
(7.13)

The results predicted that linear effect changes in injection

pressure, blend and the quadrtic effect of the blend were primary determing
factors on the brake thermal efficiency as these had the largest coefficients.

The quadratic effect of CR2 and the interaction effect, CR*IT,

CR*B were secondary determining factors and that other terms of the model
showed no significant effect on BTE. Positive coefficients, as with IT and B2,
enhance the yield. However, all the other terms had negative coefficients.
From Table 7.12 it can been seen that the quadratic model developed for BTE
is significant because P-value is less than 0.0001 for all input parameters.
 174

Table 7.12 ANOVA response For BTE quadratic model

Sum of Mean F p-value

Source Squares df Square Value Prob > F
Model 409.894 12 34.16 31044.46 0.0001
A-CR 10.295 1 10.30 9357.27 0.0001
B-IP 0.262 1 0.26 238.32 0.0001
C-IT 0.069 1 0.07 63.47 0.0001
D-B 121.382 1 121.38 110318.62 0.0001
AB 0.002 1 0.00 2.06 0.1577
AC 0.399 1 0.40 362.69 0.0001
AD 2.128 1 2.13 1934.75 0.0001
BC 2.94 1 2.94 2672.02 0.0001
BD 0.000 1 0.00 0.39 0.5343
CD 0.430 1 0.43 391.18 0.0001
A2 8.680 1 8.68 7889.19 0.0001
D2 137.535 1 137.54 124999.22 0.0001
Residual 0.056 51 0.00
Cor Total 409.950 63

Model equation for brake specific fuel consumption: The

predicted model for BSFC is

BSFC = 0.306 + 0.014 CR + 0.034IP -0.023 IT+0.027 B- 0.001

P -0.008 CR*IP -0.019 CR*IT+0.026IP*IT+0.017 IP2-0.001 B2 (7.14)

It can be seen that the regression coefficients of the linear terms for
compression ratio, injection pressure, injection timing and blend the quadratic
terms in IP2 and B2, and the interaction terms in CR*IP,CR*IT and IP*IT had
significant effects on the yield ( P value <0.05).
 175

The results predicted that linear effect changes in compression

ratio, injection pressure ,blend and the quadrtic effect of the injection
pressure were primary determing factors on the Brake specific fuel
consumption as these had the largest coefficients.

The quadratic effect, B2 and the interaction effect CR*IT, CR*IP

were secondary determining factors and that other terms of the model showed
no significant effect on BSFC. Positive coefficients, as with IP and CR,
enhance the yield. However, all the other terms had negative coefficients.
From Table 7.13 it can been seen that the quadratic model developed for
BSFC is significant because P-value is less than 0.0001 for all input
parameters.

Table 7.13 ANOVA response For BSFC quadratic model

Sum of Mean F p-value

Source Squares df Square Value Prob > F
Model 0.0590 12 0.0049 67.6165 0.0001
A-CR 0.0035 1 0.0035 48.71318 0.0001
B-IP 0.0001 1 0.0001 1.625575 0.21
C-IT 0.0001 1 0.0001 0.779007 0.38
D-B 0.0138 1 0.0138 190.0739 0.0001
AB 0.0000 1 0.0000 0.515429 0.48
AC 0.0002 1 0.0002 3.207439 0.0001
AD 0.0000 1 0.0000 0.070687 0.79
BC 0.0008 1 0.0008 11.22489 0.0001
BD 0.0000 1 0.0000 0.000327 0.99
CD 0.0001 1 0.0001 1.167089 0.29
A2 0.0000 1 0.0000 0.085905 0.77
D2 0.0190 1 0.0190 261.7446 0.0001
Residual 0.0037 51 0.0001
Cor Total 0.0627 63
 176

Model equation for carbon monoxide (CO): The predicted model for CO is

CO=-0.078-0.092CR+0.76IP-0.078IT-0.071B+0.098CR*IP

- 17CR*IT+0.58 IP*IT+0.081 IP*B-0.12IT*B (7.15)

The results predicted that linear effect changes in injection

pressure,the interaction effect IP*IT and the quadratic effect CR2 were
primary determing factors on as these had the largest coefficients.

The interaction effect CR*IT, CR*IP and CR*B were secondary

determining factors and that other terms of the model showed no significant
effect on BSFC. Positive coefficients, as with IP, IP*IT and CR2, enhance the
yield. However, all the other terms had negative coefficients.

Table 7.14 ANOVA response For CO quadratic model

Sum of Mean F p-value

Source Squares df Square Value Prob > F
Model 20.53176 12 1.71098 92.71264 0.0001
A-CR 0.162971 1 0.162971 8.830868 0.0001
B-IP 0.060174 1 0.060174 3.260666 0.0001
C-IT 0.000648 1 0.000648 0.035099 0.8521
D-B 0.09446 1 0.09446 5.118499 0.0280
AC 0.019072 1 0.019072 1.033453 0.3141
AD 0.001081 1 0.001081 0.058591 0.8097
BC 0.392704 1 0.392704 21.27941 0.0000
BD 0.004301 1 0.004301 0.233036 0.6313
CD 0.009373 1 0.009373 0.507882 0.4793
A2 0.936056 1 0.936056 50.72196 0.0001
D2 0.687893 1 0.687893 37.27478 0.0001
Residual 0.941187 51 0.018455
Cor Total 21.47294 63
 177

Model equation for Hydrocarbon (HC): The predicted model for HC is

HC = 126.418+18.09 * CR -166.49 * IP+197 * IT+14.139*B- 36.65CR*IP

+60.39CR*IP -5.95CR*B -31.5IP*IT- 25.96IP*B+44.28IT*B
+15CR2 (7.16)

The results predicted that linear effect changes in Compressio

ratio,injection pressure ,injection timing ,blend, the interaction effects CR*IP
CR*IT, CR*B, IP*IT, IT*B, IT*B were primary determing factors on the
NOx emissions as these had the largest coefficients.

Positive coefficients, as with CR, IT, B CR*IT, B*IT enhance the

yield. However, all the other terms had negative coefficients

Model equation for Nitrogen oxides (NOx): The predicted model for NOx
is

NOx = 1168.218 -51.39 CR- 66.70 IP + 152.08 IT -37.55 B+0.11P+ 0.175

CR*IP +116.503 CR*IT+98.12 CR*B-67.21IP*IT-55.325 IT*B+22.35CR2 -
28.48 B2 (7.17)

The results predicted that linear effect changes in Compressio

ratio,injection pressure ,injection timing ,blend, the interaction effects
CR*IT,CR*B,IP*IT , IT*B and the quadrtic effect of the CR2, were primary
determing factors on the NOx emissions as these had the largest coefficients.

The interaction effect, CR*IP, IP*B were secondary determining

factors and that other terms of the model showed no significant effect on
NOx. Positive coefficients, as with IT, CR*IT,CR*B and CR2, enhance the
yield. However, all the other terms had negative coefficients.
 178

Table 7.15 ANOVA response for NOx quadratic model

Sum of Mean F p-value

Source Squares df Square Value Prob > F
Model 1065769.80 12 88814.15 1148122.05 0.0001
A-CR 50435.10 1 50435.10 651986.78 0.0001
B-IP 462.63 1 462.63 5980.47 0.0001
C-IT 2485.81 1 2485.81 32134.63 0.0001
D-B 26714.70 1 26714.70 345347.38 0.0001
AB 0.02 1 0.02 0.26 0.610
AC 9192.28 1 9192.28 118830.85 0.0001
AD 187289.29 1 187289.29 2421134.00 0.0001
BC 5355.09 1 5355.09 69226.60 0.0001
BD 0.01 1 0.01 0.13 0.720
CD 2040.57 1 2040.57 26378.95 0.0001
A2 6320.25 1 6320.25 81703.40 0.0001
D2 10198.93 1 10198.93 131844.07 0.0001
Residual 3.95 51 0.08
Cor Total 1065773.75 63

7.4.4.2 Evaluation of the model

The stability of the models was validated using Analysis of

variance (ANOVA). The output showed that the model was significant with p
values less than 0.0001. The reference limit for p was chosen as 0.05. The
regression statistics goodness of fit (R2) and the goodness of prediction
(Adjusted R2 ) were shown in Table 7.16 for all the responses. The R2 value
indicates the total variability of response after considering the significant
 179

factors. The (Adjusted R2 ) value accounts for the number of predictors in the
model. Both the values indicate that the model fits the data very well.

Table 7.16 Response surface model evaluation

Model BSFC
BTE % CO % HC PPM NOx PPM
kg/kWh
Mean 33.85 0.268 0.098 31 884
Std.Deviation 0.036 0.030 0.158 0.035 0.27
Model degree Quadratic Quadratic Quadratic Quadratic Quadratic
R 0.9999 0.9591 1.0000 0.9392 1.000
Adj.R2 0.9998 0.9400 1.000 0.9307 1.000
Pred.R2 1.000 0.9249 1.000 0.9147 1.000

The analysis of variance revealed that BTE model was adequate to

express the actual relationship between the response and significant variables,
with a satisfactory coefficient of determination (R2=0.999), which indicated
99.9 % of the variability in the response could be explained by the 2nd-order
polynomial predictive equation (7.13).

The analysis of variance revealed that BSFC model was adequate to

express the actual relationship between the response and significant variables,
with a satisfactory coefficient of determination (R2=0.94), which indicated 94
% of the variability in the response could be explained by the 2nd-order
polynomial predictive equation (7.14).

The analysis of variance revealed that CO model was adequate to

express the actual relationship between the response and significant variables,
with a satisfactory coefficient of determination (R2=1.000), which indicated
 180

100 % of the variability in the response could be explained by the 2nd-order

polynomial predictive equation (7.15).

The analysis of variance revealed that HC model was adequate to

express the actual relationship between the response and significant variables,
with a satisfactory coefficient of determination (R2=0.93), which indicated
93.9 % of the variability in the response could be explained by the 2nd-order
polynomial predictive equation (7.16).

The analysis of variance revealed that NOx model was adequate to

express the actual relationship between the response and significant variables,
with a satisfactory coefficient of determination (R2=1.000), which indicated
100 % of the variability in the response could be explained by the 2nd-order
polynomial predictive equation (7.17).

7.4.5 Interactive Effect of Input Parameters on Responses

7.4.5.1 Brake thermal efficiency (BTE)

The percentage change in the BTE is shown in Figures 7.11-7.19.

As seen in the figures ,the BTE usually increases with the increase in
biodiesel percentage in the fuel blend. The maximum brake thermal efficiency
is 35% for the compression ratio 17.9 and fuel blend between B10-
B20,whereas low brake thermal efficiency lies in the region around 17.7 CR
and fuel blend between B30-B40. Thus,the primary reason for the decrease in
the BTE of biodiesel is the higher BSFC in spite of lower LHV of biodiesels
The effects of the variation in CR on the BTE indicate that at higher CRs
improve the engine efficiency.This can be attributed to better combustion and
higher lubricity of biodiesel.As seen in Figure 7.11,the increased CR
increased the BTE by 2% for B10 compared to the results of original CR
(17.5).
 181

During advancement of injection timing BTE emission increased.

This could be due to the following fact: in-cylinder charge temperature and
pressure decreased with an advancement of the injection timing resulting in
extended ignition delay of the injected fuel.

Figure 7.11 The BTE variations against compression ratio and injection
pressure
 182

Figure 7.12 The BTE variations against compression ratio and Injection
timing

Figure 7.13 The BTE variations against compression ratio and Power
 183

Figure 7.14 The BTE variations against Compression ratio and fuel
Blend

Figure 7.15 The BTE variations against blend and injection timing
 184

Figure 7.16 The BTE variations against fuel blend and power

Figure 7.17 The BTE variations against fuel blend and injection pressure
 185

Figure 7.18 The BTE variations against fuel blend and power

Figure 7.19 The BTE variations against injection pressure and injection
timing

Simultaneously, the penetration of fuel spray enhanced reaction

between fuel and air improved and ultimately resulted in premixed or rapid
combustion phase of the Combustion process.
 186

Increasing the injection pressure from 190bar to 230bar increased

BTE. Increasing the injection pressure beyond 220 bar, the inverse trend was
noticed at all injection timings. The above result could be due to the following
fact: With increase in injection pressure, better atomization of the fuel
resulted in the smaller droplet size; faster evaporation of fuel sprays; and
improved reaction between fuel and air. These resulted in comparatively
better combustion and contributed for higher BTE at all injection timings.
Beyond 220 bar of injection pressure, faster velocity of the fuel jets caused
most fuel particles to hit the wall of combustion chamber where the fuel
particles got cooled and not participated in the combustion process effectively
thus resulting in incomplete combustion The above discussions revealed that a
compression ratio 17.9, injection pressure of 220 bar combined with advanced
injection timing 27 ο bTDC produced highest BTE.

7.4.5.2 Brake specific fuel consumption (BSFC)

The change in the BSFC is observed in Figures 7.20- 7.27.As

shown in Figure 7.21 the BSFC generally increased with the increase in
biodiesel percentage in the fuel blend.It can be considered that the decrease in
the lower heating value of the the blends by adding biodiesel,this requires
more fuel to be injected into the cylinder to get the same power output,leads
to the increase in the BSFC As shown in Figure 7.20,increased in CR,the
maximum cylinder pressure increases due to the fuel injected in hotter
combustion chamber and this leads to higher effective power.Therefore,fuel
consumption per output will decrease.

As the BSFC is calculated on weight basis obviously higher

densities resulted in higher values for BSFC.As density of karanja biodiesel
was higher than that of biodiesel for the same fuel consumption on volume
basis,pure biodiesel yield higher BSFC.The higher densities of biodiesel
blends caused higher mass injection for the same volume at the same injection
 187

pressure.The calorific value of the biodiesel is less than diesel.Due to these

reason,the BSFC for the other blends were higher than that of diesel.Similar
trends of decrease in the BSFC value with increasing load for different
biodiesel were also reported by other researchers (Baiju et al .2009) while
testing biodiesel obtained from Karanja.

Figure 7.20 The BSFC variations against Injection timing and

Compression ratio
 188

Figure 7.21 The BSFC variations against Biodiesel fuel fraction and
Compression ratio

Figure 7.22 The BSFC variations against Power and Compression ratio
 189

Figure 7.23 The BSFC variations against injection timing and injection
pressure

Figure 7.24 The BSFC variations against Biodiesel fuel fraction and
injection pressure
 190

Figure 7.25 The BSFC variations against Biodiesel fuel fraction and
injection timing

Figure 7.26 The BSFC variations against Power and Injection timing
 191

Figure 7.27 The BSFC variations against power and Biodiesel fuel
fraction

7.4.5.3 Engine Emissions

The engine operating parameters such as air-fuel equivalence ratio,

fuel type, combustion chamber design and atomization ratio affect all
emissions emitted by internal combustion engines.Especially,emissions of CO
and unburned HC in the exhaust are very important since they represent the
low chemical energy that cannot be totally used in the engine. Aksoy et al
(2011) studied and found that emissions such as CO2, NOx emitted by diesel
engine have important effects on ozone layer The engine emissions with
karanja biodiesel have been evaluated in terms of CO,HC and NOX at various
CR,IP,IT,Blend at different loading conditions of the engine.

7.4.5.3.1 Carbon monoxide(CO)

The variation in CO of the engine is shown in Figures 7.28-7.35 As

viewed in Figure 7.28 increased CR at pressure 210 bar decreased the CO
 192

emissions by 37.09% and reduced CR and higher injection pressure 230 bar
increased CO emissions by 9.67 % compared to the results of original CR
(17.5) for B 50. Cenk sayin et al. (2011) found out at lower CR ,insufficient
heat of compression delays ignition and so CO emissions increase.Raheman
et al .(2008) investigated that the possible reason for this trend could be that
the increased CR actually increases the air temperature inside the cylinder
therefore reducing the ignition lag causes better and more complete burning
of the fuel .It is seen in Figure 7.28 that the percentage of CO is less than
0.3% at CR 17.7,B20 and maximum percentage of 1.9% at CR 17.5 ,B50 .The
lower CO emissions of biodiesel blends may be due to their more complete
oxidation as compared to diesel.Some of the CO produced during combustion
of biodiesel might have been converted into CO2 by taking up extra oxygen
molecule present in the biodiesel chain and thus reduces CO formation.

Figure 7.28 The CO variations against injection pressure and

compression ratio
 193

Figure 7.29 The CO variations against injection timing and compression

ratio

Figure 7.30 The CO variations against Biodiesel fuel fraction and

compression ratio
 194

Figure 7.31 The CO variations against Power and compression ratio

Figure 7.32 The CO variations against injection timing and pressure

 195

Figure 7.33 The CO variations against Biodiesel fuel fraction and

pressure

Figure 7.34 The CO variations against Biodiesel fuel fraction and

injection timing
 196

The maximum and minimum CO produced is 0.02gm/kWh and

0.002gm/kWh, which is much less than, mentioned in EURO– IV Norms
(max 1.5gm/kWh). It is an indication of the complete combustion of biodiesel
being an oxygenated fuel.

7.4.5.3.2 Hydrocarbon (HC)

It is seen in Figures 7.35-7.42 that there is a significant decrease in

the HC emission level with karanja oil as compared to pure diesel. The HC
emission is minimum 33 PPM occurring at CR (18.1) and blend B30,so the
value of HC reduces as compression ratio and fuel blend increases . HC
concentration decreases with biodiesel addition and this suggests that adding
oxygenate fuels can decrease HC from the locally over rich mixture.
Furthermore,oxygen enrichment is also favorable to the oxidation of HC in
the expansion and exhaust process(Huang H etal; 2005).As confirmed in the
Figure 7.35 increased CR reduced the HC emissions by 4 % and reduced CR
increases the HC emissions.At lower CR,insufficient heat of compression
delays ignition and so HC emissions increase(Jindal etal .2010). These
reductions indicate that more complete combustion of the fuels and thus,HC
level decreases significantly.The reduction in HC emission was linear with
the addition of biodiesel for the blends.

During advancement of injection timing the exhaust emissions HC

were reduced this could be due to the following fact: in-cylinder charge
temperature and pressure decreased with an advancement of the injection
timing resulting in extended ignition delay of the injected fuel.
Simultaneously, the penetration of fuel spray enhanced, reaction between fuel
and air improved and ultimately resulted in premixed or rapid combustion
phase of the combustion process. Increasing the injection pressure from 190
bar to 230 bar reduction in HC Occurs. The above result could be due to the
following fact: With increase in injection pressure, better atomization of the
 197

fuel resulted in the smaller droplet size; faster evaporation of fuel sprays; and
improved reaction between fuel and air. These resulted in comparatively
better combustion and contributed for with lesser HC emissions at all
injection timings. Beyond 220 bar of injection pressure, faster velocity of the
fuel jets caused most fuel particles to hit the wall of combustion chamber
where the fuel particles got cooled and not participated in the combustion
process effectively thus resulting in incomplete combustion. From the
experiments is very clear that HC emissions reduces for biodiesel blends.

Figure 7.35 The HC variations against Biodiesel fuel fraction and

compression ratio
 198

Figure 7.36 The HC variations against power and compression ratio

Figure 7.37 The HC variations against Injection timing and Pressure

 199

Figure 7.38 The HC variations against Biodiesel fuel fraction and

Pressure

Figure 7.39 The HC variations against power and Pressure

 200

Figure 7.40 The HC variations against Biodiesel fuel fraction and

injection timing

Figure 7.41 The HC variations against power and injection timing

 201

Figure 7.42 The HC variations against power and Biodiesel fuel fraction

Figure 7.43 The HC variations against compression ratio and injection

timing
 202

7.4.5.3.3 Nitrogen oxides (NOX )

The NOx values for different fuel blends at various compression

ratio is shown in Figure 7.44-7.49 The amount of NOx produced for B10-
B50 is the range of 720-1300 ppm as compared to diesel varying from 300-
900 ppm.It can been seen that the increasing proportion of biodiesel in the
blends increases NOx as compared with diesel.This could be attributed to the
increased exhaust temperatures and the fact that biodiesel had some oxygen
content in it faclitated NOx formation..Since the size of injected particles of
vegetable oils is bigher than that of diesel fuel, combustion efficiency and
maximum combustion temperatures with vegetable oils were lower.
Therefore, NOx emissions were lower (Ramadhas et al 2004). As illustrated
in the Figure 7.44 and Figure 7.45, increased CR increased the NOX
Emissions By 10% and reduced CR decreased NOX emissions by 12%
compared to the results of original CR for B50.Reduced CR is to reduce the in
- cylinder temperatures and thus flame temperatures during the combustion to
supress NOx emissions(Raheman etal. 2008).NOx emissions were also higher
at part loads for biodiesel.This is probably due to higher bulk modulus of bio-
diesel resulting in a dynamic injection advance apart from static injection
advance provided for optimum efficiency.Excess oxygen(10%) present in bio-
diesel would have aggravated the situation(Pradeep et al. 2007). When
ignition time has bben increased NOx formation increases, Huang et al.(2005)
explained by the increase of ignition delay with increased IT, because of low
gas temperature in the case of early fuel injection, and this long ignition delay
would increase the amount of fuel burned in the premixed burning phase,
resulting in a high temperature and high cylinder gas pressure, finally
increasing the NOx concentration .

Gumus (2008). found out that when the IP was decreased, NOx
emissions diminished for all the fuel mixtures. Increase in IP decreased the
 203

particle diameter and caused the biodiesel-diesel fuel spray to vaporize

quickly. So, higher IP initially generates faster combustion rates, resulting in
higher temperatures. As a consequence, NOx concentrations start to increase.

For B20 blend the maximum and minimum NOx produced is

2.41g/kW-hr and 1.3g/kW-hr, that is much less than, mentioned in
EURO – IV Norms (max 3.5gm/kW-hr).

Figure 7.44 The NOx variations against Pressure and compression ratio
 204

Figure 7.45 The NOx variations against Biodiesel fuel fraction and
compression ratio

Figure 7.46 The NOx variations against Power and compression ratio
 205

Figure 7.47 The NOx variations against injection timing and Pressure
 206

Figure 7.48 The NOx variations against Biodiesel fuel fraction and
injection timing

Figure 7.49 The NOx variations against Biodiesel fuel fraction and
Pressure

7.4.5 Optimization

The criteria for the optimization such as the goal set for each
response lower and upper limits used, weight used and importance of the
factors are presented in Table 7.13. In desirability based approach, different
best solutions were obtained. The solution with high desirability is preferred.

Different possible solutions were analyzed using the experimental

results. First the target is fixed and the combined desirability of 0.78 is
obtained and out of 64 experimental results 30 solutions were found.The
solutions obtained is given in Tables 7.17 and 7.18.
 207

Table 7.17 Optimization criteria and desirability response

lower upper
Name Goal Weight Weight Importance
Limit Limit
CR target=17.70 17.5 18.1 1 1 5
IP target = 230.00 190 230 1 1 5
IT target = 27.00 23 27 1 1 5
Blend target = 20.00 10 50 1 1 5
Power target = 3.64 3.64 5.2 1 1 5
BTE maximize 26.12 35.42 1 1 5
BSFC minimize 0.23 0.35 1 1 5
CO minimize 0.02 2.28 1 1 5
HC minimize 31 298 1 1 5
NOx minimize 726 1335 1 1 5

Table 7.18 The values of the response

S.No. Response Prediction

1 BTE 31.02212
2 BSFC 0.286227
3 CO 0.469985
4 HC 96.19798
5 NOx 1110.492

The target is fixed in range and the combined desirability of 0.80 is

obtained and out of 64 experimental results, 79 solutions were found and the
solutions obtained are CR- 17.9,IP-200 bar, IT 23ο bTDC,Blend B 10,Power
4.2 kW. The solutions obtained is given in Table 7.19 and 7.20.
 208

Table 7.19 Optimization criteria and desirability response

Lower Upper Lower Upper

Importance
Name Goal Limit Limit Weight Weight
CR is in range 17.5 18.1 1 1 5
IP is in range 190 230 1 1 5
IT is in range 23 29 1 1 5
B is in range 10 50 1 1 5
P is in range 3.64 5.2 1 1 5
BTE maximize 26.12 35.42 1 1 5
BSFC minimize 0.23 0.35 1 1 5
CO minimize 0.02 2.28 1 1 5
HC minimize 31 298 1 1 5
NOx minimize 726 1335 1 1 5

Table 7.20 The values of the response

S.No. Response Prediction

1 BTE 33.84566 %
2 BSFC 0.268502 kg/kW hr
3 CO 0.098169 %
4 HC 31.00062 PPM
5 NOx 954.6132 PPM

When the goal is fixed as maximum value for the parameters

CR -18.1,IP230 bar, IT 29ο bTDC, B 50,P-5.2 kW the solutions obtained are
shown in Table 7.21 and 7.22 and and the combined desirability of 0.76 is
obtained.
 209

Table 7.21 Optimization criteria and desirability response

Lower Upper Lower Upper

Importance
Name Goal Limit Limit Weight Weight
A:CR maximize 17.5 18.1 1 1 5
B:IP maximize 190 230 1 1 5
C:IT maximize 23 29 1 1 5
D:B maximize 10 50 1 1 5
E:P maximize 3.64 5.2 1 1 5
BTE maximize 26.12 35.42 1 1 5
BSFC minimize 0.23 0.35 1 1 5
CO minimize 0.02 2.28 1 1 5
HC minimize 31 298 1 1 5
NOx minimize 726 1335 1 1 5

Table 7.22 The values of the response

S.No. Response Prediction

1 BTE 32 %
2 BSFC 0.279 kg/kWhr
3 CO 0.25 %
4 HC 128 PPM
5 NOx 1118 PPM

Now the taget is fixed as CR -17.7,IP-220 bar,IT-25ο bTDC

B-30,P-3.64. 38 solutions were found and the desirability factor is 0.70 the
solutions obtained are shown in Table 7.23 and 7.24.
 210

Table 7.23 Optimization criteria and desirability response

Lower Upper Lower Upper

Importance
Name Goal Limit Limit Weight Weight
CR target = 17.70 17.5 18.1 1 1 5
IP target= 220.00 190 230 1 1 5
IT target = 25.00 23 27 1 1 5
B target = 30.00 10 50 1 1 5
P target = 3.64 3.64 5.2 1 1 5
BTE maximize 26.12 35.42 1 1 5
BSFC minimize 0.23 0.35 1 1 5
CO minimize 0.02 2.28 1 1 5
HC minimize 31 298 1 1 5
NOx minimize 726 1335 1 1 5

Table 7.24 The values of the response

S.No. Response Prediction

1 BTE 31.2685
2 BSFC 0.296934
3 CO 0.267428
4 HC 10.5028
5 NOx 1147.283

Maximum desirability of 0.995 was obtained at an compression

ratio 17.9, injection pressure 200 bar,injection timing 23 oBtdc, 20% of fuel
blend and 3.64 kW of power that could be considered as the optimum
parameters for the test engine having 5.2kW as rated power at 1500 rpm.
 211

Table 7.25 Optimization criteria and desirability response

Lower Upper Weight

Source Importance Goal Desirability
limits Limits Upper Lower

Compression
17.5 18.1 1 1 5 Target 1
Ratio
Injection
190 230 1 1 5 Target 1
pressure
Injection
23 27 1 1 5 Target 1
timing
Biodiesel fuel
10 50 1 1 5 Target 1
fraction
Power 3.64 5.2 1 1 5 Target 1
BTE 26.12 35.42 1 0.1 5 Maximize 0.998
BSFC 0.234 0.358 0.1 1 5 Minimize 1.000
CO 0.028 2.287 0.1 1 5 Minimize 0.983
HC 31.73 298 0.1 1 5 Minimize 1.000
NOx 725.6 1334.7 0.1 1 5 Minimize 1.000
Combined 0.995

7.5 MODEL EQUATION FOR THE RESPONSE USING

REGRESSION AND C ++ PROGRAM DEVELOPED

A relationship between the chosen design and control parameters

and the responses variables BTE, BSFC, NOx, HC and CO are established
through multiple regression modeling. These regression equations are very
much useful to predict the response variable without experimental testing and
optimization techniques. A good agreement is observed between the measured
output variables and the predicted values. A C++ program is developed using
this model which can be used to predict the responses for different input
parameters.
 212

The empirical model developed through regression for BTE is

given in equation (7.18)

ηBTH = −1.353CR - 0.032IP - 0.301IT - 0.1278B- 1.586P+ 79.60 (7.18)

From the developed regression model,the BTE ,value can be

calculated for the sixteen experiments and compared with them.It is shown in
Table 7.26. From the Table 7.26 it is obseved that the equation is valid for
BTE as error % is within limits.

The empirical model developed through regression for BSFC is

given in equation (7.19). From the developed regression model,the BSFC can
be calculated for the sixteen experiments and compared with them.It is shown
in Table 7.27. From the table it is obseved that the equation is valid for BSFC
as error % is within limits.

Table 7.26 The BTE value for the Model developed

BTE Regression
Expt. No. BTE Experimental Error
model
1 34.41 34 -0.41
2 32.03 33 0.97
3 29.65 28 -1.65
4 26.32 27 0.68
5 28.51 24 -4.51
6 27.70 29 1.3
7 32.16 34 1.84
8 32.96 33 0.04
9 26.73 30 3.27
10 29.83 26 -3.83
11 30.77 34 3.23
12 33.75 35 1.25
13 29.69 30 0.31
14 30.24 31 0.76
15 28.54 29 0.46
16 30.70 27 -3.7
 213

In the model equation developed compression ratio and power

factor has more influence on BSFC,injection pressure has less effect.

BSFC = 0.0344CR + 0.00032IP + 0.0033IT + 0.0014B + 0.0202P - 0.59 (7.19)

Table 7.27 The BSFC value for the Model developed

Experiment BSFC BSFC Error

Regression Experimental
1 0.25 0.25 0.00
2 0.28 0.275 -0.005
3 0.30 0.31 0.01
4 0.34 0.34 0.00
5 0.32 0.39 0.07
6 0.33 0.31 -0.02
7 0.28 0.26 -0.02
8 0.27 0.27 0.00
9 0.34 0.31 -0.03
10 0.31 0.37 0.06
11 0.30 0.27 -0.03
12 0.27 0.26 -0.01
13 0.31 0.3 -0.01
14 0.31 0.3 -0.01
15 0.33 0.32 -0.01
16 0.31 0.35 0.04

The empirical model developed through regression for Emission

NOx,, HC, CO is given in equations (7.20), (7.21) and (7.22).

From the developed regression model, the Emission of NOx,

HC,CO can be calculated from the sixteen experiments and compared with
them, is shown in Table 7.28. It is obseved that the equation is valid for NOx
and CO, as error % is within limits.
 214

In the model equation developed for NOx, compression ratio,

power has more influence on NOx and Injection timing has less influence.

NOX = −121.50 CR + 22.07 IP + 1.74 IT - 4.77 B + 99.95P - 1764(7.20)

HC = 40.5CR - 3.12 IP -1.64 IT - 0.170B + 43.9 P - 115 (7.21)

CO = −0.27 CR - 0.055IP + 0.58 IT - 0..002B + 1.963 (7.22)

Table 7.28 The Emission value for the Model developed

CO CO NOx
Exp. No.
Regression Experiment Regression Experiment
1 0.075407 0.04 1541.988 1572
2 0.260186 0.25 1329.042 1300
3 0.444966 0.49 1116.096 1057
4 0.654746 0.6 634.75 690
5 0.684476 0.8 1581.69 1606
6 0.571196 0.52 1210.136 1300
7 0.167957 0.32 1138.702 1122
8 0.036677 0.02 594.148 627
9 0.293447 0.16 1361.522 1258
10 0.013688 0.04 1187.728 1100
11 0.716466 0.72 1160.21 1133
12 0.261687 0.13 666.042 662
13 0.194958 0.11 1483.802 1720
14 0.617237 0.6 1411.27 1300
15 0.146918 0.3 840.368 800
16 0.551197 0.59 594.836 625
 215

The empirical model developed through regression modeling is

used as an input to the C++ PROGRAM written for this model given in
Annexure I. Using this program we can calculate the value of BTE, BSFC,
CO, NOx and HC for any number of inputs.

The optimum value is obtained using this program. Thus a simple

execution method is shown below.

Step 1: Open an exe(ex – Again_modified.exe) file.

Step 2: Please enter number of levels.

Example – Enter number of levels: 1

Step 3: Enter Values for levels:1

Enter values for level

Enter value for Compression Ratio (CR) : 17.9

Enter value for Static Injection Pressure(bar)(IP) : 200

Enter value for Injection Timing( οbTDC)(IT) : 23

Enter value for Biodiesel fuel fraction (%)(B) : 20

Enter value for Power (P) kW : 3.64

Step 4: Select your choice:

If we want to calculate all the response enter: 6

If we want to calculate BTE enter :1

If we want to calculate BSFC enter :2

If we want to calculate NOx enter: 3

If we want to calculate HC enter: 4

If we want to calculate CO enter: 5

 216

If we want to calculate all the response enter: 6

Not Interested enter :0

Enter your OPTION: 6

The results obtained is shown below

Maximum value of BTE is: 33.74 %

Minimum value of BSFC is: 0.267 kg/kWh

Minimum value of NOx is :783 ppm

Minimum value of HC is: 104ppm

Minimum value of CO is:0.016 %

Step 5: Enter 0 to exit.

Using these program 1024 combinations of input parameters can be

given and the responses can be obtained.

A mathematical equation for performance and emission parameters

of the diesel engine was developed. It may not always be convenient to make
experimentation every time while switching over from one blend to another.
Mathematical equations are thus helpful and it allows active optimization of
engine performance without experimentation.

7.6 VALIDATION OF OPTIMUM

After selecting the optimal levels of the engine, the final step is to
verify the results using the optimum design parameter levels in comparison
with standard engine parameters with biodiesel fuel. A confirmation test for
the combined objective is conducted by choosing the five design and control
parameters as found in multi objective optimization. An experiment was
 217

conducted with the optimized parameters and the results are shown in
Table 7.29.

7.6.1 Validation for Taguchi Method

For the combined objective with equal weight age for all the
responses, the optimum setting is found out.

CR-17.9, IP-200 bar, IT-23° bTDC, B20. P -3.64kW is the optimal

combination that gives maximum efficiency and minimum fuel consumption
and emissions.

The Taguchi method of DOE analysis is showing good

reproducibility and the values of the BTE, BSFC, HC, CO and NOx were
found to be 34.75%, 0.2663 kg/kWh, 22 ppm, 0.18% and 598 PPM
respectively.

The confirmation test was conducted with optimized parameters

and the results are shown in the Table 7.29.The experimental value is close to
predicted value. Thus Taguchi optimization is valid.

Table 7.29 The comparison of the prediction and confirmation using

experimental value

Predicted Value at Confirmation test

Response
S.No. optimum condition value at optimum
variable
using Taguchi condition
1. BTE 34.75 % 33 %
2. BSFC 0.239 kg/kW-hr 0.26 kg/kWhr
3. CO 0.015 % 0.02 %
4. HC 22 PPM 35 PPM
5. NOx 598 600 PPM
 218

7.6.2 Validation of Optimized result for RSM Method

In order to validate the optimized result, the experiments were

performed thrice at the optimum compression system parameters. For the
actual responses, the average of three measured results was calculated.
Table 7.30, summarizes the average of experimental values and predicted
values. The validation results indicated that the model developed were quite
accurate as the percentage of error in prediction were in a good agreement.
The results of the predicted value and the experimental values as shown in
Table 7.30 indicate that the equations developed and the optimum thus
obtained, are able to predict the performance and emission characteristics
accurately.

Table 7.30 The comparison of the prediction and confirmation using

experimental value

Predicted Value at Confirmation test

Response
S.No. optimum condition value at optimum
variable
using RSM condition
1. BTE 33.84 % 33 %
2. BSFC 0.268 kg/kWhr 0.26 kg/kWhr
3. CO 0.098 % 0.02 %
4. HC 31 PPM 35 PPM
5. NOx 884 PPM 600 PPM

7.6.3 Comparision of the Optmization Results

The values obtained from Taguchi,RSM and Model regression

equation were compared with the confirmation results.The values are shown
in Table 7.31.
 219

Table 7.31 Comparison between Taguchi ,RSM. Model equation,

Confirmation test

IP Biodiese Power BTE BSFC HC

IT NOx
CO
S.No. Value CR l fuel
bar bTDC fraction kW % Kg/kWh ppm % ppm

1. RSM 17.9 200 23 20 3.64 33.84 0.268 31 0.098 884

2. Taguchi 17.9 200 23 20 3.64 34.75 0.239 22 0.015 598

Model 0.267
3. 17.9 200 23 20 3.64 33.74 104 0.016 783
equation

Experiment
4. 17.9 200 23 20 3.64 33.5 0.26 35 0.020 600
results

7.6.3.1 Brake Thermal Efficiency

From Figure 7.50 it is observed that Response surface methodology

and Model equation give value closer to experiment results, whereas Taguchi
method has slight deviation. The maximum BTE is 34.75%.

35
34.8
34.6
34.4
34.2
BTE (%)

34
33.8
33.6
33.4
33.2
33
32.8
RSM Taguchi Model Experimental
BTH 33.8 34.75 33.74 33.5

Figure 7.50 Comparison of brake thermal efficiency

 220

7.6.3.2 Brake specific fuel consumption

From Figure 7.51 the experimental value for the brake specific fuel
consumption for RSM and empirical model values are nearer to experimental
results. Whereas Taguchi optimization gives more deviation result.

0.275
0.27
0.265
0.26
bsfc (kg/kWh)

0.255
0.25
0.245
0.24
0.235
0.23
0.225
0.22
RSM Taguchi Model Experimental
bsfc 0.268 0.239 0.267 0.26

Figure 7.51 Comparison of brake specific fuel consumption

7.6.3 .2 Hydro Carbon Emissions

From Figure 7.52 the experiment result shows the HC emission is

35PPM that is close to RSM. Whereas the model equation results are high and
Taguchi results are low.

For the given parameters the maximum and minimum HC produced

is 0.02g/kWh and 0.004 g/kWh that is very less compared to EURO – IV
Norms (max 0.55 g/kW-hr). The shorter ignition delay associated with
biodiesel higher Cetane number could also reduce the over mixed fuel which
is the primary source of un-burnt hydrocarbons.
 221

120

100

80
HC (ppm)

60

40

20

0
RSM Taguchi Model Experimental
HC 31 22 104 35

Figure 7.53 Comparisons of the Hydrocarbon Emissions

7.6.3.4 NOx Emissions

For NOx emission, the experimental value for the optimized results
is 600 PPM. Taguchi and model equation values are closer to experimental
values whereas RSM optimization techniques result is not valid. The value of
NOx emission is 884 PPM when RSM optimization is used.

The maximum and minimum NOx produced is 1.61g/kW-h and

0.99g/kW-h, that is much less than, mentioned in EURO – IV Norms (max
3.5g/kW-h).
 222

1000
900
800
700
Nox (ppm)

600
500
400
300
200
100
0
RSM Taguchi Model Experimental
NOx 884 598 783 600

Figure 7.53 Comparisons of the NOx Emissions

7.6.3.5 CO Emission

From Figure 7.54 the experiment value of CO emissions is 0.02 %

which is very close to the value obtained from taguchi and model equation,
But RSM value is too high.

The maximum and minimum CO produced is 0.01g/kW-h and

0.00165 g/kWh, which is much less than, mentioned in EURO - IV Norms
(max 1.5g/kWh). It is an indication of the complete combustion of biodiesel
being an oxygenated fuel.
 223

0.12

0.1

0.08
CO ( %)

0.06
CO %
0.04

0.02

0
RSM Taguchi Model Experiment
CO % 0.098 0.015 0.016 0.02

Figure 7.54 Comparisons of the CO Emissions

7.7 CONCLUSION

In this work, the influence of compression ratio, injection pressure,

injection timing and power on the performance and emission characteristics
for various karanja biodiesel- diesel blends (10, 20, 30 and 50 %) has been
investigated by conducting experiments in a naturally aspirated four
stroke,1500 rpm compression ignition direct injection of 5.2 kW water-cooled
diesel engines. Using this experimental data, functional relationships for BTE,
BSFC, CO, HC and NOx, have been formulated in terms of compression
ratio, injection pressure, injection timing, blend ratio and power. These
functional forms were then used to accomplish a multi-objective optimization.
The engine operating parameters considered are to maximize brake thermal
efficiency and to minimize the fuel consumption and emissions.

Based on the results of this study ,the following conclusions were

drawn, in terms of fuel properties and exhaust emission characteristics.
 224

Taguchi optimization method, is more valid for single objective

optimization only .

The maximum BTE of 39.5 % was obtained when CR 17.9, IP

210bar, IT 23 ο Btdc, blend 10 % and Power 4.16 kW.It is only predicted
value.

The minimum value of BSFC (0.4074 kg/kW hr) was obtained at

CR 18.1, IP 220bar, IT 29 ο Btdc, blend 30 % and Power 5.2 kW.

The minimum value of CO (1.075 % )was obtained at CR 17.9, IP

230bar, IT 29ο Btdc, blend 10 %, and Power 3.64 kW.

The minimum value of HC 122 ppm was obtained at CR 17.7, IP

230bar, IT 23 ο Btdc, blend 20 % and Power 3.64 Kw.

The emissions NOx (318 ppm )was minimum at CR 17.9, IP

190bar, IT 27 ο Btdc, blend 50 % and Power 3.64 kW.

It is very clear using single objective optimization using taguchi we

obtain only different results. So we have to go for multiobjective
optimizations.

To obtain an optimal combination of engine parameters considering

performance and emissions multiple objective optimizations is used.

A weightage of 0.2 was given to all responses in multi objective

optimization; the optimum setting is found out, CR-17.9, IP-200 bar, IT-23°
bTDC, B20. P -3.64 kW and the values of the BTE, BSFC, HC, CO and
NOx were found to be 34.75%, 0.2393 kg/kWh, 22 ppm, 0.015% and 598
PPM respectively.
 225

RSM optimization was used in multi objective optimization to

optimize the parameters. The study intended to make use of the RSM process
to maximize the efficiency and minimize emissions. It was a desire to develop
a mathematical model which would describe the relationships between the
variables, using this model we can predict the results without doing
experiments.

To study the interactions effect between the input parameters and

for combined objective. (to maximize efficiency and minimize fuel
consumption ,emissions), RSM optimization is very effective method.

This experimental design, considerably reduced the time required

by minimizing the number of experiments to be performed and provided
statistically proven models for all response.

The ranking of parameters based on their influence on brake

thermal efficiency fuel consumption and emissions and the combined effect
are found out.

When the blend is low (B 10),the BTE is high 33.9 % and fuel
consumption is less and it is minimum when the blend is maximum (B 50)
this is due to the fact at 50 % blend the viscosity and density is is high and
HHV is less.

When the injection pressure is maximum (230 bar),the BTE is high

34.02 % this is due to the fact when the injection pressure is high because of
better combustion due to finer breakup of fuel droplets providing more
surface area and better mixing with air. In summary of this part of study,
RSM proved to be a powerful tool for the optimization of engine input
parameters. A second-order model was successfully developed to describe the
relationships between BTE, BSFC, CO, HC, NOx and (CR,IT,IP,B,P) The
 226

optimal conditions for the maximum BTE and minimum BSFC, HC, CO,
NOx with Desirability approach of 0.995 was obtained at the optimum engine
parameters of compression ratio of 17.9,injection pressure 200 bar, injection
timing 23 o bTDC, fuel blend B20 and 3.64kW power, where the values of the
BTE,BSFC,CO,HC and NOx were found to be 33.85%, 0.268 kg/kWh, 0.09
%, 31 ppm, 884 ppm respectively.

Using multiple regression a mathematical model was developed for

all the response.The mathematical model was useful to calculate the output
response without doing experiments.

Validation experiment is done for the optimized input

parmeters,that gives the results with very little error percentage.

The optimum combinations recommended are valid within

the selected range of input parameters, the optimum setting are CR-17.9, IP-
200 bar, IT-23° bTDC, B20. P -3.64kW is the optimal combination to achieve
multiple–performance characteristics of the engine.

From the comparison of the above three methods with the

confirmation test for the selected parameters, the responses obtained are very
close to values for all the methods, only there is a deviation in NOx emission
in RSM optimization. Hence, it can be concluded that for the present study in
a water cooled direct injection single cylinder 5.2 kW naturally aspirated
diesel engine, all the three methods give good results, that is close to the
optimized results obtained through confirmation results, for a compression
ratio of 17.9,injection pressure of 200 bar, injection timing of 23° bTDC,
power 3.64 kW and 20% karanja biodiesel blend, the output responses were
BTE -33.5 %, BSFC- 0.261 kg/kW h, HC-35 PPM,CO -0.02 % and NOx-600
PPM.