Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

CiiT International Journal of Artificial Intelligent Systems and Machine Learning, Vol 4, No 11, November 2012 612

Waste Cooking Oil Bio Diesel Performance Analysis


in Variable Compression Ratio Diesel Engine Using
Functional Back Propagation Algorithm
M. Suresh, Maheswar Dutta and S. Purushothaman

Abstract---This paper presents the implementation of functional producing fatty acid esters. The various multipurpose oils are
back propagation algorithm (FUBPA) for estimating the power, also used as biofuel such as Castor oil, Coconut oil (copra oil),
torque, specific fuel consumption and presence of carbon monoxide, Colza oil, Corn oil, Cottonseed oil, False flax oil, Hemp oil,
hydrocarbons in the emission of a direct injection diesel engine. Mustard oil, Palm oil, Peanut oil, Radish oil. Rapeseed oil,
Experimental readings were obtained using the biodiesel prepared
from the waste cooking oil collected from the canteen of Sri Sairam
Ramtil oil, Rice bran oil, Safflower oil, Salicornia oil,
Engineering College, India. This waste cooking oil was due to the Soybean oil, Sunflower oil, Tigernut oil, Tung oil, are lists of
preparation of varieties of food (vegetables fried and non vegetarian). vegetable oils that are suitable for biodiesel. Similarly,
To obtain the biodiesel, transesterification was done in chemical lab inedible oils are used only or primarily as biofuel such as
for more than a week, and the biodiesel was obtained. The biodiesel Copaiba, Honge oil, Jatropha oil, Jojoba oil, Milk bush, Nahor
was mixed in proportions of 10%, 20 % , 30%,40%, 50% with oil, Paradise oil, Petroleum nut oil.
remaining combinations of the diesel supplied by the Indian
government. Variable compression ratio (VCR) diesel engine with
Vegetable oils are evaluated for use as a biofuel based on:
single cylinder, 4 stroke diesel type was used. a) Suitability as a fuel, based on flash point, energy content,
The outputs of the engine as power, torque and specific fuel viscosity, combustion products and other factors b) Cost,
consumption were obtained from the computational facility attached based in part on yield, effort required to grow and harvest, and
to the engine. The data collected for different input conditions of the post-harvest processing cost.
engine was further used to train FUBPA. Alternative fuels for diesel engines are becoming
The trained FUBPA network was further used to predict the increasingly important due to diminishing petroleum reserves
power, torque and SFC for different speed, biodiesel and diesel
and the environmental consequences of exhaust gases from
combinations and full load conditions. The estimation performance of
the FUBPA network is discussed.
petroleum fuelled engines. A number of studies have shown
that triglycerides hold promise as alternative diesel engine
fuels. So, many countries are interested in that.
Keywords---Functional Back Propagation Algorithm, Waste
Cooking Oil, Biodiesel
II. EXPERIMENTAL INVESTIGATION
I. INTRODUCTION The setup consists of single cylinder, four stroke, VCR
N this paper, performance of a diesel engine and exhaust (Variable Compression Ratio) diesel engine connected to eddy
I emission content of the diesel engine when using biodiesel
blended with diesel has been analyzed. Data collected from
current type dynamometer for loading. The compression ratio
can be changed without stopping the engine and without
the engine for various loads / speed were used to train altering the combustion chamber geometry by specially
functional back propagation (FUBPA) neural networks.. designed tilting cylinder block arrangement. Setup is provided
Subsequently, the FUBPA was used to estimate the with necessary instruments for combustion pressure
performance of the diesel engine and estimate the quality of measurements. The setup has stand-alone panel box consisting
the exhaust gas for different loads / speeds and combinations of air box, two fuel tanks for duel fuel test, manometer, fuel
of biodiesel fuel other than that used for training the FUBPA. measuring unit, transmitters for air and fuel flow
Biodiesel refers to a vegetable oil or animal based diesel measurements, process indicator and engine indicator.
fuel consisting of long chain (methyl, propyl or ethyl) esters. Rotameters are provided for cooling water and calorimeter
Biodiesel is typically made [1-2], by chemically reacting lipids water flow measurement.
(eg. Vegetable oil, animal fat, tallow) with an alcohol The setup enables study of VCR engine performance for
brake power, indicated power, frictional power, brake mean
effective pressure (BMEP), indicated mean effective pressure
Manuscript received on November 05, 2012, review completed on
November 17, 2012 and revised on November 27, 2012. (IMEP), brake thermal efficiency, indicated thermal
M. Suresh, Assistant Professor, Mechanical Engineering, Sri Sai Ram efficiency, mechanical efficiency, volumetric efficiency,
Engineering, College, Chennai- 44 specific fuel consumption, Air/Fuel ratio and heat balance.
Maheswar Dutta, Professor and Principal, M.N.R Engineering , College,
Hyderabad Labview based Engine Performance Analysis software
S. Purushothaman, Professor and Dean, Mechanical Engineering, Udaya package ―EnginesoftLV‖ is provided for on line performance
School of Engineering, India-629204. E-Mail: drsppuru@gmail.com
Digital Object Identifier No: AIML112012003.

0974-9667/CIIT–IJ-3910/06/$20/$100 © 2012 CiiT Published by the Coimbatore Institute of Information Technology


CiiT International Journal of Artificial Intelligent Systems and Machine Learning, Vol 4, No 11, November 2012 613

evaluation. The various measurements that are used in the F3,F4 Calorimeter water flow kg/hr
experimental measurements are: T3 Calorimeter water inlet temperature oK
Brake power (BP)= 2 * π * n T/ (60 * 1000) T2,T4 Calorimeter water outlet temperature oK
Brake specific fuel consumption (Kg/kwh)= Fuel flow in T5 Exhaust gas to calorimeter inlet temp. oK
kg / hour / BP T6 Exhaust gas from calorimeter outlet temp. oK
Specific fuel consumption (SFC): Brake specific fuel There are two fuel tanks, one is for diesel fuel and the other
consumption and indicated specific fuel consumption, for fuel blends. The engine under study is a VCR, water
abbreviated BSFC and ISFC, are the fuel consumptions cooled single cylinder, in-line, naturally aspirated, Kirlosker
on the basis of Brake power and Indicated power diesel engine. The test engine was coupled to an electric eddy
respectively. current dynamometer. A vehicle gas analyzer model was used
for measuring CO and HC emissions. Engine was run at
A. Biodiesel preparation several speeds at full load and power, torque, fuel
In the present investigation, biodiesel [4-8] was produced consumption and emissions was measured. Table 1 presents
from waste cooking oil. Two gram Alkali catalyst and 35 cc experimental data obtained. Table 2 presents the CO and HC
methanol (as an alcohol) was applied for 150 gram waste emissions value obtained from exhaust gas analyzer.
cooking oil in this reaction. Biodiesel production reaction time
was two hours with stirring and with moderate heat. Upto two TABLE 1
EXPERIMENTAL DATA OBTAINED
weeks time is needed for separation. The waste cooking oil

Litre/ KW-hour
Waste cooking

Speed(rpm) of

consumption
Power (KW)

Specific fuel
methyl ester was added to diesel fuel in 10 to 50 percent ratios

(Biodiesel)

the engine
Full load

Torque
(N-M)
Diesel

(SFC)
S.No.
and then used as fuel for one cylinder diesel engine.

oil
B. Experimental set up and test procedure
The experimental setup consists of single cylinder diesel 1 1 0 1 1200 6.2 48 0.32
engine, an engine test bed and a gas analyzer. The engine 2 1 0 1 1600 9.2 54 0.33
setup is shown in Figure 1. The schematic of the experimental 3 1 0 1 2000 12.3 57 0.34
setup is shown in Figure 2. 4 1 0 1 2400 16.0 63 0.33
5 1 0 1 2800 17.6 53 0.33
6 1 0 1 3200 17.7 51 0.34
7 1 10 90 1200 7.0 54 0.34
8 1 10 90 1600 9.8 57 0.35
9 1 10 90 2000 12.0 56 0.33
10 1 10 90 2400 15.2 62 0.30
11 1 10 90 2800 16.1 55 0.31
12 1 10 90 3200 16.3 48 0.37
13 1 20 80 1200 6.6 51 0.33
14 1 20 80 1600 9.2 53 0.33
15 1 20 80 2000 12.8 55 0.30
16 1 20 80 2400 16.3 58 0.29
17 1 20 80 2800 16.8 54 0.32
18 1 20 80 3200 18.0 52 0.33
19 1 30 70 1200 6.8 47 0.32
20 1 30 70 1600 9.6 51 0.31
21 1 30 70 2000 12.4 57 0.29
22 1 30 70 2400 15.0 64 0.34
23 1 30 70 2800 16.8 59 0.33
24 1 30 70 3200 17.4 48 0.36
Fig 1 Variable Compression Ratio Diesel Engine (Apex Innovations) 25 1 40 60 1200 6.0 52 0.32
26 1 40 60 1600 9.6 56 0.31
27 1 40 60 2000 12.4 58 0.27
28 1 40 60 2400 15.0 59 0.31
29 1 40 60 2800 18.0 56 0.32
30 1 40 60 3200 17.0 53 0.34
31 1 50 50 1200 6.2 48 0.31
32 1 50 50 1600 9.0 53 0.32
33 1 50 50 2000 12.4 56 0.33
34 1 50 50 2400 15.8 59 0.32
35 1 50 50 2800 17.0 58 0.33
36 1 50 50 3200 16.8 50 0.36

Fig 2 Schematic Layout of the Setup (Apex Innovations)


Where
F1 Fuel consumption kg/hr
F2 Air consumption kg/hr

0974-9667/CIIT–IJ-3910/06/$20/$100 © 2012 CiiT Published by the Coimbatore Institute of Information Technology


CiiT International Journal of Artificial Intelligent Systems and Machine Learning, Vol 4, No 11, November 2012 614

TABLE 2 D = | Desired output-Network output | (3)


EXHAUST GAS ANALYZER OUTPUT
Fuel blend HC C
If ‗nm‘ is empty, i.e. Not even one node in the outer layer
B0 32 0.48 does satisfy Equation (2), go to step 2.
B10 18 0.49 Step 5: If ‗nm‘ is not empty, the objective function
B20 16 0.46 ‗j‘ is computed by
B30 5 0.45
B40 7 0.4 J= 1
2 D2 (4)
B50 7 0.38 Xi nm
Step 6: The weights and thresholds are updated.
III. FUNCTIONAL UPDATE BACK PROPAGATION Step 7: The steps (2-6) are repeated, until the total MSE of
ALGORITHM all the patterns is below a specified value. The main advantage
The concept of functional update method indicates that of FUBPA is that it will stop as soon as the misclassified set is
there will be weight updating of the connection between layers empty.
only if at least one node in the output layer is misclassified. A. Advantages of Fubpa Over Conventional Update
The meaning of misclassification is valid when the difference Methods
between the target and the actual value of node in the output
When the network is trained with analog data by
layer is more than 0.5. By converting the analog patterns into
conventional update method the number of iterations is large
binary patterns, the convergence and classification behavior
for the objective function (J), to reach the desired MSE. The
change. When binary patterns are presented in BPA with
objective function does not reach the desired MSE due to
modification in the convergence condition, the algorithms is
some local minima, whose domains of attractions are as large
called FUBPA.
as that for the global minimum. The network converges to one
The flowchart for BPA with functional update is given in
of those local minima or the network diverges. The updating
Figure 3. The number of layers and the number of nodes in the
of the weights will not stop, unless every input is outside the
hidden layers are decided. The weights between layers, for
significant update region (0.1 to 0.9) and the outputs of the
hidden and output layers are initialized. A training pattern is
network will be approaching either 0 or 1. This requires much
presented to the input layer of the network and the difference
iteration for the network to converge. To overcome these
between the network‘s output and the target output is
difficulties a functional criterion that results in faster
calculated for each node in the output layer. If the data is
convergence of the network is used.
analog, the number of nodes in the output layer is 1. If the data
is binary, the number of nodes in the output layer is more
IV. RESULTS AND DISCUSSION
depending upon binary conversion. If the difference obtained
for each node is greater than the value of functional criteria, a In Figure 4, the various parameters used for training the
counter is incremented and the weights and thresholds are FUBPA ANN is shown. The parameters are, 1) load, 2)
updated. If the difference of not even one node is greater than biodiesel, 3) diesel and 4) speed of the engine. The targets
0.5 then the weights and thresholds are not updated. The MSE used in the output layer are power of the engine, Torques of
of the network for each pattern is calculated, when at least one the engine, brake SFC and quality of exhaust.
node in the output of the network is misclassified. Remaining A. Properties of waste cooking oil biodiesel fuels
training patterns are presented to the network. Training of the Property Biodiesel
network is stopped, when a performance index of the network Flash point, closed cup 182 °C
is reached. Pour point -3°C
The algorithm for the functional update is as follows: Kinematical viscosity, 40°C 4.15 mm2/s
Total Sulfur 0.0018 wt. %
Step 1: The weights of the network are initialized. Copper strip corrosion 1a
Step 2: The inputs and outputs of a pattern are Cloud point 0 °C
presented to the network.
Step 3: The output of each node in the successive B. Torque and Power
layers is calculated by Fuel rack is placed in maximum fuel injection position for
Xi = 1/ (1+exp (- Wij Xi)) (1) full load conditions. The engine is loaded slowly. The engine
Where speed is reduced with increasing load. Range of speed was
i=1 to 3 selected between 1200 – 3600 rpm. Engine test results with
Xi represents no. of nodes in the o/p layer net diesel fuel showed that maximum torque was 64.2 Nm
Step 4: The number of nodes in the output layer, which occurred at 2400 rpm. The maximum power was 18.12
which are misclassified, kW at 3200 rpm. Power and torque for fuel blends at full load
are denoted by ‗nm‘. A node is misclassified, if it does not is shown in Table 1. Power and torque performance with fuel
satisfy the equation. blends, are similar to net diesel fuel. The training and
1- > D 0.5 (2) (2) estimation performance of FUBPA for torque is presented in
Where Figure 5a-c and for Torque in Figure 6a-c.
is the value fixed by the programmer, and

0974-9667/CIIT–IJ-3910/06/$20/$100 © 2012 CiiT Published by the Coimbatore Institute of Information Technology


CiiT International Journal of Artificial Intelligent Systems and Machine Learning, Vol 4, No 11, November 2012 615

2
C. Brake specific Fuel consumption 4X3X1

Fuel consumption curves of net diesel fuel at full load are


1.5
shown in Fig. 7c. The curves show that fuel consumption at
full load condition and low speeds is high. Fuel consumption

MSE
first decreases and then increases with increasing speed. The 1

estimate by FUBPA is shown in blue color.


0.5

0
0 500 1000 1500 2000 2500 3000
Iterations
(5a)
100
4X3X1

Percentage Classifications
80

60

40

20

0
0 500 1000 1500 2000 2500 3000
Iterations
(5b)
2

1.5
Power, KW Target
FUBPA Estimate

0.5
0 5 10 15 20 25 30 35 40
Patterns

(5c)
Fig.5a-c Estimation of Power
0.16
4X3X1
0.14

0.12

0.1
MSE

0.08

0.06

0.04

0.02
1 2 3
Iterations

(6a)
Fig 3 Flowchart for FUBPA

(6b)
Fig 4 FUBPA Network for Predicting Engine Performance

0974-9667/CIIT–IJ-3910/06/$20/$100 © 2012 CiiT Published by the Coimbatore Institute of Information Technology


CiiT International Journal of Artificial Intelligent Systems and Machine Learning, Vol 4, No 11, November 2012 616

6.5
Target work can include further experiments that can be conducted
FUBPA Estimate
with different operating conditions.
6

Torque
5.5
REFERENCES
[1] Alamu O.J., Waheed M.A., Jekayinfa S.O. and Akintola T.A., Optimal
Transesterification Duration for Biodiesel Production from Nigerian
5
Palm Kernel Oil, Agric Eng‘g Int: the CIGR E journal, Manuscript EE
07 018., Vol. 9, p.1-11 , 2007.
4.5 [2] Al-Widyan, M.I.; Al-Shyoukh, A.O., Experimental evaluation of the
0 5 10 15 20 25 30 35 40
Patterns transesterification of waste palm oil into biodiesel Bioresource
(6c) Technology, 2002, Vol. 85, Issue 3, pp. 253-256.
Fig. 6a-c Estimation of Torque [3] Andrade, J. B., Pinto, A. C., Guarieiro, L. L. N., Rezende, M. J. C.,
Ribeiro, N. M., Torres,E. A., Lopes, W. A., Pereira, P. A. P.
Biodiesel:An Overview – J. Braz. Chem. Soc., Vol. 16, No. 6B, p. 1313-
0.5 1330, 2005.
4X3X1
[4] Bento, F.M.; Camargo, F.A.O., Okeke, B.C., and Frankenberger, W.T.,
0.4 Comparative bioremediation of soils contaminated with diesel oil by
natural attenuation, biostimulation and bioaugmentation ,Bioresource
0.3
Technology, 2005, Vol. 96, Issue 9, pp. 1049-1055.
[5] Bugaje I.M and Mohammed I.A., Biofuel Production Technology, Sci.
MSE

and Tech. Forum (STF) Zaria, Nigeria 1 st Edition, p.25-200, 2008.


0.2
[6] Canakc M I, Erdil A, Arcaklioglu E., Performance and exhaust
emissions of a biodiesel engine. Applied Energy, 2006, Vol.83, pp.594–
0.1
605.
[7] Chinmoy Baroi, Ernest K. Yanful and Maurice A. Bergougnou,
0 Biodiesel production from Jatropha curcas oil using potassium
0 500 1000 1500 2000 2500 3000 3500 4000 4500
Iterations carbonate as an unsupported catalyst, Int. J. of Chem. Reaction
(7a) Eng‘g.,Vol. 7 ,No.72, p.1-20 , 2009.
[8] Demirbas A. Importance of biodiesel astransportation fuel - Energy
Policy 35, p. 4661–4670,2007.
[9] Demirbas A., Biodiesel from Vegetable Oils via Transesterification in
Supercritical Methanol, Energy Conversion and Management, Vol.43,
p.2349-56, 2002.
[10] Jefferson S. de Olivera, Polyanna M. Leite., Lincoln B. de Souza,
Vinicius M. Mello, Eid C. Silva, Joel C. Rubim, Simoni M.P.
Meneghethi, Paulo A.Z Suarez. Characteristics and composition of
Jatropha gossypiifolia and Jatropha curcas L. oils and application for
biodiesel production, Biomass and Bioenergy, Vol.33, p.449–453 ,
2009.
[11] Karmee, S.K.; Chadha, A., Preparation of biodiesel from crude oil of
(7b) Pongamia pinnata, Bioresource Technology, 2005, Vol. 96, Issue 13,
pp.1425-1429.
0.38
[12] Lang, X., Dalai, A.K., Bakhshi, N.N., Reaney, M.J., Hertz, P.B.,
Target Preparation and characterization of bio-diesels from various bio-oils,
0.36
FUBPA Estimate
Bioresource Technology, 2001, Vol. 80, Issue 1, pp. 53-62.
0.34 [13] Lima, Daniela G., Valerio C.D., et al., Diesel-like fuel obtained by
pyrolysis of vegetable oils, J Anal Appl Pyrolysis, Vol.71,p. 987–96,
B-SFC

0.32
2004.
0.3
[14] Linus N. and Okoro et.al, Synthesis, Calorimetric and Viscometric
Study of Groundnut oil Biodiesel and Blends, Res. J. Chem. Sci.,
0.28 Vol.1,No.3, p. 49-57 ,2011.
[15] Murugesan, A., Umarani, C.,Subramanian, R., Nedunchezhian, N. Bio-
0.26
1 2 3 4 5 6 7 8 9 10 diesel as an alternative fuel for diesel engines—Areview - Renew
Patterns
Sustain Energy Rev (2008),doi:10.1016/j.rser.2007.10.007.
(7c) [16] Noureddini, H.; Gao, X.; Philkana, R.S, Immobilized Pseudomonas
Fig. 7a-c Estimation of Brake Specific Fuel Consumption cepacia lipase for biodiesel fuel production from soybean oil,
Bioresource Technology, 2005, Vol. 96, Issue 7, pp. 769-777.
[17] Palmarola-Adrados, B.; Choteborska, P.; Galbe, M.; Zacchi, G, Ethanol
V. CONCLUSION production from non-starch carbohydrates of wheat bran, Bioresource
The engine has been tested under same operating conditions Technology, 2005, Vol. 96, Issue 7, pp.843-850.
[18] Scholz, V., Silva, J. N. Prospects and risks of useof castor oil as a fuel
with diesel fuel and waste cooking biodiesel fuel blends. The Biomass and Bioenergy 32, p.95-100, 2008.
results were found to be very comparable. The maximum [19] Shay E.G., Diesel fuel from vegetable oils: status and opportunities,
power and torque produced using diesel fuel was 18.2 kW and Biomass Bioenergy,Vol. 4, p. 227–242, 1993.
[20] Singh S.P. and Singh D., Biodiesel production through the use different
64.2 Nm at 3200 and 2400 rpm respectively. By adding 20%
sources and characterization of oils and their esters as the substitute of
of waste cooking oil methyl ester, the maximum power and biodiesel: A review. Renewable and Sustainable Energy Reviews,
torque increased by 2.7% and 2.9% respectively. The Vol.12, p.200-16 , 2010.
concentration of the CO and HC emissions were significantly [21] Tint T.K. and Mya M., Production of Biodiesel from Jatropha oil
(Jatropha curcas) in Pilot Plant, World Academy of Science,
decreased when biodiesel was used (Table 2). FUBPA was Engineering and Technology, p.477-480 , 2009.
used with the collected data of this work from VCR. Future [22] Veljkovic´ V.B., Lakicevic S.H., Stamenkovic O.S., Todorovic Z.B.
and Lazic K.L., Biodiesel production from tobacco (Nicotiana tabacum

0974-9667/CIIT–IJ-3910/06/$20/$100 © 2012 CiiT Published by the Coimbatore Institute of Information Technology


CiiT International Journal of Artificial Intelligent Systems and Machine Learning, Vol 4, No 11, November 2012 617

L.) seed oil with a high content of free fatty acids, Fuel, Vol.85,
p.2671–2675, 2006.
[23] Volkhard Scholz, Jadir Nogueira da Silva, Prospects and risks of the use
of castor oil as fuel, Biomass and Bioenergy, Vol.32, p.95-100, 2008.
[24] Zhang Y., Dube M.A., Mclean D.D. and Kates M., Biodiesel production
from waste cooking oil: Bioresource Technology, Vol.90, p.2225–2240,
2003.
[25] Zullaikah, S.; Lai, C.C.; Vali, S.R.; Ju, Y.H., A two-step acid-catalyzed
process for the production of biodiesel from rice bran oil, Bioresource
Technology, 2005, Vol. 96, Issue 17, pp.1889-1896.

M. Suresh completed B.E.. in 1988 at Government


College of Engineering, Salem. He Completed M.E. ,
Thermal power Engineering in 1995, Annamalai
Univesity. Indiustrial experience two and half years in
project at Madras Fertilizer Ltd, Chennai. He has 20
years of teaching experience.At present working as
senior faculty in Sri Sai Ram Engineering College,
Chennai,Tamilnadu.His research area –I.C.Engine
alternate fuel using Artifical neural network

Dr.D. Maheswar completed his PhD in JNT University,


Hyderabad in the area of alternate fuels in I.C Engines.
At present he is working in M.N.R college of
Engineering at Hyderabad as principal and professor of
Mechanical Engineering. At present, he has 12 PhD
research scholars under his guidance. His two scholars
have been awarded PhD. He has published 58
international papers to his credit

Dr.S. Purushothaman completed his PhD from Indian


Institute of Technology Madras, India in 1995. He has
109 publications to his credit. He has 19 years of
teaching experience. Presently he is working as
Professor and Dean in Udaya School of Engineering,
India

0974-9667/CIIT–IJ-3910/06/$20/$100 © 2012 CiiT Published by the Coimbatore Institute of Information Technology

You might also like