ADVANCES

IN
ENGINEERING
TECHNOLOGY
Volume - 1

Chief Editor
Dr. Jaivir Singh
Associate Professor, Department of Engineering, SVPUAT Mpdipuram,
Meerut, Uttar Pradesh, India

AkiNik Publications
New Delhi
Published By: AkiNik Publications

AkiNik Publications
169, C-11, Sector - 3,
Rohini, Delhi-110085, India
Toll Free (India) – 18001234070

Chief Editor: Dr. Jaivir Singh

The author/publisher has attempted to trace and acknowledge the materials

reproduced in this publication and apologize if permission and
acknowledgements to publish in this form have not been given. If any material
has not been acknowledged please write and let us know so that we may rectify
it.

© AkiNik Publications
Publication Year: 2018
Pages: 109
ISBN: 978-93-5335-124-3
Price: ` 525/-
 Contents

Chapters Page No.

1. Significance of Product Vending Machine 01-20
(C.S Sundar Ganesh, Selva Girishwaran N and Vishwavijith A)

2. Research Aspects of Bio-Energy Generation from Waste

Using MFC 21-39
(Dr. K. Senthilkumar and M. Naveenkumar)

3. Comparative Study of Cementitious Coated RCC Roof and

Green Roof 41-56
(Raunak Katiyar and Akhilesh Kumar Chauhan)

4. Influences of Engine Speed Variations on Single Cylinder

Direct Injection Diesel Engine 57-67
(Upendra Rajak and Tikendra Nath Verma)

5. Regression and Classification in Machine Learning 69-94

(Anandan R and Kalaivani K)

6. Experimental Investigation of Performance and Emission of

Lal Ambari Biodiesel and Its Blends in a Compression
Ignition Engine 95-109
(Pankaj Shrivastava and Tikendra Nath Verma)
 Chapter - 1
Significance of Product Vending Machine

Authors
C.S Sundar Ganesh
Assistant Professor, Department of Robotics and Automation
Engineering, PSG College of Technology, Coimbatore, Tamil
Nadu, India
Selva Girishwaran N
UG Student Department of Robotics and Automation
Engineering, PSG College of Technology, Coimbatore, Tamil
Nadu, India
Vishwavijith A
UG Student Department of Robotics and Automation
Engineering, PSG College of Technology, Coimbatore, Tamil
Nadu, India

Page | 1
Page | 2
 Chapter - 1
Significance of Product Vending Machine
C.S Sundar Ganesh, Selva Girishwaran N and Vishwavijith A

Abstract
Product Vending machine is an automated machine that dispenses
products such as snacks, beverages etc. to consumers based on cash or
cashless transactions. Different products can be stacked in each tray of the
vending machine. In the modern vending machines heavy items cannot be
stacked as the item falls from a higher level into a lower compartment, so the
main objective of the product is to take the item from the level at which it is
placed and dispenses it into the delivery compartment which makes it easier
to hold heavier items as well. There are two problems in the current market,
one is the Single axis movement which makes it difficult to hold delicately
packed food items and the other one is the Collection Bin which at the
bottom of the machine makes it difficult for elderly people to bend down and
pick the product.
The product vending machine which is now been developed eliminates
the problem of the product falling from a higher level as this machine has
both X and Y axis movements. The delivery bin moves near the item which
is selected by the user and delivers it safely. This ensures that the delicately
packed items are dispensed with no real damage. The main advantage from
the other machine is that the collection bin is kept at a certain height from
the bottom and that makes the machine user friendly.
In the near future cashless transaction methods can include Paytm,
Debit card/Credit card transactions etc. The display used here is an LCD
display and there can be improvements in the future by replacing the LCD
display with a touch screen display. Android apps can also be used for the
transaction. Voice recognition technology can also prove to be a big
breakthrough in the field of vending machines.
Keywords: Product vending machine, embedded system, digital banking,
security system

Page | 3
1. Introduction
A vending machine is an automated machine that provides items such
as snacks, beverages etc. to consumers after money, a credit card, or
specially designed card is inserted into the machine. The first modern
vending machines were developed in England in the early 1880s and
dispensed postcards. Vending machines exist in many countries, and in more
recent times, specialized vending machines that provide less common
products compared to traditional vending machine items have been created
and provided to consumers The first modern coin-operated vending
machines were introduced in London, England in the early 1880s,
dispensing postcards, a German chocolate manufacturer, was selling its
chocolate in 15,000 vending machines. It set up separate companies in
various territories to manufacture vending machines to sell not just
chocolate, but matches, chewing gum and soap products. The first vending
machine in the U.S. was built in 1888 by the Thomas Adams Gum Company
selling gum on New York City train platforms. The idea of adding games to
these machines as a further incentive to buy came in 1897 when the Pulver
Manufacturing Company added small figures, which would move around
whenever somebody bought some gum from their machines. This idea
spawned a whole new type of mechanical device known as the "trade
stimulators. After payment has been tendered, a product may become
available by:
 The machine releasing it, so that it falls in an open compartment at
the bottom, or into a cup, either released first, or put in by the
customer, or
 The unlocking of a door, drawer, or turning of a knob.
Some products need to be prepared to become available. For example,
tickets are printed or magnetized on the spot, and coffee is freshly concocted.
One of the most common form of vending machine, the snack machine,
often uses a metal coil which when ordered rotates to release the product.
The main example of a vending machine giving access to all
merchandise after paying for one item is a newspaper vending machine (also
called vending box) found mainly in the U.S. and Canada. It contains a pile
of identical newspapers. After a sale the door automatically returns to a
locked position. A customer could open the box and take all of the
newspapers or, for the benefit of other customers, leave all of the newspapers
outside of the box.

Page | 4
2. Literature Survey
During the literature survey being reported here, many papers reporting
such researches were gathered from databases. The highlights of the
researches reported in these papers are described in the following parts of the
section.
Asmita P. Bodhale et al. [1] has discussed different beverage machines
each includes unique operation like vend beverage when user will insert coin
or design of vending machine depends on FSM etc. Vending Machine is a
system which can vend or offer different products normally installed in
supermarkets like household goods, railroad lines, offices, institutes and
various other public region. Dispense different products is an automated
process, without man power easy to save precious time of human normally
seen in fast moving cities because of fast paced life. This paper compares
different aspects or response like space, time, speed, power dissipation of a
vending machine that is automated product to create our day to day life more
suitable and to cover market viability.
Mohamed Aamir. M et al. [2] in this research has inevitable and its
demand is increasing steadily particularly in the places such as educational
institutions, government offices, etc. At the same time, time is a precious
thing that one does not want to waste in any way. In stationary shops it is
quite difficult to buy papers during rush time period and the counting of the
paper depending on the requirement would cause further time delay and
there is a chance for the error in the manual counting of paper. To avoid
these problems, this project titled “Automatic Paper Vending Machine” is
proposed to deliver the paper to the public by using the sensors and
microcontrollers based on the Mechatronics principles. It will be more cheap
and economic for the bulk production and it will be very useful for the
college and school students. Here it is designed to deliver sheets by inputting
the respective coin in the system.
Dishita Makwana et al. [3] this is a project on vending machine. A
vending machine is a machine that gives out different kinds of products
when a person inserts a coin into it; therefore it is a coin based vending
machine. These machines can be implemented using various methods but in
this project it has been implemented using 8051 microcontroller. Vending
machines makes it easier for making small purchases, it occupies less space
and it doesn’t need any continuous monitoring.
Charal W. Percy et al. [4] in his patent, they focused on a modular
vending machine for dispensing a variety of different sized products with at

Page | 5
least one array received within a cabinet, wherein the array is made up of a
plurality of storage chambers with a dispensing mechanism disposed at the
bottom of each chamber. The cabinet can receive any number of arrays
which in turn can receive any number of storage chambers. The modular
vending machine further includes a sensing device received within a
receiving trough for confirming that a product has actually been dispensed.
The dispensing mechanism includes a solenoid actuated plunger wherein the
plunger controls the movement of a toggle member which is movable
between a blocking position and a dispensing position so that only a single
product is dispensed.
3. Architecture
In this chapter detailed explanation about the architecture of the vending
machine and the various interface options are discussed in detail. The
vending machine has a controller in which two communication protocols
were used namely RS 232 and USB. Fig. 1 shows the architecture of vending
machine.

Fig 1: Architecture

Page | 6
4. User Interface
4.1 Matrix Keypad

Fig 2: Matrix keypad

From the circuit it is seen that when one of the 16 buttons are pressed, a
pair of pins are connected together. This feature is used to detect the button
that was pressed. Initially all switches are assumed to be released. So there is
no connection between the rows and columns. When any one of the switches
are pressed, the corresponding row and column are connected (short
circuited). This will drive that column pin (initially high) low. Using this
logic, the button press can be detected. Typically, one port pin is required to
read a digital input into the controller. When there are a lot of digital inputs
that have to be read, it is not feasible to allocate one pin for each of them.
This is when a matrix keypad arrangement is used to reduce the pin count.
4.2 Payment Methods
4.2.1 Bank Note Validator
The MSM bill validator consists of two main parts
 The bill validator itself and
 The lockable removable cassette.
The MSM bill validator has a centering mechanism with a self-
adjustable bill path. The width of the path is automatically adjusted to

Page | 7
accommodate each bill. The MSM is used for currencies where the width of
the bill changes with the denomination. The various sensors used are listed
below:
 Six multi coloured optical sensors collect images from both sides of
the bill.
 Patented inductive sensors evaluate magnetic properties of
specialized ink at the bill.
 Patented dielectric sensors detect authenticity of bill paper and
some special protective features of the bill. All sensors have auto-
calibration and do not require any manual adjustment. As a result,
the validator keeps the same high acceptance level during its
lifetime.
 An additional sensor allows reading of bar-coded coupons widely
used in gaming applications. The highest security level is provided
by an anti-stringing sensor that can detect any sort of string, thread
or film attached to the bill.
When new bills are issued in circulation, or when new counterfeits are
found the MSM bill validator has a possibility of a fast and easy software
update with the Cash Code Memory stick. The update can be performed in
seconds, in the location where the validator is installed. The procedure does
not require high-qualified personnel, validator disconnection, or any tool or
equipment. The MSM bill validator operates at a twice faster speed than the
previous ST. Fig. 3 shows the note validator.

Fig 3: Note Validator

Page | 8
 The MSM bill validator has all the required features that allow it to be
used in any application. It can be installed in the STACKER UP or the
STACKER DOWN position. It means that the cassette is located above the
bill entrance (which is more common for vending equipment) or below the
bill entrance (which often used in amusement machines). The MSM fits into
the most common cutout dimensions at the front door. The MSM supports
wide variety of interfaces – MDB, CC serial, CCNET, ID003. Typically, all
connectors are situated at one side of the validator In case, when the space in
the machine at this side is limited, the customer may order the MSM with the
“left-side” connector. This option is available for MDB interface only, The
MSM bill validator operates with the Lockable Removable cassette. These
cassettes are not exactly the same that were used with the previous Cash
Code ST and SM bill validators. All accepted bills are stored in the cassette
as a stack. The cassette cover can be locked with locks of different styles,
limiting access to the stored bills. Cassettes are not differ in the width of the
bill path. Widths are available – 62…77mm. The maximum length of the bill
that can be stored into the cassette is160 mm. depending of available space
in the machine, the cassette of proper size can bechosen. The cassettes are
available in following sizes – 300; 500; 1000 bills. Please keep in mind than
when the cassette capacity is mentioned – it means the space inside the
cassette and how many brand new bills can be stores in this space. Street
grade bills require more space and as a result, fewer bills may be stored. For
additional security the cassette cover can be equipped with the tab for
applying a seal.
4.2.2 Coin Validator
The currenza C2 coin changer makes the vending machine more
profitable & reliable. This is the next cycle coin changer and recycler. Fig.4
shows the coin validator.

Fig 4: Coin Validator

Page | 9
  6 tube cassette system for maximum coin capacity and flexible tube
configurations.
 3 independent payout motors for highest payout security.
 4 human interfaces fulfill individual requirements.
I) Reliability
 Latest NRI validation technology, with patented new sensor system
and 40 measurements of coin parameters
 Unique Payout Redundancy Protection (PRP)-System: 3
independent payout motors and the 6 tube cassette system offer
maximum payout security and minimises machine down time;
because 1 motor controls 2 tubes – in the unikey case of a motor
jam there are still 4 coin tubes for payout available
 Extremely reliable optical sensor system with industry leading
operating temperature of -25 to +80°C guarantees reliable operation
independent of environmental conditions
II) Profitability
 3 independent motors, with payout of up to 3 coins simultaneously
at up to 6 coins/sec. Minimizes transaction times
 4 human interfaces fulfilling individual requirements – c2 blue:
with display and keys, c2 green: with keys, c2 white: without keys
or display, c2 orange: without keys and display but with internal
IrDA
 HENRI setting module with SIM Card slot for simple and quick
programming.
 6 self-filling tubes reduce refilling needs; c2 optimizer provides
optimum set-up of payout configuration & float levels for less cash
stock in the c2.
III) Flexibility
 Cassette system with 6 tubes
 Acceptance & payout of all the world’s circulation coins
 Additional maximum capacity of low replenishing coin
nominations
 Huge range of potential payout configurations
 22% higher change capacity then other 6 tube coin changers
 User friendly intuitive menu structure for individual configuration

Page | 10
IV) Additional Features
 Robust tube cassette, can be replaced even while completely filled
 Water penetration protection of internal components
Easy & fast access to sorting area for easy cleaning
4.3 Communication Protocols
This chapter describes the protocol and commands which is used by the
YHY502CTG to communication with host
Table 3.2: Command list

Code Command Description

0x01 Module Type Read Module Type
0x02 Module Serial Number Read Module Serial Number
0x03 Power Down Set Module Power Down
0x10 Firmware version Read Module Firmware Version
0x11 Antenna control Set Module Antenna On Or Off
0x12 Card IDLE Set Card IDLE
0x13 Seek Set Auto Search Card
0x14 Beep Set Buzzer ON/OFF
0x15 Beep interval Set Buzzer Beep Interval Time
0x16 Output 1 Set Output 1
0x17 Output 2 Set Output 2
0x19 Card type Read Card Type
0x20 Card serial number Read Card Serial Number
0x21 Block read Read Card Block data, 16 bytes
0x22 Block write Write Card Block Data, 16 bytes
0x23 Initialize ePurse Initialize One Block into epurse value
0x24 Value read Read ePurse Value, 4 bytes
0x25 Increment Increase ePurse Value, 4 bytes
0x26 Decrement Decrease purse Value, 4 bytes
0x32 Read E2 Read Module EEPROM
0x33 Write E2 Write Module EEPROME

4.4 Thermal Printers

Thermal printing (is a digital printing process which produces a printed
image by selectively heating coated thermochromic paper, or thermal
paper as it is commonly known, when the paper passes over the thermal print
head. The coating turns black in the areas where it is heated, producing an
image. Two-colour direct thermal printers can print both black and an

Page | 11
additional colour (often red) by applying heat at two different temperatures.
A two inch thermal printer is used to print the receipt of the completed
transaction.
(I) Design
A thermal printer comprises these key components:
 Thermal head: generates heat; prints on paper
 Platen: a rubber roller that feeds paper
 Spring: applies pressure to the thermal head, causing it to contact
the thermosensitive paper
In order to print, thermo-sensitive paper is inserted between the thermal
head and the platen. The printer sends an electric current to the heating
elements of the thermal head, which generate heat. The heat activates the
thermo-sensitive coloring layer of the thermosensitive paper, which changes
color where heated. Such a printing mechanism is known as a thermal
system or direct system. The heating elements are usually arranged as a line
of small closely spaced dots. The paper is impregnated with a solid-state
mixture of a dye and a suitable matrix; a combination of a fluoran leuco
dye and an octadecyl phosphonic acid is an example. When the matrix is
heated above its melting point, the dye reacts with the acid, shifts to its
colored form, and the changed form is then conserved in metastable state
when the matrix solidifies back quickly enough (a process known
as thermochromism). Fig 5 shows the thermal printer of vending machine.

Fig 5: Thermal Printer

4.5 Controller
The controller which used here is selected based upon the specification
make the STM32F103xx High-density performance line microcontroller
family suitable for a wide range of applications:
 Motor drive and application control
 Medical and handheld equipment

Page | 12
  PC peripherals gaming and GPS platforms
 Industrial applications: PLC, inverters, printers, and scanners
 Alarm systems, Video intercom, and HVAC
4.5.1 Features
 Core: ARM 32-bit Corte X
 M3 CPU
 72 MHz maximum frequency, 1.25 DMIPS/MHz (Dhrystone
2.1) performance at 0 wait state memory access
 Single-cycle multiplication and hardware division
 Memories
 256-to-512 Kbytes of Flash memory
 up to 64 Kbytes of SRAM
 Flexible static memory controller with 4 Chip Select. Supports
Compact Flash, SRAM, PSRAM, NOR and NAND memories
 LCD parallel interface, 8080/6800 modes
 Clock, reset and supply management
 2.0 to 3.6 V application supply and I/Os
 POR, PDR, and programmable voltage detector (PVD)
 4-to-16 MHz crystal oscillator – Internal 8 MHz factory-
trimmed RC
 Internal 40 kHz RC with calibration – 32 kHz oscillator for
RTC with calibration
 Low power
 Sleep, Stop and Standby modes
 V BAT supply for RTC and backup registers
 3 × 12-bit, 1 µs A/D converters (up to 21 channels)
 Conversion range: 0 to 3.6 V
 Triple-sample and hold capability
 Temperature sensor
 2-channel 12-bit D/A converter
 DMA: 12-channel DMA controller

Page | 13
  Supported peripherals: timers, ADCs, DAC, SDIO, I2Ss, SPIs,
I2Cs and USARTs
 Debug mode
 Serial wire debug (SWD) & JTAG interfaces
 Cortex-M3 Embedded Trace Macrocell.
 Up to 112 fast I/O ports
 51/80/112 I/Os, all mappable on 16 external interrupt vectors,
all 5 V-tolerant except for analog inputs
 Up to 11 timers
 Up to four 16-bit timers, each with up to 4 IC/OC/PWM or
pulse counter
5. Types of Vending Machine
The various types of vending machines available in the market is
discussed. The various models available were studied carefully and the best
features were considered for the design of the vending machine constructed.
5.1 Change Machine
A change machine is a vending machine that accepts large
denominations of currency and returns an equal amount of currency in
smaller bills or coins. Typically these machines are used to provide coins in
exchange for paper currency, in which case they are also often known as bill
changers.
5.2 Bulk Candy and Gumball Vending
The profit margins in the bulk candy business can be quite high-
gumballs, for instance, can be purchased in bulk for around 2 cents per piece
and sold for 25 cents in gumball machines in the U.S., and other countries.
Gumballs and candy have a relatively long shelf life, enabling vending
machine operators to manage many machines without too much time or cost
involved. In addition, the machines are typically inexpensive compared to
soft drink or snack machines, which often require power and sometimes
refrigeration to work. Many operators donate a percentage of the profits to
charity so that locations will allow them to place the machines for free. Bulk
vending may be a more practical choice than soft drink/snack vending for an
individual who also works a full-time job, since the restaurants, retail stores,
and other locations suitable for bulk vending may be more likely to be open
during the evening and on weekends than venues such as offices that host

Page | 14
soft drink and snack machines. The Bulk vending Machines of today provide
many different vending choices with the use of adjustable gumball and
candy wheels. Adjustable gumball wheels allow an operator to not only offer
the traditional 1-inch gumball, but they can also vend toy capsules, bouncy
balls and larger gumballs. Adjustable candy wheels on the other hand allow
an operator to offer a variety of pressed candies, jelly candy, and even nuts.
Fig 6 shows the bulk candy vending machine.

Fig 6: Bulk Candy Vending Machine

5.3 Newspaper Vending Machine

A newspaper vending machine or newspaper rack is a vending machine
designed to distribute newspapers. Newspaper vending machines are used
worldwide, and they can be one of the main distribution methods for
newspaper publishers. According to the Newspaper Association of America,
in recent times in the United States, circulation via newspaper vending
machines has dropped significantly: in 1996, around 46% of single-sale
newspapers were sold in newspaper boxes, and in 2014, only 20% of
newspapers were sold in the boxes.
5.4 Photo Booth
A photo booth is a vending machine or modern kiosk that contains an
automated, usually coin-operated, camera and film processor. Today, the
vast majority of photo booths are digital. Traditionally, photo booths contain
a seat or bench designed to seat the one or two patrons being photographed.
The seat is typically surrounded by a curtain of some sort to allow for some
privacy and help avoid outside interference during the photo session. Once
the payment is made, the photo booth will take a series of photographs and
the customer is then provided with prints. Older photo booth vending
machines used film and involved the process of developing the film using
liquid chemicals. Fig 7 shows the photo booth vending machine.

Page | 15
 Fig 7: Photo Booth Vending Machine
5.5 Stamp Vending Machine
A stamp vending machine is a mechanical, electrical or electro-
mechanical device which can be used to automatically vend postage
stamps to users in exchange for a pre-determined amount of money,
normally in coin.
5.6 Ticket Machines
A ticket machine is a vending machine that produces tickets. For
instance, ticket machines dispense train tickets at railway stations, transit
tickets at metro stations and tram tickets at some tram stops and in some
trams. The typical transaction consists of a user using the display interface to
select the type and quantity of tickets and then choosing a payment method
of either cash, credit/debit card or smartcard. The ticket or tickets are then
printed and dispensed to the user.
5.7 Automobile Vending Machine
In November 2013, online auto retailer Carvana opened the first car
vending machine in the U.S. located in Atlanta.
In late 2016, Autobahn Motors, a car dealership in Singapore, opened a
15-story-tall luxury car vending machine containing 60 cars,
dispensing Ferrari and Lamborghini vehicles.
5.8 Bait Vending Machine
A bait machine is a vending machine that dispenses live fishing bait,
such as worms and crickets, for fishing. Fig 8 shows the bait vending
machine.

Fig 8: Bait Vending Machine

Page | 16
5.9 Book Vending Machine
Book vending machines dispense books, which may be full-sized. Some
libraries use book vending machines. Go Library is a book lending vending
machine used by libraries in Sweden and the U.S. state of California.
5.10 French Fry Vending Machine
A French fry vending machine is a vending machine that dispenses hot
French fries, also known as chips. The first known French fry vending
machine was developed circa 1982 by the defunct Precision Fry Foods Pty
Ltd. in Australia. A few companies have developed and manufactured
French fry vending machines and prototypes. Furthermore, a prototype
machine was also developed at Wageningen University in the Netherlands.
5.11 Pizza Vending Machine
Let's Pizza is the name of a vending machine that makes
fresh pizza from scratch. It was developed in 2009 by Italian company Sitos
srl. The machine combines water, flour, tomato sauce, and fresh ingredients
to make a pizza in approximately three minutes. It includes windows so
customers can watch the pizza as it is made. The pizza is cooked in
an infrared oven. The device was invented by Claudio Torghele, an
entrepreneur in Rovereto, Italy. The vending machine began in Italy and is
now spreading into the United Kingdom and becoming popular there. Fig 9
shows the pizza vending machine.

Fig 9: Pizza Vending Machine

6. Construction and Working

The construction, working and hardware components of the vending
machine is discussed in detail. Selection of these components is based on the
requirement and objective of the application.

Page | 17
6.1 Construction
 20mm dia circular rod is used for the linear movement of the
collection bin. Fig 10 shows the circular rod used in the vending
machine.

Fig 10: Circular Rod

 Total weight of the machine is 100kg (approx)

 30x30 square pipe is used for the construction of the outer frame.
 Timing belt is preferred for the lifting and lowering of the collection
bin. Fig.11 shows timing belt used in vending machine.

Fig 11: Timing Belt

 48 DC motors are used for the movement of the spring.

 Factor of safety for the bin is 1.8.
 For the fine positioning, a timing belt with 5mm pitch and 12mm
width is used.
 The drive ratio for the pulley is 1:1.
 For the construction of the frame mild steel is used.
 Stepper motor is used for the collection bin movement.
 2 mm Mild steel square pipe is used. Fig. 12 shows the square rod
used in the vending machine.

Page | 18
 Fig 12: Square Pipe

 Pre-Coated sheets are used so the powder coating cost is reduced.

 Fabrication cost is reduced.
7. Conclusion
It is an attempt to make an import substitution product in India.
Currently such products made in India prove to be very expensive so people
prefer to import used machines from UK and USA because they find it
cheaper than the machines indigenously manufactured in India. Some of the
pros of the modern day vending machine are buying and selling becomes
easy. They accept both the cash and the credit cards for the electronic
transactions. You can shop for your intended product on a 24 hour,
throughout the year. Most vending machines are convenient & they offer
time saving for the customers There are cons as well, namely fixed prices
apply and this may be unfriendly both to the customer and the owner. The
heavy taxation levied on the sites of location may be the discouragement for
this capital intensive investment. There are great losses to faulty
programming such as the continuous dispensing of product due to the
technical errors Keeping this in mind, we have tried to keep the BOM as low
as possible so that the company providing internship for us stand to benefit
by being the most competitive manufacturer in India.
In the near future cashless transaction methods can include Paytm, Debit
card/Credit card transactions etc. The display used here is an LCD display
and there can be improvements in the future by replacing the LCD display
with a touch screen display. Android apps can also be used for the
transaction. Voice recognition technology can also prove to be a big
breakthrough in the field of vending machines.
References
1. Asmita Bodhale P, Kulkarni JS et al. Case Study on Different Vending
Machines, International Research Journal of Engineering and
Technology, 2017; 4(4):31-35.

Page | 19
2. Mohamed Aamir M, Kamalanathan P, Irshath Ahmed R, Kalaiselvan P
et al. Automatic Paper Vending Machine, International Journal of
Science, Engineering and Technology Research, 2015:4(4):634.
3. Shweta Dour, Rishikesh Shukla, Dishita Makwana, Neelima Satuluri,
Lakshmi Mahiwal et al. Vending Machine Using 8051 Micro
Controller, International Journal of Advance Research in Science and
Engineering. 2017; 6(5):181.
4. Charal Percy W, Alvin Russell V, Modular Vending Machine, US
Patent. 1997; 5(651):476.
5. Sasikala G, Kuldipsing Rajput, Sarfaraj Hussain, Aastha Shrivastava et
al. Real Time Embedded Based Drinking Water Wending Machine,
Asian Journal of Science and Technology. 2014; 5(1):804-809.
6. Carl Nelson A, Hossian SGM, Ala’a Al-okaily, Jason Ong et al. A novel
vending machine for supplying root canal tools during surgery, US
National Library of Medicine National Institutes of Health, 2012, 36(2).
7. Automatic-paper-vending-machine, 2016. http://gyti.techpedia.in. 8
January.
8. Types of Vending Machines, 2017. http://smallbusiness.chron.com, 11
September.

Page | 20
 Chapter - 2
Research Aspects of Bio-Energy Generation from
Waste Using MFC

Authors
Dr. K. Senthilkumar
Associate Professor, Department of Chemical Engineering,
Kongu Engineering College, Erode, Tamil Nadu, India
M. Naveenkumar
Research Scholar, Department of Chemical Engineering,
Kongu Engineering College, Erode, Tamil Nadu, India

Page | 21
Page | 22
 Chapter - 2
Research Aspects of Bio-Energy Generation from Waste
Using MFC
Dr. K. Senthilkumar and M. Naveenkumar

Abstract
The high energy requirement of conventional industrial treatment
systems are demanding for the alternative treatment technology which will
require less energy for its efficient operation and recover useful energy to
make this operation sustainable. Traditional treatment processes have only
focused on degrading organic pollutants and have ignored the extraction of
chemical energy from the pollutants. Microbial fuel cells are capable to
provide clean energy, apart from effective treatment of wastewater.
Microbial fuel cell (MFC) is a device which converts chemical energy to
electrical energy during substrate oxidation with the help of microorganisms.
This book chapter describes elaborately about the conventional treatment
methods & its necessity and demerits, basics and applications of MFCs, its
design & configurations, research opportunities in which the key factors
influencing bio-energy production on MFCs and its future perspective.
Keywords: Microbial fuel cell, bio-energy, wastewater treatment, waste to
wealth
Introduction
Because of genesis of the process industries in the growing world, where
the conventional forms of energy are moving towards depletion and are
making generously to universal concerns such as global warming and
greenhouse effect, the necessary to innovate and employ alternate energy
sources has become more essential for the survival of a future. Waste-to-
Energy is the production of energy in the form of heat or electricity from
waste. Every day millions of tons of waste is produced which causes huge
environmental impact in terms of ecosystems, wildlife and to human health.
Keeping this in the mind, numerous new waste treatment plants have come
up and have developed new ways to generate energy from waste.
For Example, the Indian distillery units are used sugarcane molasses as a

Page | 23
raw material because of its huge availability for the production of alcohol.
There are more than about 450 sugar mills and 300 distilleries in India.
Currently about 48.56 million m3 of spent wash is annually generated from
distillery units alone in India. The spent washes are highly variable because
of its production and characteristics, and also dependent on feed stocks and
various aspects of the ethanol production process. The spent wash is acidic
(pH 3.90 - 4.25) dark brown liquid with high BOD (46000–110000 mg/Lit)
and COD (80000– 220000 mg/lit) and emits obnoxious odor. Its discharge
without any treatment brings about immediate discoloration and depletion of
dissolved oxygen in the receiving water streams, posing serious threat to the
aquatic flora and fauna.
In India bulk of the alcohol is being produced from sugar cane molasses.
Molasses is a thick viscous byproduct of the sugar industry which is acidic in
nature, rich in salts, dark brown in colour and it also contains sugar which
could not be crystallized. For manufacturing alcohol, the molasses is diluted
with water into a solution containing 14-15% of sugars. This solution is then
inoculated with yeast strain and is allowed to ferment at room temperature.
To obtain alcohol at adequate/ requisite strength and quality/specification,
the fermented wash is distilled in a series of distillation columns. This
alcohol is used for various purposes including potable and industrial
applications. For manufacture of alcoholic beverages, the alcohol is, if
required, matured and blended with malt alcohol (for manufacture of
whisky) and diluted to requisite strength to obtain the desired type of liquor/
Indian Made Foreign Liquor (IMFL).
For treating distillery effluent some of the membrane based separation
processes like Ultra filtration (UF), Nano filtration (NF), Reverse osmosis
(RO) and Membrane Bioreactor (MBR) are used. Distilleries are one of the
17th most polluting industries listed by the Central Pollution Control Board
(CPCB) of India (CPCB, 2011). Primary, Secondary and Tertiary treatment
is used for the distillery wastewater treatment processes in India. The typical
treatment sequence is screening and equalization, followed by
biomethanation; ferti-irrigation and bio composting with sugarcane press
mud are the most widely used options for effluent disposal. However, these
methods are highly energy intensive and hence quite expensive. The
distillery spent wash is characterized as one of the caramelized and
recalcitrant wastes containing extremely high COD, BOD, inorganic solids,
color and low in pH. These disadvantages emphasized the need for further
research using novel treatment techniques.
Bio-Energy Potential from Distillery Effluents
In India distillery industries producing more than 3.20 billion litres of

Page | 24
alcohol generating 46 billion litres of wastewater annually. The enormous
distillery wastewater has potential to produce 1200 million cubic meters of
biogas. The post methanation wastewater if used carefully for irrigation of
agricultural crops can produce more than 85000 tonne of biomass annually.
This biogas normally contains 60% methane gas, which is a well-recognized
fuel gas with minimum air pollution potential (Zhuwei Du et al. 2007). If
this source of energy is tapped, it will fetch additional energy units worth 5
trillion-kilo calories annually.
Besides, the Post Methanation Effluent (PME) can provide 2,45,000
tons of potassium, 12,500 tons of nitrogen and 2,100 tons of phosphorus
annually. Thus the manorial potential of effluent can be measured by the fact
that one year’s effluent can meet the potassium requirement of 1.55 million
hectare land, nitrogen requirement of 0.13 million hectare land and
phosphorus requirement of 0.025 million hectare land if two crops are taken
in a year.
Effect of Distillery Wastewater on Environment
In a distillery, sources of wastewater are stillage, fermenter, condenser
cooling water and fermented wastewater. The liquid residues from the
process units during the industrial phase of the production of alcohol or
liquor, sugar cane washing water and from the cleaning of the equipment,
apart from other residual water. This extract is extremely polluting as it
contains approximately 5% organic material and fertilizers such as
potassium, phosphorus and nitrogen. The amount of water used in this
process is large, generating a high level of liquid residues (Qing wen et al.
2009). The annual bio energy potential of distillery effluent in various States
of India stated in Jadhav et al. (2010). The effluent contains large amounts of
dark brown colored molasses spent wash because of effluents from molasses
based distilleries. Molasses spent wash (MSW) creating problem to
environment in two ways. One is the highly colored nature of MSW can
block out sunlight from rivers and streams, thus reducing oxygenation of the
water by photosynthesis and hence becomes problem to aquatic life. Another
one, it has a high pollution load which would result in eutrophication of
contaminated water sources. Due to the presence of putrescible organics like
skatole, indole and other sulfur compounds, the MSW that is disposed in
canals or rivers produces obnoxious smell.
Water Pollution
 Agrochemicals and sediments can pollute nearby water sources.
 Siltation and eutrophication of nearby surface water bodies due to

Page | 25
 soil erosion due to flood irrigation and over irrigation (Qing Wen et
al. 2010).
Water Scarcity
 Sugarcane is a deep-rooted crop and can have a great impact on
river flow as it reduces run-off from the catchment into rivers and
draws heavily on ground water resources (Qing Wen et al. 2009).
Air Pollution
 Burning of cane to speed harvest causes air pollution and increases
erosion.
 Bagasse, which is commonly used as fuel in boilers, produces fly
ash, which escapes to the atmosphere and can affect the population
with irritation in eyes, nose, throat and lungs, and can damage crops
(Qing Wen et al. 2010).
Field Level Impact
 Erosion is a significant issue in areas under sugar cane or beet
cultivation, since erosion rates in tropical agro-ecosystems are
usually greater than the rate of soil formation.
 Cane harvesting can cause a significant removal of soil with the
roots. Declining soil quality is associated with cane and beet
production, due to soil compaction, loss of organic matter,
salinization and acidification. 10-30% of the total beet harvest
weight is soil (Deepak Pant and Alok Adholeya, 2007).
Impact on Soil Health
 Heavy infield transport machinery is most commonly associated
with soil compaction problems.
 Soil compaction decreases porosity and water infiltration rate,
restricting the rooting ability of the crop.
 Conventional tillage commonly promotes erosion by exposing soil
aggregates to rainfall and also drastically changes soil structure.
 Acidification is also more prevalent in cane than beet growing
areas, largely due to the use of inorganic nitrogenous fertilizers such
as urea and ammonium sulphate.
 Most of the farmers still use the flood irrigation pattern which
results in huge wastage of water, electricity and results in
salinization of the soil which is another important cause of lower
productivity.

Page | 26
  Productivity of the sugarcane in the region is going down owing to
excessive use of chemical fertilizers/herbicides (Deepak Pant and
Alok Adholeya 2007).
Solid Waste Management
 The bagasse when used as fuel in boilers produces particulate
matter, nitrogen oxide and sulphur. If pollution control equipment is
not installed, fly ash escapes to the atmosphere and can affect the
population with a number of health related problems (Senthil kumar
et al. 2008).
Conventional Treatments Methods
In most of the molasses based distillery units anaerobic treatment is
implemented as an accepted practice and various types of high rate anaerobic
reactors have been tried at pilot and full-scale operation. But anaerobically
treated effluent still contains high concentrations of organic pollutants and as
such cannot be discharged into the soil. Physicochemical treatment with
active activated carbon was investigated to removal the color causing
components (mainly melanoidins) from biomethanated distillery spent wash
but was ineffective for application to large scale as it requires a pretreatment
with sulphuric acid, phosphoric acid, nitric acid and hydrochloric acid for the
adsorbent (Qing Wen et al. 2010). Traditional methods require more
electrical energy to digest the dissolved matter. The proven technology is
anaerobic digestion but the treatment process faces several problems (Bruce
E Logan 2010) such as energy cost, footprint, investment costs, sludge
treatment costs and reliability. In recent years a number of studies have
conducted towards utilization of microbial activities for treating industrial
effluents along with generation of Bioenergy from waste. The proposed
chapter describes the principle and applications of microbial fuel cell for the
treating the effluents and also generating bio-energy from effluents. It is one
of the emerging and promising technologies for the wastewater treatment.
About Microbial Fuel Cell
Microbial fuel cell (MFC) technology is an emerging technology that
treating wastewater and converts its chemical energy into electrical energy
using microorganisms as catalysts. Microbial fuel cell (MFC) is a device
which converts chemical energy into electrical energy during oxidation of
organic and inorganic matters with the help of microorganisms. MFC’s can
be grouped into two general categories, those that use a mediator and those
that are mediator-less. The first MFC’s demonstrated in the early 20th
century used as a mediator and a chemical that transfers electrons from the

Page | 27
bacteria in the cell to the anode. Mediator-less MFC’s are a more recent
development dating to the 1970s; in this type of MFC the bacteria typically
have electrochemically active redox proteins such as cytochromes on their
outer membrane transfer electrons directly to the anode. Since the turn of the
21st century MFC’s have started to find a commercial use in the treatment of
wastewater (Nastro 2014). Microbial fuel cells (MFC’s) are devices that use
bacteria as the catalysts to generate energy as electricity utilizing organic and
inorganic matter. Electrons generated by the microorganisms from these
substrates are transferred to the anode (negative terminal) and flow to the
cathode (positive terminal) linked by a conductive material containing a
resistor, or operated under a load. By convention, positive current flows from
the positive to the negative terminal, a direction opposite to that of electron
flow. The fuel cell device must be able to oxidizing the substrate in the
anode compartment, either continuously or intermittently; otherwise, the
system is considered to be a bio battery. Electrons can be transferred to the
anode by electron mediators by direct membrane associated electron transfer,
or by other as yet undiscovered means. In most MFC’s the electrons that
reach the cathode combine with protons that diffuse from the anode through
a separator and oxygen provided from air; resulting product is water.
However, there is no net carbon emission because carbon dioxide in the
renewable. Biomass originally comes from the atmosphere due to
photosynthesis process. A component of the fuel cell was described in Table 1.
Table 1: Basic Components of Microbial Fuel Cells

Items Materials
Anode Graphite, graphite felt, carbon paper, carbon-cloth
Cathode Graphite, graphite felt, carbon paper, carbon-cloth, black
Anode chamber Glass
Cathode chamber Glass
Proton exchange Ultrex, polyethylene. Poly, salt bridge, porcelain septum,
membrane system solely electrolyte, Nafion
Electrode Catalyst Pt, Pt black, MnO2, Fe3+, polyaniline, electron mediator
immobilized on anode

MFC Configurations and Designs

In order to increase the power output of MFCs, many studies have
focused on altering their designs to overcome the barriers to electron and
proton transport and to enhance the surface area and reactivity of the
electrodes. The essential physical components of MFCs are the anode,
cathode and proton exchange membrane (PEM) as shown in figure 1 (In S.
Kim et al. 2008). Electrochemically active bacteria in the anode chamber

Page | 28
oxidize the substrate and separate the electrons from protons. These e- and
H+ travel to the cathode through an external circuit, and the latter diffusing
through the electrolyte and a PEM. Protons and electrons are subsequently
combined at the cathode with oxygen, aided by a catalyst such as platinum to
form water. A number of configurations for MFCs have been developed.
These are run in batch, fed-batch or continuous mode and have a variety of
shapes. A two-chambered configuration is a typical MFC, which consists of
an anode and a cathode chamber separated by a PEM such as Nafion or
Ultrex or even by a salt bridge. Recently, Single-Chambered MFC (SC-
MFC) has been considerable attention on a treatment, which does not require
a cathode compartment because the cathode is exposed directly to air
(Shaoan Cheng et al, 2011). Single chambered MFCs are quite attractive for
increasing the efficiency of power production because they can be run
without artificial aeration in an open air cathode systems and can reduce the
internal resistance by avoiding the use of a catholyte as a result of combining
two chambers. However, many single-chambered MFCs still use Nafion (or
PEM) as a physical separator, where Nafion is either hot-pressed directly
onto the cathode or works as a membrane electrode assembly (Xu Wang et
al. 2016). The undesirable back diffusion of oxygen from the cathode to the
anode and microbial contamination of the cathode are the two major
concerns with single-chambered MFCs.
A substantial increase in oxygen transfer into the anode chamber in the
absence of a PEM was reported. On the other hand, even with a relatively
low power output, two-chambered MFCs can have a wide variety of
applications (Akshay et al. 2016). Cathodic denitrification (nitrate removal
as an opportunity for cathodes) or external power assisted bio-hydrogen
production is good example. There is still considerable room for a two
chambered system. This design is striking for wastewater treatment due to its
easy scale-up and lack of PEM associated limitations, such as proton transfer
problems. However, oxygen back diffusion is a critical drawback, which is
more severe than in membrane used MFCs, demonstrated successful
generation of electricity using an open air bio-cathode system with
microorganisms that act as biocatalysts for oxygen reduction instead of
chemical catalysts.

Page | 29
 Fig 1: Schematic of microbial fuel cell

Advantages of MFC
 MFC is ability to directly convert substrate into electricity.
 It allows wastewater processes to reduce activated sludge compared
to anaerobic digestion and conventional aerobic activated sludge
processes.
 It is insensitivity to the operational environment.
 It can be used widely in locations where there is insufficient
electrical infrastructure.
 Its do not consume much energy in comparison of other methods
and do not require highly regulated distribution systems like the
ones needed for Hydrogen fuel cells.
 Its operate efficiently at ambient temperature with safe and good
performance.
 Its do not require gas treatment and not generate more CO2 than
typical biological wastewater treatment processes.
 MFC technology holds promise towards sustainable power
generation and wastewater treatment.
 It minimum amount of sludge may be produced while treating the
wastewater. The treated water may be used for irrigation purposes
and used for industrial cleaning processes.

Page | 30
Research Opportunities in MFC
Electrodes in MFC
Electrode material is the main component in deciding the performance
and cost of MFC. Electrode design is the greatest challenge in making
MFC’s a cost-effective and scalable technology. Recently, area of
motivation is increased in the electrode material and their configurations of
MFCs are studied. Over the past decade, a variety of electrodes have been
extensively explored for MFC’s. Surface area of electrode plays a vital role
in the generation of electricity. In MFC, the electrode materials are having
different ability and also its self-characteristic. For all the types of
electrodes, their base materials must generally be of good conduction, good
chemical stability, high mechanical strength, and low cost (Yongtae Ahn et
al. 2014). Carbon materials and non-corrosive metals, which can basically
meet the general requirements above, are currently the most-widely used
base materials. In addition, there are some specific requirements for each
group of electrodes. For high bio-catalytic activity, high surface roughness,
good biocompatibility, and efficient electron transfer between bacteria and
electrode surface are essential. In order to improve bacterial adhesion and
electron transfer, electrode surface modification has become a new topic of
interest in the research field of MFC’s. The air-cathode electrode material is
composed of a base material, a catalyst, a binder, and a waterproof coating.
The graphite brush anode is an ideal electrode that achieves high surface
area, high porosities, and efficient current collection. The use of a brush
anode was first reported by Logan et al. (2007). In their studies, the brushes
were made of carbon fibers cut to a set length and wound into a twisted core
consisting of two conductive but noncorrosive titanium wires. Two brush
sizes were used in this study: the smaller brush, about 2.5 cm in diameter and
2.5 cm long, had an estimated surface area of 18,200 m2/m3-brush volumes
and 95% porosity, while the larger brush, about 5 cm in diameter and 7 cm
long, produced 7,170 m2/m3-brush volumes and 98% porosity. The cube
MFC’s containing the smaller brush reached a maximum power density of
2,400 mW/ m2 (normalized to the cathode projected surface area), and a
maximum columbic efficiency (CE) of 60%. Bottle MFC’s with the larger
brush anode produced a maximum power density of 1,430 mW/ m2, versus a
600 mW/ m2 plain carbon paper anode. The performance of brushes with
different masses of fibers were also tested, but the lack of a clear trend in
power per mass loading suggested that the clumping of fibers was a problem
that hindered bacterial access to the fiber surfaces, as well as the diffusion of
substrate into the brush interior (Logan et al. 2007).

Page | 31
Effect of Spacing between Anode and Cathode on Power Production
The total power production is increased when the spacing between the
anode and cathode electrodes are at less distance. Maximum power density
of 10.9, 8.6, and 7.4mW/ m2 was observed at electrode spacing 20, 24 and 28
cm, respectively (Yongtae Ahn et al. 2014). The maximum power density
was observed at larger spacing. Maximum power density was observed at
external resistance between 900Ω and1200 Ω. Decrease in power density was
observed with increase in resistance beyond 1200 Ω.
Effect of Electrode Surface Area on Power Production
The experimental set up consist of three electrodes (area 70.21 cm2,
140.43 cm2 and 210.64 cm2 respectively) which are connected in series to
the circuit in order to evaluate effect of the anode surface area on power
production. The experimental results and the literature data reveals that area
of anode increases power density decreases. Thus it indicates that, in a
comparatively bigger reactor, area of anode is utilized less efficiently than
those of smaller reactors, used as BOD sensors (Sarah Hays et al, 2011). In
this Mediator less microbial fuel cell (ML–MFC), when areas of anode
electrodes were 70.21 cm2, 140.43 cm2 and 210.64cm2. Maximum power
densities observed were 10.13mW/m2, 6.45mW/m2 and 4.66mW/m2
respectively (Yongtae Ahn et al. 2014). The use of third electrode had very
little effect on enhancement of power production as compared to two
electrodes, and it improved net power production (mW) only by 8.9%.
However, increase in net power production was observed by 28.6% when
two electrodes were used as anode instead of single electrode. Although, as
the area of anode was made twice and thrice, the increase in net power
production was 28.6% and 40%, respectively; indicating that power
generation was limited by some other factor, apart from anode surface area.
The electron transfer from microbial cell to the anode surface through
probable limitation is occurring due to the absence of addition of external
mediator.
Influence of Microorganisms in MFC
The literature evidenced that numerous microorganisms have the
capability to transfer the electrons derived from the metabolism of organic
matters to the anode (Kun Guo et al, 2012). Marine sediment, soil,
wastewater, fresh water sediment and activated sludge are all rich sources for
these microorganisms (Jae Kyung Jang et al, 2013). A plenty of recent
publications discussed the screening and identification of microorganisms
and the building of a chromosome library for microorganisms that are able to

Page | 32
generate electricity from degrading organic matters (Logan et al., 2007;
Merina Paul Das 2015; Nasser Mehrdadi et al, 2016). The anodic electron
transfer mechanism in MFC is a key issue in understanding the theory of
how MFC’s work. Microorganisms transfer electrons to the electrode
through an electron transport system to the cathode chamber. Under
anaerobic conditions in soils and sediments, Geobacter belongs to
dissimilatory metal reducing microorganisms, which produce biologically
useful energy in the form of Adenosine Triphosphate (ATP) during the
dissimilatory reduction of metal oxides. The electrons are transferred to the
final electron acceptor (Shakunthala C et al, 2015). In dissimilarity metal
reducing microorganisms the chemical compounds proposed to be involved
in the electron transportation from electron carriers (anode) (Anand prakash
2016). The microorganisms such as S. putrefactions, G. sulfurreducens and
G. metallireducens transfer electrons to the solid electrode (anode) by the
similar method.
Challenges in MFC
The MFC technology has to compete with the mature methanogenic
anaerobic digestion technology that has seen wide commercial applications
because they can utilize the same biomass in many cases for energy
productions. MFCs are able to converting biomass at low temperature (20°C)
and with low substrate concentrations, both of which are problematic for
methanogenic digesters. In order to improve the power density output, new
anodophilic microorganisms that enormous improve the electron transport
rate from the biofilm covering an anode to the anode are much needed
(Stirling JL et al, 2009). It is claimed that a MFCs current flow could
increase by four orders of magnitude if Geobacter transports electrons to the
anode at the same rate as it does to its natural electron acceptor. Mutagenesis
and even recombinant DNA technology can conceivably be used in the
future to obtain some “super bugs” for MFCs. Microorganisms may be used
as a pure culture or a mixed culture forming a synergistic microbial
consortium to give a better performance. Microbial consortium may provide
electron mediators that are used by another type of bacterium to transport
electrons more efficiently to an anode. It is possible in the future that an
optimized microbial consortium can be obtained to operate an MFC without
extraneous mediators while achieving superior mass transfer and electron
transfer rates (Xu wang et al, 2016).
MFCs can be used in various applications as aforementioned. When
used in wastewater treatment, a large surface area is needed for bio-film to
build up on the anode. To resist fouling by creating inexpensive electrodes in

Page | 33
MFCs are to be need a termination. The power density output from an MFC
to match that of conventional chemical fuel cell such as a hydrogen-powered
fuel cell is unrealistic. The fuel in an MFC is dilute biomass (as in
wastewater treatment) in the anodic chamber that has a limited energy
(reflected by its BOD). Coulombic efficiency over 90% has been achieved in
some cases; it has little effect on the crucial problem of low reaction rate.
Although some basic knowledge has been gained in MFC research, there is
still a lot to be learned in the scale up of MFC for large-scale applications
(Bruce E Logan. 2010; Pandey B K et al, 2011). Besides, even with the
remarkable improvements in power density, the large-scale application of
MFC’s has yet to be implemented due to low yields of power generation and
high cost.
Applications of Microbial Fuel Cell
Electricity Generation
MFC’s are capable of converting the chemical energy stored in the
chemical compounds in a biomass to electrical energy with the aid of
microorganisms. Because chemical energy from the oxidization of fuel
molecules is converted directly into electricity instead of heat, the Carnot
cycle with a limited thermal efficiency is avoided and theoretically a much
higher conversion efficiency can be achieved (>70%) just like conventional
chemical fuel cells. Electricity generation with an electron yield by
Ferrireducens as high as 80% was reported and higher electron recovery as
electricity up to 89% was also reported (Zhuwei Du et al., 2007). An
extremely high Columbic efficiency of 97% was reported during the
oxidation of Formate with the catalysis of platinum (Pt) black. However,
MFC power generation is still very low, that is the rate of electron
abstraction is very low. One feasible way to solve this problem is to store the
electricity in rechargeable devices and then distribute the electricity to end-
users.
Bio Hydrogen
MFC’s can be readily modified to produce hydrogen instead of
electricity; protons released by the anodic reaction migrate to the cathode to
combine with oxygen to form water due to normal operating conditions. The
protons and the electrons produced by the metabolism of microbes in an
MFC are thermodynamically unfavorable. It applied an external potential to
increase the cathode potential in a MFC circuit and thus overcame the
thermo dynamic barrier. Protons and electrons produced by the anolyte
reaction are combined at the cathode chamber to form hydrogen in this

Page | 34
mode. The external potential for an MFC is theoretically requires 100 mV,
much lower than the 1110 mV required for direct electrolysis of water at
neutral pH. This is because of some energy comes from the biomass
oxidation process in the anodic chamber. MFC’s can potentially produce
about 8–9 mol H2/mol glucose compared to the typical 4 mol H2/mol glucose
achieved in conventional fermentation. In bio hydrogen production using
MFC’s, oxygen is no longer needed in the cathodic chamber. Thus, oxygen
leak to the anodic chamber is no longer an issue in improved efficiency of
MFCs. Another advantage is that hydrogen can be accumulated and stored
for later usage to overcome the inherent low power feature of the MFC’s.
Therefore, MFC’s provide a renewable hydrogen source that can contribute
to the overall hydrogen demand in a hydrogen economy.
Wastewater Treatment
The MFC’s are considered to be used for treating wastewater. Municipal
wastewater contains a multitude of organic compounds that can fuel MFC’s.
The amount of electric power generated by MFC’s during the wastewater
treatment process can potentially halve the electricity needed in a
conventional treatment process that consumes a lot of electric power aerating
activated sludge’s. MFC’s yield 50–90% less solids to be disposed of
furthermore, organic molecules such as acetate, propionate, and butyrate can
be thoroughly broken down to CO2 and H2O. A hybrid incorporating both
electrophiles and anodophilic are especially suitable for wastewater
treatment because more organics can be biodegraded by a variety of
organics. MFC’s using certain microbes has a special ability to remove
sulfides as required in wastewater treatment. MFC’s can enhance the growth
of bio electrochemically active microbes during wastewater treatment thus
they have good operational stabilities. Continuous flow and single-
compartment MFC’s and membrane-less MFC’s are favored for wastewater
treatment due to concerns in scale-up. Sanitary wastes, food processing
wastewater, swine wastewater and corn Stover are all great biomass sources
for MFC’s because they are rich in organic matters. Up to 80% of the COD
can be removed in some cases and a Columbic efficiency as high as 80% has
been reported.
Biosensor
Another potential application of the MFC technology is to use it as a
sensor for pollutant analysis and in situ process monitoring and control Apart
from the aforementioned applications. The correlation between the
Coulombic yield of MFCs and the strength of the wastewater make MFCs

Page | 35
possible biological oxygen demand (BOD) sensors. An accurate method to
measure the BOD value of a liquid stream is to calculate its Coulombic
yield. A number of works showed a good linear relationship between the
Coulombic yield and the strength of the wastewater in a quite wide BOD
concentration range. However, a high BOD concentration requires a long
response time because the Coulombic yield can be calculated only after the
BOD has been depleted unless a dilution mechanism is in place. Efforts have
been made to improve the dynamic responses in MFCs used as sensors. A
low BOD sensor can also show the BOD value based on the maximum
current. During this stage, substrate concentration is limiting the reaction of
anodic chamber. This monitoring mode can be applied to real-time BOD
determinations for either surface water, secondary effluents or diluted high
BOD wastewater samples. MFC-type of BOD sensors is advantageous over
other types of BOD sensor because they have excellent operational stability
and good reproducibility and accuracy. An MFC-type BOD sensor
constructed with the microbes enriched with MFC can be kept operational
for over 5 years without extra maintenance, far longer in service life span
than other types of BOD sensors reported.
Artificial Wastewater
The wastewater which is artificially produced for research works is
called artificial wastewater. It is also called as synthetic wastewater. If using
the natural wastewater; it should be stored in the cold condition; it shows
different characteristics while testing it. So, artificial wastewater is employed
for avoid the change in characteristics of the wastewater and to maintain the
same characteristics throughout the research work.
Conclusion and Future Perspectives
The microbial fuel cell inoculated with mixed anaerobic sludge
demonstrated its effectiveness as a wastewater treatment process along with
electricity production, without incorporating any costly component such as
electrode and membrane. The existence of Microbial Fuel Cell technology
has surfeit of applications in the day today lives as it is environmental
friendly, eco-friendly and more importantly green technology. There are
more challenges left over for the complete utilization of MFCs, to find ways
to make it cost effective, and to fabricate the innovative MFC bioreactors for
industrial effluent treatment. Identification of new microorganisms which
can be directly transfer electrons to or from an electrode is essential to treat
contaminated effluent with generation of electricity. There is a broad scope
for design and development of these reactors as the power density is too low

Page | 36
for consumption in various industrial applications. Besides, the organism
may be genetically altered in order to form high reducing microbial strains
with wide range of MFC applications. Future researches are necessary to
minimize the internal resistance and corrosion related problems in MFC.
MFCs can also have utilization in army applications in order to power up
remote surveillance and communication gears for use in unmanned
applications. Further, potential researches on optimization of the electricity
production from the two chambered MFC are also necessary. With further
improvements and optimization, it could be possible to increase power
generation and researchers are working to enhance it for the scale up &
commercial applications. Also MFC as a continuous reactor may also be
studied. Further research towards the conditions to be maintained in the
reactor, aerated condition in order to produce more electricity may be
needed. Thus, the combination of wastewater treatment along with electricity
production might help in compensating the cost of wastewater treatment.
However, the Microbial Fuel Cell technology is still in initial stage and
needs special attention in future research attempts.
References
1. Zhuwei Du, Haoran Li, Tingyue GU. A state of the art review on
microbial fuel cells: A promising technology for wastewater treatment
and Bioenergy, Biotechnology Advances. 2007; 25:464-482.
2. Qing Wen, Ying Wu, Lixin Zhao. Production of electricity from the
treatment of continuous brewery wastewater using a microbial fuel cell.
Fuel, 2010, 1381-1385.
3. Qing Wen, Ying Wu, Dianxue Cao. Electricity generation and modeling
of microbial fuel cell from continuous beer brewery wastewater.
Bioresource Technology. 2009; 100:4171-4175.
4. Jadhav RN, Narkhede SD, Mahajan CS, Khatik VA, Attarde SB.
Treatment and disposal of distillery spentwash. Asian Journal of
Environmental Science. 2010; 5(1):75-82.
5. In Kim S, Kyu-Jung Chae, Mi-Jin Choi, Willy Verstraete. Microbial
Fuel Cells: Recent advances, bacterial communities and application
beyond electricity generation. Environmental Engineering Research.
2008; 13(2):51-65.
6. Logan BE, Cheng S, Watson V. Graphite fiber brush anodes for
increased power production in air-cathode microbial fuel cells.
Environmental Science and Technology. 2007; 41(9):3341-3346.

Page | 37
7. Stirling JL, Bennetto HP, Delaney GM, Mason JR, Roller SB, Tanaka K
et al. Microbial fuel cells. Biochemical Society Transactions. 2009;
11(4):451-453.
8. Kun Guo, Daniel Hassett J, Tingyue Gu. Microbial Fuel Cells:
Electricity Generation from Organic Wastes by Microbes. Chapter 9, in
Microbial biotechnology: Energy and Environment Edited by Arora R,
CAB International, Oxon, UK, 2012, 162-189.
9. Senthilkumar K, Sivakumar V, Akilamudhan P. Experimental studies on
disposal of various industrial solid wastes. Modern Applied Science.
2008; 2(6):128.
10. Yongtae Ahn, Marta Hatzell C, Fang Zhang. Different electrode
configurations to optimize performance of multi electrode microbial fuel
cells for generating power or treating domestic wastewater. Journal of
Power Sources. 2014; 249:440-445.
11. Shaoan Cheng, Bruce Logan E. Increasing power generation for scaling
up single-chamber air cathode microbial fuel cells. Bioresource
Technology. 2011; 102(6):4468-4473.
12. Nastro RA. Microbial Fuel Cells in Waste Treatment Recent Advances.
International Journal of Performability Engineering. 2014; 10(4):367-
376.
13. Merina Paul Das. Bioelectricity production using algae in microbial fuel
cell. Der Pharma Chemica. 2015; 7(11):8-10.
14. Xu Wang, Enren Zhang, Eileen Hao Yu. Low Cost Materials for the Air
Cathodes in Single-Chamber Microbial Fuel Cells-A Mini Review.
Chemical Engineering Transactions. 2016; 51:37-42.
15. Shakunthala C, Dr. Surekha Manoj. Microbial Fuel Cells (MFC) for
Energy Production from Wastewater-A Review. International Journal of
Science and Research. 2015; 2319-7064.
16. Akshay Tharali D, Namrata Sain, Jabez Osborne W. Microbial fuel cells
in bioelectricity production. Frontiers in Life Science. 2016; 9(4):252-
266.
17. Nasser Mehrdadi, Gholamreza Nabi-Bidhendi, Hamid Reza Tashauoei.
Producing Renewable Energy from Municipal Wastewater Treatment
Using a Bio-electrochemical System. Journal of Safety Environment and
Health Research. 2016; 1(1):23-26.
18. Pandey BK, Mishra V, Agrawal S. Production of bio-electricity during

Page | 38
 wastewater treatment using a single chamber microbial fuel cell.
International Journal of Engineering Science and Technology. 2011;
3(4):42-47.
19. Sarah Hays, Fang Zhang, Bruce Logan E. Performance of two different
types of anodes in membrane electrode assembly microbial fuel cells for
power generation from domestic wastewater. Journal of Power Sources.
2011; 196(20):8293-8300.
20. Bruce Logan E. Scaling up microbial fuel cells and other
bioelectrochemical systems. Applied Microbiology Biotechnology.
2010; 85(6):1665-1671.
21. Deepak Pant, Alok Adholeya. Biological approaches for treatment of
distillery wastewater: A review. Bioresource Technology. 2007;
98(12):2321-2334.
22. Anand Prakash. Microbial fuel cells: A source of energy. Journal of
Microbial and Biochemical Technology. 2016; 8:247-255.
23. Jae Kyung Jang, Jinjun Kan, Orianna Bretschger. Electricity Generation
by Microbial Fuel Cell Using Microorganisms as Catalyst in Cathode.
Journal of Microbial and Biochemical Technology. 2013; 23(12):1765-
1773.

Page | 39
Page | 40
 Chapter - 3
Comparative Study of Cementitious Coated RCC
Roof and Green Roof

Authors
Raunak Katiyar
Department of Mechanical Engineering, Kamla Nehru Institute
of Technology, Sultanpur, Uttar Pradesh, India
Akhilesh Kumar Chauhan
Department of Mechanical Engineering, Kamla Nehru Institute
of Technology, Sultanpur, Uttar Pradesh, India

Page | 41
Page | 42
 Chapter - 3
Comparative Study of Cementitious Coated RCC Roof and
Green Roof
Raunak Katiyar and Akhilesh Kumar Chauhan

Abstract
To investigate the feasibility of a green roof in the region of Sultanpur,
U.P., India, a prototype of a room was fabricated with reinforced cement
concrete (RCC) roof. A green roof was developed over RCC roof in order to
investigate the reduction in the inside room temperature. The geographical
position of Sultanpur is: longitude: 82o4'58.4''E, latitude: 26o17'17.41''N,
altitude: 93m amsl. The experiment was performed in the month of March in
two stages. In the first stage, experiment was conducted with traditional
(RCC) roof with a cementitious coating and the second stage was with the
developed green roof. The temperatures of top surface, inside roof surface,
inside air and ambient air were recorded at an interval of 15 minutes for both
the roofs. From the experiments, it was observed that there is decrease in
inside temperature of the room due to green roof as compared to RCC roof.
This reduction in temperature is due to evapotranspiration and lower thermal
conductivity of green roof.
Keywords: Absorptivity, evapotranspiration, green roof, RCC roof and
thermal conductivity.
1. Introduction
From the last decade, it is observed that the population of India is
increasing with 17.7% as shown in Table 1 [1]. This explosion of population
is increasing the demand of basic needs such as food, shelter, clothes,
electricity etc. For achieving such demands, there is a rapid increase in
concrete structures, resulted in decrease in the vegetation area, water level of
underground fresh water and increase in urban heat island (UHI) effect. The
phenomenon of increase in urban air temperature and entrapment of heat due
to the urban morphology is termed as UHI. The demand of electricity is
being fulfilled at the cost of our natural resources (coal, petrol, and diesel).
The use of these natural resources accelerate the emission of CO2, which is a

Page | 43
green house gas and is responsible for increasing the global mean
temperature of environment and deteriorating our comfort conditions. Thus,
it becomes necessary for the sustainable development that we should
complete our demands without compromising with our comfort conditions
and environment.
Table 1: Decadal Increase in total, rural and urban population from 2001-2011 [1]

Decadal
Percentage Percentage Percentage
Change Total Rural Urban
Total Rural Urban
2001-2011
Persons 181959458 90973022 90986436 17.7 12.3 31.8
Males 90965182 46030080 44935102 17.1 12.1 29.9
Females 90994276 44942942 46051334 18.3 12.5 34.0

From the literature, it is observed that indoor thermal comfort condition

and energy demand can be achieved by using some alternatives such as
installing the P-V module for production of electricity and green roof for
thermal insulation. Though the installation of solar panel is the best
alternative but it has large installation and maintenance cost and becomes in-
effective towards recovery of our environment as it is not able to decrease
the CO2 emission, particulates, noise pollution, storm water retention etc.
Another alternative is green roof. The green roof not only reduces the
electricity demand but also helps in saving of our environment.
The concept of green roof was first presented in the hanging gardens of
Babylon which was constructed around 500 BC [2]. Gradually people started
sod roofs for the purpose of insulation as this sod roof provides cold space
during summer and hot space during winter. Modern green roof was
developed in Germany in 1960s. It is seen that tropical zone is the favorable
conditions for developing green roofs. One of the best examples of the
advantages of green roof is city of Singapore which is near the equator.
Wong et al. [3] performed a field measurement from Oct 25 to Nov 10, 2003
in Singapore with a green roof which had 400 mm soil substrate with six
vegetation species (Heliconia, Spider lily, Ophiopogon, Rhapis Palm,
Pandanus, Erythrina) and categorized the effects in direct and indirect
effects. They observed that the maximum temperature of hard surface
reaches around 57oC while the maximum temperature for the bare soil
surface was around 42oC and the maximum surface temperature for planted
roof was not more than 36oC. They also concluded that bare hard surface
shows heat gain of 366.3 kJ/m2, bare soil gave heat gain of 86.6 kJ/m2 while
bare soil with turf showed heat gain of 29.2 kJ/m2. They further reported that
maximum temperature decrease caused by plants was around 30oC.

Page | 44
 Wong et al. [4] performed DOE-2(Design of Experiments) energy
simulation program for determination of effect of roof top garden on annual
energy consumption, cooling load and roof thermal transfer value for a five
story hypothetical commercial building with clay soil (dry, 40% wet) with
different thickness and three types of plants (shrubs, trees, turf). They
concluded that shrubs are dominant with respect to trees and turf while dry
clay is dominant with respect to 40% wet clay with 300 mm soil thickness
for the objective of reducing heat gain, cooling cost, energy consumption etc.
According to the conservation law of energy, total energy absorbed by
the top surface must be equal to the sum of energy dissipated to the
surrounding from this surface through convection and radiation; and transfer
of heat inside the room. Lazzarin et al. [5] have reported that in case of green
roof the heat dissipated to the surrounding are convection, radiation and
evapotranspiration. The part of heat used to evaporate the moisture content
in the green roof is called evapotranspiration. Lazzarin et al. [5] performed
their experiment in the summer of 2002, 2003 and winter of 2004 on the roof
of S. Bortolo Hospital in Vicenza, Italy with dry soil and insulating layer.
They developed an empirical method for estimation of actual
evapotranspiration which is expressed as:
ET = KcET0 … (1)
They found that with dry soil, green roof reduces thermal heat gain up to
60% as compared with traditional roof with an insulating layer.
Celik et al. [6] performed experimental and theoretical analysis for
insulation properties of different green roof systems. They used 12 different
samples in which there were three species of sedum and four types of
growing media. The sedum species were sedum kamtchaticum, sedum
spurium, and sedum sexangulare while growing Medias were Arkalyte,
Pumice, Lava, Haydite.
They concluded that Haydite and sedum sexangulare combination shows
the best insulation. As most of the part of India belongs to tropical region, so
this can be quite beneficial if green roofs are proposed over the RCC roof to
reduce the electricity consumption and save the environment. Thus in this
paper an effort has been done to investigate the feasibility of green roof in
Indian climate.
2. Problem Formulation
On the basis of literature, it has been observed that the tropical region is
the best zone for the green roof approach. The place of research is Sultanpur,

Page | 45
Uttar Pradesh, India, which belongs to tropical region. The most of the
population of Sultanpur is facing bad comfort condition in the summer,
particularly in the absence of electricity which availability is very poor in
this region. If the green roofs are made in this region then the human comfort
may be increased even in the less availability of electricity. In order to check
the feasibility of green roof in Sultanpur, Uttar Pradesh, India, an
experimental study is required. The green roof will be constructed on a
prototype of a room on the top floor of KNIT Sultanpur, Uttar Pradesh,
India.
3. Methodology
An experimental set up has been prepared which involves fabrication of
a model of a room, fabrication of green roof (soil selection, development of
vegetation), measurement of temperature and solar radiation. In this
investigation following assumptions has been considered:
 Heat flow is uni-directional that is heat is flowing only in vertical
direction. For this purpose, all the side walls are insulated with the
help of thermocol sheet.
 Steady state heat flow that is heat is flowing steadily. In actual, the
heat flow is not flowing steadily as solar radiation is varying all the
time so an average of all the quantities have been taken from the
reading of that day for 10:00AM- 4:00PM.
3.1 Fabrication of Model of a Room
A prototype of a room was constructed whose dimensions are
1m×0.5m×0.9m.
Side Walls
There are three side walls in which one is in the east side, other one is in
the west side and back wall is in the south while the door is in the north.
These side walls are formed with lateral single brick with mixture of cement,
maurang in the ratio of 1:3 along with water. A layer of plaster of thickness 1
cm was used over the brick walls with the same mixture.
RCC Roof
This layer is formed by the mixture of cement, maurang, and aggregate
in the ratio of 1:2:2 along with water. 5.8 kg of TMT rods were used in the
roof. This roof is shown in Fig 1.
Thermocol Cover
The side walls were covered with Thermocol cover internally as well as

Page | 46
externally to eliminate the heat gain or heat loss from walls as shown in the
Fig. 2.

Fig 1: RCC Roof with Fig 2: Side Walls with

Cementitious Coating Thermocol Covers

Fig 3: Soil of KNIT, Sultanpur

Fig 4: Sowed Grass and Soil

Fig 5: Fabricated Green Roof

Page | 47
3.2 Fabrication of Green Roof
The fabrication of green roof was done with locally available materials.
Green roof was fabricated in the following steps:
Selection of Soil
A sandy soil was selected for the green roof as shown in the Fig. 3.
Development of Vegetation
A layer of grass was sowed over the 10 cm thick layer of soil with cow
dung in the ratio of 2:1 over a plastic sheet in a steel container as shown in
the Fig 4. This whole layer of soil was covered by grass within 23 days as
shown in Fig 5.
3.3 Measurement of Temperature, Ambient Air Velocity and Solar
Radiation
A pyranometer is available at the top floor of the Department of
Mechanical Engineering, KNIT Sultanpur as shown in the Fig. 6. The
reading of this pyranometer is recorded automatically at Solar Radiation
Resource Assessment (SRRA) Chennai. Solar radiation, velocity of
atmospheric air and ambient temperature were taken from SRRA. The
specification of sensors and their positions are given in the Table 2 and 3
respectively.

Fig 6: SRRA Station, KNIT Sultanpur, Uttar Pradesh, India

Table 2: Quantities and Specification of Sensor

Sensor/
S. No. Specification Quantity
Instrument
Surface Gadget Hero's Digital Mini LCD Temperature
1. 4
temperature Thermometer with Probe/Sensor

Page | 48
 Table 3: Position of Sensors

S. No. Symbols Means Position

1. T1 Top surface temperature At the top surface of model
T2 Inside roof surface temperature At the bottom surface of roof of
2.
model
3. Ti Inside room air temperature Hanged inside the model
4. To Outside room air temperature Hanged outside the model

4. Results and Discussion

Temperatures of top surface (T1), inside roof surface (T2), inside air for
RCC roof (Tic), inside air for green roof (Tig) and ambient air were recorded
from 10:00 AM to 4:00 PM in the interval of 15 minute. For RCC roof the
temperature were recorded on 01/03/2016, 04/03/2016 and 05/03/2016 and
those for green roof were on 07/03/2016, 10/03/2016 and 11/03/2016. The
Solar radiation (S) and velocity of atmospheric air (V) were taken from
SRRA station KNIT, Sultanpur.
Fig. 7 shows the variation of solar radiation recorded from SRRA KNIT
Sultanpur for 6 days. From Fig. 7, it is seen that the maximum solar radiation
was between 11:00 AM to 1:00 PM on each day. The pattern of variation in
solar radiation is nearly same for all the days. It increases from morning to
noon then it starts decreasing. There is a rapid change in S5 at 10:30 which is
due to the sudden appearance of clouds for less than 30 minutes. The
ambient temperatures (To) of air for these 6 days were recorded from SRRA
and the variation of ambient temperature with time is shown in Fig. 8.
Further the pattern of variation of ambient temperature with time is almost
same for all the days.

Fig 7: Variation in solar radiation with time on respective days

Page | 49
 Fig 8: Variation in ambient air temperature with time on respective days
The temperature (T1) of the top surface of the roof was plotted with
respect to time (Fig.9). From Fig. 9, it is observed that, the top surface
temperature was more in case of reinforced cement concrete roof (RCC) than
that for green roof (GR). The some of the absorbed energy (solar radiation)
by the RCC roof was convected and radiated to the atmosphere and
remaining is conducted through the roof to the room. Lazzarin et al. [5] has
investigated that 25% of the solar radiation was used to evaporate the
moisture in green roof in addition to convection, radiation and conduction as
in the case of RCC roof. Therefore, the temperature of top surface of green
roof is lower than that of RCC roof. From Fig. 9, it is also seen that the top
surface temperature on March 11, 2016 was greater than those on March 7
and 10, 2016, which may be due to higher ambient temperature on March 11,
2016 (Fig. 8).

Fig 9: Variation in top surface roof temperature with time on respective days
From top surface of the roof, heat is conducted through the roof to inside
surface of the roof and then it is convected and radiated inside the room. The
inside roof surface temperature (T2) was plotted with time for all the 6 days

Page | 50
as shown in Fig.10. From Fig. 10, it is seen that inside roof surface
temperatures (T2) were lower for green roof than those for RCC roof. The
rate of increase of temperature is very less in case of green roof. The inside
room temperatures (Tic and Tig) are plotted with time (Fig. 11). From Fig. 11,
it is observed that inside temperatures are increasing with time with nearly
same rate. However, the temperatures for green roof (Tig) were lower than
those for RCC roof (Tic).

Fig 10: Variation in inside roof surface temperature with time on respective days

Fig 11: Variation in inside room temperature with time on respective days
The temperature is one of the parameters which must be lower for better
comfort inside the room. The inside room temperature is mainly due to
convection and radiation from the inside surface of the roof i.e. it can be said
that inside room temperature is function of inside surface temperature of the
roof (T2). For green roof the slopes of the inside surface temperature (T2)
with time is smaller than those for RCC roof, which indicates that in case of
green roof the environment inside the room is more comfort than in case of
RCC roof.

Page | 51
4.1 Theoretical Validation
The absorbed energy by the roof is function of the solar radiation which
is equal to the product of absorptivity (α) and solar radiation (S) i.e. (α×S).
Ignoring the thermal accumulation in the roof, the heat absorbed will be
equal to heat transfer from the roof. If qtotal be the heat transfer from the roof,
α×S = qtotal … (2)
In case of RCC roof, part of the absorbed energy will return back to the
atmosphere through convection (qconv1) and radiation (qrad1) and remaining
will transfer to inside surface of the roof through conduction (qcond).
Therefore,
qtotalR = qrad1 + qconv1 + qcond … (3)
The heat transfer through convection is expressed as per Newton’s law
of cooling which is given in equation 3a.
qconv1 = h0 × (T1 - T0) … (3a)
Where, ho is the heat transfer coefficient for top surface.
The heat transfer through radiation is expressed as per Stefan
Boltzmann’s law of radiation which is given in equation 3b.
qrad1 = ϵ × σ × (T14-T04) … (3b)
Where, ϵ is the emissivity of RCC roof and σ is Boltzmann constant.
The value of heat transfer coefficient can be determined from equation [7]
h = 2.8 + 3.0V W/m2K … (4)
Where, V is the air velocity in m/s.
The energy at the inside surface of the roof (qcond) is further transfer in to
the room through convection (qconv2) and radiation (qrad2). Thus,
qcond = qconv2 + qrad2 … (5)
Where,
qconv2 = hi × (T2-Tic) … (5a)
qrad2 = ϵ × σ × (T2 -Tic )
4 4
… (5b)
From equations (3) and (5), we get;
qtotalR = qrad1 + qconv1 + qconv2 + qrad2 … (6)
The modes of heat transfer from the green roof are same as in the case
of RCC roof with an additional mode of heat transfer which is
evapotranspiration. If qtotalG be the total heat transfer from the green roof,

Page | 52
 αs×S = qtotalG … (7)
And
qtotalG = qrad1G + qconv1G + qevapo + qcond … (8)
qrad1G is the heat transfer through radiation, qconv1G is the heat transfer
through convection, qevapo is the heat transfer through evapotranspiration [5]
from the top surface of green roof which are expressed as:
qrad1G = ϵs × σ × (T14-T04) … (8a)
qconv1G = h0 × (T1-T0) … (8b)
qevapo = 0.25 × S … (8c)
Where, ϵs is the emissivity of soil on the top surface of green roof.
qcond is the heat transfer through conduction from top of green roof to the
inside surface of the roof. Further the heat flows from inside surface of roof
to the room through convection (qconv2G) and radiation (qrad2G). Thus,
qcond = qconv2G + qrad2G … (9)
Where-
qconv2G = hi × (T2-T0)
The value of hi is determined from equation (4). Since the velocity
inside the room was observed to be zero, thus hi is equal to 2.8 W/m2K. So
qconv2G = 2.8 (T2-T0) … (9a)
And
qrad2G = ϵ × σ × (T24-Tic4) … (9b)
From equations (8) and (9), we get;
qtotalG = qrad1G + qconv1G + qevapo + qconv2G + qrad2G … (10)
In case of absorptivity of cementitious coated RCC roof and green roof,
neither any direct method is known nor it’s confirmed value is available in
literature. Lazzarin et al. [5] suggested that absorptivity of traditional roof
(RCC) and wet green roof are 0.86 and 0.39 respectively while Luuk
Gunnink [8] suggested that reflectance of cementitious coated roof varies
from 0.5 to 0.7 which means absorptivity value will vary in between 0.3 to
0.5 when transmissivity is taken zero. Therefore, in this investigation
absorptivity of green roof (αs) has been taken equal to that of RCC roof (α)
which is 0.3 [8]. The value of emissivity for RCC roof [9] is taken as 0.91
while that for soil of green roof [10] is taken as 0.96.

Page | 53
 Energy balances for the RCC roof and green roof are shown in the
Tables 4 and 5 respectively.
Table 4: Energy Balance Table for the RCC Roof

Day T1 T2 Tic T0 S V ho qconv1 qrad1 qconv2 qrad2 qtotalR αS

1 38.2 31.4 24.7 29.3 568.5 1.57 7.52 67.0 53.0 18.8 37.7 176.1 170.6
4 40.4 33.6 26.1 31.8 589.8 2.03 8.89 76.5 52.4 21.0 43.0 192.9 177.0
5 40.7 32.8 226.2 31.6 550.8 1.24 6.51 59.2 55.5 18.5 37.7 170.9 165.2

Table 5: Energy Balance Table for the Green Roof

qconv1G

qconv2G
qrad1G

qrad2G

qtotalG
qevapo
Day T1 T2 Tig S To V ho α sS

7 33.0 23.9 23.0 624.4 30.4 1.41 7.02 18.3 16.0 156.1 2.5 4.8 197.8 187.3
10 32.7 24.7 23.1 693.6 30.2 3.75 14.05 35.1 15.4 173.1 4.5 8.6 237.0 208.1
11 35.9 25.3 24.6 659.9 31.8 3.06 12.0 49.1 25.8 165.0 2.0 3.8 245.7 198.0

From Tables 4 and 5, it is quite obvious that in both the cases the total
heat transfer from the top surface (qtotalR or qtotalG) is more than the heat
absorbed in respective roofs. This is due to the selection of absorptivity of
roof. Therefore, it is possible that the absorptivity of RCC and green roof
may be greater than 0.3. Further, it is also observed that the absorptivity of
green roof will be more than that of RCC roof for thermal energy balance.
However, this interpretation is contrary to the findings of Lazzarin et al. [5].
4.2 Estimation of Thermal Conductivity
The thermal conductivities for RCC and green roofs are determined by
considering the heat transfer through conduction from equations (5) and (9).
The values of thermal conductivities for RCC and green roofs are shown in
Tables 6 and 7 respectively. From Tables 6 and 7, it is observed that the
thermal conductivity for green roof is less than that of RCC roof. The
thermal conductivity determined for RCC roof is 0.83 which is less than that
of available value for RCC roof (1.58) [11]. In equation (5), it is considered
that the heat transfer through conduction (qcond) is equal to the sum of heat
transfer through convection (qconv2) and heat transfer through radiation (qrad2).
But it may be possible that the heat transfer through conduction (qcond) is
more than the above to heat transfers because there will be some
accumulation of thermal energy inside the roof. Therefore, the actual value
of thermal conductivity may be higher than that of the calculated value.
From the above discussions, it can be proposed that one additional mode
of heat transfer (evapotranspiration) and lower value of thermal conductivity

Page | 54
of green roof are mainly responsible for lower inside temperature for green
roof as compared to RCC roof (Fig. 11).
Table 6: Thermal conductivity of RCC roof

Date qcond = qconv2 + qrad2 Thickness (l) K = qcond× l /(T1- T2)

01/03/2016 56.83 0.1 0.85
04/03/2016 63.66 0.1 0.93
05/03/2016 56.11 0.1 0.71
Mean 0.83

Table 7: Thermal conductivity of green roof

Date qcond = qconv2 + qrad2 Thickness (l) K = qcond× l /(T1- T2)

07/03/2016 7.52 0.2 0.17
10/03/2016 13.20 0.2 0.33
11/03/2016 5.69 0.2 0.11
Mean 0.31

The decrease in temperature of the inside room can be further enhanced

by preparing the green roof with lower thermal conductivity. However, from
this investigation it is proposed that green roof is developed over the RCC
roof, the environment inside the room can be maintained comfortable.
5. Conclusions
In this investigation, a model of room with RCC roof was made. A
green roof was developed on the RCC roof. It was found out that there is
decrease in temperature of the inside room when green roof was present.
From this experiment it is advisable to develop green roof on the upper
surface of the RCC roof to maintain the lower temperature inside the room
which is one of the parameters required for comfort environment.
References
1. Karna Binay, Kumar Sanjeev, Mukul Manwendra, Suman RP, Sinha
Renu. “Lucent’s general knowledge”, Lucent Publication, 5th edition,
2014, 663.
2. Magill Jhon D, Midden Karen, Groninger Jhon, Therrell Mathew. A
History and Definition of Green Roof Technology with
Recommendations for Future Research, Southern Illinois University
Carbondale Open SIUC, 2011.
3. Wong NH, Chen Y, Ong CL, Sia A. Investigation of Thermal Benefits
of Rooftop Garden in the Tropical Environment, Building and
Environment. 2003; 38:261-270.

Page | 55
4. Wong NH, Cheong DKW, Yan H, Soh J, Ong CL, Sia A. The Effects of
Roof Top Garden on Energy Consumption of a Commercial Building in
Singapore, Energy and Buildings. 2003; 35:353-364.
5. Lazzarin RM, Castellotti F, Busato F. Experimental Measurements and
Numerical Modeling of a Green Roof, Energy and Buildings. 2005;
37:1260-1267.
6. Celik S, Morgan S, Retzlaff W, Once O. Thermal Insulation
Performance of Green Roof System, International Green Energy
Conference-vi. 2005; 2011:076:232-238.
7. Tiwari GN. Solar Energy, Fundamentals, Design, Modeling and
Applications, Narosa Publication, Revised Edition, 2013, 285.
8. Gunnink luuk. Heat Flux Roof Effect Colour Rooftop on Heat Flow
through Roof during summer, Building Engineering, TU Delft, Feb,
2012.
9. Fronapfel Edward l, Stuls Bradley J. Emissivity Measurement of
Common Construction Material, Inframation Proceedings, 2006.
10. Cengel Yunus A. Heat and Mass Transfer, a practical approach,
McGraw-Hill Education, India, (table 1.6), 2007, 28.
11. Conductivity of RCC Roof, APPENDIX IV.1, Properties of Building
Materials, http://mnre.gov.in/solar-energy/ch4.pdf [Accessed on
18/04/2016].

Page | 56
 Chapter - 4
Influences of Engine Speed Variations on Single
Cylinder Direct Injection Diesel Engine

Authors
Upendra Rajak
Department of Mechanical Engineering, National Institute of
Technology, Imphal, Manipur, India
Tikendra Nath Verma
Department of Mechanical Engineering, National Institute of
Technology, Imphal, Manipur, India

Page | 57
Page | 58
 Chapter - 4
Influences of Engine Speed Variations on Single Cylinder
Direct Injection Diesel Engine
Upendra Rajak and Tikendra Nath Verma

Abstract
In this paper, the effects of speed on the physical behaviour in combustion
cylinder process in a diesel engine fueled with diesel. A numerical study was
performed using the CFD software fluent at different engine speeds. All input
factors such as compression ratio, spray angle, injection timing and input mass
flow rate have been unchanged, and variable factors are the engine speed to a
diesel engine. The results show that, at high engine speed, achieve a more
cylinder pressure and temperature, which is beneficial to the in-cylinder
combustion processes. The show's growth in the rate of the heat releasing with
increasing engine speed. The found that better engine speed on a 1500 rpm for
the four strokes, diesel engine.
Keywords: Numerical study; speed; diesel engine
1. Introduction
Currently, unadventurous diesel engines and gasoline engines face the
challenge of global warming and renewable energy resources [1]. In this
regards, many researchers presented the different way of clean combustion
and reduction emission by using alternative fuel [2]. However, previous study
devising various methods to the reduction in emissions such as CO (carbon
monoxide), HC (unburned hydrocarbons), smoke, nitrogen oxides and PM
emissions have rendered the need for the application of chamber design,
nozzle, injection, swirl [3-5].
Tay K.L et al. [6] investigated the effect of triangular and ramp injection
rate-shapes on the engine characteristics using kerosene-diesel through
numerical simulations. Results show that rate-shaping affect combustion
control and the significant reduction in exhaust gas emission. Hosseini S.M.
and Ahmadi R. [7] studied the effect of hydrogen-diesel fuel on a diesel engine
characteristics at full load with 1600 rpm for simulation. The simulated results
show that reduction in emissions of NOX by 8%, HC by 54%, soot by 14%,

Page | 59
CO by 70% by and CO2 by 14%. Indicated power higher adding hydrogen in
combustion chamber and advance in the rise of pressure and also heat release
rate. Yousefi A. and Birouk M. [8] studied the influence of natural energy and
fuel injection timing at unchanged total energy input on the engine
characteristics under low load condition. The better result shows that injection
timings (12 and 20 b TDC), the maximum ITE (26.7%) occurs at 50% natural
gas energy fraction.
In present simulation is the effect of different engine speed with the use
diesel fuel into combustion cylinder. Further precisely, the effect of engine
speed (1200 rpm to 1800 rpm) and constant injection timing (21.5º b TDC) on
combustion parameters will be investigated at a full engine load (100% load).
To achieve the objectives of this simulation of CFD model, which has been
developed is used for simulation work.
2. Modeling and Validation
2.1 CFD Model
Simulation for fuel diesel, new fuel combustion is completed CFD-Fluent
software package [10]. To reduce cost of computational, meshing is performed
just for 1/6 of the bowl shape piston geometrical symmetry by module, and it
is depicted in Fig. 1 and mesh size is about 1 mm [5]. The number of meshes
at TDC was about 11,972 as shown in Fig. 1 and cylinder region as shown in
Fig. 2.
Numerical simulation of initial boundary conditions of the 3D model are
given in Table 1 and wall temperature is shown in Table 2. The engine speed
was1500 rpm, 1800 rpm and 2100 rpm in the numerical study and grid
independence test are conducted at 1500 rpm due to it is a commonly used
condition of the single-cylinder direct injection diesel engine in general for
three different number of mashes (15544, 11972 and 9616) and selected 11972
for simulation as shown in Fig. 3. For the numerical simulation, the k-Ɛ
Turbulence model was selected [7].

Fig 1: Mesh specification of

the model at (TDC) Fig 2: Region of cylinder

Page | 60
 Fig 3: Grid independency test at 1500 rpm
Table 1: Basic specifications of engine

S. No. Engine Type Single Cylinder

1. Compression ratio 17.5
2. Speed 900-2100 rpm
3. Cooling system water
4. Fuel diesel
5. Power 3.7
6. Bore & stroke 80 mm &110 mm
7. Connecting rod length 235.0 mm
8. Number of valve 4
9. Number of holes 6
10. Swirl ratio 1.3
11. Injected duration (CA) 21.5º b TDC
12. Spray angle 70.0º

Table 2: Temperature boundary conditions at full load

Name of Boundary Type of Boundary Temperature (K)

Inlet of intake port inlet/outlet 300
Intake port wall 602
Cylinder head wall 567
Cylinder liner wall 567
Piston top movement 645

3. Results and Discussion

3.1 Combustion Parameter
3.1.1 Cylinder Pressure
Fig. 3. Cylinder pressure summarizes the effects of varying the engine
speed versus crank angle. The result of combustion pressure increase with the

Page | 61
increase in engine speed all speed for single cylinder four strokes direct
injection diesel engine [14]. In this study observed 1500 rpm diesel engine
smooth running as compared to 1200 rpm and 1800 rpm. Initially, the rate of
combustion depends on cylinder peak pressure of diesel engine and further it
is affected by fuel which is involved in premixed combustion phase. At
CR17.5, the cylinder peak pressure (CPP) of diesel was found to be 15.2 MPa
for 1200 rpm, 15.6 MPa for 1500 rpm and 15.98 MPa for 1800 rpm at full load
condition.

Fig 4: Pressure versus crank angle

Fig 5: Heat release rate versus crank angle

3.1.2 Cylinder Heat Release Rate

The diesel depicts a variation of the cylinder heat release rate concerning
the different engine speeds as shown in Fig. 5 at CR17.5. The cylinder peak

Page | 62
heat release rate (CPHRR) was found to be 20.82 J/CA for1200 rpm, 21.97
J/CA for 1500 rpm, 17.4 J/CA for 1800 rpm. As can be seen in this figure
CPHRR was evaluated, and better performance was found to be at 1500 rpm
engine speed which is higher as compared to other speeds diesel.
3.1.3 Cylinder Temperature
Fig. 6 shows the cylinder temperature versus crank angle at the different
speed for diesel fuel. As can be seen in this figure, cylinder temperature
increase with increasing engine speed. Cylinder temperature higher central
region for higher NOX produced due to higher flame temperature. The cylinder
temperature directly control on emission and knocking tendency in engine [9,
13, 14, 15, 16]
. The cylinder peak temperature (CPT) was found to be 2918.9 K
for1200 rpm, 3278.34 K for 1500 rpm, 3324.66 K for 1800 rpm. As can be
seen in this figure CPT was evaluated higher for 1800 rpm engine speed.

Fig 6: Cylinder temperature versus crank angle

3.1.4 Temperature Contours

Fig. 7 shows the contour of temperature distribution in the combustion
chamber at a different crank angle with various engine speed. All images were
given in a range of temperature 500–3440 K. Higher combustion cylinder
temperature occurred near about in the bowl wall region due to oxygen is
regularly present region [11]. The temperature contours show that bulky
volumes of energy in the outer area of the burning chamber, representing an
inclination for gap flow. This effect is estimated, meanwhile the circular feast
from the piston bowl [12]. The higher temperature occurred at 736 crank angle
in all engine speed. Higher temperature occurred at fuel injection into the
combustion cylinder.

Page | 63
Fig 7: Combustion chamber temperature contour changes in various engine speed at
full load condition with crank angle and when piston move TDC to BDC

3.1.5 Velocity Contours

Fig. 8 shows the contour of velocity distribution in the combustion
chamber at the different crank angle with various engine speed. All images
were given in the range of temperature 4.06–81.3 m/s. Higher combustion
cylinder velocity occurred near about in the nozzle region and increases with
engine speed and decrease with crank angle. The higher velocity occurred at
736 crank angle in all engine speed at fuel injection into the combustion
cylinder.

Page | 64
Fig 8: Combustion chamber velocity contour changes in various engine speed at full
load condition with crank angle and when piston move TDC to BDC

4. Conclusion
The following decisions are perceived of the combustion behavior of
engine:
 From the different speed was obtained superior running with 1500
rpm.
 The combustion pressure, maximum temperature and heat release
rate increase with higher engine speed.

Page | 65
  The results deliver numerical simulation and graphical evidence
about burning behaviour in the diesel engine.
 End the compression stroke was obtained highly homogeneous fuel-
air mixtures at each engine speed. The piston bowl shape to be
obtained maximum temperature and velocity at fuel injection at 736
CA with each speed.
 Predicted peak pressure of 15.98 MPa for 1800 rpm, heat release rate
of 21.97 J/CA for 1500 rpm and temperature of 3324.66 K for 1800
rpm.
Reference
1. Wang Z, Li L, Wang J, Reitz RD. Effect of biodiesel saturation on soot
formation in diesel engines. Fuel. 2016; 175:240-248. https://doi.org/
10.1016/j.fuel.2016.02.048
2. An H, Yang W, Li J, Maghbouli A, Chua KJ, Chou SK. A numerical
modeling on the emission characteristics of a diesel engine fueled by
diesel and biodiesel blend fuels. Applied Energy. 2014; 130:458-465.
https://doi.org/10.1016/j.apenergy.2014.01.004
3. Jafarmadar S, Nemati P. Numerical investigation of the effect split
injection scheme on exergy terms in an IDI (indirect injection) diesel
engine by three dimensional modeling. Energy. 2015; 93:2280-2291.
https://doi.org/10.1016/j.energy.2015.10.104
4. Song E, Liu Z, Yang L, Yao C, Sun J, Dong Q. Effects of nozzle structure
on the gas mixture uniformity of marine gas engine. Ocean Engineering.
2017; 142:507-520. https://doi.org/10.1016/j.oceaneng.2017.07.011
5. Li X, Zhou H, Ming L, Su L, Xu H, Liu F. Effect of split injections
coupled with swirl on combustion performance in DI diesel engines.
Energy Conversion and Management. 2016; 129:180-188.
https://doi.org/10.1016/j.enconman.2016.09.011
6. Tay KL, Yang W, Zhao F, Yu W, Mohan B. Effects of triangular and
ramp injection rate-shapes on the performance and emissions of a
kerosene-diesel fueled direct injection compression ignition engine : A
numerical study. Applied Thermal Engineering. 2017; 110:1401-1410.
https://doi.org/10.1016/j.applthermaleng.2016.09.072
7. Hosseini SM, Ahmadi R. Performance and emissions characteristics in
the combustion of co-fuel diesel-hydrogen in a heavy duty engine.
Applied Energy. 2017; 205:911-925. https://doi.org/10.1016/j.apenergy.
2017.08.044.

Page | 66
8. Yousefi A, Birouk M. Investigation of natural gas energy fraction and
injection timing on the performance and emissions of a dual-fuel engine
with pre-combustion chamber under low engine load. Applied Energy.
2017; 189:492-505. http://dx.doi.org/10.1016/j.apenergy.2016.12.046
9. Rajak U, Nashine P, Subhaschandra T, Nath T. Numerical investigation
of performance, combustion and emission characteristics of various
biofuels. Energy Conversion and Management. 2018; 156:235-252.
https://doi.org/10.1016/j.enconman.2017.11.017
10. Coskun G, Soyhan HS, Demir U, Turkcan A, Ozsezen AN, Canakci M.
Influences of second injection variations on combustion and emissions of
an HCCI-DI engine: Experiments and CFD modelling. Fuel. 2014;
136:287-294. https://doi.org/10.1016/j.fuel.2014.07.042
11. Kiat H, Gan S, Ng J, Mun K. Simulation of biodiesel combustion in a
light-duty diesel engine using integrated compact biodiesel-diesel
reaction mechanism. Applied Energy. 2013; 102:1275-1287.
https://doi.org/10.1016/j.apenergy.2012.06.059
12. Keskinen K, Kaario O, Nuutinen M, Vuorinen V, Künsch Z, Ola L et al.
Mixture formation in a direct injection gas engine : Numerical study on
nozzle type, injection pressure and injection timing effects. Energy. 2016;
94:542-556. https://doi.org/10.1016/j.energy.2015.09.121
13. Rajak U, Verma TN. Effect of emission from ethylic biodiesel of edible
and non-edible vegetable oil, animal fats, waste oil and alcohol in CI
engine, Energy Conversion and Management. 2018; 166:704-718.
https://doi.org/10.1016/j.enconman.2018.04.070
14. Rajak U, Verma TN. Spirulina microalgae biodiesel-A novel renewable
alternative energy source for compression ignition engine, Journal of
Cleaner Production, 201:343-357. https://doi.org/10.1016/j.jclepro.
2018.08.057
15. Rajak U, Nashine P, Verma TN. Assessment of diesel engine
performance using spirulina microalgae biodiesel. Energy, 2018.
https://doi.org/10.1016/j.energy.2018.10.098
16. Rajak U, Verma TN. Comparative assessment of the emission
characteristics of first, second and third generation biodiesel as fuel in a
diesel engine. Journal of Thermal Engineering, (Accepted Manuscript),
2018.

Page | 67
Page | 68
 Chapter - 5
Regression and Classification in Machine
Learning

Authors
Anandan R
Department of Computer Science and Engineering, Vels
Institute of Science, Technology and Advanced Studies,
Chennai, Tamil Nadu, India
Kalaivani K
Department of Computer Science and Engineering, Vels
Institute of Science, Technology and Advanced Studies,
Chennai, Tamil Nadu, India

Page | 69
Page | 70
 Chapter - 5
Regression and Classification in Machine Learning
Anandan R and Kalaivani K

Abstract
Regression analysis is a set of statistical processes used for estimating
the relationships among independent and dependent variables. There are
seven types of regression analysis like Linear Regression, Logistic
Regression, Polynomial Regression, Stepwise Regression, Ridge Regression,
Lasso Regression and Elastic Net Regression. These types are classified
based on three metrics or indices namely the number of independent
variables, type of dependent variables and shape of regression line.
Classification is the problem of identifying to which of a set of
categories a new observation belongs, on the basis of a training set of data
containing instances whose category membership is known. Classification is
categorized into two types based upon the identified problems as binary
classification and multiclass classification. In binary classification, the task
is well defined and only two classes are involved, whereas multiclass
classification involves assigning an object to one of several classes. Since
many classification methods have been developed specifically for binary
classification, multiclass classification often requires the combined use of
multiple binary classifiers
Keywords: Regression, Classification, Linear Regression, Logistic
Regression, Polynomial Regression, Stepwise Regression
1. Regression Analysis or Regression Algorithm
Regression analysis is a form of predictive modeling technique or
method, which integrates the relationship between Explanatory or dependent
variable and independent variable. In this chapter, let us discuss in detail
about four regression models commonly used such as linear regression,
logistic regression, polynomial regression and stepwise regression. It is
mainly used to determine the strength of predicators or results.
Some of the applications of Regression analysis are as follows:

Page | 71
  To predict the salary of an employee in a particular company based
on their age.
 To forecast an effect of Rainfall in a particular season
 To find the sales of a product based on the marketing ability of
employee.
 It also helps to predict or determine the predicator as shown below:
 How much additional income do I get when the turnover is more
than 100$?
 What will be the price of an item for next 6 months?
 What will be the dollar rate for next successive years?
1.1 Linear Regression
Linear Regression is also known as line of best fit, here the data is
modeled using a straight line and it always uses a continuous variable. The
accuracy can be measured by loss or by R square method or by adjusted R
square method. The figure (Fig.1) below shows linear Regression model.

Fig 1: Linear Regression

From the above figure, it is inferred that it follows a mathematical

equation as Y = mx + c, where Y represents Independent variable, m –
represents slope, x – represents Dependent variable and c – represents the
intercept.
1.1.1 Selection of Regression Model
Linear Regression model is always used to find that there is a
continuous mapping of x variable to continuous y variable. This model is not
best suited for classification.

Page | 72
 There is important query among readers to know when or under what
circumstances we should follow linear regression. When we have to predict
the future of continuous variable, the option is linear regression and here the
task or challenge is to fit a straight line based on the data set.
1.1.2 Data Quality
To maintain the quality of data, it is always important to denote the
missing value by removing the data points away from regression line.
1.1.3 Computational Complexity
This linear Regression model is not computationally complex compared
to decision tree and clustering methods. The complexity depends on
dependent and independent variables x and y. Let us discuss in detail about
some real time instances to understand the breath of algorithm.
1. Evaluating the Trends and Sales Estimate
In the above example our goal is to find the relationship between the
monthly sale of goods and time. The independent variable is monthly sale
and dependent variable is time and the graph (Fig 2) below denotes the
regression analysis.

Fig 2: Sales Estimate

From the above graph we infer that as time increases there is a steady
increase in monthly sales.
2. Analyzing the Impact of Price Change
In this example the impact of linear Regression can be studied based on
quality and price of an item. Here Independent variable is sales and
dependent variable is price. The figure (Fig 3) below shows the regression
analysis.

Page | 73
 Fig 3: Impact of price change

From the above graph, we infer that as price increases the quality of an
item also increases or vice versa.
3. Assessment of Risk in Financial Service and Insurance Domain
From the prima facie of the problem, we can come to a conclusion that
there is a relationship between Number of customers and Age of the
customer. Here Independent variable is Age and dependent is customers. The
figure (Fig. 4) below shows the model.

Fig 4: Risk Assessment

From the above graph, we infer that the insurance claim is

comparatively more from old customers.
1.2 Types of Linear Regression
There are basically two types of linear regression namely positive
regression and negative regression.

Page | 74
1.2.1 Positive Regression
In positive regression, the data point in dependent variable x increases
and independent variable y is kept constant and moves gradually as shown in
figure (fig. 5) below.

Fig 5: Positive Regression

There is a positive regression line with a mathematical notation Y = mx

+c
1.2.2 Negative Regression
In negative regression, the data point in independent variable y
decreases and dependent variable x is constant as shown in figure (fig. 6)
below.

Fig 6: Negative Regression

It is noticed from figure that there is a negative regression line of Y = -

mx + c

Page | 75
1.3 The Mathematical Procedure of Linear Regression Model
To implement a mathematical model, let us discuss with an example.
Our motive is to find a Regression model for a traveler based on speed and
distance covered by a vehicle. Let us assume a dependent variable x as speed
and independent variable y as distance covered. The steps to be followed for
building a mathematical model are as follows:
 Find the Independent and dependent variable and denote it as x & y.
 Find the mean of x and y.
 Determine the linear Regression equation as Y = mx + c
̅][𝒚− 𝒚
∑[𝒙−𝒙 ̅]
 Find the slope m from above equation as m = ̅]𝟐
∑[𝒙− 𝒙

 Find the constant c from above equation.

 From the given values of m and c, determine the predicted values of
y.
 Evaluate the distance between actual and predicted value.
 Finally find the goodness of fit using R2 method.
For example if speed x is {10, 20, 30, 40, 50} and distance y is {30, 40,
20, 40, 50}, Let us proceed step by step to build a mathematical model. The
mean value of x is 30 and y is 36.
x y ̅
𝒙−𝒙 ̅
𝒚− 𝒚 ̅)𝟐
(𝒙 − 𝒙 ̅) (𝒚 − 𝒚
(𝒙 − 𝒙 ̅)
10 30 -20 -6 400 120
20 40 -10 4 100 -40
30 20 0 16 0 0
40 40 10 4 100 40
50 50 20 14 400 280
̅)(𝒚− 𝒚
∑(𝒙−𝒙 ̅)
Now let us find slope m = ̅ )𝟐
∑(𝒙−𝒙

From the above table, we infer y = 36, x = 30 and m = 0.4

y = mx + c
36 = 0.4(30) + c
c= 24.
Let us predict the value of y for different values of x
y = 0.4 x + 24
The values of x is 10, 20, 30, 40, 50

Page | 76
 y = 0.4(10) + 24 = 28
y = 0.4(20) + 24 = 32
y = 0.4(30) + 24 = 36
y = 0.4(40) + 24 = 40
y = 0.4(50) + 24 = 44
The predicted values of y are 28, 32, 36, 40, 44
x y ̅
𝒚− 𝒚 ̅)𝟐
(𝒚 − 𝒚 yp yp - y (yp – y)2
10 30 -6 36 28 -8 64
20 40 4 16 32 -4 16
30 20 16 256 36 0 0
40 40 4 16 40 4 16
50 50 14 196 44 8 64

Now we have to find the difference between observed value and

predicted value
To determine this, we have to find the goodness of fit by R2 method.
The formula to find R2 is as follows
𝒑 ̅)2
∑(𝒚 − 𝒚
R2 = ∑(𝒚−𝒚
̅)𝟐

R2 = 16/52 = 0.3
From the value of R2, we infer that the value is very low and it does not
fit, since the data points are far away. When we increase the value of R, it
comes closer. If the value is 0.9 it’s closer and if it is 1, it exactly matches.
2. Logistic Regression
Logistic Regression or logit regression or logit model is a regression
model, where the dependent variable is categorical and the probability of
some obtained event is represented as a linear function of a combination of
predictor variable, the output is probability of occurrence and event. The
output value is discrete like 0 or 1, yes or no and goodness of fit can be
obtained by accuracy, precision, recall, f score, ROC curve or by confusion
matrix.
Some of the applications of logistic regression are as follows
 It is used to predict which team will win in a particular match.
 It is used in Image segmentation.
 It is used in geographic processing.

Page | 77
2.1 Logistic Regression Model
To determine the regression analysis, it is predominant to design a
model. The figure (fig. 7) below shows the logistic regression model.

Fig 7: Logistic Regression

It consists of a data set, model and decision based on the function in a

model.
Suppose, if we consider our Target is to develop a logistic model for a
football team and to predict which team wins from opponent. We have to
calculate the probability of success. If the probability is above 0.5, consider
as win and the value is one and if it is less than 0.5, then it is failure and the
value is assumed as zero. The equation for straight line in linear regression is
Y = mx + c. This can be extended for a set of values as Y = c + b 1x1 + b2x2 +
b3x3+… where y ranges from minus infinity to plus infinity.
Now, let us try to reduce the logistic regression from the above equation
and if we set a value of y from zero to infinity then it’s transformation is
Y/1-y, y=0 when range is 0 and y=1 when range is infinity on further
transformation for a range from minus infinity to plus infinity
log [y/1-y] = c+b1x1+b2x2+………
This equation can be denoted as hø (x) = g (øtx)
G (z) = 1/1+e-z.

Page | 78
This equation implementation is shown in figure (fig. 8) below.

Fig 8: Selection based on Threshold

Following are the steps to develop a logistic regression model:
 Develop a regression model
 Discover approximately what credit score is associated with
probability 50%
 Input the observed value.
 Determine the optimal credit score.
Now let us discuss the logistic regression model with an example.
Consider a bank is sanctioning a loan to its customer based on their credit
score. Their respective credit score and the result (i.e. the status of approval)
is given in table below
S. No. Credit Score Status of Approval
1 655 0
2 692 0
3 681 0
4 663 1
5 688 1
6 693 1
7 699 0
8 699 1
9 683 1
10 698 0
11 655 1
12 703 0
13 704 1
14 745 1
15 702 1

Page | 79
 Here the maximum value of credit score n = 1000 and it is assumed that
if the status is zero the loan is rejected and if the value is one then the loan is
approved. Based on the data given in table, a graph is plotted as shown in
below figure (fig. 9).

Fig 9: Status of Approval for sanctioning Bank Loan

From the graph above,
 A probability model is created for an event occurring depending on
the values of independent variables, which can be categorical or
numerical.
 The probability is estimated.
 The effect of a series of variables on a binary response variable is
predicted.
 The observations are classified by estimating the probability on
observation of a particular category.
From the table we get the approval for following applications
663,688,699,683,704,745, 702 and the rejected applications are 655,
692,681,699,698,703.
Suppose, if we want to calculate the logistic model of an application
having a score of 670 and to determine whether the loan is sanctioned or
rejected.
This can be attained by the equation
𝑦 𝑒 𝛽0 +𝛽1𝑥1+ 𝛽2𝑥2
𝑙𝑜𝑔 = =y
1−𝑦 1+ 𝑒 𝛽0 +𝛽1𝑥1+ 𝛽2𝑥2

The status of approval (y) for different credit score were determined
using the above equation and the following graph (fig. 10) is developed

Page | 80
 Fig 10: Status of Approval using logistic regression

The goodness of fit for the logistic coefficient can be achieved by

Gradient descent method or by stochastic gradient descent method. The
figure (fig. 11) below shows one such model.

Fig 11: Gradient Descent

3. Polynomial Regression
The polynomial Regression is mainly used to find the relationship
between the dependent variable and independent variable. A model is said to
be linear when it is linear in parameters.

Page | 81
The model
Y =a0+a1 x +a2 x2+ε
and
y =a0+a1 x1+a2 x2+a11 x12+a22 x22+a12 x1 x2+ε
are the linear models. In fact, they are the second order polynomials in
one and two variables respectively. The polynomial models can be used in
situations where the relationship between study and explanatory variables is
curvilinear. Sometimes a nonlinear relationship in a small range of
explanatory variable can also be modeled by polynomials. The figure (fig.
12) below shows the polynomial Regression model.

Fig 12: Polynomial Regression

Let us illustrate the polynomial Regression model by an example, now

consider a problem to find the yield of rice crops at 33 degree centigrade.
Given a data of temperature for 5 months with successive yield in tons per
hectare as shown in table below.
Month Temp in 0 C - T Yield in tons/hectare - Y
January 30 4.94
February 31 3.29
March 34 3.33
April 36 3.12
May 38 3.00

From the above table, we understand that we have go for second order
polynomial since we have two variables in this case. Let us consider

Page | 82
Temperature T as dependent variable along x axis and yield Y as
independent variable along y axis. The second order polynomial equation is
Y = 𝑎0 +𝑎1 𝑇𝑖 +𝑎2 𝑇𝑖2
The second order polynomial equation can be solved by differentiation
of second order and we get a coefficient matrix as follows.
𝑛 ∈ 𝑇𝑖 ∈ 𝑇𝑖2 𝑎0 ∈ 𝑌𝑖
⌊ ∈ 𝑇𝑖 ∈ 𝑇𝑖2 ∈ 𝑇𝑖 ⌋ ⌊ 1 ⌋= ⌈ 𝑌𝑖 𝑇𝑖 ⌉
3 𝑎 ∈
∈ 𝑇𝑖2 ∈ 𝑇𝑖3 ∈ 𝑇𝑖4 𝑎2 ∈ 𝑌𝑖 𝑇𝑖2
The left hand side is the coefficient matrix, is the polynomial variables
and the right hand side is the vector and n is the number of items in the
record. This equation can be solved by Naive Gaussian Elimination method
N=5
∈ 𝑇𝑖 = [30+31+34+36+38] = 169
∈ 𝑇𝑖2 = [(30)2+(31)2+(34)2+(36)2+(38)2] = 5757
∈ 𝑇𝑖3 = [(30)3+(31)3+(34)3+(36)3+(38)3] = 197623
∈ 𝑇𝑖4 = [(30)4+(31)4+(34)4+(36)4+(38)4]= 6834609
∈ 𝑌𝑖 = [4.94+3.29+3.33+3.12+3.00] = 17.68
∈ 𝑌𝑖 𝑇𝑖 = [4.94(30)+3.29(31)+3.33(34)+3.12(36)+3.00(38)] =
589.73
∈ 𝑌𝑖 𝑇𝑖2 = [4.94(30)2+3.29(31)2+3.33(34)2+3.12(36)2+3.00(38)2 =
19832.69
5 169 5757 𝑎0 17.68
⌈ 169 5757 197623 ⌉ ⌊𝑎1 ⌋ = ⌈ 589.73 ⌉
5757 197623 6834609 𝑎2 19832.69
By forward Elimination method the coefficient matrix is solved as
follows
169÷5 = 33.8
The value 33.8 is multiplied with first row of the matrix
[5 169 5757 17.68] x [33.8]
169 5712.2 194586.6 597.584
Now, we have to subtract the resultant with matrix 2 nd row
169 5757 197623 589.73
(-) 169 (-) 5712.2 (-) 194586.6 (-) 597.584
0 44.8 3036.4 -7.854

Page | 83
 5 169 5757 𝑎0 17.68
⌈ 0 44.8 3036.4 ⌉ ⌊𝑎1 ⌋ = ⌈ −7.854 ⌉
5757 197623 6834609 𝑎2 19832.69
Similarly column 3 of first order derivative coefficient matrix is solved
by forward Elimination method and the resultant matrix is
5 169 5757 𝑎0 17.68
⌈0 44.8 3036.4 ⌉ ⌊𝑎1 ⌋ = ⌈−7.854⌉
0 −3036.4 −205999.2 𝑎2 524.06
Similarly the second order matrix is
5 169 5757 𝑎0 17.68
⌈0 44.8 3036.4 ⌉ ⌊𝑎1 ⌋ = ⌈−7.854⌉
0 0 −192.0 𝑎2 −8.280
By substituting the values in the matrix. The values of 𝑎0 ,𝑎1 and 𝑎2 are
derived as follows:
-192.0 𝑎2 = -8.28
𝑎2 = -8.28 ÷ -192.0 =0.043
44.8𝑎1 + 3036.4𝑎2 = -7.854
44.8𝑎1 +3036.4(0.043) = -7.854
𝑎1 = -3.09
5𝑎0 +169𝑎1 + 5757𝑎2 = 17.68
5𝑎0 +169(-3.09) + 5757(0.043) = 17.68
𝑎0 = 58.5
Second order polynomial equation is
Y = 𝑎0 +𝑎1 𝑇𝑖 +𝑎2 𝑇𝑖2
Temperature at 33°C the yield is calculated as
Y = 58.5 + (-3.09) (33) + (0.043) (33)2
Y = 3.357
At temperature 33°C the yield is 3.357ton/hectare.
4. Stepwise Regression
Stepwise regression is a variable selection process for an independent
variable. It consists of a series of steps designed to find the most useful X
variables to include in a regression model. At each step of the process each
X variable is evaluated using a set of criterion to see if it should remain in
the model. The selection could be based on either.

Page | 84
  Choosing a variable that satisfies a stipulated criterion or
 Removing a variable that least satisfies the criterion.
 The criterion selection is based on the time value of the data set.

Fig 13: Scatter plot of Stepwise Regression

Now let us discuss in detail about the stepwise regression using an

example
X1 – No. of transaction by NEFT
X2 – No. of Gift cards given to customer
X3 – No. of items returned to customer
X4 – No. of unbilled transactions
X5 – No. of cheques processed

Page | 85
 X6 – No. of items under defect
X7 – No. of items with discount
Y – No. of hours worked by an individual
S. No. Y X1 X2 X3 X4 X5 X6 X7
1 128.5 7781 100 886 235 644 56 737
2 113.6 7004 110 962 388 589 57 1029
3 146.6 7267 61 1342 398 1081 59 830
4 124.3 2129 102 1153 457 891 57 1468
5 100.4 4878 45 803 577 537 49 335
6 119.2 3999 144 1127 345 563 64 918
7 109.5 11777 123 627 326 402 60 335
8 128.5 5764 78 748 161 495 57 962
9 131.2 7392 172 876 219 823 62 665
10 112.2 8100 126 685 287 555 86 577
11 95.4 4736 115 436 235 456 38 214
12 124.6 4337 110 899 127 573 73 484
13 103.7 3079 96 570 180 428 59 456
14 103.6 7273 51 826 118 463 53 907
15 133.2 4091 116 1060 206 961 67 951
16 111. 4 3390 70 957 284 745 77 1446
17 97.7 6319 58 559 220 539 41 440
18 132.1 7447 83 1050 174 553 63 1133
19 135.9 7100 80 568 124 428 55 456
20 131.3 8035 115 709 174 498 78 968

The task is to find the regression between the dependent (X1, X2, X3,
X4, X5, X6, X7) and independent variable Y.
The significance level for a given hypothesis test is a value for which a
P-value less than or equal to is considered statistically significant.
Typical values are 0.1, 0.05, and 0.01. These values correspond to the
probability of observing such an extreme value by chance.
For this problem we consider the level of significance between 0.01>
P<0.1
When we say p value as 0.001, it means the term significance level
(alpha) is used to refer to a pre-chosen probability and the term "P value" is
used to indicate a probability that you calculate after a given study.
Conventionally 5% (less than 1 in 20 chance of being wrong), 1% and 0.1%
(P < 0.05, 0.01 and 0.001) levels have been used.

Page | 86
 1. A small p-value (typically ≤ 0.05) indicates strong evidence against
the null hypothesis, so you reject the null hypothesis.
2. A large p-value (> 0.05) indicates weak evidence against the null
hypothesis, so you fail to reject the null hypothesis.
The t-statistic is the coefficient estimate divided by the standard error.
The standard error is an estimate of the standard deviation of the coefficient,
the amount it varies across cases. It can be thought of as a measure of the
precision with which the regression coefficient is measured. The regression
for "large" sample (30 or more observations), a t-statistic greater than 2 (or
less than -2) indicates the coefficient is significant with >95% confidence.
T – Statistics (t) for Single Population is defined as

T – Statistics (t) for two Populations is defined as

A t-test is usually performed to find the significant difference between

population’s means (2-sample t) or between the populations mean and a
hypothesized value (1-sample t).
Step 1
Fit all 2 variable model S: B0 + B1 X?
Y, Regressed Only on: X1 X2 X3 X4 X5 X6 X7
b1 -0.001 0.128 0.030 0.013 0.049 0.556 0.019
t-statistic -0.054 2.165 3.679 0.602 5.131 4.072 3.559
P value 0.957 0.035 0.001 0.550 0.000 0.000 0.001
# X5 has the largest absolute t-value. It is therefore selected in the first run
# from this first run, the resulting model is defined as: X = B0 + B1X5
Step 2
Search for the best. 3-variable model: Y = B0 + B1X5 +B2 X?
Y, Regressed
X5+X1 X5 +X2 X5 + X3 X5 + X4 X5 +X6 X5 + X7
on:
b2 0.001 0.132 0.014 -0.032 0.358 0.007
t-statistic 1.422 2.817 1.680 -1.742 2.762 1.248
P-value 0.162 0.007 0.099 0.087 0.008 0.2179
# X2 has the largest absolute t-value. It is then added to the model. Thus, the
model so

Page | 87
 Far, the model is
Y = B0 +B1X5+ B2X2
Step 3
Searching for the best 4-variable model:
Y = B0 + B1X5 + B2X3 +B3X?
Y, Regressed on: X5+X2+X1 X5+X2+X3 X5+X2+X4 X5+X2+X6 X5+X2+X7
b3 0.0014 0.0083 -0.0347 0.2582 0.0048
t-statistic 1.5004 0.9744 -2.0238 1.9001 0.8854
P value 0.1401 0.3347 0.0486 0.0634 0.3804
# X4 has the largest absolute t-value. It is then added to the model. Thus, the model
is:
Y= B0+B1X5+B2X2+B3X4
Step 4
Searching for the best 5-variable model:
Y= B0+B1X5+B2X2+B3X4+B4X?
Y, Regressed on: X5+X2+X4+X1 X5+X2+X4+X3 X5+X2+X4+X6 X5+X2+X4+X7
b4 0.0015 0.0158 0.2523 0.0051
t-statistic 1.7579 1.8066 1.9156 0.9664
P-value 0.0853 0.0772 0.0615 0.3388

5. Classification Algorithms in Machine Learning

Classification can be done on either structured or unstructured data. It is a
method used to categorize data into number of classes. The main objective of
a classification problem is to assign a class or category for all observations in
a dataset. Classes are sometimes called as targets or labels. Classification
algorithms are widely used in various applications such as handwriting
recognition, speech recognition, biometric identification, face detection, fault
detection, medical diagnosis, spam detection, customer segmentation, etc.
The terms used in the classification techniques are as follows:
Feature – A measurable property of a data being observed.
Classifier – An algorithm that maps the input data to a specific class or
category. A classifier uses training data to understand the relationship
between input variables and the class. If the classifiers are trained properly, it
can be used well for predicting the class / label for test / unknown data.
Classification model – A model that predicts the class label or category

Page | 88
for the new data using the training dataset. The model is built to approximate
a mapping function (fn) from input variable (x) to discrete output variables
(y).
The classification process (fig. 14) involves the following stages:

Fig 14: Stages of Classification process

There are broadly three different types of classification. They are binary,
multiclass and multilabel classification. Binary or binomial Classification
(fig.15) is a task of classifying the elements of a given dataset into two groups
based on the classification rule. E.g.: Gender classification (male/female),
Email filtering (spam/not spam), test results (pass/fail), etc.
Multiclass or multinomial classification (fig. 15) is the process of
classifying elements into one of three or more classes. In this classification,
each element is assigned only one target label. E.g.: An animal can be a cat or
dog but not both at the same time, type of fruit, etc.

Fig 15: Binary and Multi-class Classification

Multilabel classification (fig. 16) is the classification task where each

element is mapped to a set of target labels. E.g.: An article can be about
sports, a person, and location at the same time.

Fig 16: Multi-Class and Multi-Label Classification

Page | 89
 Different classification algorithms that are used in machine learning are
Linear Regression, Logistic Regression, Naive Bayes classification, Support
Vector Machine, Stochastic Gradient Descent, Decision Tree, Random Forest,
Neural Networks and Nearest Neighbour. The choice of a classification
algorithm depends on the application and the type of dataset.
5.1 Support Vector Machine
It is a supervised learning algorithm that can be used for classification
and regression problems. The objective of SVM is to generate an optimal
hyperplane that best divides the data in to two classes. The SVM algorithm
works as follows:
 Plot the training data on a plane.
 Check whether the data is linear or non-linear.
 For linearly separable data, draw left and right separating hyperplane
close to the boundary data points. The boundary data points are
called support vectors. The separating hyperplanes are drawn in such
a way that the distance (margin) between them is maximum.
 Generate the decision boundary hyperplane in the mid of two
separating hyperplanes.
 For non-linear separable data, use kernel transform to convert into
higher dimensions. Then a linear decision boundary hyperplane can
be generated.
In general, SVM (fig. 17) provides precise prediction.

Fig 17: Support Vector Machine

The choice of kernel parameters decides the efficiency of the algorithm.

Hence more number of possible values are to tested, which directly increases
the computation time.

Page | 90
5.2 Decision Tree
It is a non-parametric supervised learning method used for both
classification and regression problems. The tree is built based on the decision
rules framed considering the data features. The rules are mostly of if-then-else
formats and the complexity of the rules can be inferred from the depth of the
tree. In general, the tree is constructed using divide and conquer approach.
The rules are learned sequentially using the training data one at a time.
This process is continued on the training set until a termination condition is
met. A tree consists of branching conditions where the value of a predictor is
compared to a trained weight. The number of branches and the values of
weights are determined in the training process. If there are too many
branches, a decision tree will become over fitted and results in anomalies
because of outliers. Pruning is done to simplify the model thereby enhancing
the performance of the model on test data. To understand the technique, let us
see an example of weather forecasting (fig. 18).

Fig 18: Decision tree of Weather Forecasting

Representation of a decision tree:
Node – Attribute or feature
Branch – Rule or decision
Leaf – outcome
The outcome of a decision tree may a categorical or continuous value.
5.3 Naive Bayes Classification
This is a probabilistic classification method assuming that the attributes
are conditionally independent. It classifies test data based on the highest
probability of its relation to a particular class. The classification model uses
Bayes theorem to derive the maximum posterior, which is the maximal of P
(x | C).

Page | 91
Where,
P (c|x) – posterior probability of class given predictor (attribute).
P (c) – prior probability of class
P (x|c) – probability of predictor given class.
P (x) – prior probability of predictor.
Naïve Bayes classification is very simple to implement, less
computational cost and easy to scale up for larger datasets. Let us take an
example of weather forecasting:
Observation Temp Humidity Windy Play
Rainy Hot High False No
Rainy Hot High True No
Overcast Hot High False Yes
Sunny Mild High False Yes
Sunny Cool Normal False Yes
Sunny Cool Normal True No
Overcast Cool Normal True Yes
Rainy Mild High False No
Rainy Cool Normal False Yes
Sunny Mild Normal False Yes
Rainy Mild Normal True Yes
Overcast Mild High True Yes
Overcast Hot Normal False Yes
Sunny Mild High True No

Calculate Posterior Probability:

 Construct Frequency table
 Transform it into Likelihood table
 Use Naive Bayes equation for each class

Page | 92
Case 1
P(x) = P (Sunny) = 5 / 14 = 0.36
P(x | c) = P (Sunny | Yes) = 3/9 = 0.33
P(c) = P (Yes) = 9 / 14 = 0.64
P (c| x) = P (Yes | Sunny) = 0.33 * 0.64 / 0.36 = 0.60
Case 2
P (x) = P (Sunny) = 5 / 14 = 0.36
P (x | c) = P (Sunny | No) = 2/5 = 0.4
P (c) = P (No) = 5 / 14 = 0.36
P (x| c) = P (No | Sunny) = 0.40 * 0.36 / 0.36 = 0.40
The class with the high posterior probability is the predicted outcome.
6. References
1. Stuart Russell, Peter Norvig. Artificial Intelligence-A Modern
Approach, 3rd Edition, Pearson Education/Prentice Hall of India, 2010.
2. Nils Nilsson J. Artificial Intelligence: A new Synthesis, Harcourt Asia
Pvt. Ltd, 2003.
3. Elaine Rich, Kevin Knight. Artificial Intelligence, 2nd Edition, Tata
McGraw-Hill, 2003.
4. Patrick Henry Winston, Artificial Intelligence, Pearson Education/PHI,
2004.
5. Ethem Alpaydin. Introduction to Machine Learning 3e (Adaptive
Computation and Machine Learning Series)‖, Third Edition, MIT Press,
2014.
6. Jason Bell. Machine learning-Hands on for Developers and Technical
Professionals‖, First Edition, Wiley, 2014.
7. Peter Flach. Machine Learning: The Art and Science of Algorithms that
Make Sense of Data‖, First Edition, Cambridge University Press, 2012.
8. Stephen Marsland. Machine Learning-An Algorithmic Perspective‖,
Second Edition, Chapman and Hall/CRC Machine Learning and Pattern
Recognition Series, 2014.
9. Tom M Mitchell. Machine Learning‖, First Edition, McGraw Hill
Education, 2013.

Page | 93
10. An Introduction to Statistical Learning with Applications in R by James
G, Witten D, Hastie T, Tibshirani R. Springer, 2013. ISBN 1461471370.
11. The Elements of Statistical Learning: Data Mining, Inference, and
Prediction, by Hastie T, Tibshirani R, Friedman J. 2nd Edition;
Springer, 2009. ISBN 0387848576.

Page | 94
 Chapter - 6
Experimental Investigation of Performance and
Emission of Lal Ambari Biodiesel and Its Blends
in a Compression Ignition Engine

Authors
Pankaj Shrivastava
Department of Mechanical Engineering, National Institute of
Technology, Imphal, Manipur, India
Tikendra Nath Verma
Department of Mechanical Engineering, National Institute of
Technology, Imphal, Manipur, India

Page | 95
Page | 96
 Chapter - 6
Experimental Investigation of Performance and Emission of
Lal Ambari Biodiesel and Its Blends in a Compression
Ignition Engine
Pankaj Shrivastava and Tikendra Nath Verma

Abstract
The increasing population growth and the quick depletion of the fossil
fuel reserves is now demanding an alternative renewable fuel that will
completely replace the diesel fuel in the upcoming future. In this research
work, first biodiesel is produced from Lal ambari oil by using
transesterification technique and different blends is prepared and their
effectiveness are evaluated experimentally under different engine load (low,
medium and full load) and with a constant 1500 rpm engine speed. The
result showed that reduction in exhaust gas temperature (EGT) and brake
thermal efficiency (BTE) by 1.08% and 1.25% for LA20 (20% Lal ambari +
80% diesel) than with diesel. The result indicate that reduction in NOx and
smoke emission by 7.78% and 4.23%, while improvement specific fuel
consumption (SFC) by 6.13% and carbon dioxide emission by 4.64% is
recorded for LA20 than with conventional fossil fuel. Thus, the experimental
results reveal that Lal ambari and their blends are one of the most suitable
alternative fuels to replace the fossil fuel.
Keywords: Diesel engine; lal ambari biodiesel; transesterification;
performance; emission
Nomenclature
BTE Brake thermal efficiency LA10 Blend with 10% Lal ambari biodiesel
bTDC Before top dead centre LA20 Blend with 20% Lal ambari biodiesel
B0 Pure diesel LA30 Blend with 30% Lal ambari biodiesel
BSN Bosch smoke number LA40 Blend with 40% Lal ambari biodiesel
CO Carbon monoxide LA100 Pure Lal ambari biodiesel
CR Compression ratio NOX Nitrogen oxide
CI Compression ignition PM Particulate matter

Page | 97
 CO2 Carbon dioxide rpm Rotation per minute
EGT Exhaust gas temperature SFC Specific fuel consumption
HC Hydro carbon

1. Introduction
The developing distress by growth of pollution causes use of
conventional fossil fuel products has prompted to find for alternative fuels
that are more environments friendly and a renewable energy source fuel.
Among different alternatives explored for fossil fuel, biodiesel fuel has been
accounted for one of the solid contenders for the reduction of exhaust
emission compared it with the diesel fuel. A few nations including India
have just started to search for the most suitable alternative renewable fuel
that will completely replace the conventional diesel fuel. It was also reported
that diesel engine parameters such as compression ratio, ignition timing and
engine load have plays an important role for the effect on the performance,
combustion and emission of the CI engine fuelled with neat biodiesel and its
blends with diesel fuel. The diesel engine is utilised in a more extensive
application like marine, transport vehicles, motors, power generating
equipment, agricultural machinery etc. as a result of lower fuel utilisation,
higher thermal efficiency and reduction of emission from the tailpipe than
with gasoline engine. The quick exhaust of unrefined petroleum holds and
increase the oil costs price, and that has attracted many researchers to work
on and search for environmentally friendly alternative fuel that will entirely
replace the diesel fuel in the future. In the on-going years, significance is to
utilized vegetable plant oils and its esters formed oil. The biodiesel fuel is
centred more on account of its ecological viewpoints and its potential as an
alternative elective fuel for compression ignition engines without any major
modification in engine setup.
Dasari et al. [1] investigate the engine characteristics by using castor
biodiesel and its different blends (5%, 10% and 15%) as a fuel in
compression ignition engine. It was found that among all combinations B10
indicates higher thermal efficiency at full load and also noticed that by
increase proportion of biodiesel in the blend decreases the HC, NO X and CO
emission than diesel. Senatore et al. [2] observed that advance heat release
take place by using rapeseed oil methyl ester than with diesel fuel. They
have also reported that fuel injection starts earlier and this leads to increase
the cylinder temperature by using biodiesel as an engine fuel. Puhan et al. [3]
investigate the engine performance by using Mahua oil ethyl ester (MOEE)
as a fuel in CI engine and recorded that an increase in specific fuel
consumption than diesel. Also, they have seen that slightly increase in

Page | 98
thermal efficiency while increase in CO2 emission and reduction in HC
emission by 63% and smoke by 70% are reported. Ramdas et al. [4] observed
that improvement in engine characteristics when engine is fuelled with
biodiesel and its blends. It was observed that thermal efficiency improved
and there is a reduction in brake specific energy consumption and smoke
emission was recorded. Schumacher et al. [5] investigate the engine
performance by using soy biodiesel and their blend as a substitute fuel in CI
engine. They have noticed that, fuelling engine with biodiesel and its blend
will slightly reduce the PM, THC and CO emission while on other side it
increases in NOx emission. Al-Widyan et al. [6] investigate by using waste
vegetable oil as feedstock oil for the production of biodiesel and analysed the
diesel engine characteristics. Masjuki et al. [7] investigated by using
preheated palm oil as a fuel in diesel engine. Puhan et al. [8] produced
biodiesel from Mahua oil by using transesterification process by using
methanol as catalyst in the presence of alkali and the biodiesel obtained was
studied for fuel properties.
Ghormade and Despande [9, 10] performed an experiment fuelled with
soybean biodiesel as a fuel in CI engine. They have concluded that slightly
change in engine thermal efficiency, but there is no changes were found for
specific fuel consumption by using different blends of biodiesel at part load
condition. Wang et al. [11] used vegetable oil and their blend as a fuel in
diesel engine. They have observed that exhaust gas temperature is somewhat
higher while there is small change in CO and NOx emission are recorded
lower than diesel. Kumar et al. [12] conducted a short term experiment with
Jatropha biodiesel and observed that HC, CO smoke emission and EGT are
somewhat higher compared with diesel fuel were recorded. Suh et al. [13]
concluded that slightly reduction in CO emission with an advanced injection
timing compared to diesel fuel. In this present research work, firstly the
biodiesel is derived from Lal ambari seeds oil by using a very common
transesterification process and then their properties were determined by
keeping the ASTM standards. Further performance and emission parameters
were analysed for Lal ambari biodiesel and their different blends (10%, 20%,
30%, 40% v/v), i.e. LA10, LA20, LA30, LA40 respectively. All experiment
test were performed at fixed injection timing (23.5° C bTDC), compression
ratio (17.5) and at different engine load low, medium and full load (25%,
50% and 100%) to check the technical feasibility by using Lal ambari
biodiesel fuelled in CI engine and compared it with diesel fuel.
2. Biodiesel Preparation
2.1 Transesterification Process
The biodiesel is derived from Lal ambari seeds that are readily available

Page | 99
in the region of Raipur, Chhattisgarh. After extracting oil from Lal ambari
seeds, a transesterification process is used to produce the biodiesel. For
creating the biodiesel from transesterification technique first triglycerides of
Lal ambari is formed, and then it reacts with the alcohol generally methanol
are used in the presence of active catalyst (sodium hydroxide and potassium
hydroxide) to create glycerol and fatty acid of the ester. Then in a separate
flask 500ml of Lal ambari oil, 220ml of pure methanol and 3-5 grams of
NaOH are taken and stirred well by using magnetic stirrer until the formation
of ester appeared in the sample. The mixture was then heated up to a
temperature between 65°-70°C and held it for 1 hour in the same phase.
Then the sample was cool under room temperature. Two layers are formed in
the sample, pure biodiesel is formed in the top side while the lower layer is
that is glycerol has appeared in, the lower side. The essential physical-
chemical properties of biodiesel and its blends are determined according to
ASTM standards. The important physical-chemical properties of Lal ambari
and its blends are shown in Table 1.
Table 1: The important physical-chemical properties of diesel, Lal ambari and its
blends

Fuel B0 LA100 LA10 LA20 LA30 LA40 Units

Density 830 - 840 878 834.8 840.1 844.9 849.7 (kg/m3)
Viscosity at 40 °C 2.5-3.11 5.66 2.96 3.22 341 3.41 (mm2/sec)
Lower heating value 42.5 38.72 42.10 41.7 41.35 40.95 (MJ/ kg)
Cetane number 48 52 48.42 48.85 49.24 49.65 ------

3. Experimental Setup
A single-cylinder, four-stroke compression ignition engine is used to
conduct the experiments with different tested fuels, and their effect is
analysed in terms of performance, emission and combustion parameters. The
engine test was carried out at different engine loads i.e. low (25%), medium
(50%) and full load (100%) with fixed injection timing of 23.5° b TDC and a
compression ratio 17.5 and compared it with mineral diesel fuel. The entire
experimental test was performed in the I.C Engine lab under the Mechanical
Engineering Department, National Institute of Technology Manipur, India.
The Schematic of the experimental setup used for investigating the engine
characteristics of different tested is presented in Fig. 1. and the detail
specification of engine test engine is shown in Table 2. Testo 350 gas
analyser is used for analysing the engine exhaust gas parameters such as
smoke, NOx and CO2. Initially, the engine started with the diesel fuel and
ran at least 20-25 minutes till it reached to steady state form and recorded the

Page | 100
baseline data and each experiment is conducted three times, and the average
of three readings were recorded for further analysis. In general, all
experimental tests have arisen some error and uncertainty from engine setup.
It may occur due to various reasons such as selection of sensor, calibration of
the engine setup, engine operating conditions, selection of test procedure and
observation reading is important factors that will create some errors in the
experimental reading. Trial estimation is subjected to some errors and
uncertainties. The summary of the apparatus utilised in the present research
and precision of the instruments is given in Table 3. Uncertainty analysis for
the experimental engine setup is necessary to reduce the error percentage as
well as verify the accuracy of the experiment. To perform the uncertainty
analysis following method is discussed by [14, 15, 16]. The overall
uncertainty analysis of the experimental was found out by using the
following equation:
Overall uncertainty (%) = square root of [(0.1)2 + (1.15)2 + (0.5)2 + (0.5)2
+ (0.2)2 + (1.0)2 + (0.1)2 + (1.0)2 + (1.0)2 + (1.0)2 + (0.5)2]
Total percentage of uncertainty = ±2.47%.

Fig 1: Experimental setup of test engine

Table 2: The detailed specification of test engine

Engine Parameters Specifications

Make Legion Brothers, India
Engine type 4- Stroke, direct injection, CI, water cooling.
Bore/Stroke 80mm/110mm
No of cylinders 1
Compression ratio 17.5
Dynamometer Eddy current dynamometer

Page | 101
Rated power@ 1500 rpm (k W) 3.5
Fuel injection pressure 500-800 bar
Dynamometer Eddy current dynamometer
Start of injection timing 23.5° before top dead centre
Connecting rod length 235 mm
Lubrication system Forced

Table 3: Uncertainty analysis of test engine

S. No Instrument Percentage Uncertainty

1 Temperature sensor ±0.1%
2 Speed sensor ±1.15%
3 Pressure sensor ±0.5%
4 Load indicator ±0.5%
5 Crank angle encoder ±0.2%
6 Smoke meter ±1.0%
7 Eddy current dynamometer ±0.1%
8 Fuel burette ±1.0%
9 Manometer ±1.0%
10 Test 350 gas analyser
CO2 ± 1.0%
NOX ± 0.5%

4. Result and Discussion

4.1 Performance Parameters
4.1.1 Brake Thermal Efficiency
Fig. 2 shows the comparison of BTE for various tested fuel with engine
load. It is seen from the figure that for all tested fuels thermal efficiency
increases with increase in engine load. It was also recorded that thermal
efficiency at full load is almost similar value. The BTE is lower for biodiesel
and its blends because of its higher viscosity, density and poor fuel
atomization of the biodiesel fuel in comparison to standard diesel fuel. The
thermal efficiency for LA20 (31.9%) is nearly close to standard diesel fuel
(32.1%). The BTE for LA10, LA20, LA30, and LA40 and LA100 was lower
by about 1.57%, 1.25%, 1.89%, 2.21% and 3.52%, than with standard diesel.
The BTE for Lal ambari blends is higher compared to neat biodiesel, and this
may because of higher viscosity of the neat biodiesel which leads to reduce
the vaporisation and atomization of the fuel and also impact on the
combustion behaviour.

Page | 102
 Fig 2: Comparison of brake thermal efficiency versus load

4.1.2 Specific Fuel Consumption

It is an important factor to analyse the how efficiently an engine fuel is
transformed in to positive work done. Fig.3 shows the comparison of
specific fuel consumption (SFC) for various tested fuels with engine load. It
is clear from the figure that specific fuel consumption decrease with
increasing in engine loading for all tested fuels. It is seen that the higher
biodiesel fuel indicated higher SFC compared to other testing fuels this may
be because of lower calorific value of the pure biodiesel. The specific fuel
consumption first increase in the part load condition and it reduces slightly at
full load conditions. The SFC for LA10, LA20, LA30, LA40 and LA100 was
found to higher by about 3.16%, 6.13%, 8.92%, 10.90% and 16.09% than
conventional diesel fuel respectively.

Fig 3: Comparison of specific fuel consumption versus load

Page | 103
4.1.3 Exhaust Gas Temperature
It is an important parameter to better understand the quality of
combustion process. Fig.4 shows the comparison of exhaust gas temperature
(EGT) for various tested fuel with engine load. It is observed from the figure
that EGT increase with increasing in engine load for all fuels [18]. It was
recorded that diesel fuel indicates higher EGT compared to other tested fuels
at 100% load condition. It is depends upon the presence of more oxygen
content in the fuel and the duration of combustion time reduces with higher
cetane value of the fuel and it burnt till the combustion phase and due to this
more heat is released in the phase. It also noticed that EGT reduces with
increasing percentage of biodiesel in blend. At 100% load condition the EGT
was found to be about 322.15°C, 321.03°C, 319.06°C, 317.13°C and
312.3°C for LA10, LA20, LA30, LA40 and LA100 compared to that of
324.51°C for mineral biodiesel respectively.

Fig 4: Comparison of exhaust gas temperature versus load

4.2 Emission Characteristics

4.2.1 NOx Emission
Fig. 5 shows the comparison of NOx emission for various tested fuels
with engine load load. It is clear from the graph that NOx emission increases
with increase in engine load for all tested fuels. The increase trend is because
of higher exhaust gas temperature at higher engine load condition [17, 19-20]. It
was found that at higher engine load NOx emission decreases for biodiesel

Page | 104
and its blends compared to diesel. The two important factors noticed from
the graph. First, the NOx emission is dependent on the temperature
prevailing inside the combustion chamber as the load of engine increases the
temperature inside the combustion chamber also increases and second factor
is by the use of biodiesel and its blends that leads to generate low
combustion temperature at 100% load. The NOx emission was lower than
diesel by about 6.69%, 7.78%, 8.53%, 9.19% and 13.0% respectively at full
load conditions.

Fig 5: Comparison of NOx emission versus load

4.2.2 Smoke Emission

The comparison of smoke emission for various tested fuels with load is
shown in Fig.6. It was seen from the figure that smoke emission increases
with increase in engine load for all tested fuels [17, 18-19]. The biodiesel fuel
which is higher oxygen content indicates reduction in smoke emission for all
operating loading conditions. The smoke emission produced because of
presence of unburns carbon particles present in the cylinder wall because of
lower temperature and fuel rich zone and it is produces because of the
improper fuel- air mixing in the rich zone. At full load condition the smoke
emission is reduced for all biodiesel and its blends and it is recorded for
LA10, LA20, LA30, LA40 and LA100 was found to be about 2.4%, 4.34%,
6.32%, 8.38% and 16.6 BSN as compared to diesel fuel by 1.68 BSN.

Page | 105
 Fig 6: Comparison of smoke emission versus load

4.2.3 CO2 Emission

The comparison of carbon dioxide emission for various tested fuel with
load is shown in Fig.7. The CO2 emission indicates the quality of
combustion and also indicates the complete burning of the fuel. It is
observed from the figure that CO2 emission for all tested fuel gradually
decreases with increasing in engine load and this leads to more fuel is
injected inside the combustion chamber. At full load condition it was seen
that LA10 blends shows lower CO2 emission compared to other biodiesel
blends. At full load condition the reduction of carbon dioxide emission for
B0, LA10, LA20, LA30, LA40 and LA100 was found to be about 682,
706.5, 715.2, 726.2, 738.9 and 775.9g/kWh respectively.

Fig 7: Comparison of CO2 emission versus load

Page | 106
5. Conclusions
In the present investigation biodiesel derived from Lal ambari oil and
check the technical feasibility of Lal ambari biodiesel and its blends as a
substitute fuel in diesel engine with different loading condition are
examined. The engine characteristics were analysed while keeping fixed
injection timing at 23.5°C b TDC, with running a constant engine speed of
1500 rpm at CR 17.5 and result were compared it with the conventional
diesel fuel. The following specific conclusions were drawn based on the
experimental study.
 Brake thermal efficiency found nearly close in low, medium and
full load for all different tested fuels and better for LA20 compared
to other biodiesel and its blends The BTE was obtained to be lower
by about 1.8% for LA20 compared to conventional diesel but it is
lower for biodiesel and its blends.
 The result shows that specific fuel consumption increased with
increase percentage of biodiesel in diesel-biodiesel blend. SFC is
higher by 16.09% for LA100 than with diesel fuel.
 At full load condition exhaust gas temperature is slightly lower for
biodiesel and its blends compared to diesel fuel.
 Smoke emission reduces by 16.7% for LA100 than with diesel and
it increases with increase in engine load.
 The CO2 emission is found to be higher for biodiesel and its blends
than with diesel.
 NOX emission is lower by 6.69%, 7.78%, 8.53%, 9.19% and
13.01% for LA10, LA20, LA30, LA40 and LA100 compared to
standard diesel fuel. It was noticed that NOX emission increase with
increase in engine load.
In view of the depletion or shortage of fossil fuel, Lal ambari biodiesel
and its blends can be an alternative fuels that will replace the diesel fuel in
the compression ignition engine.
Reference
1. Dasari SR, Chaudhary AJ, Goud VV, Sahoo N, Kulkarni VN. In-situ
alkaline transesterification of castor seeds: Optimization and engine
performance, combustion and emission characteristics of blends. Energy
conversion and management. Jun 15. 2017; 142:200-14.
2. Senatore A, Cardone M, Rocco V, Prati MV. A comparative analysis of

Page | 107
 combustion process in DI diesel engine fueled with biodiesel and diesel
fuel. SAE Technical Paper; Mar 6, 2000.
3. Puhan S, Vedaraman N, Sankaranarayanan G, Ram BV. Performance
and emission study of Mahua oil (Madhuca indica oil) ethyl ester in a 4-
stroke natural aspirated direct injection diesel engine. Renewable
energy. Jul 1. 2005; 30(8):1269-78.
4. Ramadhas AS, Jayaraj S, Muraleedharan C. Use of vegetable oils as IC
engine fuels-a review. Renewable energy. Apr 1. 2004; 29(5):727-42.
5. Schumacher LG, Borgelt SC, Fosseen D, Goetz W, Hires WG. Heavy-
duty engine exhaust emission tests using methyl ester soybean oil/diesel
fuel blends. Bioresource Technology. Jul 1. 1996; 57(1):31-6.
6. Al-Widyan MI, Tashtoush G. Utilization of ethyl ester of waste
vegetable oils as fuel in diesel engines. Fuel Processing Technology.
May 20. 2002; 76(2):91-103.
7. Masjuki H, Abdulmuin MZ, Sii HS. Investigations on preheated palm
oil methyl esters in the diesel engine. Proceedings of the Institution of
Mechanical Engineers, Part A: Journal of Power and Energy. Apr. 1996;
210(2):131-8.
8. Puhan S, Vedaraman N, Ram BV, Sankarnarayanan G, Jeychandran K.
Mahua oil (Madhuca Indica seed oil) methyl ester as biodiesel-
preparation and emission characterstics. Biomass and bioenergy. Jan 1.
2005; 28(1):87-93.
9. Ghormade TK, Deshpande NV. Soybean oil as an alternative fuels for
IC engines. Proceedings of recent trends in automotive fuels, Nagpur,
India, 2002.
10. Agarwal AK, Rajamanoharan K. Experimental investigations of
performance and emissions of Karanja oil and its blends in a single
cylinder agricultural diesel engine. Applied Energy. Jan 1. 2009;
86(1):106-12.
11. Wang YD, Al-Shemmeri T, Eames P, McMullan J, Hewitt N, Huang Y
et al. An experimental investigation of the performance and gaseous
exhaust emissions of a diesel engine using blends of a vegetable oil.
Applied Thermal Engineering. Oct 1. 2006; 26(14-15):1684-91.
12. Kumar MS, Ramesh A, Nagalingam B. Experimental Investigations on
a Jatropha oil methanol dual fuel engine. SAE Technical Paper; Mar 5,
2001.

Page | 108
13. Suh HK, Roh HG, Lee CS. Spray and Combustion Characteristics of
Biodiesel/Diesel Blended Fuel in a Direct Injection Common-Rail
Diesel Engine. Journal of Engineering for Gas Turbines and Power. May
1. 2008; 130(3):032-807.
14. Singh TS, Verma TN, Nashine P, Shijagurumayum C. BS-III Diesel
Vehicles in Imphal, India: An Emission Perspective. In Air Pollution
and Control Springer, Singapore, 2018, 73-86.
15. Rajak U, Verma TN. Spirulina microalgae biodiesel-A novel renewable
alternative energy source for compression ignition engine. Journal of
Cleaner Production. Nov 10. 2018; 201:343-57.
16. Rajak U, Nashine P, Singh TS, Verma TN. Numerical investigation of
performance, combustion and emission characteristics of various
biofuels. Energy Conversion and Management. 2018; 15;156:235-52.
17. Rajak U, Verma TN. Effect of emission from ethylic biodiesel of edible
and non-edible vegetable oil, animal fats, waste oil and alcohol in CI
engine, Energy Conversion and Management. 2018; 166:704-718.
https://doi.org/10.1016/j.enconman.2018.04.070
18. Rajak U, Nashine P, Verma TN. Assessment of diesel engine
performance using spirulina microalgae biodiesel. Energy. 2019;
166:1025-1036. https://doi.org/10.1016/j.energy.2018.10.098
19. Rajak U, Verma TN. Comparative assessment of the emission
characteristics of first, second and third generation biodiesel as fuel in a
diesel engine. Journal of Thermal Engineering, (Accepted Manuscript),
2018.
20. Singh TS, Verma TN. Impact of tri- fuel on compression ignition engine
emission: Blends of waste frying oil- alcohol-diesel. Springer
Singapore, 2019, 135-156. https://doi.org/10.1007/978-981-13-3287-
6_7.

Page | 109