1. MACHINE LAYOUT DESIGN & OPTIMIZATION
A PROJECT REPORT
Submitted by
NARESH KUMAR.K (070111303029)
NIVAS.S (070111303033)
SENTHIL NATHAN.R (070111303050)
In partial fulfillment for the award of the degree
of
BACHELOR OF ENGINEERING
IN
MECHANICAL ENGINEERING
INSTITUTE OF ROAD AND TRANSPORT TECHNOLOGY
ERODE-638316
ANNA UNIVERSITY OF TECHNOLOGY,
COIMBATORE 641047
APRIL 2011
i
2. TABLE OF CONTENTS
CHAPTER NO TITLE PAGE NO
ABSTRACT vii
LIST OF TABLES viii
LIST OF FIGURES x
LIST OF ABBREVIATIONS xi
1 MACHINE LAYOUT AND OPITIMIZATION
1.1 INTRODUCTION 2
1.2 INTRODUCTION TO MACHINE LAYOUT 3
1.3 BASIC LAYOUT TYPES
1.3.1 PROCESS LAYOUT 3
1.3.2 CELL LAYOUT 4
1.3.3 PRODUCT LAYOUT 4
1.4 CYCLE TIME 5
1.5 HIERARCHY OF MACHINE LAYOUT DATA 5
2 INTRODUCTION
2.1 COMPANY PROFILE 8
2.2 CNC MACHINE SHOP 9
2.3 STEERING KNUCKLE 10
2.4 OPERATIONS PERFORMED 11
2.5 TIME STUDY FOR ALL COMPONENTS 13
3 INTRODUCTION TO GENETIC ALGORITHM
3.1 DEFINITION OF GENETIC ALGORITHM 15
3.2 BASIC GENETIC ALGORITHM 16
3.3 OTHER SEARCH TECHNIQUES
ii
3. 3.3.1 HILL CLIMBING 17
3.3.2 ENUMERATIVE 17
3.3.3 RANDOM SEARCH ALGORITHM 18
3.3.4 RANDOMIZED SEARCH TECHNIQUES 18
3.4 THE DIFFERENCE BETWEEN GENETIC ALGORITHM
AND TRADITIONAL METHODS 18
3.5 BASIC GENETIC ALGORITHM OPERATIONS
3.5.1 REPRODUCTION 19
3.5.2 CROSS OVER 20
3.5.3 MUTATION 22
3.6 POWER OF GENETIC ALGORITHM 23
4 LAYOUT MODELLING USING EXCEL
4.1 PROBLEM STATEMENT 25
4.2 APPLICATION OF GENETIC ALGORITHM 25
4.3 ASSUMPTIONS 26
4.4 COMPONENT DETAILS 26
4.5 MATHEMATICAL MODEL 29
4.6 EXISTING LAYOUT WITH STEERING KNUCKLE FLOW 30
4.7 CENTROID CALCULATION 31
4.8 PART ROUTING MATRIX 32
4.8.1 COST MATRIX 32
4.8.2 DISTANCE MATRIX 33
4.8.3 FLOW MATRIX 35
4.9 MATERIAL HANDLING COST FOR EXISTING LAYOUT 35
5 LAYOUT OPTIMIZATION
5.1 GA PARAMETERS 39
5.2 OPTIMIZER OUTPUT 39
5.2.1 PROPOSED LAYOUT #1 (NEW) 40
iii
5. ABSTRACT
The basic layout problem is the arrangement of the departments according to flow of
materials between them. The design criterion routinely used in most of the layout deign
procedures - a measure of long-term material handling efficiency, fails to capture the impact of
layout configuration on operational performance measures such as cycle time, queue times at
processing departments, throughput rates etc.
As a result, layout performance tends to deteriorate significantly with fluctuations in
product volumes, mix, or routings. In this project, an approach that combines meta-heuristic
algorithm with simulation to optimize the layout for manufacturing effectiveness and evaluate
the same based on operational performance measures is proposed.
Application of meta-heuristic algorithms like Simulated Annealing, Genetic Algorithm and
Hybrid algorithms are helps us in reaching near optimal or optimal solutions for the medium,
large size facility layout problems without much computation difficulties. Due to the advances in
computer technology, simulation become more dominant tool for analyzing manufacturing
systems based on quantitative and qualitative criteria. In view of this a combined approach of
Meta-heuristic algorithms and system simulation to solve facility layout problems is proposed
here.
v
6. LIST OF TABLES
TABLE
TABLE NAME PAGE NO
NO
SEQUENCE OF OPERATION AND NO OF MACHINES USED
2.1 11
FOR COMPONENT #1
SEQUENCE OF OPERATION AND NO OF MACHINES USED
2.2 12
FOR COMPONENT #2
SEQUENCE OF OPERATION AND NO OF MACHINES USED
2.3 12
FOR COMPONENT #3
SEQUENCE OF OPERATION AND NO OF MACHINES USED
2.4 13
FOR COMPONENT #4
MACHINES INVOLVED IN PRODUCTION OF STEERING
4.1 25
KNUCKLE
4.2 SEQUENCE OF MACHINES USED FOR 4 COMPONENTS 26
MACHINING TIME AND SEQUENCE OF OPERATION FOR
4.3 27
COMPONENT #1
MACHINING TIME AND SEQUENCE OF OPERATION FOR
4.4 27
COMPONENT #2
MACHINING TIME AND SEQUENCE OF OPERATION FOR
4.5 28
COMPONENT #3
MACHINING TIME AND SEQUENCE OF OPERATION FOR
4.6 28
COMPONENT #4
CALCULATION OF AREA REQUIRED FOR EACH
4.7 31
MACHINES
4.8 BATCH SIZE OF 4 COMPONENTS 32
4.9 COST MATRIX OF EXISTING LAYOUT 33
4.10 DISTANCE MATRIX OF EXISTING LAYOUT 34
4.11 FLOW MATRIX OF EXISTING LAYOUT 35
vi
7. MATERIAL HANDLING COST MATRIX OF EXISTING
4.12 36
LAYOUT
5.1 MATERIAL HANDLING COST OF ALL SOLUTION LAYOUTS 39
COMPARISION OF ALTERNATIVE LAYOUT
5.2 47
CONFIGURATIONS
vii
8. LIST OF FIGURES
FIGURE
FIGURE NAME PAGE NO
NO
1.1 PROCESS LAYOUT 4
1.2 HIERARCHY OF MACHINE LAYOUT DATA 5
DISTANCE TRAVELLED BY PARTS REDUCED BY CHANGING
1.3 6
MACHINE LAYOUT
2.1 STEERING KNUCKLE 10
2.2 STEERING KNUCKLE 10
3.1 BASIC GENETIC ALGORITHM - FLOW CHART 16
4.1 EXISTING LAYOUT WITH STEERING KNUCKLE FLOW 30
PROPOSED LAYOUT #1 (NEW) WITH
5.1 40
STEERING KNUCKLE FLOW
PROPOSED LAYOUT #2 (NEW) WITH
5.2 41
STEERING KNUCKLE FLOW
PROPOSED LAYOUT #3 (NEW) WITH
5.3 42
STEERING KNUCKLE FLOW
PROPOSED LAYOUT #1 (MODIFIED) WITH
5.4 43
STEERING KNUCKLE FLOW
PROPOSED LAYOUT #2 (MODIFIED) WITH
5.5 44
STEERING KNUCKLE FLOW
GRAPH (NO OF TRIALS Vs VALUES) OF MODIFIED LAYOUT
5.6 45
#1 AND MODIFIED LAYOUT #2
viii
9. LIST OF ABBREVIATIONS
GA ----- GENETIC ALGORITHM
SACL ----- SAKTHI AUTO COMPONENTS LIMITED
CNC ----- COMPUTER NUMERICAL CONTROL
SLA ----- SHORT LONG ARM
ix
11. 1.1 INTRODUCTION
The traditional facility layout problem in a manufacturing setting is defined as the
determination of relative locations for, and allocation of, the available space among a given
number of workstations. Although most facility layout solutions have, in the past, focused
on minimizing the amount of transportation, the effect of a given layout design on the
production function of a manufacturing system is much more than just the cost of material
handling. While material handling cost remains critical, shorter cycle times have become
much more important in today’s manufacturing systems.
Rapid developments in new products, coupled with short delivery times demanded
by customers, are the bases of the time-based competitive strategies rapidly being adopted
by inventory and short manufacturing cycle times are practical considerations that have
strong impacts on the layout design and should be incorporated into the layout design
process as genuine concerns. But, the difficulty in linking the layout configurations and
operational performance measures via mathematical or analytical models has been recorded
in the literature by various researchers and practitioners for the past few years.
However, we require new design models and solution procedures that account for
uncertainty and variability in design parameters such as product mix, production volumes,
and product life cycles, for complex manufacturing system analysis and rational decision
making while handling
12. 1.2 INTRODUCTION TO MACHINE LAYOUT
One of the most important factors to consider in designing the manufacturing
facilities is finding an effective layout.
Laying out a factory involves deciding where to put all the facilities, machines,
equipment and staff in the manufacturing operation.
Layout determines the way in which materials and other inputs (like people and
information) flow through the operation. Relatively small changes in the position of a
machine in a factory can affect the flow of materials considerably. This in turn can affect the
costs and effectiveness of the overall manufacturing operation. Getting it wrong can lead to
inefficiency, inflexibility, large volumes of inventory and work in progress, high costs and
unhappy customers. Changing a layout can be expensive and difficult, so it is best to get it
right first time.
1.3 BASIC LAYOUT TYPES
Once the type of operation has been selected (jobbing, batch or continuous) the basic
layout needs to be selected. There are three basic types:
Process layout
Cell layout
Product layout
1.3.1 PROCESS LAYOUT
In process layout, similar manufacturing processes (cutting, drilling, wiring, etc.)
are located together to improve utilisation. Different products may require different
processes so material flow patterns can be complex.
13. 1.3.2 CELL LAYOUT
In cell layout, the materials and information entering the operation are pre-selected to
move to one part of the operation (or cell) in which all the machines to process these
resources are located. After being processed in the cell, the part-finished products may go
on to another cell. In effect the cell layout brings some order to the complexity of flow that
characterises process layout.
1.3.3 PRODUCT LAYOUT
Product layout involves locating the machines and equipment so that each product
follows a pre-arranged route through a series of processes. The products flow along a line of
processes, which is clear, predictable and relatively easy to control.
To design a process layout, the designer needs to know:
The area required by each work centre.
The constraints on the shape of the area allocated for each work centre.
The degree and direction of flow between each work centre (for example
number of journeys, number of loads, cost of flow per distance travelled).
The desirability of work centres being close together.
14. 1.4 CYCLE TIME
The cycle time of a product layout is the time between completed products emerging from
the operation. Cycle time is a vital factor in the design of product layouts and influences most
other detailed design decisions. It is calculated by considering the likely demand for the products
over a period and the amount of production time available in that period.
1.5 HIERARCHY OF MACHINE LAYOUT DATA:
15. Machine Machine Machine Machine
9 11 14 10
15
15 7 3
5
Machine Machine 3 Machine 3 Machine
12 5 2 7
15 1
7
Machine 1 Machine Machine Machine
8 6 3 1
23
23
7
Machine Machine
4 13
Part 1: 4-5-7 Part 15: 6-9-8-14
Part 3: 2-10-3-11 Part 23: 4-5-13
Part 5: 8-9
Part 7: 1-12-7-13
Movements of parts at Generation 1 : Distance Travelled = 234
CELL 1 CELL 2
Machine Machine Machine 1 Machine
14 9 4 23 5
15
15 5 23
15
1
Machine Machine Machine Machine
7
6 8 7 13
7
CELL 3
Machine 3 Machine Machine Machine
10 2 12 7 1
3
Machine 3 Machine
11 3
Part 1: 4-5-7 Part 15: 6-9-8-14
Part 3: 2-10-3-11 Part 23: 4-5-13
Part 5: 8-9
Part 7: 1-12-7-13
Movement of parts at Generation 100: Distance Travelled by parts: 109
17. 2.1 COMPANY PROFILE
SAKTHI AUTO COMPONENTS LIMITED (SACL)
Sakthi Auto Component Limited is one among the MULTI FACETED Sakthi Group
situated at Mukasi Pallagoundenpalayam, Erode District, Tamilnadu State, India, established
in the year 1983. Presently the Sakthi Auto has a capacity to produce 24000 Tonnes /
annum of S.G.IRON Castings, on a 100 Acre Land with all amenities for Workmen and
Officers like Housing, Transport etc. Sakthi Auto is one of the major producers of S.G.Iron
Castings, meeting the needs of most of the Automotive and other general Engineering
Industries
Sakthi Auto Component Limited is a major supplier of critical components to
passenger car manufacturers. The components are Steering knuckles, Brake drums, Brake
discs, Hubs , Brake calipers, Carriers, Differential cases and Manifolds etc. Presently the
supplies of these components are made to Maruti Udyog Ltd., Hyundai, Ind Auto Ltd., Ford,
Honda Siel Cars and Tractors and farm Equipment Ltd. etc,. Castings meant for trucks and
refineries are exported to USA. The quantum of exports per month ranges between 250 MT
to 500 MT. It is likely to go up to 1000 MT in near future
Supplying most CRITICAL COMPONENTS like STEERING KUNCKLE, BRAKE
DRUMS and MANIFOLDS for all Suzuki Vehicles Manufactured in India by M/s. Maruti
Udyog Limited at New Delhi & to many leading passenger car manufacturers in fully
machined condition.
R&D Lab is attached to our Sakthi Auto with modern computerised equipments like
Direct Reading Spectrometer, Carbon Sulphur determination, Universal Testing Machine,
Scanning Electron Microscope, Industrial X-RAY Scanner etc.
Sakthi Auto is equipped with DISAMATIC FOUNDRY with the state of the art
manufacturing technology which is regarded as the best anywhere in the World. And
equipped with many sophisticated special purpose and CNC machines to produce precision
oriented component for passenger car and automobile industries.
18. The sakthi group of companies performs, contributes and touches the lives of
many with operation in the fields of sugar, alcohol, tea, soft drinks, soya foods, synthetics,
gems, textiles, transport, retreading, finance and foundry. The company has strategically
invested in the most modern foundry facility and looks forward to set the pace for the
industry in the years to follow...
Auto and engineering component slack adjusters, wing nuts and unions, steering
knuckles in machined condition auto and engineering component slack adjusters, wing nuts
and unions, steering knuckles in machined condition automotive parts, component.
Technical know how from Georg Fischer is expected that the unit will double its
output by Foundry Systems. This has helped meeting June 2004 and further look at some
expansion in increasing demands from indigenous and 2005. This is due to the fact that the
Indian auto overseas original equipment manufacturers, market is growing at more than 20%
and the especially in automobile sector.
Other facilities global players like Delphi, Visteon, Rover and include engineering
workshop, testing Haldex have approached the company for laboratories, spectrometer, X-
ray scanner, etc. further components.
2.2 CNC MACHINE SHOP:
SACL is the sole vendor for many critical components like steering knuckles, brake
CNC DIVISION drums, brake discs, exhaust manifolds and case The CNC machine division
of SACL has imported differentials for leading manufacturers in India equipments for
machining rough castings to like Maruti, Suzuki, Huyndai, FIAT and Delphi. exacting
standards of dimensional specifications. www.sakthiauto.com SACL has also received a
purchase order for 2.5 million dollars per annum from Delphi and has begun shipping the
components to the US.
SACL is one of the first units in the Asia Pacific zone to export castings to the
Delphi north American markets. Delphi is in the process of negotiating a new purchase order
for about 18 million dollars per annum. It is expected that this order will be received by the
first quarter of 2004. At present the domestic and export enquiries at the plant are for about
150% of the capacity.
19. 2.3 STEERING KNUCKLE
A forging that usually includes the spindle and steering arm, and allows the front
wheel to pivot. The knuckle is mounted between the upper and lower ball joints on a SLA
suspension, and between the strut and lower ball joint on a MacPherson strut suspension.
20. There are four different type of steering knuckle components are manufacturing in
CNC machine shop, SACL. There are given below:
J200 Knuckle
MCI Knuckle
GIO Knuckle
MXI Knuckle
2.4 OPERATIONS PERFORMED:
The operations performed for these components are given by,
COMPONENT 1:
J200 KNUCKLE:
The sequence of operations and no of machines used of component 1 are given
below,
NO OF
Operation MACHINE
OPERATION NAME MACHINES
NO NO
USED
1 Turning 2 1&2
SBA Milling Drilling, Caliber arm Milling,
2 1 3
Drilling & Coverhole tapping
3 Kingpin arm milling Drilling 1 4
Milling, slitting, drilling, tie rod arm milling,
4 1 5
drilling
5 ABS milling, Drilling, Tapping 1 6
21. COMPONENT 2:
MCI KNUCKLE:
The sequence of operations and no of machines used of component 2 are given
below,
NO OF
Operation MACHINE
OPERATION NAME MACHINES
NO NO
USED
1 Turning 1 7
2 Caliber arm Milling, Drilling 1 8
SBA milling drilling,Tie rod arm milling
3 drilling,Kingpin arm milling Drilling, 1 9
Tapping
COMPONENT 3:
GIO KNUCKLE:
The sequence of operations and no of machines used of component 3 are given
below,
NO OF
Operation MACHINE
OPERATION NAME MACHINES
NO NO
USED
1 Turning 1 10
SBA Milling Drilling, Mounting hole drilling,
2 1 11
Tapping
Tie rod arm milling drilling, Taper reaming,
3 1 12
Kingpin arm milling Drilling
4 Kingpin arm drilling slitting milling 1 13
22. COMPONENT 4:
MXI KNUCKLE:
The sequence of operations and no of machines used of component 4 are given
below,
NO OF
Operation MACHINE
OPERATION NAME MACHINES
NO NO
USED
1 Turning 2 14 & 15
2 Caliber arm Milling,Drilling 1 16
3 SPI milling 1 17
4 Tie rod arm milling 1 18
Kingpin arm machining, Tie rod
5 1 19
arm milling
SBA drilling, Kingpin arm drilling
6 1 20
& slitting
2.5 TIME STUDY FOR ALL COMPOENTS:
The time required to produce a steering knuckle can be obtained by the following
table:
Component Component Component Component
#1 #2 #3 #4
Machining
19 min 37 sec 11 min 49 sec 12 min 33 sec 12 min 3 sec
Time
Loading Time 2 min 25 sec 1 min 15 sec 2 min 33 sec 1 min 55 sec
Unloading
1 min 50 sec 55 sec 2 min 18 sec 1 min 46 sec
Time
TOTAL
23 min 52 sec 13 min 59 sec 17 min 24 sec 15 min 44 sec
TIME
24. 3.1 DEFINITION OF GENETIC ALGORITHM:
“Genetic algorithms are search algorithms based on the mechanics of natural
selection and natural genetics”
Bauer gives a similar definition as follows:
“Genetic algorithms are software , procedures modelled after genetics and
evolution”
GA exploits the idea of the survival of the fittest and an interbreeding population to
create a novel and innovative search strategy.A population of strings, representing solutions
to a specified problem , is maintained by the GA. The GA then iteratively creates the new
populations from the previous population by ranking and interbreeding the fittest to create
new strings, which are closer to the optimum solution to the problem.
GA is a form of randomized search,in that way in which strings are chosen and
combined is a stoichastic process. This is a radially different approach to the problem
solving methods, which are tends to be more deterministic in nature.
The idea of survival of the fittest is of great importance to genetic algorithms. GAs
use what is termed as a fitness function in order to select the fittest string that will be used to
create new and better populations of strings. The fitness function takes a string and assigns a
relative value to the string. The method and the nature of the fitness value does not matter.
The fitness function must do is to rank the strings by producing the fitness value. These
values are then use to select the fittest strings.
25. 3.2 BASIC GENETIC ALGORITHM
The following flowchart shows the iterative cycle of a basic genetic algorithm.
Firstly, an initial population of strings is created. The process then iteratively selects
individuals from the population that undergo some form of transformation (via the
recombination step) to create new population. The new population is then tested to see if it
fulfills some stopping criteria. If it does, then the process halts, otherwise iteration is again
performed.
26. 3.3 OTHER SEARCH TECHNIQUES:
We will look at some of the other, more traditional, optimization techniques, and
show both their strengths and shortcomings when compared with GAs.
3.3.1 Hill climbing:
Hill climbing optimization techniques have their roots in the classical mathematics
developed in the 18th and 19th centuries. In essence, this class of search methods finds an
optimum by following the local gradient of the function (they are sometimes known as
gradient methods). They are deterministic in their searches. They generate successive results
besed solely on the previous results.
There are several drawbacks to hill climbing methods. Firstly, they assume that the
problem space being searched is continuous in nature. In other words, derivative of the
function representing the problem space exists. This is not true of many real world problems
where the problem space is noisy and discontinuous.
Another major disadvantage of using hill climbing is that hill climbing algorithm
only find the local optimum in the neighbourhood of the current point. They have no way of
looking at the global picture in general. However, parallel methods of hill climbing can be
used to search multiple points in the problem space. This still suffers from the problem that
there is no guarantee of finding the optimum value, especially in very noisy spaces with a
multitude of local peaks or troughs.
3.3.2 Enumerative:
The basis of Enumerative techniques is simplicity itself. To find optimum value in a
problem space (which is finite), look at the function values at every point in the space. The
problem here is obvious. This is horribly inefficient. For very large problem spaces, the
computational task is massive, perhaps intractably so.
27. 3.3.3 Random search algorithms:
Random searches simply perform random walks of the problem space, recording the
best optimum values discovered so far. Efficiency is a problem here as well. For large
problem spaces, they should perform no better than enumerative searches. They do not use
any knowledge gained from previous results and thus are both dumb and blind.
3.3.4 Randomized search techniques:
Randomized search algorithms use random choice to guide themselves through the
problem search space. But these are not just simply random walks. These techniques are not
directionless like the random search algorithms. They use the knowledge gained from
previous results in the search and combine them with some randomizing features. The result
is a powerful search technique that can handle noisy, multi model search spaces with some
relative efficiency. The two most popular forms of randomized search algorithms are
simulated annealing and genetic algorithms.
3.4 THE DIFFERENCE BETWEEN GENETIC ALGORITHM AND
TRADITIONAL METHODS:
The following list is a very quick look at the essential differences between GAs and
other forms of optimization.
Genetic algorithms a coded form of the function values (parameter set),
rather than with the actual values themselves, So, for example, if we want to
find the minimum of the function f(x)=X3+X2+5, the GA would not deal
directly with X or Y values, but with strings that encode these values. For
this case, strings representing the binary X values should be used.
Genetic algorithms use a set, or population, of points to conduct a search, not
just a single point on the problem space. This gives GAs the power to search
noisy spaces littered with local optimum points. Instead of relying on a single
point to search through the space, the GAs looks at many different areas of
the problem space at once, and uses all of this information to guide it.
28. GAs are probabilistic in nature, not deterministic. This is a direct result of the
randomization techniques used by GAs.
GAs are inherently parallel. Here lies one of the most powerful features of
genetic algorithms. GAs, by their nature, is very parallel, dealing with a large
number of points (strings) simultaneously. Holland has estimated that a GA
processing n strings at each generation, the GA in reality processes n3 useful
substings.
GA use only payoff information to guide themselves through the problem
space. Many search techniques need a variety of information to guide
themselves. Hill climbing methods require derivatives, for example. The only
information a GA needs is some measure of fitness about a point in the space
(sometimes known as an objective function value). Once the GA knows the
current measure of ―goodness‖ about a point, it can use this to continue
searching for the optimum.
3.5 BASIC GENETIC ALGORITHM OPERATIONS:
There are three basic operators found in every genetic algorithm. (Although some
algorithms may not employ the crossover operator, we shall refer to them as evolutionary
algorithms rather than genetic algorithms)
1. Reproduction
2. Crossover
3. Mutation
3.5.1 Reproduction:
The reproduction operator allows individual strings to be copied for possible
inclusion in the next generation. The chance that a string will be copied is based on the
string’s fitness value, calculated from a fitness function. For each generation, the
reproduction operator chooses string that are placed into a mating pool, which is used as the
basis for creating the next generation.
29. There are many different types of reproduction operators. One always selects the
fittest and discards the worst, statistically selecting the rest of the mating pool from the
remainder of the population. There are hundreds of variants of this scheme. None are right
or wrong. In fact, some will perform better than others depending on the problem domain
being explored.
3.5.2 Crossover:
Once the matting poll is created, the next operator in the GA’s arsenal comes into
play. Remember that crossover in biological terms refers to the blending of chromosomes
from the parents to produce new chromosomes for the offspring. The analogy carries over to
crossover in GAs.
The GA selects two strings at random from the mating pool. The strings selected
may be different or identical, it does not matter. The GA then calculates whether crossover
should take place using a parameter called the crossover probability. This is simply a
probability value p and is calculated by flipping a weighted coin. The value of p is set by the
user, and the suggested value is p=0.6, although this value can be domain dependent.
If the GA decides not to perform crossover, the two selected strings are simply
copied to the new population (they are not deleted from the mating pool. They may be used
multiple times during crossover).If crossover does takes place, then a random splicing point
is chosen in a string, the two strings are spliced and the spliced regions are mixed to create
two (potentially) new strings. These child strings are then placed in the new population.
As an example, say that the strings 10000 and 01110 are selected for crossover and
the GA decides to mate them. The GA selects a spacing point of 3.the following then occurs
100 00 100 10
011 10 011 00
Crossover in Action
The newly created strings are 10010 and 01100.
30. Crossover is performed until the new population is crested. Then the cycle starts
again with selection. This iterative process continues until any user specified criteria are met
(for example, fifty generations, or a string is found to have a fitness exceeding a certain
threshold).
Single point crossover - one crossover point is selected, binary string from beginning of
chromosome to the crossover point is copied from one parent, the rest is copied from the
second parent
11001011+11011111 = 11001111
Two point crossover - two crossover point are selected, binary string from
beginning of chromosome to the first crossover point is copied from one parent,
the part from the first to the second crossover point is copied from the second
parent and the rest is copied from the first parent
11001011 + 11011111 = 11011111
Uniform crossover - bits are randomly copied from the first or from the second
parent
11001011 + 11011101 = 11011111
31. Arithmetic crossover - some arithmetic operation is performed to make a new
offspring
11001011 + 11011111 = 11001001 (AND)
3.5.3 Mutation:
Selection and crossover alone can obviously generate a staggering amount of
differing strings. However, depending on the initial position chosen, there may not be
enough variety of strings to ensure the GA sees the entire problems space. Or the GA may
find itself converging on strings that are not quite close to the optimum it seeks due to a bad
initial population.
Some of these problems are overcome by introducing a mutation operator into the
GA. The GA has a mutation probability, m, which dictates the frequency at which mutation
occurs. Mutation can be performed either during selection or cross over. For each string
element in each string in the mating pool, the GA checks to see if it should perform a
mutation. If it should , it randomly changes the element value to a new one. In our binary
strings, 1s are changed to 0s and 0s to 1s.For example, the GA decides to mutate bit position
4 in string 10000:
Mutate
10000 10010
The resulting string is 10010 as the fourth bit in the string is flipped. The mutation
probability should be kept very low ( usually about 0.001% ) as a high mutation rate will
destroy fit strings and degenerate the GA algorithm into a random walk, with all the
associated problems.
32. But the mutation will help prevent the population from stagnating, adding ― fresh
blood‖, as it were, to a population. Remember that much of the power of a GA comes from
the fact that it contains a rich set of strings of great diversity. Mutation helps to maintain that
diversity througthout the GA s iterations.
Bit inversion - selected bits are inverted
11001001 => 10001001
3.6 POWER OF GENETIC ALGORITHM:
Selection + crossover = innovation
- Selection gives us a population of the strongest individuals
- Crossover attempts to combine parts of good individuals to make even better new
ones
Selection + Mutation = Stochastic Hill Climbing
- Mutation makes slight alternations to these
- We essentially have the equivalent of stochastic hill climbing
All put together we get,
Selection + Crossover + Mutation = The Power of GA
Add crossover to that, and we have stochastic hill climbing with a means of jumping
to potentially ―interesting‖ parts of the search space.
34. 4.1 PROBLEM STATEMENT
To minimize the material handling cost by the optimal arrangement of machines in
the shop floor.
4.2 APPLICATION OF GENETIC ALGORITHM
Genetic algorithm search technique is applied to the above problem in order to find
the minimum material handling cost in the production of Knuckle Joint in the CNC Machine
shop.
The following table gives the list of various machines involved in the production of
Knuckle Joint.
1 Turning machine
2 SBA milling drilling
3 Caliber arm milling drilling
4 Drilling and cover hole tapping
5 King pin arm milling drilling
6 Tie rod arm milling drilling
7 Slitting machine
8 Tapping machine
9 Drilling machine
10 Taper reaming machine
11 Mounting hole drilling tapping
12 King pin arm drilling and slitting
13 Tie rod arm milling machine
14 SPI milling machine
35. 4.3 ASSUMPTIONS
The work areas of the work stations are rectangular in shape and their orientations
are known.
Lot size does not change with the distance of travel between the machines that it
connects.
Every workstation works only one part at a time.
Every transporter carries only one type of part at a time.
The operating sequences of tasks are the same for the same part types.
Transportation cost between facilities is assumed to be unit/m/part.
4.4 COMPONENT DETAILS
There are 20 machines involved in the machining of 4 steering knuckle components.
In these machines, 12 machines are Vertical Machining Centre, 6 machines are Turning
machines, 2 machines are Milling machines.
Machines which are used for 4 components and sequence of machines are given
below:
COMPONENT SEQUENCE
LH—1-3-4-5-6
COMPONENT 1
RH—2-3-4-5-6
COMPONENT 2 7-8-9
COMPONENT 3 10-11-12-13
COMPONENT 4 14/15-16-17-18-19-20
36. COMPONENT 1
J200 KNUCKLE:
The machining time and the sequence of operation for component 1 are as follows:
Operation MACHINE TIME
OPERATION NAME
NO NO (Min)
1 4 min 17 sec
1 Turning
2 4 min 48 sec
SBA Milling Drilling,Caliber arm
2 3 3 min
Milling,Drilling & Coverhole tapping
3 Kingpin arm milling Drilling 4 1 min 38 sec
Milling,slitting,drilling,tie rod arm
4 5 3 min 41 sec
milling,drilling
5 ABS milling,Drilling,Tapping 6 2 min 13 sec
COMPONENT 2
MCI KNUCKLE:
The machining time and the sequence of operation for component 2 are as follows:
Operation MACHINE TIME
OPERATION NAME
NO NO (MIN)
1 Turning 7 3 min 5 sec
2 Caliber arm Milling, Drilling 8 4 min 16 sec
SBA milling drilling,Tie rod arm milling
3 drilling,Kingpin arm milling Drilling, 9 4 min 28 sec
Tapping
37. COMPONENT 3
GIO KNUCKLE:
The machining time and the sequence of operation for component 3 are as follows:
Operation MACHINE TIME
OPERATION NAME
NO NO (MIN)
1 Turning 10 2 min 57 sec
SBA Milling Drilling, Mounting hole
2 11 3 min 53 sec
drilling, Tapping
Tie rod arm milling drilling, Taper
3 12 3 min 39 sec
reaming, Kingpin arm milling Drilling
4 Kingpin arm drilling slitting milling 13 2 min 4 sec
COMPONENT 4
MXI KNUCKLE:
The machining time and the sequence of operation for component 4 are as follows:
Operation MACHINE CYCLE TIME
OPERATION NAME
NO NO (MIN)
14 3 min 13 sec
1 Turning
15 4 min 5 sec
2 Caliber arm Milling,Drilling 16 2 min
3 SPI milling 17 1 min 5 sec
4 Tie rod arm milling 18 1 min 35 sec
Kingpin arm machining, Tie rod arm
5 19 2 min 13 sec
milling
SBA drilling, Kingpin arm drilling &
6 20 1 min 57 sec
slitting
38. 4.5 MATHEMATICAL MODEL
The single row layout problems for facilities with unequal lengths (Heragu, 1997)
can be formulated as follows,
n-1 n
Minimize ---------------------------------- (1)
cij fij xi-xj
Subject to xi - xj ½ (li+lj) + dij i = 1, 2,3,….,n-1; j = i+1,….,n --------(2)
Where
n = no. of facilities
cij = cost of moving a standard unit by a unit distance between facilities i and j
fij = number of trips between facilities i and j
li = length of the horizontal side of facility i
dij = minimum distance by which facilities i and j are to be separated horizontally
xi = distance between the center of facility i and the vertical reference line
The material handling cost is calculated using the above mathematical model. Two
loops are formed which calculates the material handling cost. The cost factor is the product
of the following three terms.
Transportation cost between machines
Quantity of material flow
Distance between facilities
40. 4.7 CENTROID CALCULATION:
Centroid of all the facilities are calculated by adding the clearances between them
with length and width in order to find the material flow distance between the facilities.
AREA
Machine Width Length CLE_X CLE_Y
Machine Name (X+Xc).(Y+Yc)
No X(m) Y(m) Xc(m) Yc(m)
(m2)
1 Turning Machine 2 4 0.5 1 12.5
2 Turning Machine 2.4 4 0.5 1 14.5
3 VMC 2.8 4.4 0.5 1 17.82
4 VMC 2.4 3.6 0.5 1 13.34
5 VMC 2.4 4 1 1 17
6 VMC 2.8 4 0.5 1 16.5
7 Turning Machine 2 4 1 1 15
8 VMC 2.8 3.6 0.5 1 15.18
9 VMC 2.4 4 1 1 17
10 Turning Machine 2.4 2.4 0.5 1 9.86
11 VMC 2.8 4 1 1 19
12 VMC 2.8 4 0.5 1 16.5
13 VMC 2.4 4 0.5 1 14.5
14 Turning Machine 2.4 2.4 0.5 1 9.86
15 Turning Machine 2 3.6 0.5 1 11.5
16 VMC 2.4 4 0.5 1 14.5
17 Milling Machine 1.2 2 0.5 1.5 5.95
18 Milling Machine 1.2 2 0.5 1.5 5.95
19 VMC 2.8 4 0.5 1 16.5
20 VMC 2.8 4 0.5 1 16.5
41. 4.8 PART ROUTING MATRIX
The part routing matrix shows the Steering Knuckle flow between the facilities. Though
the batch size is different for all 4 components. There are given below
COMPONENT BATCH
MACHINES USED
NO SIZE
#1 1 2 3 4 5 6 402
#2 7 8 9 207
#3 10 11 12 13 234
#4 14 15 16 17 18 19 20 294
With the help of the values from the part routing matrix, the centroid values and the
cost factor the following matrices are formed. The product of the following matrices gives
the material handling cost.
Cost matrix
Flow matrix
Distance matrix
4.8.1 COST MATRIX
The cost matrix is formed in order to know the transportation cost between various
facilities. In our problem, we had assumed an amount of one unit per meter per component
or part. Since we are having the distance matrix values in meter the cost matrix values will
be in 0.1 units.
43. 4.8.3 FLOW MATRIX
The flow matrix is formed by taking in to account the flow of components between the
facilities where one to many relationship is followed. From the matrix it is clear that the
diagonal values are zero, since there will be no material flow within the same machine. The
remaining half of the matrix will have the mirror image values of the first half.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
1 0 0 201 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
2 0 201 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
3 0 402 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
4 0 402 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
5 0 402 0 0 0 0 0 0 0 0 0 0 0 0 0 0
6 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
7 0 207 0 0 0 0 0 0 0 0 0 0 0 0
8 0 207 0 0 0 0 0 0 0 0 0 0 0
9 0 0 0 0 0 0 0 0 0 0 0 0
10 0 234 0 0 0 0 0 0 0 0 0
11 0 234 0 0 0 0 0 0 0 0
12 0 234 0 0 0 0 0 0 0
13 0 0 0 0 0 0 0 0
14 0 0 147 0 0 0 0
15 0 147 0 0 0 0
16 0 294 0 0 0
17 0 294 0 0
18 0 294 0
19 0 294
20 0
4.9 MATERIAL HANDLING COST FOR EXISTING LAYOUT:
For the existing layout of the facilities in the CNC Machine shop involved in the
production of Steering Knuckle , the distance matrix, flow matrix and the cost matrix are
formed as above. Now, the material handling cost spent for the existing layout is calculated
by multiplying the three matrices as said before and is shown in the following table.
44. Thus the objective function value, which is the material handling cost for the existing
layout is found to be 14256.42 Rupees. In order to have the objective function a maximum
value the fitness value is calculated. The relation between the objective function value and
the fitness value is as follows.
Fitness value = (1 / Objective Function Value)
46. 5.1 GA PARAMETERS:
The following are the parameters that were used in the genetic algorithm optimizer in
order to get the optimum solution.
Population Size : 100
Cross Over : 0.6
Mutation : 0.2
No of trials : 3000
Random seed : 01
String representation : Single String
5.2 OPTIMIZER OUTPUT:
With the above parameter the GA optimizer was made to run by selecting strings at
random and the following results were obtained within a computational time of 28 seconds.
Material handling cost for the existing layout = Rs. 14256.42
MATERIAL
S.NO LAYOUT SEQUENCE
HANDLING COST
First Row: 12-11-3-4-5-6-7-8-9
1 Rs.1944.87
Second Row: 10-2-1-13-14-15-16-17-18-19-20
First Row: 12-11-3-4-5-6-7-8-9
2 Rs.1750.17
Second Row: 10-2-1-13-14-15-16-17-18-19-20
First Row: 12-11-3-4-5-6-7-8-9
3 Rs.1685.13
Second Row: 10-2-1-13-14-15-16-17-18-19-20
53. 5.3 SELECTION OF LAYOUT:
By using the genetic algorithm optimizer three feasible layouts were obtained. The
relocation cost of machines is to be taken in to account when the layout is to be changed.
With respect to the following comparison, the most feasible layout can be selected.
NEW LAYOUT #1
12 11 3 4 5 6 7 8 9
10 2 1 13 14 15 16 17 18 19 20
NEW LAYOUT #2
12 11 3 4 5 6 7 8 9
10 13 1 2 14 15 16 17 18 19 20
NEW LAYOUT #3
11 12 2 4 5 6 7 8 9
10 13 1 3 14 15 16 17 18 19 20
MODIFIED LAYOUT #1
13 20 19 18 17 16 1 2 9 14
11 12 15 6 5 4 3 8 7 10
MODIFIED LAYOUT #2
13 20 19 18 17 16 1 2 9 8
12 11 14 15 3 4 5 6 7 10
54.
55. When the optimizer solutions are compared with the existing layout, the solution
with machine sequence as per solution layout #3 is likely to be the most feasible one. The
reason for selecting solution layout #3 is as follows:
When compared with the other 2 solution layouts, the distance travel by the
component is less, so the material handling cost is low than the other 2 solution layouts.
But So compared this two layouts modified layout #1 is likely to be most feasible
one. . The reason for selecting modified layout #1 is as follows:
When modified layouts #1 and #2 are compared with existing layout, 11 of the
machines in the modified layout #1 and #2 need not to be altered, where as in the new
solutions #1,#2 and #3 there is no machines retains the same place. But comparing modified
layout #1 and #2 , modified layout #1 has minimum distance travelled, material handling
cost and less backflow.
Thus, from the GA optimizer output it is clear that any of the layout solution obtained can
be used with respect to the relocation cost involved.
57. CONCLUSION:
This project presented an approach for solving facility layout design problems with the
consideration of material handling cost. The proposed approach integrates Genetic Algorithm to
assist the end user in solving combinatorial optimization problems, and modeling and evaluating the
performance of complex manufacturing systems. This shows that the approach can be used to solve
single row unequal area layout problems effectively.
The results of the project work carried out in the CNC Machine Shop at Sakthi Auto
Components Limited, Erode for steering knuckle are given below.
1. Machining relayout of Steering Knuckle manufacturing section.
Expected reduction in travel distance = (Total distance travelled in existing layout) -
(Total distance travelled in proposed layout)
= 647m – 64.9m
= 582.1m
Expected reduction in material Handling cost = (Total cost in existing layout) -
(Total cost in proposed layout)
= Rs. 14,256.42 – Rs. 1685.13
= Rs. 12,571.29
(Keeping the material flow and transportation cost / unit as constant)
From the results, it is clear that the above practices will help the management in improving
the productivity. Hence, we recommend the above practices for implementation in CNC Machine
Shop with consideration of additional work in this area like,
1. Economic analysis of proposed layout like relocation cost, payback period etc.,
2. Performance analysis of the proposed layout based on operational parameters like
work-in-progress, queue length etc.,
59. REFERENCES:
1. Goldberg, David E. (1989). Genetic Algorithms in Search Optimization and Machine
Learning. Addison Wesley. pp. 41.
2. Fraser, Alex; Donald Burnell (1970). Computer Models in Genetics. New York:
McGraw-Hill.
3. Crosby, Jack L. (1973). Computer Simulation in Genetics. London: John Wiley &
Sons.
4. Syswerda, G. (1989). "Uniform crossover in genetic algorithms". In J. D.
Schaffer. Proceedings of the Third International Conference on Genetic Algorithms.
Morgan Kaufmann.
5. Srinivas. M and Patnaik. L, "Adaptive probabilities of crossover and mutation in
genetic algorithms," IEEE Transactions on System, Man and Cybernetics, vol.24,
no.4, pp.656–667, 1994.
WEB SITES:
1. www.globalsecurity.com
2. www.solver.com
3. www.cs.orchester.edu