`1

CHAPTER 1
INTRODUCTION

`2

INTRODUCTION

Joints of dissimilar metal combinations are employed in different applications requiring certain special combination of properties as well as to save cost incurred towards costly and scarce materials. Conventional fusion welding of many such dissimilar metal combinations is not feasible owing to the formation of brittle and low melting intermetallics due to metallurgical incompatibility, wide difference in melting point, thermal mismatch, etc. Solid-state welding processes that limit extent of intermixing are generally employed in such situations. Friction welding is one such solid-state welding process widely employed in such situations. The joining of materials by conventional welding techniques becomes difficult if the physical properties such as melting temperature and thermal expansion coefficient of the two materials differ a lot, as it is necessary to have controlled melting on both sides of weld joints simultaneously. Even if this criterion is met, it may not be possible to have an appropriate joint when the two materials are metallurgically incompatible. At the same time, the welding process, generally involves many input and output parameters. To produce the joints with highest quality, the proper process parameters have to be selected. The suitable process parameters, to obtain the required output, need many experiments and thus make the process to consume more time and money. Here the interest is to screen the experiments using the computational methods to analyze and optimize friction welding and other welding process parameters respectively.

`3

1.1.

FRICTION WELDING PROCESS

Friction welding is a class of solid-state welding processes that generates heat through mechanical friction between a moving work piece and a stationary component, with the addition of a lateral force called "upset" to plastically displace and fuse the materials. Technically, because no melt occurs, friction welding is not actually a welding process in the traditional sense, but a forging technique. However, due to the similarities between these techniques and traditional welding, the term has become common. Friction welding is used with metals and thermoplastic in a wide variety of aviation and automotive applications.

1.1.1 PROCESS MECHANISM

The mechanism of friction welding is schematically illustrated in Fig. 1.2 for better understanding of the process. As the pieces are initially contacted, micro asperities come into contact area. i. One component is rotated while the other is advanced into pressure contact with it. ii. Heat is produced at the faying surfaces. Overheating of metals cannot occur as the weld zone temperature is always stabilized below melting point. iii. Softened material begins to extrude in response to the applied pressure, creating an annular upset. iv. Heat is conducted away from the interfacial area for forging to take place.

`4

v. Rotation is stopped and a forge force is applied to complete the weld. vi. The joint undergoes hot working to form a homogenous, full surface, full diameter, high-integrityweld.

Fig 1.1. Process mechanism of the friction welding process.

1.2.

INTRODUCTION TO OPTIMIZATION:

Optimization is getting the best under the given circumstances. Optimization is the process of getting the maximum or minimum value of the function. (S.S. Roa, 2009). Best in the sense may be maximum or minimum under the given circumstances. One may need the maximum salary or the minimum expense. Optimization can be defined as the process of finding the conditions that give the maximum or minimum value of a function.

`5

1.2.1. STATEMENT OF AN OPTIMIZATION PROBLEM:

Optimization problem may be stated as follows (S.S.Roa, 2009): Find X={ x1, x2, x3, x4 xn } which minimize or maximize the function f (X) , Subjected to constraints, gi (X) 0, i= 1, 2, 3, 4, n li (X) = 0, Where, X is an n dimensional design vector, f (X) is the objective function, gi (X) is the inequality constraints lj (X) is the equality constraints The above is said to be unconstraint optimization problem. The constrained optimization technique is stated as follows: Find X={ x1, x2, x3, x4 xn } which minimize or maximize the function f(X) j = 1, 2, 3,4., n

1.2.2. TYPES OF OPTIMIZATION PROBLEM:

The various methods of optimization are defined as the follows: i. ii. iii. iv. Calculus methods Calculus of variations Nonlinear programming Geometric programming

`6

v. vi. vii. viii. ix. x. xi.

Quadratic programming Linear programming Dynamic programming Integer programming Stochastic programming Separable programming Multi objective programming The above is said to be traditional optimization technique. Following are some of the modern optimization techniques:

i. ii. iii. iv. v. vi.

Genetic algorithms Simulated annealing Ant colony optimization Particle swarm optimization Neural networks Fuzzy optimization.

1.2.3. APPLICATION OF OPTIMIZATION TECHNIQUE:

Optimization techniques are used in the application such as design of aircraft and aerospace structures for minimum weight design of civil engineering structures such as frames, foundations, bridges, towers, chimneys, and dams for minimum cost, optimum design of linkages, cams, gears, machine tools, and other mechanical components, Selection of machining conditions in metal-cutting processes for minimum production cost, design of material handling equipment, such as conveyors, trucks,

`7

and cranes, for minimum cost, design of pumps, turbines, and heat transfer equipment for maximum efficiency, shortest route taken by a salesperson visiting various cities during one tour, optimal production planning, controlling, and scheduling, analysis of statistical data and building empirical models from experimental results to obtain the most accurate, optimum design of control systems.(S.S. Roa, 2009).

`8

CHAPTER 2
GENETIC ALGORITHM

`9

GENETIC ALGORITHM

An algorithm is a series of steps for solving a problem. A genetic algorithm is a problem solving method that uses genetics as its model of problem solving. Its a search technique to find approximate solutions to optimization and search problems. Basically, an optimization problem looks really simple. One knows the form of all possible solutions corresponding to a specific question. The set of all the solutions that meet this form constitute the search space. The problem consists in finding out the solution that fits the best, i.e. the one with the most payoffs, from all the possible solutions. Each solution is represented through a chromosome, which is just an abstract representation. Coding all the possible solutions into a chromosome is the first part, but certainly not the most straightforward one of a Genetic Algorithm. A set of reproduction operators has to be determined, too. Reproduction operators are applied directly on the chromosomes, and are used to perform mutations and recombination over solutions of the problem. Appropriate representation and reproduction operators are really something determinant, as the behavior of the GA is extremely dependant on it. Frequently, it can be extremely difficult to find a representation, which respects the structure of the search space and reproduction operators, which are coherent and relevant according to the properties of the problems. Selection is supposed to be able to compare each individual in the population. Selection is done by using a fitness function. Each chromosome has an associated value corresponding to the fitness of the solution it represents. The fitness should correspond to an evaluation of how good the candidate solution is. The optimal solution is the one, which maximizes the

`10

fitness function. Genetic Algorithms deal with the problems that maximize the fitness function. But, if the problem consists in minimizing a cost function, the adaptation is quite easy. Either the cost function can be transformed into a fitness function, for example by inverting it; or the selection can be adapted in such way that they consider individuals with low evaluation functions as better. Once the reproduction and the fitness function have been properly defined, a Genetic Algorithm is evolved according to the same basic structure. It starts by generating an initial population of chromosomes. This first population must offer a wide diversity of genetic materials. The gene pool should be as large as possible so that any solution of the search space can be engendered. Generally, the initial population is generated randomly. Then, the genetic algorithm loops over an iteration process to make the population evolve. Each iteration consists of the following steps:

i.

SELECTION: The first step consists in selecting individuals for reproduction. This selection is done randomly with a probability depending on the relative fitness of the individuals so that best ones are often chosen for reproduction than poor ones.

ii.

REPRODUCTION: In the second step, offspring are bred by the selected individuals. For generating new chromosomes, the algorithm can use both recombination and mutation.

iii.

EVALUATION: Then the fitness of the new chromosomes is evaluated.

`11

iv.

REPLACEMENT: During the last step, individuals from the old population are killed and replaced by the new ones.

The general optimization procedure using a genetic algorithm is Step 1: Choose a coding to represent problem parameters, a selection operator, a crossover operator and a mutation operator. Choose population size n, crossover probability pc, and mutation probability pm. Initialize a random population of strings of size.. Choose a maximum allowable generation number tmax. Set t=0. Step 2: Evaluate each string in the population.

Step 3: If t > tmax or other termination criteria is satisfied, terminate. Step 4: Perform reproductions on the population.

Step 5: Perform crossovers on pair of strings with probability pc. Step 6: Perform mutations on strings with probability pm.

Step 7: Evaluate strings in the new population. Set t = t + 1 and go to Step 3.

`12

The flowchart showing the process of GA is as shown in Fig. 2.1

START

CREATE INITIAL RANDOM POPULATION

EVALUATE FITNESS FOR EACH POPULATION

STORE BEST INDIVIDUAL

CREATING MATING POOL

CREATE NEXT GENERATION BY APPLYING CROSSOVER

OPTIMAL OR GOOD SOLUTION FOUND?

STOP

REPRODUCE AND IGNORE FEW POPULATIONS

YES
PERFORM MUTATION

NO

FIG 2.1. FLOW CHART FOR GENETIC ALGORITHM

`13

In short, the basic four steps used in simple Genetic Algorithm to solve a problem are, The representation of the problem The fitness calculation Various variables and parameters involved in controlling the algorithm The representation of result and the way of terminating the algorithm 2.1. COMPARISON OF GENETIC ALGORITHM WITH OTHER OPTIMIZATION TECHNIQUES:

The principle of GAs is simple: imitate genetics and natural selection by a computer program: The parameters of the problem are coded most naturally as a DNA-like linear data structure, a vector or a string. Sometimes, when the problem is naturally two or three-dimensional also corresponding array structures are used. A set, called population, of these problem dependent parameter value vectors is processed by GA. To start there is usually a totally random population, the values of different parameters generated by a random number generator. Typical population size is from few dozens to thousands. To do optimization we need a cost function or fitness function as it is usually called when genetic algorithms are used. By a fitness function we can select the best solution candidates from the population and delete the not so good specimens. The nice thing when comparing GAs to other optimization methods is that the fitness function can be nearly anything that can be evaluated by a computer or even something that cannot! In the latter case it might be a human judgment that cannot be stated as a crisp program, like in the case of eyewitness, where a human being selects among the alternatives generated by GA. So, there are not any definite mathematical restrictions on the properties of the fitness function. It may be discrete, multimodal etc. The

`14

main criteria used to classify optimization algorithms are as follows: continuous / discrete, constrained / unconstrained and sequential / parallel. There is a clear difference between discrete and continuous problems. Therefore it is instructive to notice that continuous methods are sometimes used to solve inherently discrete problems and vice versa. Parallel algorithms are usually used to speed up processing. There are, however, some cases in which it is more efficient to run several processors in parallel rather than sequentially. These cases include among others such, in which there is high probability of each individual search run to get stuck into a local extreme. Irrespective of the above classification, optimization methods can be further classified into deterministic and non-deterministic methods. In addition optimization algorithms can be classified as local or global. In terms of energy and entropy local search corresponds to entropy while global optimization depends essentially on the fitness i.e. energy landscape. Genetic algorithm differs from conventional optimization techniques in following ways:

i.

GAs operate with coded versions of the problem parameters rather than parameters themselves i.e., GA works with the coding of solution set and not with the solution itself.

ii.

Almost all conventional optimization techniques search from a single point but GAs always operate on a whole population of points(strings) i.e., GA uses population of solutions rather than a single solution from searching. This plays a major role to the robustness of genetic algorithms. It improves the chance of reaching the global optimum and also helps in avoiding local stationary point.

`15

iii.

GA uses fitness function for evaluation rather than derivatives. As a result, they can be applied to any kind of continuous or discrete optimization problem. The key point to be performed here is to identify and specify a meaningful decoding function.

iv.

GAs use probabilistic transition operates while conventional methods for continuous optimization apply deterministic transition operates i.e., GAs does not use deterministic rules.

2.2. ADVANTAGES OF GENETIC ALGORITHM

The advantages of genetic algorithm includes, Parallelism Liability Solution space is wider The fitness landscape is complex Easy to discover global optimum The problem has multi objective function Only uses function evaluations. Easily modified for different problems. Handles noisy functions well. Handles large, poorly understood search spaces easily Good for multi-modal problems Returns a suite of solutions. Very robust to difficulties in the evaluation of the objective function. They require no knowledge or gradient information about the response surface

`16

Discontinuities present on the response surface have little effect on overall optimization performance They are resistant to becoming trapped in local optima. They perform very well for large-scale optimization problems Can be employed for a wide variety of optimization problems

2.3. LIMITATION OF GENETIC ALGORITHM

The limitation of genetic algorithm includes, The problem of identifying fitness function Definition of representation for the problem Premature convergence occurs The problem of choosing the various parameters like the size of the population, mutation rate, cross over rate, the selection method and its strength. Cannot use gradients. Cannot easily incorporate problem specific information Not good at identifying local optima No effective terminator. Not effective for smooth unimodel functions Needs to be coupled with a local search technique. Have trouble finding the exact global optimum Require large number of response (fitness) function evaluations Configuration is not straightforward

`17

2.4. APPLICATIONS OF GENETIC ALGORITHM

Genetic algorithms have been used for difficult problems (such as NP-hard problems), for machine learning and also for evolving simple programs. They have been also used for some art, for evolving pictures and music. A few applications of GA are as follows: Nonlinear dynamical systemspredicting, data analysis Robot trajectory planning Evolving LISP programs (genetic programming) Strategy planning Finding shape of protein molecules TSP and sequence scheduling Functions for creating images Controlgas pipeline, pole balancing, missile evasion, pursuit Designsemiconductor layout, aircraft design, keyboard configuration, communication networks Schedulingmanufacturing, facility scheduling, resource allocation Machine LearningDesigning neural networks, both architecture and weights, improving classification algorithms, classifier systems Signal Processingfilter design Combinatorial Optimizationset covering, traveling salesman (TSP), Sequence scheduling, routing, bin packing, graph coloring and partitioning.

2.5. TERMINOLOGIES AND OPERATORS OF GA Genetic Algorithm uses a metaphor where an optimization problem takes the place of an environment and feasible solutions are considered as

`18

individuals living in that environment. In genetic algorithms, individuals are binary digits or of some other set of symbols drawn from a finite set. As computer memory is made up of array of bits, anything can be stored in a computer and can also be encoded by a bit string of sufficient length. Each of the encoded individual in the population can be viewed as a representation, according to an appropriate encoding of a particular solution to the problem. For Genetic Algorithms to find a best optimum solution, it is necessary to perform certain operations over these individuals. This part of the chapter discusses the basic terminologies and operators used in Genetic Algorithms to achieve a good enough solution for possible terminating conditions.

2.6. INDIVIDUALS

An individual is a single solution. Individual groups together two forms of solutions as given below: 1. The chromosome, which is the raw genetic information (genotype) that the GA deals. 2. The phenotype, which is the expressive of the chromosome in the terms of the model. A chromosome is subdivided into genes. A gene is the GAs representation of a single factor for a control factor. Each factor in the solution set corresponds to gene in the chromosome\

2.7. GENES Genes are the basic instructions for building a Generic Algorithms. A chromosome is a sequence of genes. Genes may describe a possible solution to a problem, without actually being the solution. A gene is a bit string of arbitrary lengths. The bit string is a binary representation of number of

`19

intervals from a lower bound. A gene is the GAs representation of a single factor value for a control factor, where control factor must have an upper bound and lower bound.

2.8. FITNESS

The fitness of an individual in a genetic algorithm is the value of an objective function for its phenotype. For calculating fitness, the chromosome has to be first decoded and the objective function has to be evaluated. The fitness not only indicates how good the solution is, but also corresponds to how close the chromosome is to the optimal one.

2.9. POPULATIONS

A population is a collection of individuals. A population consists of a number of individuals being tested, the phenotype parameters defining the individuals and some information about search space.

2.10. DATA STRUCTURES

The main data structures in GA are chromosomes, phenotypes, objective function values and fitness values. This is particularly easy implemented when using MATLAB package as a numerical tool. An entire chromosome population can be stored in a single array given the number of individuals and the length of their genotype representation. Similarly, the design variables, or phenotypes that are obtained by applying some mapping from the chromosome representation into the design space can be stored in a single array. The actual mapping depends upon the decoding scheme used.

`20

The objective function values can be scalar or vectorial and are necessarily the same as the fitness values. Fitness values are derived from the object function using scaling or ranking function and can be stored as vectors.

2.11. SEARCH STRATEGIES

The search process consists of initializing the population and then breeding new individuals until the termination condition is met. There can be several goals for the search process, one of which is to find the global optima. This can never be assured with the types of models that GAs work with. There is always a possibility that the next iteration in the search would produce a better solution. In some cases, the search process could run for years and does not produce any better solution than it did in the first little iteration. Another goal is faster convergence. When the objective function is expensive to run, faster convergence is desirable, however, the chance of converging on local, and possibly quite substandard optima is increased. Apart from these, yet another goal is to produce a range of diverse, but still good solutions. When the solution space contains several distinct optima, which are similar in fitness, it is useful to be able to select between them, since some combinations of factor values in the model may be more feasible than others. Also, some solutions may be more robust than others.

2.12. ENCODING

Encoding is a process of representing individual genes. The process can be performed using bits, numbers, trees, arrays, lists or any other objects. The encoding depends mainly on solving the problem. For example, one can encode directly real or integer numbers

`21

2.12.1. BINARY ENCODING

Each chromosome encodes a binary (bit) string. Each bit in the string can represent some characteristics of the solution. Every bit string therefore is a solution but not necessarily the best solution. Another possibility is that the whole string can represent a number. The way bit strings can code differs from problem to problem Binary encoding gives many possible chromosomes with a smaller number of alleles. On the other hand this encoding is not natural for many problems and sometimes corrections must be made after genetic operation is completed. Binary coded strings with 1s and 0s are mostly used. The length of the string depends on the accuracy.

In this, Integers are represented exactly Finite number of real numbers can be represented Number of real numbers represented increases with string length The other encoding methods are i. ii. iii. iv. v. Octal Encoding Hexadecimal Encoding Permutation Encoding Value Encoding Tree Encoding

2.13. BREEDING

The breeding process is the heart of the genetic algorithm. It is in this process, the search process creates new and hopefully fitter individuals. The breeding cycle consists of three steps:

`22

a. Selecting parents. b. Crossing the parents to create new individuals (offspring or children). c. Replacing old individuals in the population with the new ones.

2.13.1. SELECTION

Selection is the process of choosing two parents from the population for crossing. After deciding on an encoding, the next step is to decide how to perform selection i.e., how to choose individuals in the population that will create offspring for the next generation and how many offspring each will create. The purpose of selection is to emphasize fitter individuals in the population in hopes that their off springs have higher fitness. Chromosomes are selected from the initial population to be parents for reproduction. The problem is how to select these chromosomes. Selection is a method that randomly picks chromosomes out of the population according to their evaluation function. The higher the fitness function, the more chance an individual has to be selected. The selection pressure is defined as the degree to which the better individuals are favored. The higher the selection pressured, the more the better individuals are favored. This selection pressure drives the GA to improve the population fitness over the successive generations. Selection has to be balanced with variation form crossover and mutation. Too strong selection means sub optimal highly fit individuals will take over the population, reducing the diversity needed for change and progress; too weak selection will result in too slow evolution. The various selection methods are discussed as follows:

`23

ROULETTE WHEEL SELECTION

Roulette selection is one of the traditional GA selection techniques. The commonly used reproduction operator is the proportionate reproductive operator where a string is selected from the mating pool with a probability proportional to the fitness. The principle of roulette selection is a linear search through a roulette wheel with the slots in the wheel weighted in proportion to the individuals fitness values. A target value is set, which is a random proportion of the sum of the fit nesses in the population. The population is stepped through until the target value is reached. This is only a moderately strong selection technique, since fit individuals are not guaranteed to be selected for, but somewhat have a greater chance. A fit individual will contribute more to the target value, but if it does not exceed it, the next chromosome in line has a chance, and it may be weak. It is essential that the population not be sorted by fitness, since this would dramatically bias the selection. The above described Roulette process can also be explained as follows: The expected value of an individual is that fitness divided by the actual fitness of the population. Each individual is assigned a slice of the roulette wheel, the size of the slice being proportional to the individuals fitness. The wheel is spun N times, where N is the number of individuals in the population. On each spin, the individual under the wheels marker is selected to be in the pool of parents for the next generation Roulette wheel selection is easier to implement but is noisy. The rate of evolution depends on the variance of fitnesss in the population. Roulette wheel selection is easier to implement but is noisy. The rate of evolution depends on the variance of fitnesss in the population.

`24

RANDOM SELECTION This technique randomly selects a parent from the population. In terms of disruption of genetic codes, random selection is a little more disruptive, on average, than roulette wheel selection.

RANK SELECTION

The Roulette wheel will have a problem when the fitness values differ very much. If the best chromosome fitness is 90%, its circumference occupies 90% of Roulette wheel, and then other chromosomes have too few chances to be selected. Rank Selection ranks the population and every chromosome receives fitness from the ranking. The worst has fitness 1 and the best has fitness N. It results in slow convergence but prevents too quick convergence. It also keeps up selection pressure when the fitness variance is low. It preserves diversity and hence leads to a successful search. In effect, potential parents are selected and a tournament is held to decide which of the individuals will be the parent. There are many ways this can be achieved and two suggestions are,

1. Select a pair of individuals at random. Generate a random number, R, between 0 and 1. If R < r use the first individual as a parent. If the R>=r then use the second individual as the parent. This is repeated to select the second parent. The value of r is a parameter to this method. 2. Select two individuals at random. The individual with the highest evaluation becomes the parent. Repeat to find a second parent. the other selection method are:

`25

Tournament Selection Boltzmann Selection Stochastic Universal Sampling

2.14.

CROSSOVER (RECOMBINATION)

Crossover is the process of taking two parent solutions and producing from them a child. After the selection (reproduction) process, the population is enriched with better individuals. Reproduction makes clones of good strings but does not create new ones. Crossover operator is applied to the mating pool with the hope that it creates a better offspring. 1. Crossover is a recombination operator that proceeds in three steps: 2. The reproduction operator selects at random a pair of two individual strings for the mating. 3. A cross site is selected at random along the string length. Finally, the position values are swapped between the two strings following the cross site. That is, the simplest way how to do that is to choose randomly some crossover point and copy everything before this point from the first parent and then copy everything after the crossover point from the other parent. The various crossover techniques are discussed as follows:

2.14.1.

SINGLE POINT CROSSOVER

The traditional genetic algorithm uses single point crossover, where the two mating chromosomes are cut once at corresponding points and the sections after the cuts exchanged. Here, a cross-site or crossover point is selected randomly along the length of the mated strings and bits next to the

`26

cross-sites are exchanged. If appropriate site is chosen, better children can be obtained by combining good parents else it severely hampers string quality.\

2.14.2.

TWO POINT CROSSOVER

Apart from single point crossover, many different crossover algorithms have been devised, often involving more than one cut point. It should be noted that adding further crossover points reduces the performance of the GA. The problem with adding additional crossover points is that building blocks are more likely to be disrupted. However, an advantage of having more crossover points is that the problem space may be searched more thoroughly.

2.14.3.

MULTI-POINT CROSSOVER

There are two ways in this crossover. One is even number of cross-sites and the other odd number of cross-sites. In the case of even number of crosssites, cross-sites are selected randomly around a circle and information is exchanged. In the case of odd number of cross-sites, a different cross-point is always assumed at the string beginning.

2.14.4.

UNIFORM CROSSOVER

Uniform crossover is quite different from the N-point crossover. Each gene in the offspring is created by copying the corresponding gene from one or the other parent chosen according to a random generated binary crossover mask of the same length as the chromosomes. Where there is a 1 in the crossover mask, the gene is copied from the first parent, and where there is a 0 in the mask the gene is copied from the second parent. A new crossover

`27

mask is randomly generated for each pair of parents. Offspring, therefore contain a mixture of genes from each parent. The number of effective crossing point is not fixed, but will average L/2 (where L is the chromosome length). The other type of cross over are: Three Parent Crossover Crossover with Reduced Surrogate Shuffle Crossover Precedence Preservative Crossover Ordered Crossover Partially Matched Crossover

2.14.5.

CROSSOVER PROBABILITY

The basic parameter in crossover technique is the crossover probability (Pc). Crossover probability is a parameter to describe how often crossover will be performed. If there is no crossover, offspring are exact copies of parents. If there is crossover, offspring are made from parts of both parents chromosome. If crossover probability is 100%, then all offspring are made by crossover. If it is 0%, whole new generation is made from exact copies of chromosomes from old population (but this does not mean that the new generation is the same!). Crossover is made in hope that new chromosomes will contain good parts of old chromosomes and therefore the new chromosomes will be better. However, it is good to leave some part of old population survive to next generation.

`28

2.15.

MUTATION

After crossover, the strings are subjected to mutation. Mutation prevents the algorithm to be trapped in a local minimum. Mutation plays the role of recovering the lost genetic materials as well as for randomly disturbing genetic information. It is an insurance policy against the irreversible loss of genetic material. Mutation has traditionally considered as a simple search operator. If crossover is supposed to exploit the current solution to find better ones, mutation is supposed to help for the exploration of the whole search space. Mutation is viewed as a background operator to maintain genetic diversity in the population. It introduces new genetic structures in the population by randomly modifying some of its building blocks. Mutation helps escape from local minimas trap and maintains diversity in the population. It also keeps the gene pool well stocked, and thus ensuring periodicity. A search space is said to be ergodic if there is a non-zero probability of generating any solution from any population state. There are many different forms of mutation for the different kinds of representation. For binary representation, a simple mutation can consist in inverting the value of each gene with a small probability. The probability is usually taken about 1/L, where L is the length of the chromosome. It is also possible to implement kind of hill-climbing mutation operators that do mutation only if it improves the quality of the solution. Such an operator can accelerate the search. But care should be taken, because it might also reduce the diversity in the population and makes the algorithm converge toward some local optima. Mutation of a bit involves flipping a bit, changing 0 to 1 and vice-versa.

`29

2.15.1.

FLIPPING

Flipping of a bit involves changing 0 to 1 and 1 to 0 based on a mutation chromosome generated

2.15.2.

INTERCHANGING

Two random positions of the string are chosen and the bits corresponding to those positions are interchanged.

2.15.3.

REVERSING

A random position is chosen and the bits next to that position are reversed and child chromosome is produced.

2.15.4.

MUTATION PROBABILITY

The important parameter in the mutation technique is the mutation probability (Pm). The mutation probability decides how often parts of chromosome will be mutated. If there is no mutation, offspring are generated immediately after crossover (or directly copied) without any change. If mutation is performed, one or more parts of a chromosome are changed. If mutation probability is 100%, whole chromosome is changed, if it is 0%, nothing is changed. Mutation generally prevents the GA from falling into local extremes. Mutation should not occur very often, because then GA will in fact change to random search.

`30

2.16.

REPLACEMENT

Replacement is the last stage of any breeding cycle. Two parents are drawn from a fixed size population, they breed two children, but not all four can return to the population, so two must be replaced i.e., once off springs are produced, a method must determine which of the current members of the population, if any, should be replaced by the new solutions. The technique used to decide which individual stay in a population and which are replaced in on a par with the selection in influencing convergence. Basically, there are two kinds of methods for maintaining the population; generational updates and steady state updates.

2.16.1.

RANDOM REPLACEMENT

The children replace two randomly chosen individuals in the population. The parents are also candidates for selection. This can be useful for continuing the search in small populations, since weak individuals can be introduced into the population.

2.16.2.

WEAK PARENT REPLACEMENT

In weak parent replacement, a weaker parent is replaced by a strong child. With the four individuals only the fittest two, parent or child, return to population. This process improves the overall fitness of the population when paired with a selection technique that selects both fit and weak parents for crossing, but if weak individuals and discriminated against in selection the opportunity will never raise to replace them.

`31

2.16.3.

BOTH PARENTS

Both parents replacement is simple. The child replaces the parent. In this case, each individual only gets to breed once. As a result, the population and genetic material moves around but leads to a problem when combined with a selection technique that strongly favors fit parents: the fit breed and then are disposed of.

2.17.

SEARCH TERMINATION (CONVERGENCE CRITERIA)

In short, the various stopping condition are listed as follows: Maximum generationsThe genetic algorithm stops when the specified numbers of generations have evolved. Elapsed timeThe genetic process will end when a specified time has elapsed. Note: If the maximum number of generation has been reached before the specified time has elapsed, the process will end.

No change in fitnessThe genetic process will end if there is no change to the populations best fitness for a specified number of generations. Note: If the maximum number of generation has been reached before the specified number of generation with no changes has been reached, the process will end. Stall generationsThe algorithm stops if there is no improvement in the objective function for a sequence of consecutive generations of length Stall generations.

`32

Stall time limitThe algorithm stops if there is no improvement in the objective function during an interval of time in seconds equal to Stall time limit. The termination or convergence criterion finally brings the search to a halt. The following are the few methods of termination techniques

2.17.1.

BEST INDIVIDUAL

A best individual convergence criterion stops the search once the minimum fitness in the population drops below the convergence value. This brings the search to a faster conclusion guaranteeing at least one good solution.

2.17.2.

WORST INDIVIDUAL

Worst individual terminates the search when the least fit individuals in the population have fitness less than the convergence criteria. This guarantees the entire population to be of minimum standard, although the best individual may not be significantly better than the worst. In this case, a stringent convergence value may never be met, in which case the search will terminate after the maximum has been exceeded.

2.17.3.

SUM OF FITNESS

In this termination scheme, the search is considered to have satisfaction converged when the sum of the fitness in the entire population is less than or

`33

equal to the convergence value in the population record. This guarantees that virtually all individuals in the population will be within a particular fitness range, although it is better to pair this convergence criteria with weakest gene replacement, otherwise a few unfit individuals in the population will blow out the fitness sum. The population size has to be considered while setting the convergence value.

2.17.4.

MEDIAN FITNESS

Here at least half of the individuals will be better than or equal to the convergence value, which should give a good range of solutions to choose from.

`34

CHAPTER 3
LITERATURE REVIEW

`35

LITERATURE REVIEW

Sathiya.A et.,al., (2008) applied successfully to process similar joints of stainless steel (AISI 304).The friction processed joints exhibited comparable strength with the base material and joint strength decreased with an increase in the friction time. The material shear flow and dimples at the fractured surface confirms the ductile mode of failure of joints during tensile testing. Hardness at the joint zone increases with the increase in friction time. The increase in hardness at the joint zone is due to the thermal history of the joining process. Friction welding method can be applied successfully to process similar joints of stainless steel (AISI 430).The processed joints exhibited better mechanical and metallurgical characteristics, than the fusion joints. Because of the solid state bonding technique, the problems associated with fusion joining are minimized in the case of friction welding. The joints exhibited 95.52% of parent materials tensile strength. The tensile specimen failures were associated primarily with the weld interface region. The fracture is predominantly associated with material (shear like) flow. The toughness of the friction welded ferritic stainless steel is comparatively higher than fusion processed joints due to the refinement of grain size at the weld zone. The ultimate capacity of the arc spot welding was modeled based on the experimental data using Artificial neural network in the literature (Abdulkadir Cevik et,al., 2008). The main objective in the study is to obtain the explicit formulation of nominal shear stress as a function of the geometric, and the mechanical properties of the spot welding. The proposed model is obtained using the Neural network (NN) tool box of MATLAB. The statistical analysis of the proposed model and the experimental is being carried out for the train sets and test sets for which the correlation coefficient is found as 0.984 and 0.969 respectively. The parametric study is made to

`36

investigate the effect of changing geometric parameters and the yield strength on the nominal shear strength of arc spot welds. An intelligent system was being developed by I.S. Kim et.al(2005).,using MATLAB/SIMULINK software. The mathematical model incorporating the different welding parameters and complex geometrical features is developed based on the regression and the neural networks. In the method of modeling using the multiple regression models, the Analysis Of Variance technique (ANOVA) is performed on the factorial design to quantify the effect of welding parameters. Then the multiple correlation coefficients and the Fratio were to measure the goodness of fit. In the method of modeling using the ANN method, Back-Propagation (BP) method is being used and at the same time the generalized delta rule is used as the learning algorithm. With the learning rate of 0.6 and the momentum term of 0.9, the network is trained for 2,00,000 iteration. The error between the desired and the actual output is less than 0.001 at the end of the training process. By comparing the above two methods, the neural network model is found to be comparatively best, which is capable of making the prediction of the experimental result with the reasonable accuracy. According to Cemal Meran, ( 2006) welding was the major joining process in the industry. Three main parameters such as welding current, welding velocity, and arc length have a big influence on the quality welding. The work by Cemal Meran(2006), deal with the use of stochastic search process, which is the basis for the genetic algorithms in developing estimation of the welding parameters for the joined brass plates. The obtained result is being compared with the experimental value for the verification, ehich showed a good agreement.

`37

In order to evaluate the effect of process parameters, such as tool rotational speed, transverse speed and axial force, on the tensile strength of friction welded RDE-40 aluminium alloy, the Taguchis parametric design and optimization approach was used( A. K. Lakshminarayanan., et.al., 2008). Through the Taguchis parametric design approach, the optimum level of the process parameter was determined. The optimal value is found to be 303MPa. Rather than the well known effect of the process parameter on the quality of the welding, the study made by Serder et.al.,(2008) focuses on the sensitivity analysis of the process parameters and fine tuning requirements of the parameters for optimum weld bead geometry. Changeable process parameters such as welding current, welding voltage and welding speed are used as design variables. Experimental work is based on the three level factorial design and the mathematical modeling is based on the multiple curvilinear regression analysis. Effect of all the three design parameters on the bead width and bead height show that small changes in theses parameters have very important role in the quality of welding operation. Conventional regression analysis were carried out by Parikshit Dutta et.al.,(2007), on some of the experimental data of a tungsten inert gas (TIG) welding process parameters to find its effect. At the same time, Artificial Neural Network (ANN) concept is used to find the effect of various process parameters, in which thousand training data were employed. Back Prorogation Neural Network (BPNN) and genetic neural system (GA-NN) are the type of ANN which is used for the modeling purpose. The performance of all those techniques were being compared and it is concluded that Neural Network (NN) based approaches were seen to be more adoptive, which may be due to the reasons that it is based on the principle of steepest descent method.

`38

An empirical model was developed by G. Padmanaban et.al., to predict the maximum tensile strength of AZ31B magnesium alloy. The process parameter includes laser power, welding speed and focal position. The response surface methodology is used as an optimization tool to predict the maximum tensile strength. The experiments were conducted based on a three factor, three level, central composite face centered matrix with full replication technique. A maximum tensile strength of 212MPa is obtained under the welding conditions in which the laser power is 2.5Kw, welding speed of 5.0m/min, and focal position of -1.5mm. Comparatively welding speed is the most factor which has the greater influence compared to others. The mathematical model was developed for the analysis and simulation of the correlation between the friction stir welding of aluminium plates and mechanical properties using Artificial Neural Network (ANN). The model can be used to calculate mechanical properties of welded Al plates as the function of weld speed and tool rotation speed. The combined effect of weld speed and tool rotation speed on the mechanical properties of welded Al plates was simulated and then the comparison between the measured and calculated data. The calculated results were in good agreement with the measured value. The works carried out by G. Mahendran et al..,(2008) developed diffusion bonding windows for joining AZ3113 Magnesium and AA 2024 Aluminium alloys. By the experimental work, the optimal process parameters were found. The constructed bonding windows may act as the reference map for selecting appropriate process parameters to obtain high strength bonds. It is concluded by the author that the highest shear strength is obtained at the bonding temperature of 4250C, the bonding pressure of 20MPa, and the holding time of 45min due to the formation of optimum thick diffusion layer at the interface of MgAl alloys.

`39

Further experimental studies were carried out by G.Mahendran et al., (2010) on Magnesium and Copper using diffusion bonding methods. Three factor five-level, central composite rotatable design matrix method is used for the purpose of designing the experiments. Empirical relationship were developed in order to predict the diffusion layer thickness, hardness and strength of the joint. The response surface methodology is used for optimization and which is found as follows: joints fabricated at the bonding pressure of 12MPa, the bonding time of 30mins, the bonding temperature of 4500C , which yields a maximum shear strength and bond strength of 66 and 81MPa respectively. The diffusion bonding process studied in Nickel alloy, Su236, by Ravishankar B et.al.,(2009) under the temperature ranges from 1123-1323k and the compressive strength of 90% of yield strength, determined the importance of the process parameters, the mechanism responsible for bonding and the joint characteristic. The experimental results were compared with the model developed by John Pilling. The mechanism of bonding was evaluated by grain growth equation. The quality of bond was assessed using optical metallographic and lap shear testing. The bonding mechanism and composition of the interface were determined using quantified EPMA line analysis. The non conventional technique is being used by S. Suresh Kumar at.al.,(2009). The ultrasonic A-scan method is used to evaluate the quality of the welded joints and to study the interface properties of the joint of diffusion bonded of Ti-6Al-4V, in that work the experiment is carried out for the given material for the pressure ranges from 1.6 4MPa, and the bonding time of 4hours. The shear strength of the bond is to be correlated with the fractional area bonded which is measured ng metallographic techniques and ultrasonic technique, which is the non-destructive method.

`40

The optimal setting of the welding process parameters was identified by Sathiya. P., et.al.,(2009) using non conventional technique such as Genitic Algorithm, Simulated Annealing, Particle Swarm Optimization. Thje mathematical model relating the process parameters is developed using Artificial Neural Network (ANN). The welding is carried out in Stainless Steel (AISI 304). The optimized value obtained through genetic algorithm closely resembles the experimental value.

TABLE

3.1.

OPTIMIZED

PROCESS

PARAMETRE

AND

EXPERIMENTAL VALUE (Sathiya. P., et.al., 2009)

Process parameter Upsetting Pressure(UP) Upsetting Time (UT) Heating Time (HT) Upsetting Time

Optimized value 17.7028 bar 4.2663 sec 35.1078 bar 4.025 sec

Experimental value 17bar 4sec 35 bar 4 sec

`41

CHAPTER 4
OPTIMIZING THE PROCESS PARAMETERS OF FRICTION WELDING PROCESS

`42

OPTIMIZING THE PROCESS PARAMETERS OF FRICTION WELDING PROCESS 4.1. PROCEDURE FOR OPTIMIZING THE PROCESS PARAMETERS

The optimum process parameter of the friction welding process, is to be found for maximizing the tensile strength of the Al/SS joint

(Purushothaman.L et.al, 2009). From the quoted literature, the experimental datas were taken since those work concentrate only on developing the experimental matrix for the various process parameters and the tensile strength of the Al/SS joint. From the experimental data the following work are being carried out using MATLAB.

4.2. DEVELOPING THE MODEL BASED ON THE MULTIPLE LINEAR REGRESSION METHOD:

4.2.1. FEASIBLE LIMITS AND THE EXPERIMENTAL DESIGN MATRIX:

The feasible working limits of friction welding process parameters from the literature is identified and presented in Tables 4.1. The experimental design matrix is presented in table 4.

`43

Table 4.1. Working Limits of friction welding Levels 0 (+1) 1.5 4.5 1.5 4.5 1.75 5.25 1.75 5.25

# 1 2 3 4

Parameter Notation Friction Pressure Frp Friction Time Frt Forging Pressure Fop Forging Time Fot

Unit Ton sec Ton sec

(-2) 1 3 1 3

(-1) 1.25 3.75 1.25 3.75

(+2) 2 6 2 6

Due to wide range of factors, it was decided to use four factors, five levels, central composite face centered design matrix to optimise the experimental conditions, which fits the second order response surfaces very accurately. Central composite rotational design matrix with the star points are at the center of each face of factorial space was used, so were coded as +2 and 2 respectively = 2. This variety

requires 5 levels of each factor. The upper limit and lower limit of a factor

`44

Table 4.2. Experimental Design Matrix Tensile strength (MPa) TS 139 108 106 88 143 114 72 104 64 186 190 112 36 67 68 67 184 104 123 123 84 185 51 86 58 62 102 184 98 180

Coded values Expt. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 1 0 1 -1 0 1 1 -1 0 0 0 0 -1 -2 0 -1 0 2 -1 1 -1 0 -1 1 1 0 1 0 -1 0 1 2 1 1 0 -1 1 1 0 0 0 0 -1 0 -2 -1 0 0 1 1 -1 0 1 -1 -1 0 -1 0 -1 0 1 0 1 1 0 1 -1 -1 -2 0 0 2 -1 0 0 1 0 0 1 -1 -1 0 -1 1 -1 0 -1 0 1 0 1 0 -1 -1 2 1 -1 1 0 0 0 0 -1 0 0 -1 0 0 1 1 1 0 -1 -1 -1 -2 1 0 1 0 A 1.75 1.5 1.75 1.25 1.5 1.75 1.75 1.25 1.5 1.5 1.5 1.5 1.25 1 1.5 1.25 1.5 2 1.25 1.75 1.25 1.5 1.25 1.75 1.75 1.5 1.75 1.5 1.25 1.5

Original values B C 5.25 1.75 6 1.5 5.25 1.75 5.25 1.5 4.5 1.5 3.75 1.75 5.25 1.25 5.25 1.25 4.5 1 4.5 1.5 4.5 1.5 4.5 2 3.75 1.25 4.5 1.5 3 1.5 3.75 1.75 4.5 1.5 4.5 1.5 5.25 1.75 5.25 1.25 3.75 1.25 4.5 1.5 5.25 1.25 3.75 1.5 3.75 1.25 4.5 1.5 3.75 1.25 4.5 1.5 3.75 1.75 4.5 1.5

D 5.25 4.5 3.75 3.75 6 5.25 3.75 5.25 4.5 4.5 4.5 4.5 3.75 4.5 4.5 3.75 4.5 4.5 5.25 5.25 5.25 4.5 3.75 3.75 3.75 3 1.75 4.5 5.25 4.5

`45

4.2.2. DEVELOPING RESPONSE SURFACE MODELS:

Response Surface Models are multivariate polynomial models. They typically arise in the design of experiments where they are used to determine a set of design variables that optimize a response.. Squared terms produce the simplest models in which the response surface has a maximum or minimum, and so an optimal response. Response surface models are multivariate polynomial models. They typically arise in the design of, where they are used to determine a set of design variables that optimize a response. The second order polynomial (regression) equation used to represent the response surface Y is given by Y = b0 + bi xi + bii xi2 + bij xi xj + er In order to estimate the regression coefficients, a number of experimental design techniques are available.All the coefficients were obtained applying central composite rotatable centered design using the MATLAB software package.

Tensile Strength = 184.83+9.29 * X1 + 10.04 * X2 + 11.96 * X3 + 20.21 *X40.062 * X1 * X2 -0.69 * X1 * X3 - 0.69 * X1 * X4 + 1.31 * X2 * X3 + 1.31 * X2 * X4- 4.31 * X3 * X4 -24.89 * X1 - 24.26 * X2 - 24.26 * X3 - 20.64 * X4
2 2 2 2

4.2.3. OPTIMIZATION USING GENETIC ALGORITHM: Genetic algorithms are computerized search and optimization algorithms based on the mechanics of natural genetics and natural selection. GA is very

`46

different from traditional search and optimization methods used in engineering problems. Because of its simplicity, ease of operation, minimum requirements and global perspective, GA has been successfully used in a wide variety of problems.

GENETIC ALGORITHM TOOLBOX IN MATLAB

In order to estimate the maximum tensile strength, the following steps are to be followed:

i.

Firstly the model which is developed using response surface method which relates the tensile stength and the process parameter is typed in editor window.

ii. iii.

Then the editor window is saved using any name. After opening the optimization toolbox, the fitness function is entered with the filename saved.

iv.

Then the number of the variables is entered and the lower and the upper bound is entered.

v.

If needed the parameters of the GA toolbox is changed, otherwise it is set as default and optimization is started to run

vi.

After executing the problem, the results will be displayed.

GENETIC ALGORITHM PARAMETERS

i. ii.

The parameters used for GA is given below. Population size = 100

`47

iii. iv. v. vi. vii. viii.

Length of chromosome = 40 Selection operator: Roulette method Crossover operator : Single point operator Crossover probability = 0.9 Mutation probability = 0.01 Fitness parameter : Tensile strength .

Fig 4.1. Editor window of the MATLAB

`48

Fig 4.2. GA TOOLBOX IN MATLAB

`49

4.3. OPTIMIZED PROCESS PARAMETERS OF FRICTION WELDING: The optimized result and the maximized tensile strength of the friction welding process parameters and the experimental results are shown as follows:

Table 4.3. Optimized Process Parameters of Friction Welding

S.N O 1

Process paramete r Friction pressure (ton) Friction time (sec) Forging pressure (ton) Forging time (sec)

Values of process parameter Predicted Experimental 1.525 1.5

Tensile strength (MPa) Predicted Experiment al

2 3

4.65 1.55

4.5 1.5

192.674

190

4.8

4.5

`50

CHAPTER 5
OPTIMIZING THE PROCESS PARAMETERS OF OTHER WELDING PROCESS

`51

OPTIMIZING THE PROCESS PARAMETERS OF OTHER WELDING PROCESS

The optimization process is being carried out in some of the welding process such as diffusion bonding and laser welding process. The empirical model relating the output variable and the input process parameter are taken from the available standard literature already published.

5.1. OPTIMIZING THE LASER WELDING PROCESS PARAMETERS

An empirical model was developed by G. Padmanaban et.al, to predict the maximum tensile strength of laser welded AZ31B magnesium alloy. The response surface methodology is used as an optimization tool to predict the maximum tensile strength. The experiments were conducted based on a three factor, three level, central composite face centered matrix with full replication technique. The process parameter includes, laser power, welding speed, focal position The empirical model developed by the authors relating the tensile strength and the process parameter that includes laser power, welding speed and focal position is shown as below. The process parameters and their working limits is shown in Table 5.1 Tensile Strength = 206.39-3.60 * X1 + 5.40 * X2 -0.80 * X3 +0.63 * X1 *
2 2 2 X2+0.88 * X1 * X3-2.37 * X2 * X3 +0.77 * X1 2.23 * X2 -19.38 * X3 .(

G. Padmanaban et.al.,)

`52

Table 5.1. Process Parameters and their Working Limits of Laser Welding Process Levels 0 (+1) 3000 5 -1.5 3500 5.5 -3

S.No Parameter Unit (-1) Laser 1 power Watts 2500 Welding 2 Speed m/min 4.5 Focal 3 position Mm 0

5.1.1. OPTIMIZATION USING GA TOOLBOX IN MATLAB

Genetic algorithms are computerized search and optimization algorithms based on the mechanics of natural genetics and natural selection. GA is very different from traditional search and optimization methods used in engineering problems. Because of its simplicity, ease of operation, minimum requirements and global perspective, GA has been successfully used in a wide variety of problems. In order to estimate the maximum tensile strength, the following steps are to be followed: i. Firstly the model which is developed using response surface method which relates the tensile stength and the process parameter is typed in editor window. ii. iii. Then the editor window is saved using any name. After opening the optimization toolbox, the fitness function is entered with the filename saved. iv. Then the number of the variables is entered and the lower and the upper bound is entered.

`53

v.

If needed the parameters of the GA toolbox is changed, otherwise it is set as default and optimization is started to run. After executing the problem, the results will be displayed.

GENETIC ALGORITHM PARAMETERS

The parameters used for GA is given below. Population size = 100 Length of chromosome = 40 Selection operator: Roulette method Crossover operator : Single point operator Crossover probability = 0.9 Mutation probability = 0.01 Fitness parameter: Tensile strength.

5.1.2. OPTIMIZED PROCESS PARAMETERS OF LASER WELDING:

The optimized result and the maximized tensile strength of the laser welding process parameters and the experimental results are shown as follows:

`54

Table 5.2. Optimized Process Parameters of Laser Welding S.NO Process parameter Laser power (Watts) Welding Speed (m/min) Focal position (mm) Values of process parameter Tensile strength (MPa)

Predicted Predicted Experimental Predicted Predicted Experimental (RSM) (GA) (RSM) (GA) 2520 2500 2500

5.24

5 212.25 213.712 212

-1.19

-1.65

-1.5

5.2.

OPTIMIZING

THE

DIFFUSION

BONDING

PROCESS

PARAMETERS

Experimental studies were carried out by G.Mahendran et al., (2010) on Magnesium and Copper using diffusion bonding methods. Three factor five-level, central composite rotatable design matrix method is used for the purpose of designing the experiments. Empirical relationship was developed in order to predict the diffusion layer thickness, hardness and strength of the joint using the response surface methodology. In this part of the work, the shear strength and the bond strength of the diffusion bonded Magnesium and Copper is to be maximized for the optimum process parameters.

`55

The mathematical model relating the shear strength, tensile strength and the process parameters is shown as below: Shear Strength = 58.97-4.33 * X1 -3.26 * X2 -3.35 * X3 +1.59 * X1 1.59 * X2 +3.42 * X3
2 2 2 2

Bond Strength = 74.46-4.58 * X1 -3.38 * X2 -3.48 * X3 -1.12 * X1 * X2 1.69 * X1 +1.87 * X2 -1.51* X3

2 2

The process parameter and their working limits are shown in table 5.3.

Table 5.3. Process Parameters and their Working Limits of Diffusion Bonding Levels S.No 1 2 3 Parameter Unit 1.682 425 4 10 -1 0 1 1.682 525 20 50

Bonding 0 C Temperature Bonding MPa Pressure Holding Min Time

450 475 500 8 20 12 30 16 40

5.2.1. OPTIMIZATION USING GA TOOLBOX IN MATLAB

In this problem the objective functions are Shear Strength and Bond Strength of the Diffusion Bonded Magnesium and Copper joints. So the problem falls under the multhiobjective optimization problem.

`56

GENETIC ALGORITHM PARAMETERS The parameters used for GA is given below. GA Solver : GAMULTIOBJ (Multi objective Optimization using GA) Population size = 60 Population type : double vector Length of chromosome = 40 Selection operator: Tournament Crossover operator : Single point operator Crossover probability = 0.8 Mutation probability = 0.01 Mutation function : Adaptive Feasible Fitness parameter : Shear strength and Bond Strength .

5.2.2. OPTIMIZED PROCESS PARAMETERS OF LASER WELDING: The optimized result and the maximized tensile strength of the laser welding process parameters and the experimental results are shown as follows: Table 5.4. Optimized Process Parameters of Friction Welding Process parameter / Unit Bonding Temperature / 0C Bonding Pressure / Mpa Holding Time / min Values of process Shear strength (MPa) Bond strength (MPa) parameter Predicted Predicted Predicted Experimental Experimental Experimental (GA) (GA) (GA) 449.37 450 69.742 66 80.083 81

S.NO

2 3

10.62 19.77

8 20

`57

CHAPTER 6
CONCLUSION

`58

CONCLUSION The following important conclusions are made from the above investigation: i. From the literature survey friction welding process, the research works of the welding are studied well. The possibilities of the research in welding process especially in the friction welding and the diffusion bonding process were identified. ii. Modeling techniques for the solid state welding processes were surveyed. Most of the studies showed that the conventional regression method is used for modeling solid state welding process. Very few literatures were published on the modeling of the welding process parameters using Artificial Neural Network (ANN). Finite element method is also used to model the mechanism of these welding processes and to simulate it. iii. Conventional optimization techniques such as Design Of

Experiments(DOE), response surface methodology, statistical methods such as Analysis of variance (ANOVA), correlation analysis were used as the optimization tool in many literatures, whereas very few literatures were interested to investigate on the non-traditional optimization techniques such as Genetic Algorithm, particle swarm optimization, simulated annealing, etc., iv. The optimization of friction welding, laser welding and diffusion bonding process parameters were carried for the specific objective functions. As a special case the multi objective optimization of diffusion bonding process parameters in order to maximize the shear strength and the bond strength of the welded joints.

`59

v.

The optimized results of the process parameters and the objective function of the friction welding, laser welding and diffusion bonding process are being compared with the experimental results and in some case it is being compared with the other Conventional optimization technique results which are all in good agreement.

`60

REFERENCES

`61

REFERENCES 1. Abdulkadir Cevik, Akif Kutuk. M, Ahmet Erlig, Ibrahim H. Guzelbey. (2008) Neural network modeling of arc spot welding Material processing technology. J. Vol 202, pp 137 144. 2. Cemal Meran, ( 2006) Prediction of the optimized welding parameters for the joined brass plates using genetic algorithm, Materials and design. J.vol 27, pp 356363 3. Hasan Okuyucu, Adem Kurt, Erol Arcaklioglu (2007)., Artificial neural network application to the friction stir welding of aluminum plates. Materials and design. J. Vol 28, pp 78-84. 4. Kim. I.S, Son.J.S, Park.C.E, Kim.I.J, Kim.H.H(2005), An investigation into an intelligent system for predicting bead geometry in GMA welding process. Material processing technology. J. Vol 159, pp 113 - 118. 5. Lakshmininarayanan. A.K., Balasubramanian. V, (2008),' Process parameters optimization for friction stir welding of RDE-40 aluminium alloy using Taguchi technique' Trans. Nonferrous Met. Soc. China, Vol 18, pp548 554. 6. Mahendran. G, Balasubramanian. V, Senthilvelan. T,(2009), Developing diffusion bonding windows for joining AZ31B Magnesium AA2024 aluminium alloys., Materials And Design, Vol. 30, pp 1240 1244. 7. Mahendran. G, Balasubramanian. V, Senthilvelan. T,(2010) Influence of diffusion bonding process parameters on bond characteristics of MgCu dissimilar joints, Trans. Nonferrous Mat. Soc. China, Vol. 20, pp 997-1005.

`62

8. Padmanaban.G , Balasubramanian.V (2010)., Optimization of laser beam welding process parameters to attain maximum tensile strength in AZ31B magnesium alloy. Optics and laser technology. J. Vol 42, pp 1253 -1260. 9. Parikshit Dutta, Dilip Kumar Pratihar, (2007), Modeling of TIG welding process using conventional regression analysis and neural network-based approaches, Material processing technology. J. Vol 184, pp 56-68. 10.Sathiya. P, Aravindan. S and Noorul Haq. N Mechanical and metallurgical properties of friction welded AISI 304 austenitic stainless steel, Int J Adv Manufacturing Technology 26 (2005) pp. 505-511 11.Sathiya.P, Aravindan. S, Noorul Haq. A, Paneerselvam. K, (2009), Optimization of friction welding parameters using evolutionary computational techniques, Materials Processing Technology. J. Vol. 209, pp 2576 2584 12.Serder Aragoglu, Abdullah Secgin., (2008), Sensitivity analysis of the submerged arc wewlding process parameter, Material processing technology. J. Vol 202, pp. 500-507. 13.Ravisankar. B, Krishnamurthy. J, Ramakrishnan. S.S, Angelo P.C. (2009), Diffusion bonding of SU 263, Material processing technology. J. Vol 209, pp. 2135 2144. 14.Suresh kumar.S, Krishna Kumar. J, Ravisankar. B, Angelo. P.C, (2009), Methodology to evaluate the quality of the diffusion bonded joints by ultrasonic method, Indian Journal of engineering and materials sciences, Vol. 16, pp 331-334.