Optimal Shape Design of Steadily Loaded Journal Bearings using Genetic Algorithms

Abstract

This thesis investigates the use of genetic algorithms to optimize the shape of an ungrooved fluid-film journal bearing under steady load and steady speed so that it can carry the maximum possible load while constrained to maintaining an adequate minimum film thickness. The shape of the bearing sleeve is assumed to vary only in the radial direction, and the shape is represented by linear interpolation of three film thickness design specifications along the bearing sleeve. Each set of film thickness specifications in the design space is encoded into a binary string called a chromosome, and a set of chromosomes makes up a current generation. The bearing load is calculated for each chromosome, and the genetic algorithm creates a new generation based on these loads using elitism selection and chromosome crossover and mutation operators. Several case studies are presented to investigate the effect of bearing geometric specifications on the resulting optimal shape. For a given bearing diameter and bearing length, a random generation of chromosomes is first constructed. Using recommended chromosome crossover and mutation probabilities, it is shown that randomly created starting sets will eventually converge to a common shape, suggesting that a global optimum may have been achieved. Additional parametric studies show that the load results are dependent upon the mutation operator to achieve global optimums. Further comparisons show that the load carried by the optimal shape bearing can be much greater than that carried by a conventional cylindrical bearing. 93 3.3.12, b Film thickness distribution with second random set of chromosomes (sv=94872) after 1000 generations, CASE #2, N_chrom = 40. 94 3.3.12, c Film thickness distribution with third random set of chromosomes (sv=30157) after 1000 generations, CASE #2, N_chrom = 40. 95 3.4. 1 Bearing load evolutions for each of three randomly selected sets (D=20.0mm, N_chrom =20), CASE #3. 96 3.4.2 Plot of Power Loss with three randomly selected sets (D=20.0mm, N_chrom=20), CASE #3. 97 3.4.3 Plot of Pmax with three randomly selected sets (D=20.0mm, N_chrom =20), CASE #3. 98 3.4.4, a Film thickness distribution with first random set of chromosomes (sv=44258) after 1000 generations, CASE #3, N_chrom = 20. 99 3.4.4, b Film thickness distribution with second random set of chromosomes (sv=28257) after 1000 generations, CASE #3, N_chrom = 20. 100 3.4.4, c Film thickness distribution with third random set of chromosomes (sv=0741 1) after 1000 generations, CASE #3, N_chrom = 20.

Geometrical Optimization of Hydrodynamic Journal Bearings with Validated Simulations and Artificial Intelligence Tools

“…The present work technique is compared with two existing methods presented in [2,6]. The reference design is calculated using the Raimondi-Boyd diagrams, this solution is very cautelative in terms of penalties; indeed, the maximum pressure and the minimum film thickness are not closer to the limits.…”

“…Previously, investigations had been done in this field by Hirani [4,5], Boedo [6] and Ghorbanian [7], but, nevertheless, significant opportunities remain to improve the optimization procedure, like introducing a faster and more precise performance function (validated) and using artificial intelligence tools for the minimization of the function. In [4], there are two objectives, non-dimensional and normalized.…”

“…The technique adopted for the minimization is the Genetic Algorithm (GA, Non-SortedGenetic-Algorithm) while the performance function is the approximated solution proposed in [5]. In [6], GA technique is adopted to minimize power loss and mass flow (in this case the objective and the performance are the same function). This work introduces a more precise and validated performance function to reduce the number of iterations.…”

A parallel computing application of the genetic algorithm for lubrication optimization

“…Recently, GAs for improved fitness are popular in many tribological optimizations [10][11][12][13][14] for their ability to obtain global optimum with high probability. None of these studies applied the GAs in parallel.…”

Journal Bearing Optimization Using Nonsorted Genetic Algorithm and Artificial Bee Colony Algorithm

“…The scaling and weighting factors are arbitrarily chosen, which determine the generation of a single objective space in which the solution is found. The last relevant work in this field, made by Boedo and Eshkabilov in [16], uses the genetic algorithm technique to optimize the shape of statically loaded journal bearings.…”