Shaft Alignment Optimization With Genetic Algorithms

Abstract

A solution to the shaft alignment problem is a set of prescribed bearing offsets that ensure an acceptable load distribution among the shaft-supporting bearings. Acceptable load distribution implies not only all positive bearing reactions under all operating conditions of the vessel but also an acceptable relative-misalignment between the shaft and the bearing. In a marine environment, the difficulty is not in finding a single suitable solution to the above criteria, but rather in defining the optimal set of solutions capable of accommodating the extreme bearing disturbances - resulting mainly from hull deflections and thermal deviation. As the problem is stochastic, with an infinite number of satisfactory bearing offsets, it is appropriate to apply the Genetic Algorithm (GA) optimization procedure to search for the optimal set of solutions, rather than rely on the plain trial and error approach or some of the step-by-step conventional search algorithms. With an ability to conduct a parallel search throughout the solution space, the GA is particularly well suited for the problem at hand, as it has the capacity to simultaneously provide multiple sets of bearing offsets that satisfy loading conditions at bearings.