Robust design optimization techniques for ultra-low flying sliders

This paper addresses the topic of global optimization as it is applied to the problem of designing slider air bearings. The techniques employed are genetic algorithms and simulated annealing. Both claim to achieve global optima in the presence of many local optima. Neither method uses calculation of gradients, numerically or otherwise, to determine search directions. These methods are described briefly, and then a typical two-dimensional problem is described and analyzed. Attainment of the optimal solution for this problem is confirmed through comparison with a mapping of the objective function over optimization search space. Finally, a five-dimensional problem is described and analyzed. In the five-dimensional problem, the optimality of the solution is determined by comparison with a manually designed slider. A 36% improvement in optimality is found in this instance