Effect of uncertainties in the estimation of dynamic coefficients on tilting pad journal bearings

It has been identified that small variations in the pad clearance and preload of a Tilting Pad Journal Bearing lead to important variations in their dynamic coefficients. Although this variation trend is already identified, a more robust statistical analysis is required in order to identify more general tendencies and quantify it. This work presents a framework that helps to identify the relation between the manufacturing tolerance of the bearing (reflected in the pad clearance and preload) and the expected variations on the dynamic coefficients. The procedure underlies the adoption of a surrogate model (based on Kriging interpolation) trained by any deterministic model available to predict dynamic coefficients. The pad clearance and preload are considered uncertain parameters defined by a proper probability density function. All statistical quantities are obtained using stochastic simulation, specifically adopting a Monte Carlo simulation employing the surrogate model. The framework is illustrated through the study of a five pad bearing.