Many industrial processes and products are affected by uncertainties arising from several factors, such as dimensional tolerances, manufacturing errors, fluctuating or unknown operational parameters. In engineering design, the uncertainties of the design parameters are transferred to the system responses, in such a way that they must be described in terms of statistical distributions rather than single deterministic values. To deal with problems affected by a large number of uncertainties and expensive simulation time, e.g. CFD analyses, it is particularly important to develop methodologies which are at the same time accurate and that can rely on a limited number of sample evaluations. In this paper, we propose an Uncertainty Quantification (UQ) method based on Reduced Order Model (ROM) and non-intrusive Polynomial Chaos Expansion (PCE) to efficiently compute the uncertainty propagation of a vectorial field of interest. The method is first applied to a RANS flow over an airfoil with uncertain angle of attack and Mach number, to evaluate its accuracy in terms of results and efficiency in terms of limited number of CFD samples evaluation, compared to other methods. The method is then applied to the Euler flow over a supersonic jet aircraft, to estimate average and standard deviation of the pressure field in each grid point, due to the propagation of uncertain operational parameters.