Over the past years, robust optimisations have become very popular and necessary. The aim of this type of optimisation is to consider the sensitivity of the output results to small variations in the operating conditions and manufacturing tolerances. To study such sensitivities, an accurate and efficient method to quantify the uncertainties in physical processes is necessary. We present here the study and design of an S-duct, suitable for distributed propulsion configurations, and we address practical considerations of the application of robust optimisation to real-world design problems under multiple uncertainties. Two different non-intrusive Polynomial Chaos techniques have been chosen to quantify the input and output uncertainties, namely the nonintrusive point collocation and the non-intrusive spectral projection. These two techniques were implemented in two different robust optimisation problems (R1D and R2D) and their optima designs were analysed. To demonstrate the effectiveness of the robust optimisation problem formulation and analysis we compared the newly discovered optimum designs with previous non-robust optimum configurations. The results are discussed in detail and have shown the superiority of the designs when uncertainty properties were considered.