UNCERTAINTY AND SENSITIVITY ANALYSIS IN A LOW-HEAD HYDRO SYSTEM

This paper presents four different methods to estimate the uncertainty in a low-head hydropower system involving the parametric study on the performance of cross-flow turbines. The study considered for uncertainty estimation here, involved the computation of maximum experimental efficiency using the measured data of shaft torque, rotational speed, water flow rate, total head, and water temperature. Commercial brochures published by the turbine manufacturers often fail to mention the uncertainty in their claimed maximum efficiency. This often leads to an overestimate in the maximum efficiency, hence the importance of estimating the uncertainty. If uncertainty is estimated by using only one method, there will not be any verification for the estimated uncertainty. Therefore, in this paper four different approaches are adopted for estimating the uncertainty so that they can be mutually verified. The four methods are: confidence limit analogy, critical path method analogy, multivariate Taylor series method, and sensitivity analysis. All the methods provide uncertainty estimates in both the positive and negative directions except the confidence limit analogy method which yields uncertainty estimate only in the negative direction. In spite of the totally different approach used in each of the methods, the uncertainty estimates compare reasonably well.