Modeling and Experimental Validation of a Pico-Scale Francis Turbine for a Self-Powered Water Disinfection System

Access to both electricity and clean drinking water is challenging in many remote communities. A self-powered water disinfection system, currently under development, can potentially address this challenge. In the proposed design, energy from water flowing through the system is harnessed using a pico turbine (nominal output power of 60 W) and used to power an electrochemical disinfection process. The characteristics of turbines at the pico-scale (less than 5kW) required for this system are not well researched, and off-the-shelf designs are either too bulky or too inefficient for this application. This paper presents a model developed to evaluate a new class of efficient pico-scale Francis turbines for this water disinfection system. A computational fluid dynamics (CFD) model of the turbine was developed in ANSYS^® CFX^® 17.1. The CFD model exploits the rotational symmetry of the turbine and draft tube fluid regions to reduce the computational cost in terms of time and memory. The turbine model is coupled with models of the electric generator and electrochemical cell to determine the balanced operating points. When validated against experimental data, the combined model showed good predictive ability despite its low computational cost: the modeled turbine efficiency is within 5% of the measured values across the operating range of the device. The current turbine design has a hydraulic efficiency above 60 % in its operating range, which is high for a compact turbine at this scale. The combined model was used with a parameterized version of the turbine geometry to identify key performance sensitivities, particularly with the blade trailing edge angle. Turbine efficiency was improved by more than 2 % across the allowable flow rates. The low computational cost of the combined model made it well suited for iterative design optimization, supplanting the need for lengthy experimental trials. Overall, the modeling approach presented here shows good promise for use in picoturbine design.