Statistical-based evaluation of direct methods application in Francis turbines design

Four applied methods in Francis turbine direct design have been surveyed in this paper. These four design methods include De Siervo, Mosonyi, Lindestrom, and Lugaresi approaches. The reliability of these methods has been evaluated by the redesign of the Francis turbine at three powerplants: Abbaspour, Masjed-e-Soleiman, and Marun. This evaluation is based on a statistical study of obtained results. The studied statical population consists of 12 performance and geometrical parameters of Francis turbine. Three performance parameters of this statical population contain output power, rotational speed, and specific speed of the turbine. The rest are geometrical parameters used for runner, spiral, and draft tube design. The obtained results show that De Siervo approach to the calculation of the 67 % parameters has precision equal to or more than the other three approaches. The maximum computational error of Mosonyi, Lindestrom, and Lugaresi approaches is greater than 150 %. Highlighted result of the current study belongs to the presentation of a new synthesized algorithm for Francis turbine direct design. This algorithm can provide the initial guess of various turbine sizes for numerical designs by an error less than 10%.