Development of the pressure-time method as a relative method

Abstract

The pressure-time method is an absolute method commonly used for flow rate measurements in hydropower plants. The method determines the flow rate by measuring the differential pressure and estimating the losses between two sections in the penstock during a closure of the guide vanes. The method has limitations according to the IEC41 standard, which make it difficult to use at low head hydropower plants. The relative method called Winter-Kennedy is usually used on low head machines to determine the step-up efficiency between the old and refurbished runner. However, due to differences of the flow field in the spiral casing induce by both runners, the Winter Kennedy method might not allow estimate the flow rate similarly and thus the correct step-up efficiency. In cases where the absolute pressure-time method cannot be used because of waterway geometry limitations, the method might be used as a relative method by measuring the pressure difference between the free surface and a section in the penstock or even a point in the spiral casing without knowing the exact geometry, i.e., pipe factor. Such measurements may be simple to perform as most of the spiral casings have pressure taps for Winter-Kennedy measurements. Furthermore, the viscous losses do not need to be accurately determined if they are handled similarly before and after the refurbishment. The pressure-time method may thus become an alternative to the Winter-Kennedy method. The present paper consists in the experimental analysis of the pressure-time method accuracy used as a relative method. The experiments are performed on Porjus U9, a Kaplan prototype turbine operated under a head of 55 m generating 10 MW at full load. The flow rate is evaluated based on pressure-time measurements with different friction models and considering or not the compressibility effect. The accuracy of the flow rate evaluation method is compared using an 8-path transit-time flow rate measurement device as reference.