Abstract:
Discharge (flow rate) measurement in hydro-power station is one complex measurement application which demands very accurate methods. For years together the accuracy of flow measurement remained unsatisfying, until the high speed digital age took over the analog methods. Latest state of art technologies like the high speed signal processing and the smart sensor design being used in the application of ultrasonic transit time flow meter for discharge measurement improved the accuracy to a good extent. The promised accuracy using an eight-path ultrasonic transit time flow meter is ± 2%.
The complexity of fluid dynamics with respect to closed pipe flow influences the measurement-of discharge. Though various correction methods have been tried over the past 20 years, yet the accuracy demands have not been met. Various errors associated with the application of ultrasonic transit time flow meter to closed pipes are Reynolds number: error, surface roughness error, installation and sure' errors, protrusion effect, roundness a ect and the numerical integration error. The focus of this dissertation work is to improve the accuracy of the ultrasonic transit time flow meter applied to closed pipes by minimizing the . numerical integration error, which influences the accuracy of flow measurement significantly.
The mathematical expressions of'nrofiles in combination with the finite element analysis techni ues have been used to study the umerical integration error as influenced by the kind of profile and the numerical integration method used. These findings formed the basis of applying artificial intelligence techniques to minimize numerical integration error. A neuro-fuzzy model is designed to work for the discharge measurement using transit time ultrasonic flow meter. The model yields a zero error for known profiles and restricted the numerical integration error for new profiles to an as low as 0.5 %.
