Investigation of Rapid Filling of Empty Pipes

Abstract
Water mains are periodically subjected to maintenance, in which drain and refill of pipeline sections is a required practice. There is the possibility of damage to the water main during such operations, even when air relief valves and vacuum breakers are properly placed or the drain and refill are conducted slowly. Current research on this and related applications has resulted in different numerical models to simulate the rapid filling process (Liou and Hunt, 1996; Izquierdo et al. 1999) based on the assumption that a vertical interface can represent the water inflow front. Models to simulate the rapid filling of stormwater sewers have also used this assumption, as exemplified by Zhou et al. (2002).
Research on the rapid filling of initially empty pipes was conducted at the National School of Engineering of Tunis (Tunisia) and the University of Michigan. This research was conducted to gain a better understanding of the filling process in closed pipes and thus try to optimize the filling operations and minimize the risk of damages. An experimental investigation was performed to characterize the inflow front and assess the validity of the vertical front assumption. In these experiments, the upstream reservoir head and the pipe slope were varied, and the inflow front was measured and characterized with the aid of digital camcorders. A piezo-resistive pressure transducer was also used to record the pressure during the filling process.
Among other findings, this investigation demonstrated that inflow fronts do not always close the pipe cross-section, invalidating the vertical interface assumption even when proposed criteria for this assumption to be applicable have been satisfied. Two different numerical models, based on the rigid column analysis proposed by Liou and Hunt (1996) and the shock-capturing technique developed by Vasconcelos et al. (2006) were used in this investigation to assess their predictive abilities in simulating the rapid filling of empty pipes. This assessment showed that while the rigid column model reproduces certain aspects of the observed flow, the shock-capturing model was capable of resolving more of the essential features of the flow.
This paper is only available in PDF Format: View full text PDF