Comparing pure CFD and 1-D solvers for the classic water hammer models of a pipe-reservoir system

Various system representations of transient events are possible, each with respective strengths and weaknesses. This paper compares results from a water hammer experiment (courtesy of Bergant et al.) to a set of simulations using a series of numerical models. The experimental rig includes two tanks connected by a long pipe under laminar flow. The water hammer event is caused by the rapid closure of a terminal valve. The transient flow is computed by means of two solvers. The first one is a basic one-dimensional (1D) Method of Characteristics (MOC) solver including unsteady friction terms based on instantaneous acceleration. The second is a 3D commercial CFD solver of the pipe. The 3D pipe model captures part of the damping caused by unsteady wall friction, but can artificially capture the pressure signal's phase shift. The challenges involved in setting up the CFD models are discussed in detail. The longer-Term goal of this work is to attempt more complex 1D-3D couplings, but there are notable challenges even in simple cases.