The ¿lling with liquid of an initially empty pipeline and its counterpart, the draining of an initially liquid-¿lled pipeline, are of great interest due to the many practical applications. Several potential problems may occur, of which water-hammer and slug impact are the most important. To investigate the ¿lling and emptying processes, di¿erent mathematical models have been proposed, in which a common assumption is that the water column evolves with unchanged front and/or tail. This is a reasonable assumption for small-scale systems, particularly in cases with relatively high upstream pressure head and low downstream resistance. However, it is not clear whether this assumption is applicable to large-scale systems. This issue is of high importance for the development of air pockets and gravity currents in pipelines during ¿lling and draining processes.
This study presents the experimental results of the ¿ow behaviour during the rapid ¿lling and emptying of a large-scale pipeline. The experimental apparatus was designed and built at Deltares, Delft, The Netherlands, as part of the EC Hydralab III project. Di¿erent from other laboratory studies, the scale of this experiment is close to the practical situation in many industrial plants. The test rig includes a variety of components (e.g. tanks, ¿ow meters, valves, pipes of di¿erent materials) and the operation procedure is rather complex. The ¿ow behaviour is measured by various instruments and hence a thorough hydrodynamic analysis is possible. All these features make the current study particularly useful as a test case for real ¿lling and draining situations.
In the ¿lling of an initially empty pipeline, the focus was on the overall behaviour of the lengthening water column and the water-air interface evolution. In the emptying of an initially water-¿lled pipeline, together with the hydrodynamics of the shortening water column, the shape and behaviour of the water tail (air-water interface) was investigated. Thirteen di¿erent combinations of initial upstream driving air pressure and downstream valve resistance were tested. The in¿uence of these two factors on the out¿ow rate is clari¿ed. It was con¿rmed that both the in¿ow front in ¿lling and the out¿ow tail in emptying do not entirely ¿ll the pipe cross section. Shape changes occur at both the water-air and air-water interfaces. Although the ¿ow regime transition is a rather complex phenomenon, certain features of the transition pattern are observed and explained qualitatively and quantitatively.