Numerical and experimental study of natural backfill of pipeline in a trench under steady currents

After a sub-sea pipeline is laid in a trench excavated on a sandy sea bed, the sand around the trench will be washed into the trench by the flow, leading to natural backfill. Natural backfill is beneficial to the stability of the pipeline and cost saving. In this study, natural backfill of pipeline trench under steady currents was investigated experimentally and numerically. Experimental tests were carried out in a water flume of a size of 0.4 m in width, 0.6 m in height and 25 m in length. The model pipeline with a diameter of 5 cm was placed in a V-shape trench. The direction of steady current was perpendicular to the pipeline. Tests were carried out in both clear-water scour and live-bed scour conditions. The bed profiles at different stages of backfill process were measured by a laser profiler. It was found that the upstream part of the trench was backfilled faster than the downstream part. In the early stage of the backfill process, sand in front of the pipeline was washed into the trench very fast. The top part of the sand behind the pipeline was washed away faster while the lower part moved towards the pipeline due to strong vortices. Two-dimensional scour model developed by Zhao and Cheng (2008) was used for simulating the backfill process numerically. This model was validated against van Rijn's (1986) navigation channel migration experiments and good agreement between experimental data and numerical results was achieved. Numerical simulation of the pipeline trench evolution was carried out under the same conditions used in the laboratory tests. The process of the backfill simulated by the numerical method agreed qualitatively with the test results. The comparison between the numerical and the test results showed that: (1) the simulated backfill rate was greater than the measured one in the upstream side of the pipeline; (2) the sand dune downstream the pipeline was washed away slower than the experimental results, and no backfill was observed. The discrepancy between the experimental and numerical results may be attributed to the fact that the empirical formulae used for predicting the bed load and the reference concentration of suspended load were derived from fully-developed straight channel flow tests, while the velocity varied dramatically along the bed in the cases of this study.