Field verification of single leak detection method based on transient pressures using optimization technique in in-situ irrigation pipeline system including branch junctions and diameter changes

Abstract

For decades, leak detection methods utilizing pressure variations (transient test-based technique: TTBT) have been recognized as convenient and accurate for pipeline leak detection. However, their effectiveness in long-distance complex structured field pipelines has not been fully demonstrated, presenting significant challenges for practical implementation. This study, for the first time, tested the effectiveness of inverse transient analysis (ITA), a type of TTBT, on detecting a single-point leak in the Ogura irrigation pipeline system in Sado, Niigata, Japan. The system spans approximately 18 km and features several branches and diameter changes. We proposed a new ITA method using the transient model and the comprehensive learning particle swarm optimization (CLPSO) as the optimization technique. In the verification experiments, pressure variations were induced by closing the downstream valve under two different initial downstream flow rates (approximately 0.032 and 0.093 m 3 /s) in scenarios without a leak, with a leak near the downstream end of the main pipeline, and with a leak on the branch pipeline. The calibration of the transient model with the wave speed distribution as parameters was conducted prior to leak detection, demonstrating pre-estimation of this wave speed distribution is essential for accurate leak detection. The application of the proposed ITA for measured pressure variations resulted in averaged estimated leak location errors along the total length of the pipeline system with the standard deviations as follows: 0.217 ± 0.038% near the downstream end of the main pipeline for a 0.032 m 3 /s downstream flow, 0.978 ± 0.456% for a 0.093 m 3 /s flow, 1.843 ± 1.815% on the branch pipeline for a 0.032 m 3 /s flow, and 0.880 ± 0.560% for a 0.093 m 3 /s flow. In cases of the leak near the downstream end of the main pipeline, smaller steady flow rate resulted in smaller variability and higher accuracy in the estimated leak location. Conversely, for the leak in the branch pipeline, larger steady flow rate was better for the estimated leak position. This suggests that the optimal steady flow rate for accurate leak detection varies depending on the leak location. The leak sizes can be expressed as ratios of the estimated leakage discharges to the actual average irrigation water flow rate, resulting in values ranging from 4.86% to 8.58%. From these results, the proposed leak detection method is expected to contribute to reducing the effort and cost for irrigation engineers to narrow down potential leak areas in pipeline systems.

Bibliographic Information