Predictive Modeling and Detection of Oil Pipeline Leakage

Oil leakage has been a part of oil industry operational problems. It   normally occurs as a result of a crack in the pipeline due to fatigue and/or corrosion, and non-detection of it early enough, has compounded the problem, leading to progression into  explosion and/or spills and occasionally fire!, with the attendant huge  costs and destruction.

Current leak prediction and detection methods include right-of-way inspection/surveillance and computational pipeline monitoring (CPM) which involves extensive field data collection and analyses. However, based on the difficult/swampy terrain of most pipeline systems and the low sensitivity of most oil field units of instrumentation, only ”worst case” scenarios could be predicted/detected, thus small/early leaks still occur undetected in most cases. Thus, in this paper, a novel approach was followed to develop simple but efficient model for early prediction/detection of oil leakage, by employing the principle of straight pipe equivalent in determining the pressure drop of pipe cracks, modeled as pipe fittings (discharge valves, constrictions, enlargements). All that is needed as inputs to the model are: the pipeline inlet (departure) and outlet (arrival) flow rates and pressures, as well as an accurate estimate of the pipeline system friction factor. With these, the oil leakage quantity, the location in the pipeline where it occurs (leakage point), and the extent of pipe damage (leakage area), can be determined by using the model, without the delays, hassles, inefficiencies, elaborate analysis and costs  associated with the current methods. It was found that to ensure high sensitivity, accuracy, reliability and robustness of the model prediction, smaller units, like Pascal, should be used to replace the current oil field units, like psi, or simply incorporated as an adjunct.