@article{scholars17087,
          volume = {248},
            note = {cited By 13},
             doi = {10.1016/j.ces.2021.117205},
            year = {2022},
           title = {Developments of leak detection, diagnostics, and prediction algorithms in multiphase flows},
         journal = {Chemical Engineering Science},
       publisher = {Elsevier Ltd},
             url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85118896502&doi=10.1016\%2fj.ces.2021.117205&partnerID=40&md5=6204b2396f291f26ea466b4884e52b7e},
        keywords = {Forecasting; Leak detection; Multiphase flow; Offshore oil well production; Offshore technology; Risk management; Water pipelines, Data driven technique; Detection and diagnosis; Detection methods; Diagnostic algorithms; Flow assurance; Flow transients; Leak detection algorithms; Leak detection and diagnose; Leaks detections; Pipe flow transient, Machine learning},
        abstract = {Leak detection, diagnostics, and prediction constitute a crucial phase of the flow assurance risk management process for onshore and offshore pipelines. There are a variety of techniques and algorithms that can be deployed to address each aspect. To date, most review papers have concentrated on steady-state and single-phase flow conditions. The goal of the current review is therefore to carry out a thorough analysis of the available leak detection and diagnosis methods by focusing on (i) multiphase flow and transient flow conditions, (ii) model-based and data-driven techniques, (iii) prediction tools, and (iv) performance measures. Detailed assessment of leak detection methods based on accuracy, complexity, data requirement, and cost of installation are discussed. Data-driven techniques are utterly dependent on qualitative and quantitative data available from pipeline systems. Contrastingly data-driven techniques, model-based techniques require less data to achieve leak detection, provided that a nearly accurate base model is available. Different methodologies and technologies can be combined in order to produce the best detection and diagnosis outputs. In many cases, statistical analysis was combined with the Real Time Transient Method (RTTM), which helped to minimize false alarms. The material in this review can be used as a robust guide for the design of diagnostic systems and further research. {\^A}{\copyright} 2021},
          author = {Kumar Vandrangi, S. and Alemu Lemma, T. and Muhammad Mujtaba, S. and Ofei, T. N.},
            issn = {00092509}
}