Model guided capacitance tomography
: a Bayesian approach to flow regime independent multiphase flow measurement

  • Ross Drury

    Student thesis: Doctoral ThesisDoctor of Philosophy

    Abstract

    Multiphase metering can provide substantial benefits to the oil and gas industry, including reduction in processing equipment and floor space at oil wells as well as improving profitability, and hence production lifetime of mature oil wells. This can be achieved by removing the need for phase separation at each oil well, which is required to perform single-phase flow measurement. If a multiphase flow measurement system can be developed, with an acceptable uncertainty, and with the capability of metering all flow regimes which occur in oil and gas pipelines, the overall extraction and distribution process can be simplified greatly. This would in turn provide significant financial benefits to the industry.

    Providing a single multiphase measurement system which can achieve these goals has proven difficult, considering the complex¬ity of the flows that occur. The three main criteria to achieve a suitable system are: a) suitability for sub-sea implementation, b) the ability to operate on all flow regimes, and c) to not be reliant on empirical cor¬relations due to their limitations. Electrical Capacitance Tomography (ECT) is an example of an instrument suited to oil and gas applications due to its non-intrusive and non-invasive nature. For slug flows, ECT is a capable cross-correlation based multiphase flow meter, but for segregated flows, such as annular, its capability is compromised, due to a heavy reliance on correlations, and by its principal of measuring fluid interfaces only, which restricts its ability to recover individual phase velocities.

    Three studies are conducted with the aim of developing ECT as a multiphase flow measurement system, focusing on providing a solution which can be applied to all flow regimes and does not rely on empirical correlations. The first study includes experiments on vertical slug and annular industrial scale flows, whereby a method is proposed that uses the similarities between wave properties measured for both flow regimes, allowing self derivation of correlation parameters using slug flow data. The results show that a power law approximation can describe similarities in wave properties, and by using the method outlined annular flow properties can be estimated within plus or minus 50%without any empirically derived parameters.

    The second study is on improving cross-correlation metering of slug flows. Experiments are conducted on horizontal oil and gas industrial scale flows. Velocity discontinuities are identified by their structure at the slug front through observation of tomographic images. Three distinct slug front structures are categorised and tested to analyse their affect on the cross-correlation measurement. Individual measurement methods are prescribed depending on slug front structure and a method of selective based cross-correlation is described. Results show that by using this method the superficial mixture velocity can be obtained within an error margin of +/− 5%, an improvement on previous studies where +/− 10% is a typical benchmark.

    The final study is an implementation of a particle filtering system, incorporating tomography and a computational fluid model. Exper¬iments are conducted on a vertical lab scale flow loop where gas is applied through a column of oil. A simulation database is created using an Euler-Euler model which is referenced through a particle filtering algorithm. Data from ECT and pressure transducers are used to update the recursion and allow an estimation of gas flowrate over time, without the use of empirical correlations, and with the potential to be applied to all flow regimes. The results showed that the mean gas superficial velocity of a range of tests can be estimated within an error margin of +/− 5%.
    Date of AwardAug 2019
    Original languageEnglish
    Awarding Institution
    • Coventry University
    SponsorsNational Engineering Laboratory
    SupervisorJames Brusey (Supervisor), Andrew Hunt (Supervisor) & Elena Gaura (Supervisor)

    Cite this

    '