Problèmes de prévision d'écoulements diphasiques dans l'exploitation des hydrocarbures souterrains

Two-phase flow prediction problems in underground hydrocarbon recovery projects

École Nationale Supérieure de Mécanique, Nantes

Abstract

Economical reasons favour two-phase transport in long lines, especially from marine gas/oil fields. The prediction of gas or liquid rate mean values as well as transient properties is required. Moreover, the liquid hold-up knowledge of some importance as its value (or variation in a transient) is needed for the design of separators, safety devices such as slug catchers and in a number of cases the flow configuration is also needed. Very few prediction models are accurate enough when used with field values of the physical properties. The main problems in the case of the extraction process (oil wells) are : - accurate prediction of head loss in vertical or inclined line (up to 45°) especially in the slug-flow regime. As the internal diameter is large, and the interfacial constant small, the Bond number ρL gD2 /s is large and the prevailing configuration is that of large gas-pockets and bubble swarms in the liquid slugs. The usual formulas found in most text books for the liquid holdup in the slug flow regime are inaccurate as they do not take into account the presence of the bubbles. - the calculation of the physical properties, usually obtained from thermodynamics analysis does not take into account the kinetics of phase-changes and can lead to errors in the flow estimates. Also, heat transfer has to be taken into account since the flow is neither isothermal nor adiabatic. In two-phase transport lines, the main difficulty is to predict the liquid holdup in the gas-lines usually in the stratified flow regime. The very few available models have been found totally inadequate when confronted with field values. Better models are possible when taking into account interfacial friction and interface curvature, and the turbulent properties of the flow near the walls and near the interface. These methods are sufficient for steady-state and slow transient prediction. Slug-flow occurs in oil-lines and in some occasions in gas-lines. Its prediction is largely related to the gas-slug velocity correlation and the influence of the bubble swarms in the liquid slug. The friction curve behaviour has still to be explained. For high velocity, high-pressure flow, droplet entrainment is an important phenomenon. Incipient entrainment gas-velocity is roughly known in most situations but entrainment rates are not and need further experimentation with field conditions.