Hydrodynamic slug flow is the prevailing flow regime in oil production, yet industry still lacks a comprehensive model, based on first principles, which fully describes hydrodynamic slug flow. This paper describes a very simple, time-dependent, two-phase (gas-liquid) model which is capable of producing hydrodynamic slugging from first principles. The slug model correctly predicts transition from stratified to slug flow via an interface instability. The model is capable of producing slug lengths and frequencies, as well as slug void fraction, from first principles. Also, flow regime transitions are effectively captured. The model also captures slug initiation on uphill pipe sections and slug decay on downhill sections. Because of its simplicity, the model runs extremely fast compared to other multiphase flow simulators Such a simple model could be a very excellent jumping off point for building more complex models capable of predicting slug distributions from first principles, without any need for additional user input. No such model currently exists.


While hydrodynamic slug flow is an inherently transient phenomenon, it has historically been modeled as an averaged ‘pseudo steady-state’ which ignores the fundamentally transient nature of the flow. Even in instances where the individual slugs are introduced and tracked in a Lagrangian frame (so-called ‘slug tracking’)1, the results have been somewhat disappointing, in that the ultimate slug distributions are heavily influenced by user input. This paper examines another approach – where the fundamental transient nature of hydrodynamic slug flow is accounted for in the model. In order to formulate a model for slug flow, we must first develop a ‘point model’ for an individual pipeline segment, or computational cell, in a pipeline. This pipeline segment must then be joined with other pipeline segments upstream and downstream of it to form a ‘steady-state’ model for the pipeline.

This content is only available via PDF.
You can access this article if you purchase or spend a download.