In this paper we compare predictions obtained using the LedaFlow Slug Capturing model with a new and unique set of slug flow experiments. The most important novelty of these experiments is that they were conducted in a very long pipe (766 meters), so that slug length evolution could be studied over long distances. The presented slug capturing model contains both improved physical models and a completely new numerical scheme compared to previous publications. These developments have led to some major breakthroughs with respect to both prediction accuracy and computational speed, and was released as part of LedaFlow in 2021.
The experiments were conducted in the Large Scale 8" loop at the SINTEF Multiphase Laboratory using oil, water and Nitrogen. The loop consisted of two main test sections, with pipe inclinations 0 and 0.5 degrees, with approximate lengths of 380 m. The experiments shown in this paper were performed at a system pressure of 45 bara, and the prevailing flow regime in most experiments was slug flow.
The experiments show that the average slug length tends to increase with distance from the inlet, and we explain some of the most important physical mechanisms that govern this evolution. We further show that by including these mechanisms in the slug capturing model, we can reproduce the observed behaviour very well. We also show that numerical accuracy is a key prerequisite for obtaining accurate results for "moderate" grid sizes.
In the model/experiment comparison, we assess the agreement between the model and the measurements in terms of average pressure drop and liquid holdup, as well as slug characteristics. We also compare the holdup/pressure drop predictions, entrained gas fraction in slugs and Taylor-bubble velocity to results predicted by the Unit-Cell Model. Finally, we examine the model's sensitivity to the grid size. The latter issue is important, as selecting an excessively coarse mesh can cause too much numerical diffusion to resolve the wave growth and initiation of slugs.
The comparisons show that the slug capturing predictions are in good agreement with the measurements in terms of pressure drop, holdup and slug characteristics, and that we obtain excellent agreement with the Unit-Cell Model. Furthermore, the slug capturing results are shown to converge well as the computational mesh is refined, which is reassuring. The results suggest that a grid size of 10 D or less is typically sufficient to resolve hydrodynamic slug flow with the new slug capturing model.
