Experimental and Numerical Study of Oil-Water Dispersions at the Inlet of a Pipe Separator

Abstract

In the production of petroleum, it is normal to not only produce oil and gas, but also to produce water. Over the lifetime of the fields, the amount of water produced will normally increase. The additional water will later have to be separated from the oil and gas, either at the surface or down on the seabed. If the processing facilities cannot handle these higher rates of water, the producing line has to be choked back, which will result in a lower production of oil. Increasing the efficiency of oil-water separators and develop subsea separation solutions will therefore allow to maintain, or increase the fields production rate, hence reducing cost. The separation performance is affected, among other things by the droplet size. This thesis is focused on studying the characteristics of oil-water flow and droplet size distribution at the inlet of an oil-water pipe separator for varying rates of oil and water.

Experimental tests were performed at the inlet section of an inline oil-water pipe separator developed at NTNU (SUBPRO project 2.9). Four main tasks were performed. First, measurements of the physical properties of the oil and water were performed. Second, the flow pattern map of the system was determined varying the flow rates and water cuts. Third, the droplet size distribution was computed by analyzing and processing pictures of the flow taken with a PVM probe. The analyzing of the images were performed manually and automatically using a computational routine written in Matlab. The droplet size distribution was quantified for several flow rate combinations, two water cuts, two probe positions and two valve openings. Additionally, the droplet size distribution data was compared with four statistical distributions and models; Log-Normal Distribution, Upper-Limit Log-Normal Distribution, and different breakage models.

It was found that the flow system with the valve fully open had larger droplet sizes for flow rates in the higher range. However, with the valve 50% open the droplet sizes got smaller for flow rates in the higher range. With more water present in the system the droplet size decreased. The droplet sizes for oil-in-water were smaller than the droplet sizes for water-in-oil. The Brauner breakage model and the Upper- Limit Log-Normal Distribution gave the best representation of the data. The experimental data gathered has a great value for future validation of multiphase and dispersion models, thus improving their predictability and quantifying and reducing their inaccuracies.