Well-test analysis of multi-phase flow from two layers communicating through the wellbore

Abstract

Conventional buildup analysis techniques are based on analytical solutions to the diffusivity equation that are subject to well-known assumptions. During actual well tests multi-phase flow in several layers with different properties is often observed. Applicability of the conventional wellness analysis technique to the interpretation of complex cases involving flow of oil and gas in two commingled layers was investigated in this research. A commercial black oil simulator with variable bubble-point was used to generate pressure and production data. A single-well radial reservoir model was utilized in the simulations. After a period of production the well was shut-in for a buildup test. A total of 283 tests was conducted in the first part of the study. A statistical analysis of results was performed. The study shows that calculated values of average effective oil permeability and skin factor depend on a wide range of parameters, such as contrast in layer properties, production mode, and production rate and time. Both average oil permeability and skin can be either over- or underestimated under multi-phase flow conditions in two commingled layers. The greater the contrast in layer properties, the less accurate results of buildup analysis we should expect. Skin was underestimated in many tests with constant oil rate production and in the majority of the tests with constant BHP production prior to shut-in. A combination of constant BHP production, high contrast in layer properties, and long production time results in overestimation of average permeability to oil. High oil rates reduce the calculated value of oil permeability. Several production forecasts were made using calculated values of permeability and skin in a single-layer reservoir model. A single-layer analogy produces acceptable results if the contrast in layer properties is not too great. If the contrast is high the forecasts generated by single-layer and two-layer reservoir models can differ significantly.