Date of Award

1987

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Abstract

The use of Systems Analysis for designing and evaluating wells flowing hydrocarbon mixtures seems to create no computational problems when considering the inflow components, i.e. reservoir and perforations. However, when a gas-condensate system is considered, the outflow component calculating the total pressure losses in the tubing creates a computational problem. This computational problem involves the estimation of the phase properties of the gas-condensate fluid. Thus, the fluid system undergoes phase changes as it moves up the tubing. Attempts by other authors have been made to overcome this by either using a single-phase or two-phase flow model. The single-phase flow model adjusts the dry gas specific gravity by adding the condensate fluid into the gas stream. This model is adequate for wells producing small amounts of condensate. At times, condensate production can be quite high implying distinctive flow regimes (slug, transition, and mist) are present. Consequently, a two-phase model is needed to adequately model these types of flow. The phase properties in the two-phase flow models already available, are computed assuming the liquid phase behaves as a "black oil". But condensate fluids and "black oils" possess distinctive differences, in that "black oils" evolve gas during pressure reduction and condensate fluids are condensed from the gas phase during pressure reduction. Thus, a method is needed for predicting pressure losses when condensate fluids flow through pipes, where PVT properties are evaluated using a condensation model. It was decided that a compositional model was the best approach to predict pressure losses for condensate fluids flowing in vertical pipes. PVT properties were computed using both the Soave-Redlich-Kwong and Peng-Robinson equations of state. Thus, a knowledge of the actual flow stream composition normalizes the equation of state used, which can then predict the liquid/gas properties at any pressure and temperature. Pressure gradients along the tubing string were approximated using existing two-phase flow correlations (Hagedorn and Brown, Duns and Ros, Orkiszewski, and Gray) with the compositional model replacing the PVT property calculations. Actual case histories were used to test the model by matching these measured flowing pressure traverses.

Pages

295

Share

COinS