In this paper we present the application of the technique of artificial neural networks to the thermodynamic characterization of reservoir fluids.The development of a demarcation system that allows the prediction of reservoir fluid type, (black oil, volatile oil or gas condensate) based on field information from production tests and validated PVT analysis is described.

Neural networks are coarse models of the human neural system an they have been used successfully in solving practical problems of pattern recognition. In this work, a carefully selected data set of eighty (80) gas-oil ratios API gravity and percentage of C7+ composition values, corresponding to black oil, gas condensate and volatile oil were used for training the neural network. The training objective was the recognition of fluid type. The trained system, was validated with two hundred and fifty (250) PVT laboratory test from eastern Venezuela showing an excellent performance in fluid type characterization. The reservoir fluid type forecasts obtained with this approach were found to be outstanding when compared to those made through the application of classifying criteria proposed by other authors. It is concluded that the proposed demarcation system is superior to the previously reported techniques.

Keywords:
artificial neural network, api gravity, neuronal network, reservoir, volatile oil, classification, upstream oil & gas, pvt measurement, application, phase diagram
Subjects:
Fluid Characterization, Formation Evaluation & Management, Information Management and Systems, Phase behavior and PVT measurements, Fluid modeling, equations of state, Artificial intelligence, Neural networks
Copyright 1994, Society of Petroleum Engineers
You can access this article if you purchase or spend a download.