MODELLING VISCOSITY BELOW BUBBLE POINT PRESSURE USING GROUP METHOD OF DATA HANDLING (GMDH): A COMPARATIVE STUDY

Below the bubble point pressure, the amount of gas dissolved in the oil increases as the pressure is increased. This causes the in-situ oil viscosity to decrease significantly. Knowledge of viscosity below bubble point is essential to many areas in the petroleum industry including reservoir and fluid production and recovery, and upgrading and transporting produced fluids. However, prediction of this parameter is difficult below bubble point pressure as the liquid undergoes a significant change in composition. These crude oils exhibit regional trends in chemical composition that categorize them as paraffinic, naphthenic, or aromatic. Because of the differences in composition, correlations developed from regional samples that are predominantly of one chemical base may not provide satisfactory results when applied to crude oils from other regions. Although some correlations show modest tolerance to assist prediction in other regions, getting accurate results with acceptable value of errors remains questionable.
The application of GMDH is not only restricted in reservoir engineering. It is critical in many areas which include accounting and auditing, finance, marketing, organizational behaviour, economics, military systems and medicine. They have several advantages compared with conventional neural networks. It has the ability to automatically organize multilayered neural networks by using the heuristic self organization method. In the GMDH-type neural networks, many types of neurons, which are polynomial type, sigmoid function type, and radial basis function type can be used to organize neural network architectures and optimum neuron architectures are selected so as to fit the complexity of the nonlinear system. The recent advancement in Soft Computing (SC) called Group Method of Data Handling (GMDH) type of Neural Networks will be able to provide a more intelligent platform for predicting viscosity below bubble point pressure with an outstanding correlation coefficient.
This paper seeks to develop a new viscosity correlation below bubble point pressure using data points taken from international oil fields. The correlation will be mapped against other existing correlations from the literature using trend analysis to verify its performance. A theoretical justification of the developed correlation will be presented. The correlation is expected to be valid for all types of crude oils within the range of data used in the study.

A series of statistical and graphical analysis relative to existing correlations will be initiated once the correlation has been formulated to provide a numerical insight on its accuracy. The comparison will validate the reliability and relevance of the proposed model to predict the viscosity below bubble point pressure.