Development of New Pvt Correlations for Indian Oil Fields Using Alternating Conditional Expectation Method

Chandra P. Verma
INTEG, GEOPIC, Dehradun, India

The objective of this study is to develop a set of new PVT correlations for estimating the properties of the Indian crude oil. This work systematically uses the method of alternating conditional expectations (ACE), a non parametric statistical regression method to alleviate the main drawback of parametric regression, i.e., the mismatch of assumed model structure and actual data. In such non-parametric regression a priori knowledge of the functional relationship between response variable and predictor variable is not required to develop a correlation. A total of 535 PVT analysis of crude oil samples from 170 different reservoirs of ONGC fields have been used to develop the new correlations to estimate oil formation volume factor ( Bo), bubble point pressure (Pb) and oil viscosity ( μ o ) and ) and solution gas oil ratio Rs in terms of reservoir pressure (Pr), temperature (T), oil density (ρo) gas specific gravity (γg ) .Such PVT correlation developed can solve problems for PVT parameters for the fields where sample is not feasible/available. It can reduce the number of samples from deep wells, which will result in lowering the cost. Average relative errors for Bo, Pb , Rs and μ o, are obtained -0.97 %,3.9% ,13 %and 2.1 % respectively Errors in the predicted Bo ,Pb , μ o and Rs are significantly low thereby rendering their applicability in practice . These correlations developed using the ACE algorithm can save calculation time in predicting Pb, Bo, Rs and μ o with more reliability than equations derived using conventional regression method.

Presentation GEO India Expo XXI, Noida, New Delhi, India 2008©AAPG Search and Discovery