关键词: 原油密度, 预测模型, 渤海油田, 井场随钻, 岩石热解参数, 多元线性回归

Abstract: The accurate judgment of crude oil properties has a guiding effect in the optimization of oil testing mode. In Bohai Oilfield with relatively complex fluid phase, well logging and mud logging lack effective evaluation means. In order to reasonably and effectively predict the oil densities of reservoirs while drilling at the well site, according to the scatter relationship between the rock pyrolysis parameters of the rock samples and the measured crude oil densities of the oil samples, the multiple linear regression analytical approach based on the least squares method has achieved the prediction of crude oil density by building the predictive models for rock pyrolysis parameters of rock samples and crude oil densities. The models have been applied to predict crude oil densities for 616 data points from 61 wells in Bohai Oilfield, among which 139 data points from 12 wells were tested. A total of 129 points had the predictive density errors within ± 0.01 g/cm³, and the coincidence rate reached 92.80%. The models built by multiple linear regression method have a wide application prospect for crude oil density prediction.

Key words: crude oil density, prediction model, Bohai Oilfield, while drilling at the well site, rock pyrolysis parameters, multiple linear regression