岩石相测井识别技术在阿姆河右岸A区的应用

由于土库曼斯坦阿姆河右岸A区卡洛夫阶-牛津阶碳酸盐岩岩石类型繁多，岩、电关系复杂，储层非均质性强，使得该区储层发育程度及横向展布的预测十分困难。为此，通过对关键井岩心、铸体薄片等地质资料的观察分析，按储层发育程度，将其岩石相细分为颗粒泥晶灰岩相、微亮晶颗粒灰岩相和生物礁灰岩相；然后确定出各岩石相的相关测井响应特征，并分析岩石相在电性上的表现特征，归纳出识别该区岩石相的测井识别标准；再采用主坐标分析、小波分析和多参数逐步判别分析等一系列数学分析技术，建立岩石相与测井相之间的数理关系模型，从而形成了一套利用测井信息自动识别岩石相的技术。结果表明：XVm层以微亮晶颗粒灰岩相和生物礁灰岩相为主，XVp层则颗粒泥晶灰岩相和微亮晶颗粒灰岩相发育，XVac层岩石相类型以颗粒泥晶灰岩相为主；储层发育程度以XVm层最好，XVp层次之，XVac层最差。根据该方法所预测的储层发育程度与测试结果吻合度较高，样本正判率在80%以上，这为该地区的油气勘探提供了技术支撑。

Application of logging lithofacies identification technology in Block A of the Right Bank of the Amu Darya River

The Callovian Oxfordian carbonate reservoirs in Block A of the Right Bank of the Amu Darya River，Turkmenistan, are featured by multiple types of carbonate rocks, strong reservoir heterogeneity, and a complex relationship between lithology and electric property, making it difficult to predict the quality and lateral distribution of the reservoirs. Based on the observation and analysis of core and cast thin section, the lithofacies were subdivided into granular micrite facies, micritic sparry grainstone facies and biohermal limestone facies according to the degree of reservoir development. The corresponding logging responses of these lithofacies were determined and their electrical properties were analyzed. Then logging identification criteria were summarized for these lithofacies. A series of mathematical analytical techniques such as main coordination analysis, wavelet analysis and multiple parameters progressive discrimination analysis were used to build a mathematical relationship model between lithofacies and logging facies, which can realize automatic identification of lithofacies with logging data. The results showed that the XVm layer is dominated by micritic sparry grainstone facies and biohermal limestone facies, the XVp layer by well developed granular micrite facies and micritic sparry grainstone facies, and the XVac layer by granular micrite facies. The XVm layer is the highest in reservoir quality, followed by the XVp and XVac layers. Reservoir quality prediction results obtained by using this method agree well with real test results with a coincidence rate over 80%.