烃源岩测井定量评价方法探讨

Logging quantitative evaluation method for hydrocarbon source rock

烃源岩评价是油气地质研究最重要的基础工作之一.岩心分析数据量往往有限,难以满足勘探的需求;测井资料经分析数据标定后能获得连续的烃源岩指标数据,应用较广泛.文中阐述了烃源岩总有机碳含量(TOC)、热解生烃潜量(S_1+S_2)和成熟度三个指标的测井定量评价方法;在孟加拉湾海域试用并对计算效果进行分析.首先详细地阐述了单因素法、多元回归法、ΔLgR法和神经网络法等4种常用TOC测井计算方法的原理及优缺点;其次介绍了TOC相关法、多元回归法和神经网络法等3种常用S_1+S_2计算方法原理及优缺点;再次介绍了镜质体反射率和剩碳率法等2种成熟度评价方法.结合研究区地质条件对定量计算效果进行分析,TOC计算结果对比表明神经网络法效果最好,多元回归法次之,ΔLgR法较差,单因素法最差.分析认为TOC较低对测井响应贡献小,单条曲线与TOC相关性低导致单因素法效果差;多元回归法考虑了多条曲线效果较好;ΔLgR法由于缺乏厚层泥岩段及热变指数难选取,效果较差;神经网络法在非线性、难用显示表达式的计算问题方面具有很大的优越性,计算效果最佳.S_1+S_2计算结果对比表明TOC相关法计算结果最优,多元回归法效果一般,神经网络法效果最差.分析认为S_1+S_2与TOC有成因联系计算效果好;S_1+S_2对测井响应贡献小,因此多元回归法和神经网络法效果都不好.成熟度评价的镜质体反射率与深度相关性好,最为常用;剩碳率法部分参数在勘探早期难取准,应用受限.

Identification and evaluation of source rock is the basis of geological study for hydrocarbon. Limited core samples restrained the demand of exploration. After calibration of analysis data,logging data can provide continuous hydrocarbon source rocks index,and it＇s widely used. Three Logging quantitative evaluation method were discussed in the paper including the content of Total Organic Carbon（ TOC）, the amount of pyrolysis hydrocarbon generation potential（ S1＋S2） and the maturity indicators. These methods were tried in Bengal bay, and combined with block geological conditions the calculation results were analyzed. First,four kinds of calculation methods for the TOC including single factor method,multiple regression method,ΔLgR method and the neural network were described in detail,as far as principles,advantages and disadvantages. What ＇s more, two methods including the vitrinite reflectance and residual carbon ratio for maturity evaluation were introduced. Finally,the quantitative calculation results in the study area were analyzed. TOC calculation results shows that the neural network was best,multivariate regression method was the second,ΔLgR method was poorer,and single factor method was the worst. Analysis shows the low TOC contributed little to the logging response,bad correlation between a single curve and TOC lead to poor effect of single factor method. As considering multiple curves,multiple regression method was good. Because of the lack of thick mudstone and thermal alteration index selection difficulty,ΔLgR method＇s effect was poorer. Neural network had great superiority when it was nonlinear and difficult to use explicit expression.Through three kinds of S1＋S2 calculation methods in the study area, showed that the TOC related method was the best,multivariate regression was the second,and neural network effect was the worst. TOC was related with S1＋S2,so its calculation effect was good. Contribution of S1＋S2 to logging response was small,so the method of multiple regression and neural network effects were bad. Maturity evaluation of vitrinite reflectance had correlation with depth, it was most commonly used. Some parameters of residual carbon ratio method were determined in the early exploration,so it had limited application.