随着信息技术在石油天然气勘探开发中应用越来越广泛,钻井、录井、井下作业等工程数字化技术得到大范围推广,页岩气水平井安全高效钻井迫切需要动态数据实时研判水平井岩屑返出有效性,为井眼清洁状况的实时分析评价和相关参数优化奠定良好的基础.文章通过调研井眼清洁分析计算模型,确定了适合实时跟踪井眼清洁程度的分析计算模型,提出了计算分析流程,制定了井眼清洁状况的预判标准;编制的井眼清洁功能模块植入工程信息平台可实现不同井斜段钻井参数和钻井液性能指标的优化模拟计算、钻井过程中井眼清洁状况实时预警,有助于减少和避免因井眼清洁不佳造成的井下复杂和卡钻风险.

As the current calculation methods for wellbore separation factor have some deficiencies, we propose and analyze a new calculation approach for wellbore separation factor based on the relative position of adjacent wellbores, named as relative position method for short. Based on the trajectory error ellipsoid model of single wellbore, the error ellipsoids model of adjacent wellbore was derived considering the correlation of trajectory errors between adjacent wells. Furthermore, the calculation formula of the separation factor based on relative position of adjacent wellbore was derived and solved with the conjugate gradient algorithm. Case study shows that the new approach is more precise and higher in applicability than the ellipsoid scaling method and the minimum distance method, it can evaluate the state of well collision more reasonably. By doing batch calculation with the new method and following the criterion of well collision avoidance, the permissible ranges of key parameters in the well design can be worked out quickly. This method has good application in the design of cluster wells and directional wells.

随钻测井作业与钻井作业同步进行,在钻井液滤液还未侵入地层或者侵入井壁很浅时获得测井资料,能准确反映原状地层特征,油气储量评价准确.图像压缩是随钻电阻率测井实时成像的关键技术,通过分析多种图像压缩方法,发现小波编码适合随钻电阻率成像测井,在此基础上,分析嵌入式零树编码和多级树集合分裂编码2种小波变换算法,进行Matlab仿真.多级树集合分裂编码方法在50倍压缩率条件下,能够保持较高的图像质量,可满足随钻电阻率测井实时成像数据压缩需要.

近年来,随着人工智能技术的发展应用,机器学习的框架及研究工具逐渐成熟.其中,TensorFlow.js是由Google的AI团队发布的一款基于硬件加速的JavaScript库,用于训练和部署机器学习,用户在浏览器端就可以利用应用程序编程接口(Application Programming Interface,API)完成机器学习的基本任务.在油气田开发领域中,产生的生产及分析数据具有数据量小、相关性强的特点,在机器学习过程中,大部分业务通过浏览器就能进行简单的数据线性回归、分类、目标识别、可视化等处理操作,具有简单、快速、易学等特点.文章以线性回归模型为例,对探井试油过程中的油管压力数据采用前端机器学习方法进行线性回归拟合,并实现压力预测.

二维地震勘探观测系统设计时,满覆盖次数理论设计应该为常数,均匀分布,但当最小偏移距大于0.5倍道距时,满覆盖区域覆盖次数理论设计可能出现变化现象.通过对此现象发生的原理分析、总结,最终得出了满覆盖区域覆盖次数理论设计稳定时影响满覆盖次数的3个关键因素,即最小偏移距、炮间距和接收道数,及其对应的关系,在实际应用中得到了较好的效果.

本文对井眼轨迹优化设计需要考虑的主要因素进行了介绍,分析了几种目前常用的井眼轨迹优化方法,并总结了地应力对井眼轨迹的影响.

地震反射波时距曲线受复杂地表地质结构的影响产生畸变而不能同相叠加,需要对地震资料做静校正处理.不同的静校正方法都有各自的适用条件:高程静校正无法消除低降速带的影响;模型静校正受限于追踪的深度和排列的长度,难以全面反映低速带底界的形态;折射静校正仅适用于地表较平缓、表层速度横向均匀性较好且有明显折射界面的地区;层析静校正适应任意表层模型的反演,但是反演结果不稳定.在地震勘探地区地貌日益复杂的今天,优选的某一种静校正方法无法有效解决所有的问题.鉴于上述问题,提出了一种适用于多种地表类型复杂区域地震资料处理的基于变差函数拟合重构的高精度三维静校正方法,该方法能够实现不同静校正量的拟合和重构,并很好地解决复杂地表条件下不同静校正方法优势拟合的难题,从而可有效提高复杂地表地形条件的三维地震资料的成像质量,保证构造形态的可靠性.

The structure and working mechanism of the rotary drilling steering tool is introduced.Combined with its design feature,a 3D solid model of the rotary drilling steering tool is set up by means of SolidWorks software,and its FEA module is established,based on the COSMOSWorks module,which is integrated in the SolidWorks system.Through FEA program,the strength of the mandrel of the drilling tool is calculated,and the rationality of the mandrel structure is verified with FEA.Moreover,the results of the virtual design and FEA make suggestions for the optimization of rotary drilling steering tool.

The lithology is rather complex and difficult to identify in ig neous reservoirs.With little formation information,traditional cross-plot and supervised neural networks(such as BP network) are restricted in identifying lithology.Therefore,in the south part of SongLiao basin,based on the principles and structure of SOM neural network,the data set of igneous samples were established by actual logging data.The cluster results were obtained by train- ing the samples with SOM network.the influence of standard means,structure parameters and log of SOM network on cluster results,which shows that good results can be achieved for lithology recognition on logging data of igneous reservoir by using normal standard method,selecting proper structure parameters and log, and taking the cluster results as the basis of classification.