基于地震敏感参数模板的储层预测研究

发布时间：2018-03-17 16:06

本文选题：储层预测　切入点：地震属性　出处：《西南石油大学》2015年硕士论文　论文类型：学位论文

【摘要】：河流相储层是我国广泛发育的储集类型,主要特点是岩性变化大,连通性差等。苏里格气田59区盒8段属于典型的河流相储层,储层内气藏的分布主要受砂岩横向展布及物性变化的影响。该工区具有储层较薄且纵向上砂体相互叠置,非均质性强,气水分布规律十分复杂等特点。通过储层预测弄清该工区的主河道空间展布及气水分布规律,划分有利勘探区,选取目标井位是对我们最大的挑战。本论文从储层预测技术的研究出发,提取了地震属性,研究了多种数学法属性优选技术及神经网络储层预测技术,并重点讨论了针对样本数分布不均或储层非均质性强等可能引起储层预测结果产生较大误差等问题。同时结合地震属性分析技术、多元属性综合分析等方法,创新性地提出了地震敏感参数模板这一全新的概念。并通过正演模拟中理论模型的检验及在实际工区中的应用,证明了该方法有助于提高储层预测精度。对于本论文研究区的目标储层,结合所掌握的有限资料,工区地质特征、储层特征及勘探情况等。首先,选择合适的时窗,通过提取相应的地震属性、应用多种方法组合优选敏感属性,并使用神经网络储层预测,初步划分出勘探有利区域。但神经网络对于样本分布有一定的要求,特别是在少井区或井间距离较远时预测精度将受到一定的影响。结合目标储层段非均质性较强,横向变化快,气水分布复杂等特点,借助地震敏感参数模板来解决神经网络在少井区的预测精度不高等问题,划分有利勘探区域,优选目标井位,提高钻井成功率,降低勘探风险。目前,该方法已经被应用于实际工区的生产研究中,并通过实钻表明该方法能够在一定程度上有效地提高储层预测精度,具有较强的实用价值和研究潜力。
[Abstract]:Fluvial reservoir is a widely developed reservoir type in China, which is characterized by great lithologic change and poor connectivity. The distribution of gas reservoirs in the reservoir is mainly affected by the transverse distribution of sandstone and the variation of physical properties. The reservoir is thin and the vertical sand bodies overlap with each other, and the heterogeneity is strong. The distribution of gas and water is very complicated. It is the biggest challenge for us to make clear the spatial distribution of the main channel and the distribution of gas and water through reservoir prediction, to divide the favorable exploration area and to select the target well location. Based on the research of reservoir prediction technology, the seismic attributes are extracted in this paper, and various mathematical attribute optimization techniques and neural network reservoir prediction techniques are studied. This paper also focuses on the problems that may result in large errors in reservoir prediction results due to the uneven distribution of samples or strong heterogeneity of reservoir. At the same time, combined with seismic attribute analysis technology and multivariate attribute comprehensive analysis methods, A new concept of seismic sensitive parameter template is put forward creatively, and it is proved that this method is helpful to improve reservoir prediction accuracy through the testing of theoretical model in forward modeling and its application in practical working area. For the target reservoir in this study area, combined with the limited data, geological characteristics, reservoir characteristics and exploration conditions. Firstly, the appropriate time window is selected and the corresponding seismic attributes are extracted. Several methods are used to select sensitive attributes, and neural network is used to predict reservoir, and the favorable exploration area is preliminarily divided. However, the neural network has certain requirements for sample distribution. In particular, the prediction accuracy will be affected when there are few well areas or longer inter-well distances. Combined with the characteristics of high heterogeneity, fast lateral change and complicated gas-water distribution in the target reservoir section, the prediction accuracy will be affected to a certain extent. By means of seismic sensitive parameter template, the problem of low prediction accuracy of neural network in the area of fewer wells is solved, the favorable exploration area is divided, the target location is selected, the success rate of drilling is improved, and the exploration risk is reduced. At present, this method has been applied to the production research of practical work area, and it is proved by real drilling that the method can effectively improve the reservoir prediction accuracy to a certain extent, and has strong practical value and research potential.
【学位授予单位】：西南石油大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P618.13;P631.4

【参考文献】