基于井下多点压力测量和数据驱动的实时井眼清洁监测新方法

A new real-time hole cleaning monitoring method based on downhole multi-point pressure measurement and data driven

在长水平井和大位移井施工中,井眼清洁不充分会导致一系列的钻井问题。传统井眼清洁分析单纯依赖理论模型或地面振动筛数据,无法准确评估井下岩屑分布状况及存在的问题。为此,引入了一种全新的研究思路,考虑了将传统钻井水力模型与人工智能方法相结合,提出基于实测数据的井下真实岩屑分布技术,并拓展井眼清洁领域通过压力反演流动特征的研究,最后提出了一种利用沿管柱测量(ASM)数据定量评估井下岩屑动态分布的新方法。研究结果表明:(1)不同井段和工况条件下井眼清洁情况与环空压耗之间的关系密切,成正比关系;(2)在给定的流动条件下,通过反推岩屑对井眼压降的影响,建立了基于压力驱动的井眼清洁模型;(3)将传统钻井水力模型与人工智能方法相结合,建立了一种可以自动修正的智能钻井水力学模型,然后将训练后的水力学模型与实测环空压力相对比得出岩屑对压力损失的影响,然后代入压力驱动的井眼清洁模型,得到实时的井下岩屑分布情况。结论认为,利用井下多点测量数据可以实现钻井过程中井眼内动态实际岩屑分布的间接测量和井眼清洁状况的准确评价,为避免和解决长水平井和大位移井的井眼清洁不充分问题提供详细的井下信息。该方法可以克服传统井眼清洁分析单纯依赖理论模型的缺陷,同时为利用智能钻杆等全井眼测量技术解决其他常见钻井问题提供支撑,有助于提高该技术的实用价值,推动该技术在油气行业的规模性应用。

During the construction of long horizontal wells and extended reach wells, inadequate hole cleaning can lead to a series of drilling problems. Traditional hole cleaning analysis is simply based on theoretical models or surface vibrating screen data, and cannot accurately assess downhole cuttings distribution and existing problems. To this end, this paper introduces a novel research idea, which considers combining the traditional drilling hydraulic model with the artificial intelligence method. Then, a downhole true cuttings distribution technology based on measurement data is put forward, and the inversion of flow characteristics from pressure in the hole cleaning field is further researched. Finally, a new method for quantitatively evaluating dynamic distribution of downhole cuttings using along-string measurement(ASM) data is proposed. And the following research results are obtained. First, the relationship between hole cleanliness and annular pressure loss in different hole sections under different working conditions is close and proportional. Second, under given flow conditions, a pressure-driven hole cleaning model is established by inferring reversely the effect of cuttings on borehole pressure drop. Third,an intelligent drilling hydraulic model that can be automatically corrected is established by combining the traditional drilling hydraulic model with the artificial intelligence method. Afterwards, the effect of cuttings on pressure loss is clarified by comparing the trained hydraulic model with the measured annular pressure, and then introduced into the pressure-driven hole cleaning model to understand realtime downhole cuttings distribution. In conclusion, downhole multi-point measurement data can realize the indirect measurement of dynamic actual cuttings distribution in the hole in the drilling process and the accurate evaluation of the hole cleaning condition and provide detailed downhole information to avoid and solve the inadequate hole cleaning of long horizontal wells and extended reach wells. This method can overcome the defects of traditional hole cleaning analysis which is simply based on theoretical models, while providing support for solving other common drilling problems by using full-hole measurement technologies such as intelligent drill pipes, which is helpful to improve the practical value of this technology and promotes the large-scale application of this technology in the oil and gas industry.