Data-Driven Solution for Downtime Identification in Unconventional Oil/Gas Reservoirs

在硬件密集型作业中，如加拿大页岩油气行业的完井作业，作业地点偏远，作业的各个部分的责任由多个供应商承担。由于这种复杂性，这些操作远远没有得到优化，并且经常由于非生产时间（NPT）、意外操作故障和安全相关事件（或未遂事故）而损失资金。随着商品价格的下降，有必要减少运营和资本支出，这可以通过流程优化来实现。**业界多次表达了对实时监控和优化系统的需求。**然而，使用昂贵的传感器（如热成像摄像机和视频分析）远程监控操作已被证明过于昂贵，并且提供的功能超出了需求。HAWK-EYE是一款价格低廉的软件系统，通过压力传感器和阀门位置传感器监测垫块的运行情况，可以先发制人地确定NPT的程度、操作故障的潜在可能性和安全事故，从而为客户提供经济高效的工艺优化解决方案。贝克休斯（BHGE）是一家领先的油气设备制造商、油田服务和数字化解决方案提供商，该公司正在寻找一种基于优化的解决方案，以降低非常规油气开采的成本。目前，传感器数据、压力测量值、阀门位置和相关的压裂相关事件在压裂作业后发送给顾问，以量化NPT。拟议的研究项目将开发一种自动数据驱动的方法，基于可用的传感器数据，对井的状态进行实时稳健估计，从而提供对井状态的实时监测。

In hardware intensive operations such as completion operations in the shale oil and gas industry in Canada,**operational sites are remote and the responsibilities for various parts of the operation are split between multiple**vendors. Due to this complexity, these operations are far from being optimized and often lose money due to**Non-Productive Time (NPT), unexpected operational failures, and safety-related incidents (or near misses). As**commodity prices drop, it becomes necessary to reduce operating and capital expenses, which may be achieved**through process optimization.**The industry has repeatedly expressed the requirement for a real-time monitoring and optimization system.**However, monitoring operations remotely with expensive sensors, such as thermal imaging cameras and video**analytics, has proven to be too costly, and provides capabilities in excess of requirements. HAWK-EYE serves**as that inexpensive software system that monitors pad operations using pressure transducers and valve position**sensors that can be used to determine the extent of NPT, operational failure potential, and safety incidents in a**pre-emptive manner, in turn, providing customers with a cost-effective solution for process optimization. Baker**Hughes, a GE company (BHGE), a leading oil and gas equipment manufacturer and oilfield services and digital**solutions provider, is looking for such an optimization-based solution to reduce costs in unconventional**hydrocarbon extraction.**Presently the sensor data, pressure measurements valve positions and associated frack-related events, are sent**to consultants after the fracking operation, in order to to quantify the NPT. The proposed research project will**develop an automatic data-driven method for robustly estimating the status of a well in real-time, based on the**available sensor data, hence providing real-time monitoring of the well status.