Predicting Mechanical Failure Using Non-Operational Data

使用非操作数据预测机械故障

• 批准号: 2279576
• 金额: --
• 依托单位: University of Oxford
• 依托单位国家: 英国
• 项目类别: Studentship
• 财政年份: 2019
• 资助国家: 英国
• 起止时间: 2019 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=studentship-2279576
• 关键词: Predicting; Mechanical; Failure; Using; Non

Summary: The Project is directed toward predicting mechanical failure using non-operational data. Background: In the process of Well Construction (planning, drilling, and completion of oil and gas wells), the mechanical tools that constitute the drilling assembly are subjected to extremely harsh conditions, experiencing high pressures and temperatures and high levels of shock and vibration. Mechanical failure of a tool during the drilling process can lead to costly and time-consuming recovery and replacement. Understanding how these tools wear and fatigue during the drilling process is critical to efficiently constructing cost-effective wells. In an ideal situation, the condition of key components of the tools would be monitored in real time, such that failure could be predicted, and maintenance performed before failure occurs. However, even if such measurements could be made, data communication rates between the subsurface and the surface are not currently sufficient to allow this type of information to be communicated in real time. A more realistic goal would be to monitor the condition of the tools both before they are put into the well, and when they are pulled out of the well. With the combination of non-destructive testing, and an appropriate (data-driven) model of how the tool will fatigue under certain drilling conditions, the probability of the tool failing during the next run could be calculated. This would allow an informed decision to be made as to whether that tool can be re-used, or whether corrective maintenance is required. Project Description: The aim of this project is to develop a system that demonstrates the ability to make such a prediction. This would combine a number of aspects, that may include:- The development of a simple model of a downhole drilling tool to investigate mechanical failures. For example, this may involve using techniques such as Finite Element Analysis to model failure modes.- The development of a laboratory scale experimental setup to recreate failures- The development of a non-destructive testing technique to assess the current condition of a tool- The use of the generated data (simulated and/or measured) to determine the probability of failure of a component within a following time period. The project will therefore allow the development of expertise in:- Advanced modelling techniques- Laboratory Scale Experiments- State of the art non-destructive testing- Data Analytics and Data Science.This project falls within the EPSRC Engineering research area.This project is an EPSRC Industrial CASE Studentship with Schlumberger Gould Research Centre.

摘要：该项目旨在使用非操作数据预测机械故障。背景资料：在井施工（油气威尔斯的规划、钻井和完井）过程中，构成钻井组件的机械工具经受极其恶劣的条件，经历高压和高温以及高水平的冲击和振动。钻井过程中工具的机械故障可能导致昂贵且耗时的恢复和更换。了解这些工具在钻井过程中的磨损和疲劳情况对于有效地建造具有成本效益的威尔斯井至关重要。在理想的情况下，工具的关键部件的状况将被真实的监控，使得可以预测故障，并且在故障发生之前进行维护。然而，即使可以进行这样的测量，地下和地面之间的数据通信速率目前也不足以允许这种类型的信息以真实的时间进行通信。一个更现实的目标是在工具被放入井中之前和从井中拔出时监测工具的状况。结合非破坏性测试和工具在某些钻井条件下如何疲劳的适当（数据驱动）模型，可以计算工具在下一次运行期间失效的概率。这将使人们能够就该工具是否可以重新使用或是否需要进行纠正性维护作出知情的决定。 项目描述：该项目的目的是开发一个系统，展示做出这样的预测的能力。这将结合联合收割机的多个方面，其可包括：-开发井下钻井工具的简单模型以研究机械故障。例如，这可能涉及使用诸如有限元分析之类的技术来对故障模式进行建模。开发实验室规模的实验装置以重现故障-开发无损检测技术以评估工具的当前状况-使用生成的数据（模拟和/或测量）来确定组件在随后时间段内的故障概率。因此，该项目将允许专业知识的发展：-先进的建模技术-实验室规模的实验-最先进的无损检测-数据分析和数据科学。该项目福尔斯EPSRC工程研究领域。该项目是斯伦贝谢古尔德研究中心的EPSRC工业案例学生奖学金。