An Integrated Approach towards Predictive Maintenance on Land Drilling Rigs
J.J. Oldenhuis (TU Delft - Mechanical Engineering)
DL Schott – Mentor (TU Delft - Transport Engineering and Logistics)
Yusong Pang – Mentor (TU Delft - Transport Engineering and Logistics)
L.R.R. Holwerda – Mentor (Huisman Equipment BV)
W. Qu – Graduation committee member (TU Delft - Transport Engineering and Logistics)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
The drilling industry faces the challenge of moving away from conventional maintenance strategies to maximise operational efficiency. This thesis introduces a holistic approach to land drilling rig Predictive Maintenance (PdM), integrating component maintenance into system-level decision-making. By taking the operations of the whole system into account, PdM can be applied more effectively, and the uptime of a drilling rig can be improved. A hierarchical Multi-Agent System (MAS) framework is proposed, employing two layers of agents for the core equipment of a drilling rig and one centralised system agent. Middle-level agents perform component diagnostics, prognostics, and operational state classification. The system agent uses expert reasoning for integrated maintenance scheduling. As a proof of concept, the proposed model is partially developed and validated, using readily available data and additional condition monitoring.
In the field research phase of this thesis, valve vibration signals were gathered from a mud pump during a geothermal drilling project in Delft. From these signals, dimensionless features were extracted and used in a Fuzzy Logic (FL) classifier combined with Weibull Accelerated Failure Time (WAFT) modelling for dynamic Remaining Useful Life (RUL) prediction. In the case study, where synthetic valve failure is induced in historical operational data, the model demonstrated its ability to predict these failures on time. Maintenance actions were suggested by the model when components reached 93% of their theoretical lifetime, preventing excessive maintenance and minimising disruption to drilling operations through integrated scheduling.