Modelling and simulation for intelligent maintenance: A case study on centrifugal gas export compressor in oil and gas transport system
Significant developments in information and communication technology (ICT) have revolutionized industry and made the fourth industrial revolution, Industry 4.0, a reality. Industry 4.0 enables a new level of organising and controlling the entire value chain within the product lifecycle by creating a dynamic and real-time understanding of cross- company behaviours. It is expected to revolutionize current maintenance practices by reaching new levels of predictive (detection, diagnosis, and prognosis processes) and prescriptive maintenance analytics through intelligent maintenance; this will minimize the operational impact by leveraging predictive maintenance (PdM) capabilities into opportunistic maintenance intervals. Although the literature is unanimous when it comes to the expected lifetime benefits of incorporating intelligent maintenance into engineering assets, maintenance management and performance are complex aspects of asset operation that are difficult to justify because of their multiple inherent trade-offs and hidden system causalities. Unfortunately, the literature involves prediction models comparable with a ‘black box’ missing the link between input data, analyses, and final predictions, which makes the industrial adaptability to such models almost impossible. In addition, there is little or no work on modelling deterioration based on loading and detection, diagnosis, and prognosis processes, but these are decisive aspects of intelligent maintenance. Therefore, this thesis adopts a six-step simulation modelling methodology to develop a novel generic multi-method simulation model that combines agent-based simulation with system dynamics using Anylogic 8; it addresses the expected lifetime benefits of incorporating intelligent maintenance by looking at a specific case study.
The case study, a centrifugal gas export compressor, is characterized as an operational bottleneck in gas transport. The six-step simulation modelling methodology reveals two additional research gaps in the literature; both are covered in this thesis. First, modelling requires an architecture for intelligent maintenance. This thesis uses a systems engineering methodology to upgrade the current Norwegian Petroleum Directorate’s (NPD) maintenance management loop based on associated needs and requirements noted by stakeholders, Industry 4.0 (technology), and standards, allocated as functions in the final architecture. Second, the simulation modelling methodology requires an analysis tool that can determine the technical specifications of PdM capabilities, including detection (capability and coverage), diagnosis (fault type, location, and severity), and prognosis (precision and predictive horizon). The proposed generic PdM assessment matrix enables the assessment of the PdM readiness of equipment that includes determining detection and prediction capabilities; this supports the assigned quantitative values in the modelled multi-method computational model.
The novel multi-method computational simulation model combines system dynamics with agent-based simulation and is simulated using two main case study scenarios: with and without intelligent maintenance. The latter scenario has four sub-scenarios to study the effectiveness of several data sources: (1) empirical case study data (experience), (2) manipulated empirical case study data, (3) offshore and onshore reliability data handbook (OREDA), and (4) mixed data input from both data sources. The simulated results clearly demonstrate lifetime benefits associated with incorporating intelligent maintenance into the specific case study. The strategy is expected to improve operational availability by 0,268%, with a reduction of 20 corrective maintenance events equalling a reduction in maintenance workload by 459 hours or 11% over an operational period of 20 years. The results also highlight the sensitivity and high degree of uncertainty associated with the mean time to repair (MTTR) values extracted from the case study company’s notification system caused by human factors.
This work is licensed under a Creative Commons Attribution 4.0 International License.