How structural digital twins can transform the maintenance and inspection of pressure vessels.
ENGINEERING simulation software has been around for decades. It has been used ubiquitously in the design of large, complex assets and is an essential tool for engineers to develop better, more efficient products. It enjoys absolute anonymity from the untrained eye, but it’s behind every asset or piece of infrastructure in our modern world. Until recently, its application has been limited to design because the numerical method behind conventional software, finite element analysis (FEA), doesn’t have the computational speed required to simulate at the pace of operations. Thanks to US Department of Defense funded research at the mechanical engineering lab at MIT, 2012 saw a breakthrough – an accelerated version of the algorithm, reduced-basis finite element analysis (RB-FEA). RB-FEA has revolutionised simulation by allowing structural integrity assessments to be run holistically on operational assets, to support remote monitoring and near real-time assessment.
At Akselos, we’ve been working with asset integrity service company, CAN Group, and the UK Oil and Gas Technology Centre (OGTC) to apply RB-FEA to the simulation of pressure vessels to create near real-time, structural digital twins. As a result, a cross-disciplinary team has developed improved processes for managing pressure vessels by combining Akselos’ new engineering simulation techniques with robotics and non-intrusive inspection (NII) techniques. Together, these techniques are expected to deliver better operational efficiency, including a 45% reduction in downtime and a 25% reduction in maintenance and inspection costs.
A 2017 OGTC survey of major upstream oil and gas operators identified that adopting non-intrusive inspection technology for process pressure vessels could deliver increased production and lower maintenance costs worth up to £242m/y (US$313m/y) to the UK Continental Shelf (UKCS). The findings showed that potentially 80% of vessels could be examined non-intrusively, without requiring a shutdown. Consequently, cost savings could reach up to 80% compared to inspections that involve entry into a vessel, and crucially, improved safety with 80% fewer confined space entries required.
The research also found that the data from ultrasounds – typically used for non-intrusive inspection – were, more often than not, difficult to process and interpret. The procedure generated tens of millions of data points but lacked context and structure, rendering them meaningless.
OGTC has been proactively seeking out new ways to make non-intrusive methods the status quo, first and foremost to reduce and eliminate risks associated with intrusive inspection and to improve the overall operational efficiency of production facilities.
When Akselos approached OGTC to test the use of a holistic simulation of the pressure vessel powerful enough to incorporate inspection data, it agreed to support what would be the industry’s first physics-based digital twin of a pressure vessel.
Our task was to make a physics-based assessment of the ultrasonic inspection data to assess the structural integrity of the vessel.
The large high pressure vessel in question had been deployed in the North Sea and used for the separation of liquids during the production process offshore, and had a history of localised corrosion. The traditional approach for assessing the status of a pressure vessel, known as intrusive inspection, includes shutdown, isolation, and visual inspection of the vessel. This can impact profitability, as it often also requires a shutdown and isolation of the production plant too. In addition, inspecting pressure vessels involves a range of hazards including confined spaces, radioactive materials and potential exposure to toxic byproducts such as mercury and pyrophoric scale.
The traditional engineering approach of the asset operator involved a staged workflow to assess the risk on the asset presented by the inspection results based on the API 579 standard. However, this process can be time consuming where defects are detected.
Akselos collaborated with CAN Group, a UKAS 17020 Accredited Inspection body which has extensive experience working on NII data visualisation projects, to create a new, API 579 compliant workflow that could be run on operational vessels. It included semi-automated inspection data that was uploaded, mapped, and assessed within the digital twin in exact detail, to enable accurate validation of the pressure retention and structural integrity, and in turn fitness for service.
The run time workflow, which Akselos has automated, can be summarised as follows (see also the diagram below):
The high-fidelity simulation model enables a Level 3 assessment by building a digital thread from robotic inspection and data gathering, mapping that data onto the digital twin, running RB-FEA tests, and providing a pass or fail based on the API 579 threshold tests.
The workflow reduced a minimum 7-day day scope to under 3 hours, without the need to take the pressure vessel offline. Not only does this significantly reduce time, cost, and risk exposure, it allows for optimisation of the inspection frequency and eliminates the need for Level 1 and 2 assessments.
Achieving an accurate Level 3 assessment result with conventional methods is no easy task, nor is it 100% reliable. The conventional approach is to manually collect inspection data and create a model using FEA. The only way to achieve an accurate understanding of the condition of the vessel would be to create a number of sub-models of particular focus areas, or make a major compromise by creating a lower fidelity holistic model, which removes the details of pitting and other anomalies through giving an average thickness measurement over critical areas. Either way, the workflow that would produce a low fidelity model would still take at least seven days (typically weeks) and a US$30,000–50,000 bill for each assessment.
Thanks to OGTC’s stewardship, the project partners came together and developed a pioneering approach to pressure vessel inspection in less than a year, utilising emerging technology from both the hardware and software worlds.
The project has shown that the use of a near real-time physics-based digital twin will:
In short, once the digital twin for the asset is set up, all the operator has to do following the upload of the c-scan data is press a button to get the latest status of the vessel and a corresponding fitness-for-service report.
We believe the additional benefits of selective inspections and fitness-for-service on demand with high-fidelity modelling, will increase the estimated £240m of annual savings to the UKCS from improved non-intrusive inspection and assessment.
As with most other projects led by OGTC, the beauty is that it won’t take long for the industry to follow suit, and for physics-based models to be the status quo for the maintenance and inspection of pressure vessels.
RB-FEA is a new approach to structural simulation which can produce results 1,000 times faster than legacy FEA. The technique can uniquely work directly with model parameters allowing variation, such as a change in thickness from one inspection to another, to be rapidly assessed. The algorithm was commercialised by Akselos and developed into engineering simulation technology that can be used to transform and optimise operations, as well as design. The technology runs holistic structural assessments in near real-time to provide asset operators with a more powerful tool for operational excellence.
The unprecedented level of speed and the parametric nature of the model are enablers of true condition-based monitoring, facilitating very fast structural analysis across an entire asset of almost any size and complexity. This approach produces simulations 1,000 times faster than legacy FEA tools for industrial-scale assets and delivers models that are hundreds of times larger and more detailed than has previously been possible. This speedup means that all structural details can be included in the model, along with inspection-based condition data such as cracks, corrosion, and damage or deformation. The extreme speed of the solution means that detailed real-world loading scenarios can also be included and also handles multi-physics effects and chemical-related parameters for corrosion and material (through partner plugins).
The speed makes it compatible with digital oilfield components such as sensors, big data and machine learning. Unlike FEA, RB-FEA is fast enough to keep up with sensor feeds and to collect statistical data to inform and enrich the model. The result of this is a precise coupling of the virtual and physical worlds via a living, learning digital twin that reflects the exact condition of its physical counterpart in real time. This improves operations, prevents downtime through its predictive capability, reduces maintenance costs, and provides data that can be used to streamline operations throughout the lifecycle of the asset.
The inspection data was collected using a combination of automated crawler and semi-automated magnetic scanning systems. The ultrasonic systems used collected high-resolution, high frequency 0° linear array phased array data; this achieved a scanning resolution of 1x1 mm. The use of high frequency probes provided improved defect detection for small pitting damage
Catch up on the latest news, views and jobs from The Chemical Engineer. Below are the four latest issues. View a wider selection of the archive from within the Magazine section of this site.