Data science for engineering structural integrity

How can data science be used to improve the assessment of structural integrity across engineering sectors?




Accurate assessment of structural integrity is vital for the safe and efficient operation of engineering assets ranging from aircraft and wind turbines to power stations and railway track. The demands on assessments are constantly increasing as new manufacturing processes and materials are developed and structures are expected to operate in more extreme environments to increase efficiency and reduce environmental impact. Advances in measurement and simulation capability offer part of the solution, but the increased volume and complexity of data brings new challenges and opportunities. The purpose of this interest group is to build a community at the intersection of engineering and data science to address these opportunities.

Explaining the science

In the context of the fourth industrial revolution, an expected capability of a digital twin is to report the current state of critical components in a physical structure. This effectively requires a structural integrity assessment to be performed on demand. To realise this vision requires the seamless integration of structural and material behaviour models with physical measurements of operating conditions, Non-Destructive Evaluation (NDE) measurements, and historical data.

A structural integrity assessment examines the mechanical state of a structure to determine if it is fit for purpose and safe to operate. The process uses available knowledge of loading and operating condition history to drive material performance models. The material performance models are built on micro/meso scale material characterisation and modelling, supported by a vast legacy of scientific literature and industrial experience. NDE is performed at manufacture to confirm the initial state of the material after the manufacturing process. Further NDE measurements are performed in-service to confirm that the evolution of the material is consistent with model predictions. If discrepancies are found, the assessment models can be updated, and new performance guidance issued. NDE therefore provides the feedback that maintains alignment between the physical and digital domains. The underlying physical modalities used for NDE are ultrasonic, radiographic, electromagnetic, and optical. Most NDE measurements are now digital and often involve arrays of sensors to perform tomographic imaging. There is also a general shift from manual to robotic inspections as well as increased use of permanently installed NDE sensors for through-life monitoring. Together this enables the acquisition of high-fidelity digital NDE data at unprecedented scales and there is a pressing need for complete automation of the data analysis workflow. Furthermore, structural integrity assessments require adaptation to the volume and sophistication of NDE data that can be generated by 21st century monitoring devices.


The Interest Group will bring together engineers with discipline-specific knowledge and data scientists who are interested in applying their skills to real-world challenges. The initial goal of the group is to develop a collaborative and trusting community with a collective vision and clear identity to perform:

  • Knowledge exchange
  • Landscape mapping to identify research requirements in data science for engineering structural integrity
  • Horizon scanning for new data science solutions.

Talking points

  • How can more information be leveraged from measured NDE data?
  • How can time-consuming human intervention from the structural integrity assessment process be minimised or eliminated?
  • How can the vast legacy of material data be digitally exploited?
  • How can measured NDE data seamlessly feed into automated structural integrity assessments?
  • How can we integrate NDE and materials modelling to create self-updating models

How to get involved

Click here to request sign-up and join


Contact info

Professor Paul Wilcox, [email protected]

Professor Mahmoud Mostafavi, [email protected]