University of Manchester

Supervisor: Prof Anthony Peyton

Industry partner: Jacobs

Industrial Motivation

Despite the advances in technologies such as Augmented Reality (AR), which enable better human performance and capabilities in other industries and sectors, NDT on the wholeis yet to make use of and incorporate these technologies within day-to-day practices. These technologies could be very beneficial, where NDT typically involves several manual tasks and behaviours which are critical to the outcomes of the inspection yetare highly influenced bytheaccuracyof the ability to position and orientate the probe relative to the test piece which include:

  • Planning, drawing and conforming to probe movement paths –poor positional accuracy could result in loss of coverage (lower POD).
  • Maintaining appropriate probe alignment and skew –poor positional accuracy could result in higher rates of false indications and a loss of coverage (lower POD).

Moreover, NDT inspections are severely limited by the ability to position the probe relative to the test piece drawings and/or CAD models. This can also affect the inspection performance in various ways including:

  • Accurately plotting indications on test piece drawings –poorpositional accuracy reduces the effectiveness of comparing drawings over asset lifetime.
  • System calibration using reference blocks –poor positional accuracy reduces the effectiveness of some system checks (e.g. beam angle measurement) and calibrations (e.g. time corrected gain).

Therefore, these provide opportunity for human error to have a significant impact on the outcomes of an individual NDT inspection as well as the longer term asset management. Position encoders provide a good measurement tool to track the differences in probe positionduring a single inspection, yetthe repeatability of zeroing position over multiple inspections is difficult to achieve.Position encoders typically also increase the size, footprint or form of probes, which reduce the versatility of the overall system and can be incompatible in some scenarios (e.g. restricted access to test piece surface).

Consequently, a probe position tracking solution which is capable of repeatable zeroing over multiple inspections would provide an elegant solution to this problem; and in doing so would significantlyreduce the severity of human error in NDT inspections.

This could be achieved using the principles of Augmented Reality (AR) photogrammetry, to accurately position probes in a repeatable manner. Other statistically independent methods such as using datumsstamped on test pieces, probe accelerometers and using real time NDT data (e.g. UT side wall reflections, eddy currentedge effects, etc.) could also be combined and incorporated to provide a highly accurate, rugged positioning method.

Furthermore, such asystem would provide a platform from which to further enhance the human capabilities and performance of the operator during inspection. Such functionalities include:

  • Rendering inspection data (C-Scans, etc.) onto test piece in real time.
  • Rendering probe path onto test piecefor operator to follow.
  • Autonomous monitoring of probe alignment and skew with arrows and guides rendered onto test piecefor operator to follow.

The ability to render NDT scans in AR would also provide a powerful new way to visualise the data and help the operator to make more informed decisions regarding indication evaluation and flaw characterisation.

The potential of such a system goes further with the remote access of data from a responsible PCN Level III, being able to remotely monitor and sign off the inspection using Virtual Reality (VR) in real time during the inspection, or alternatively view the captured data / scans rendered in VR, etc.

As such, this proposal describes a paradigm shift in the way in which NDT probes are positioned; and from such, an array of additional, powerful capabilities are enabledwhich also aligns to the vision of NDT4.0.

Aims & Objectives
  • Primary –A demonstrable inspection system with VR-enabled positioning and acquired UT data visualisation on a rectangular test piece in real time (lab conditions, ~TRL 4).
  • Secondary –
    • Adapt system for curved test piece or nozzle.
    • Record data, renders, etc. in non-proprietary file formats to align with NDT4.0 vision.
    • Fuse other NDT data sources (e.g. eddy current testing) with UT data and render in AR.
    • Incorporate Artificial Intelligence to provide additional support and functionality (e.g. improved flaw characterisation or sizing).
Breakdown of Key Tasks / Deliverables
  • Literature survey regarding:
    • The principles and theoretical capabilities of augmented, mixed and virtual reality, including position tracking.
    • The use of augmented, mixed and virtual reality in NDT and wider engineering.
    • The use of other Internet-of-Things applications in NDT.
    • Coordinate system transforms.
    • Other topics considered applicable.
  • Use statistically independent positioning techniques to achieve millimetre probe-test piece accuracy suitable for NDT inspection in field.
  • Transform NDT flaw detector data output coordinate system to be compatible with AR projection / rendering.
  • Construct interface between AR “controller”, test piece drawing / CAD model and NDT flaw detector.
  • Visualise 3D UT data using inspector-test piece orientation to provide perspective to aid interpretation.
  • Progress reports –first year MPhil/PhD transition report, thesis, RCNDE/FIND CDT review(s).