REVIEW AND ANALYSIS 43 Fig 2: Digital reality capture followed by automatic segmentation, defect detection and classification Fig 3: Labelling of defects for machine learning algorithms Spatial data capture Segmentation Defect detection Defect classification Deposit → → → Crack Delamination Peeling Mechanical damage Deposit Encrustation Blistering Disintegration Alveolisation Perforation Discolouration Biological colonisation Bursting Fragmentation Erosion Crust Efflorescence Combination reducing carbon through targeting maintenance and repair interventions. In addition, the HDS project developed the use of artificial intelligence or, more specifically, machine learning algorithms for the automatic detection and classification of defects in masonry in accordance with the ICOMOS stone defects glossary (see fig 2 and 3). While the AI algorithms detect and classify defects, it is not envisaged that they replace surveyors. These technologies simply help redirect a surveyor’s time towards ‘value added’ activities as opposed to relatively mundane operations. It is important to set these forms of operations in context: the scale of masonry repair required to our traditional built environment is significant, with an estimated annual cost of £30 million for repairs in the Glasgow region alone. This financial cost cannot be decoupled from carbon cost, with every intervention expending CO2. BIM AND DIGITAL TWINS Turning to progress in the wider sector, the deployment of digital applications such as building information modelling (BIM) is growing rapidly. However, the use of BIM in maintenance is often misunderstood, and ultimately the development of a visual model may not be required to attain digital efficiency. What may be of greater importance is the organisation of digital data (in an ontologically structured way) to enable its use by multiple stakeholders with a wide range of professional and specialist perspectives. As with more conventional forms of project information, the ability to share and use digital data is critical for effective project delivery. In building conservation, concepts of BIM and digital twins are becoming more prevalent. In digital twinning, the existing physical building (or infrastructure asset) is linked with a semantically rich digital model. Sensor technology, which is increasingly being used in existing buildings, can feedback real time digital data to the model, allowing it to assess building performance and recommend actions. (In some cases they may even trigger actuators to execute the action required.) In this way digital twins can aid in structural health monitoring, evaluation and optimisation of environmental conditions (such as moisture and relative humidity), and energy efficiency. The integration of data on the performance of key components can highlight deterioration risks and wider performance issues, triggering intervention with great immediacy. These might include thermal efficiency of fabric and hygrothermal problems, or the progressive movement in structures
RkJQdWJsaXNoZXIy MjgyMjA=