18 C O N T E X T 1 6 8 : J U N E 2 0 2 1 sections of loadbearing masonry wall had been removed and subsequently re-infilled with tim- ber studwork. The most important lesson was to anticipate the unknowns, to use the model as a tool to identify where further surveys would be needed, and to record this data. It would theoretically be possible to discover all the layers of a building’s construction behind the veneer, and to input them into a BIM model with all the correct attributes assigned to each element. However, this would be an exhausting exercise which would quickly cost more to produce in time and cost than the benefits of using such a model would ever realise within the life expectancy of a typical project. An appreciation of where there is real value in the data highlights the greatest area of potential for BIM models: as a record of an asset and works undertaken for future conservation, repair and maintenance. We have all had experience of working on a building and discovering an unknown, whether a structural alteration, repair or service, or of the challenge of trying to identify the specification of a product used in the past. Searching through archived records may reveal some insights, and buildings with more recent interventions may have more extensive records, such as operations and maintenance manuals. However, as a building passes from one generation to the next, often the most valuable source of information, that which is held in the memory of the building’s custodians, is lost. The greatest potential of technology lies in the ability to collate live, dynamic records of buildings, by tagging data such as notes, photographs, specifications and other records, to a 3D model. This is in effect creating a 3D index for building records. Companies such as Matterport, which offers the creation of an immersive tour of buildings, by using special- ist hardware to scan and survey, creating an immersive ‘point-cloud’ of photogrammetric data, are starting to offer such asset tagging as a feature for estate agent viewings. Those who have invested in specialist headsets such as Magic Leap, Facebook’s Oculus or Microsoft’s Hololens can already bring immer- sive scans of existing buildings into virtual reality. However, the real transformation will occur with the superimposition of this 3D data into a user’s field of vision, using ‘augmented reality’. Companies such as Google and Apple are investing a vast amount of money in this, and the much-rumoured Apple Glasses are eagerly awaited by the tech sector. The applications for historic buildings are vast, not just in operation and maintenance, but also for education and interpretation, and even wayfinding and safety. Imagine being able to walk through a building, selecting an icon in your viewer, hovering over a feature such as door or window, and having important data pop-up for you to review. There is enormous potential to collect all relevant data, archive it and use this simple visual index to assist with retrieval. It puts the value and decision making back into the hands of the building stakeholders, avoiding unnecessary time being spent creating a complex BIM model for no purpose. The benefit lies in focusing the human effort into the analysis of a building for specific purposes. This could be in recording the thermal performance of the building fabric to help analyse its energy use, or even the embodied carbon tied up in its materials and construction.The real technologi- cal advances in the sector will come when soft- ware developers can start to write algorithms to train computers (machine learning) to recognise certain property features, to assist in analysis and predictions. This is the approach that my team is pursuing with our new company Ambue. We have recog- nised the difficulties that residential property owners have to simply and inexpensively explore the potential options to retrofit their homes. We have built a simple software platform with tools to help them create a basic 3D model of their home. They can use this to test various retrofit options, get instant results and make informed investment decisions. The real revelation has come with testing this tool on historic properties. We built a bespoke model of a Grade II listed cottage in South Oxfordshire to run through our thermal modelling analysis. The home was heated by an oil-fired boiler, a converted oil-fired Aga and open fireplaces. Our software analysis showed that the annual carbon dioxide emissions were equivalent to over 17 tons, compared to the UK domestic average of around five tons. The analysis showed a range of potential measures to reduce the emissions, ranging from internal wall insulation, floor heating, secondary glazing and loft insulation. Certain measures are problematic, in terms not only of disruption but also of visual intrusion and impact on the historic fabric. But we were able to recommend a combination of non-obtrusive loft insulation, secondary glazing to selected windows and an air-source heat pump, which could reduce emis- sions by nearly half. As more properties and datasets are modelled, collected and analysed, we will be able to train the software to look at the best-practice solutions from comparable buildings, with performance data collected from real installations. This will be used to refine the advice and accuracy of the energy-saving predictions, assisting more building owners in making the best-informed decisions about their properties. Hamish McMichael, an RIBA conservation architect, is former director at BGS Architects and co-founder of Ambue.