A Review of AI in the Construction Industry
The construction industry is facing some big challenges. Underneath some of the superficial issues caused by COVID such as material shortages, delivery delays impacting on projects, there is something far worse. Labour-productivity growth for the construction industry has only grown by 1% over the past two decades, compared to 2.8% productivity growth for the whole world economy. Other sectors are rapidly growing, such as retail which has reinvented itself for the digital area and manufacturing which has thrived off the back of industrial automation and digitisation. One of the reasons for this lag is the construction industry is slow to adopt new technologies and invest in digitisation. Undergoing this transformation can be daunting, in part because it’s not always clear what the innovation opportunities are or which will deliver the greatest operational and commercial value.
Artificial Intelligence (AI) is a term used to describe when machines simulate human problem solving by identifying patterns and trends within data in order to perform tasks that normally require human intelligence. Their diverse range of automation capability means that it can simplify and streamline the planning, design and build stages in construction. AI has been successfully implemented in many other industries and could increase the construction sectors productivity by up to 60%, adding $1.6 trillion in value, approximately 2% of the global economy. Some companies have seen the operational, environmental and commercial benefits of AI and are already begun to take action to adopt this capability.
Taking a leaf out of the success stories in the retail and finance industries, construction firms should invest in AI and ML to reduce costs, improve productivity, process and people efficiency and grow margins.
Generative Design
Building Information Modelling (BIM) is a process and tools for generating and managing digital representations of buildings. BIM is readily available and being used by many construction companies. The 3-D model-based software gives architecture, engineering and construction (AEC) professionals insights into efficiently planning, designing, and constructing buildings and infrastructure. Autodesk Revit allows users to design whole buildings and internal components related to work schedules and specifications. The system is first given clearly specified design goals, after which it explores multiple variations of potential solutions to find the optimal design scenario. The challenge for construction firms using Revit is to ensure that different models are compatible and don’t clash when other teams’ designs are added on. Building System Planning has launched an add-on to Autodesk Revit, GenMEP, which solves this issue. The software focuses on the mechanical, electrical and plumbing (MEP) aspects of BIM to route electrical systems within the model whilst considering the complexity of shapes and geometries to ensure too many cables don’t end up on the same route. Another subset to complement GenMEP developed by BuildingSP is ClashMEP which detects clashes in real-time between different models, such as wall and electrical cabling. AI in this aspect can simplify and accelerate operations by optimising designs and ensuring cross-model compatibility. This would streamline the design phase of construction, aiding to increased productivity and process efficiency.
Robotics
Robotics, powered by ML, is an excellent tool to improve site control and management by helping with on-site tasks and monitoring buildings upon completion. Doxel uses robots and drones with cameras and LiDAR sensors to monitor and scan worksites, collecting data that can be used to improve construction productivity. The data is processed using algorithms to measure the currently installed quantities of a design and the rate of production of construction, by comparing to the planning and design parameters of the client. Komatsu also uses cameras and drones to provide accurate and real time insights of project progress and site conditions, measure on-site productivity, and provide improvements in the efficiency, accuracy, and quality in the final build. A conventional survey of a 20,000 m² job site could take up to 3 days, but with Komatsu’s Drones, a 3D point cloud can be available within 30 minutes.
Building Robotics has created Comfy which collects data from existing Building Management Systems (BMS) and considers users requests for hot or cool air to moderate and optimize temperatures in different parts of a building. Over time, the app identifies patterns and preferences based on time of day to automatically adjust temperature based on these trends. This increases cost savings and improves energy efficiency in commercial spaces to reduce carbon emissions and environmental impact.
Optical Character Recognition (OCR)
Optical Character Recognition is the electronic conversion of images of handwritten or printed text into machine-encoded text. Instead of manually typing or re-drawing documents, OCR technology can convert images into editable and searchable data. This makes it faster and easier to search databases for documents and drawings and make edits, saving both time and reducing costs. One application, Egnyte9, uses OCR to scan, number and name sheets, and link related sheets together, saving construction businesses considerable time and cost, and making it easier to group together projects and find archived projects. Considering the abundance of unstructured text data created throughout a construction project, OCR tools such as these can help simplify and streamline project management by putting accessibility to relevant information at the touch fingertips.
Analytics and Insights
AI can identify characteristics of previously successful applications and replicate these to increase a firm’s success rate in the tender process. The algorithms analyse the elements of successful previous bids and compiles data which can be used to predict the likelihood of scenarios, increase margins and drive project value. Dodge Data & Analytics has used natural language processing (NLP) to assess previous projects and data instead of having to start from scratch each time, creating projects that are most likely to be successful in increasing margins and profits.
Another example is Dassault Systèmes, which has incorporated analytics in two ways — EXALEAD and NETVIBES — to drive enhanced operations in construction. EXALEAD aggregates data from multiple sources and amalgamates it for us in 3D models, whilst NETVIBES takes data from media and news articles and tailors it to the organisation’s operations to create design strategies. These solutions take large volumes of data and sorts it into meaningful information that can be used to deeply understand and improve product development processes, and analyse industry trends. Both solutions offer intelligent data analysis and insights generation to effectively execute a design strategy, hence enhancing the operations in construction.
Carbon Calculators
Carbon calculators are a compelling new technology that predict the embodied carbon cost of a project. Using a scheme’s BIM and historic embodied carbon data from previous construction projects, Winvoc and UWE Bristol currently testing an AI system for Predicting Embodied Carbon in Construction (ASPEC). ASPEC performs analytics to identify hypothetical ‘what if’ design specifications that would lower the embodied carbon of a project through the selection of different materials. This would in turn lead to reduced environmental impact, material wastage, and improve project delivery speed.
Elliott Wood and the Institution of Structural Engineers (IStructE) has created an excel-based carbon calculator applicable to any sized project. This tool can work out the carbon footprint of a beam or column within minutes, or a basic structure within half an hour, by inputting material specifications and quantities. The calculator then provides where carbon hotspots are in a project and target those hotspots for alternative approaches either in design or material specification, to limit the carbon footprint of a project.
The carbon calculator tested by Balfour Beatty works on a more long-term scale compared to Winvoc and IStructE, by allowing users to input information from environmental product declarations sheets and verified and registered documents about the environmental impact throughout the life cycle of a product or material. This allows optimal design choices based on the carbon output throughout the lifecycle of a building, rather than just the construction phase. Currently in the beta testing phase, the initial tests show potential embodied carbon savings of up to 14% through better-informed design choices.
Carbon Re has used carbon calculators combined with feed rates, sensor data and control parameters to provide quantified recommendations to reduce the mass of CO2 emitted per useful heating value (kgCO2/UHV), resulting in lower fuel costs and lower emissions.
Final thoughts
There are a variety of AI and ML software and projects available and being developed in the construction industry that can be used to increase productivity, lower costs, and provide sustainable development and growth. The breadth and complexity of the construction sector and the many manual tasks involved makes it a prime candidate for automation, hosting benefits of overcoming labour shortages, increasing staff productivity, and accelerating the design stage. As pressure for high quality and affordable housing grows in the future in response to population growth and increased demand, AI and ML will play a key role in meeting these expectations and challenges, as well as reduce costs, improved margin, and shorten project delivery times.
The current mindset of the construction industry is resistant to investing in these technologies and prefers to embrace traditional ways of constructing. Fear of the unknown, security concerns and resistance to change are currently barriers to implementing AI in construction, but as technologies continue to advance, the use of AI becomes more prominent as a requirement for industries that want to grow. This obstacle can be overcome by recognising that automation is something to complement and augment humans and helping to alleviate some of the smaller time-consuming tasks, to increase workforce productivity, but there is a need for a greater culture and mindset shift towards openness, experimentation and acceptance.
