AI and data analytics – building the backend of construction

There is no single way to use data analytics and Artificial Intelligence (AI) in construction and to think of it as such will only limit its future power. 

With Hinkley Point C using a geographic information system (GIS) to map out its construction site, to HS2 using building information model (BIM) processes to simulate and alter the design, data analytics and AI have already arrived in construction. 

Their expansive presence in all stages of development, from design to pre-construction, all the way to maintenance, shows that these technologies are as essential to construction as a spreadsheet and a hard hat. 

However, the purpose of AI and data analytics in the industry cannot be boiled down to one reason, as we have not explored all use cases for them yet, and not all projects are the same. 

In this SPECIAL REPORT, we look at how AI and data are used in construction projects and the roadmap for the technologies. 

The problem 

“From an engineering perspective, we use Excel for everything,” said Sam Hardy, the digital consultant to Europe at international construction management company AECOM. 

“We extract the information from BIM models to Excel, from costing software, the quantities and takeoff software, everything into Excel – and then share it around within an Excel spreadsheet.” 

Sam and AECOM’s data, analytics and AI director, Andreas Galatoulas, are working to build on the use cases of these technologies and provide a more interconnected process for engineering and construction. 

From a reality perspective, a project has many moving parts and dozens of data points can emerge. 

For the project team, using different programmes or holding information in individual spreadsheets can waste time and leave members at other points in development, which could cause further health and safety risks on site. 

Fortunately, we can use AI to organise information into a single source of information that all project members can access and see. 

“[With AI] it is more about making sure everything is connected at the back-end so that information can move around smoothly, rather than there being 1000 different systems and everyone having a different version of the truth,” said Galatoulas. 

Building the system

“I want to highlight that we try to avoid that typical approach most people have of ‘let’s do some AI”, you cannot just do AI if you don’t have all the groundwork ready,” said Galatoulas. 

Building AI that is fit for purpose requires the collation and analysis of data, from the designer’s and engineers’ site plan to the sensors on the site itself; that data improves the ability of project managers to identify risks and developments. 

These data points can then be used to make predictive models that can inform better-suited design choices for a given project. 

“From a health and safety point of view, project teams are able to visualise the walk around on site as it currently is, without being on-site, thus removing that actual risk of being on construction sites, and those who haven’t had experience on construction sites can be remotely walking around and inspecting things,” said Hardy. 

“And, from a client perspective, they may not want to walk around on site all day; when there are large amounts of work happening, they can be in a meeting room and visualise the entire site remotely.” 

Standardisation?

With collating data, there is no one size fits all approach, but using processing algorithms like Optical character recognition (OCR) to develop metadata can remove the tedious practice of an analyst having to sit down and go through each document related to a single project. 

“I want to move away from a standardised approach with data and AI. The only reason we standardise data is so that we can understand them, not the computer. It’s a very human-centric approach to collecting data rather than a more-centric one,” said Galatoulas. 

From a financial perspective, cost plays a significant role in AI adoption. According to their recent financial accounts, companies like Balfour Beatty and Laing O’Rourke are putting millions of pounds of investment into their digital transformation budgets. 

Substantial budgets are necessary because firms must sponsor the data and the additional work of building specialised AI systems. 

In the long haul, AI aims to make construction more efficient by not using as many materials for design and engineering and minimising health and safety risks and on-site changes. 

The engineer of tomorrow 

With AI, data, and analytics comes the expansion of what modern engineers can create with their projects. 

“It’s gonna be an engineer who, rather than drawing on a pen and paper, will be able to focus on automating their design,” said Galatoulas. 

This transition into a more digitally literate engineer will take time, as technology changes rapidly and education and degree courses have yet to catch up to what is happening on site. 

To reduce this gap in expertise, AECOM launched its data academy to help professionals in the industry come to grips with emerging technologies and diversify their skill sets. 

“[The academy] is a bespoke approach for every single engineer based on the role they want to do and where they want to go,” said Galatoulas. 

“It focuses on upskilling on specific areas of interest, from visualisation, basic data analytics, Machine Learning (ML) and AI, all which lead to the learner achieving a master’s degree.”

Hardy added: “We also provide a digital career pathway, where someone can come in and say, “I’m an engineer with an engineering degree” but then they spark an interest in data science and data analytics. The academy allows those engineers to learn more about data science and analytics and boost their career and expertise in that space.” 

Future impact 

AECOM’s analytics and AI team is working on a fully automated AI project for ancillary works with an undisclosed water company. 

“We are moving from standardised components and computational design to a full AI-driven automatic design, which means utilising and combining rule-based modelling with previous historic design and trying to create a design that the engineers and an architect will make minimal changes to,” said Galatoulas. 

“We are not at the point that I would feel comfortable releasing something full-on AI in the design without using a designer because it’s a partnership between designers and engineers, data analysts, data scientists and AI experts.”

Once AI programmes can be entirely sufficient, without the need for data analysts, then comes the next step in the adaptation of having fully-automated infrastructure and design. 

“The main goal is to develop a good flow of data for data engineering purposes, taking the design through to construction to operations,” said Hardy. 

“We can use information that is developed throughout design and construction, to then input that into a smart building per se, or digital twin, and have a visual print form to visualise on a mobile phone or any device, anywhere. We can pull up the design and look at how the operating side works. We’ve done that on certain projects, where we can inspect mechanical equipment, plant air conditioning, and temperatures.” 

AI is inherently linked to the future as it is used on construction projects and informs the approaches of projects and schemes that come after it. 

From collecting the data to building the AI, the more information streamlined into software, the more it can predict and contextualise what a specific project should look like.

Through further specialisation, modern buildings become future-proofed, reducing the need for contractors to go back and fix problems they didn’t see during construction. 

If you found this article interesting, we recommend The case for AI in construction – key players and Benefits.