Calling all building designers and contractors! How AI can improve your process
When it comes to construction projects, planning is everything — whether it’s a new hospital or an eight-unit condo.
“Any small mistake at the design phase has a domino effect and is expensive to fix once construction starts,” says Sobhan Kouhestani (MASc 19), a recent graduate of Concordia’s Gina Cody School of Engineering and Computer Science.
It’s an issue that Kouhestani wanted to address.
He created a way to improve the design phase by using process mining, a form of machine learning and artificial intelligence (AI).
“The result is better resource planning, reduction of time-consuming reworks, better monitoring of ongoing projects and improved collaborative design,” says Mazdak Nik-Bakht, assistant professor in the Department of Building, Civil and Environmental Engineering.
Nik-Bakht and Kouhestani recently co-authored a study published in Automation in Construction, a highly respected journal in the field.
Sobhan Kouhestani (MASc 19).
Bridging the gap between BIM and process mining
Their story starts at the digital drafting table.
In the design phase, many architects and engineers are using digital authoring software to make building information models (BIM) — “smart” 3D models that contain information like costs, scheduling and materials, such as the type of windows and number of floor tiles.
“It’s called digital twinning,” explains Nik-Bakht, who is also communication and outreach director of Concordia’s new Centre for Innovation in Construction and Infrastructure Engineering and Management (CICIEM).
“With this study, we were able to bridge the gap between BIM and AI by using process mining so that BIM managers get more functionality out of the digital twin. It’s one of a cluster of projects I proposed in 2017 that got funding from NSERC” — the Natural Sciences and Engineering Research Council of Canada.
Mazdak Nik-Bakht, assistant professor in the Department of Building, Civil and Environmental Engineering.
How it works
It’s already possible to take “snapshots” of the BIM as it evolves over time, using digital authoring software. When viewed as stand-alone files, these snapshots may not provide much information about workflow. However, when the files are compared automatically and continuously over time, they increase a project manager’s understanding of the workflow involved, thus adding value.
Kouhestani and Nik-Bakht saw an opportunity to improve BIM capability by automating an archiving and analyzing process.
“We created an algorithm to make ‘event logs’ that track changes in consecutive files,” says Kouhestani. “Once we created the event logs, we used them as input for process mining.”
Process mining adds functionality
Quickly and efficiently, process mining identifies a chain of events (also called process) by following the day-to-day activities (design) performed by the actors involved — in this case, designers. Accordingly, deviations and discrepancies in the design, workflow bottlenecks, material shortages and more can be identified automatically.
“We make the digital design process modular and track the changes over time,” Kouhestani reports. “Then AI flags the discrepancies.”
By analyzing the event logs, process mining also evaluates team and individual performance.
“BIM managers can use this information to identify, then break down existing silos and improve collaboration among team members,” adds Nik-Bakht. “Does it look like the mechanical engineer isn’t talking to the electrical engineer? Now they know.”
A new corporate archive
By automating the archiving process, construction companies have a handy corporate record of each project that doesn’t walk out the door when employees leave or retire.
“Organizational memory is like wisdom,” Nik-Bakht says. “It’s expensive to acquire in the construction industry and companies want to build on what they learned from previous projects. That’s hard to do without a digital archive of what happened, when.”
Both researchers stress that their study focuses exclusively on the design phase, but their methods are applicable to later phases in the construction process.
“And it’s scalable, so if you have 15 or 150 people on the team, it still works,” Kouhestani says. He’s one of several students working with Nik-Bakht who are currently looking at applications in the construction phase.
Funding for their study came from the Natural Sciences and Engineering Research Council of Canada.
Read the cited study: IFC-based process mining for design authoring.
Learn more about the Centre for Innovation in Construction and Infrastructure Engineering and Management (CICIEM) at Concordia’s Gina Cody School of Engineering and Computer Science.
