AI in the construction industry
Published 12.6.2024
How far can we get with AI in the construction of large works?
Hello and welcome to our 7th newsletter!
Many of the ideas that science fiction laid before us between the end of the 19th century and the end of the 20th, have become reality: man landed on the moon, we have holograms, drones and even robots themselves.
The current context is reminiscent of Isaac Asimov and his Three Laws of Robotics, which proposed a set relationship between man and robots, machines equipped with artificial intelligence. Did these visionary writers imagine that only a few decades later, the universe they had dreamed of would come true?
Although AI has been under development for several years, 2023 will be remembered as the year when this (r)evolutionary technology hit the scene, irrevocably changing the way things are done in industry and in much of society.
What it can do, and the expectations about how it can be applied, have given it a universal reach, perhaps as important as the steam engine or even Internet itself. We have all seen how what’s called generative artificial intelligence can be applied to create images, improve texts or translations and even make songs, videos and animations. In fact, we’re using it ourselves, and not to write songs, but to ask questions about our customers’ contracts.
Engineering and construction are not immune to the advances of AI. Whether it’s at a gas treatment plant in Vaca Muerta or a gas pipeline in Córdoba (both in Argentina), or at a water desalination project in Chile and even at a refinery in Mexico, we’re already using this technology to great effect. Here’s how we’re doing it.
In Vaca Muerta, Argentina, we are trying out an app to measure physical progress using multidimensional virtual models, made from laser scans and processed with Artificial Intelligence. A camera scans everything, from people and scaffolding to signage cones and structures, and the inbuilt software learns to recognize what it sees through a process of categorization.
Then, it produces an accurate three-dimensional model on the basis of a file containing the coordinates of that object and compares it with its real-world status: the tool compares the points of coincidence between the engineering model and the actual results. Using geospatial 3D point cloud technology to create precise models of reality, the tool gives us a much more accurate view of the physical progress of the work. And it optimizes measurement and planning, enabling us to save time as well as gain efficiency and practicality.
At the same time, in Córdoba, Argentina, we are testing an app to help us reduce the time we spend in manually redrawing the piping and instrumentation diagrams (P&ID) we receive from our customer, which often lack any attributes or metadata and are thus considered “not smart”.
Using Artificial Intelligence, the app processes diagrams with no attributes or metadata to generate a large percentage of the diagram in just a few seconds (approximately 70% complete), thus saving the many hours of manual work often spent on these processes.
At a project in northern Chile, we’ve started testing a tool that also works with AI and a 3D point cloud, representing the current context of the work in graphic form. This result provides greater precision when it comes to manufacturing the spools (which takes place in Santiago before transport to the construction site) and, by association, improves the assembly process.
The application measures the relationship between each part with millimeter precision, enabling us to rethink the work before executing it on location, as instead of estimating the space between each flange, for instance, it enables us to have a precise scan of each spool.
Finally, at our projects in Chile and Mexico, we are applying machine learning to the company’s expediting model, which will help us to obtain a far more accurate timetable when planning the arrival of materials. The model is using previous supplier delivery-date data together with historical commodity data to calculate the deltas for delivery dates and, consequently, improve project planning.
How far will AI go?
The future is coming towards us in giant steps, just as Asimov predicted or as Turing envisaged. We still don't know if AI will be able to build our next gas pipeline in the Andean mountain range, but we do know that it is already making a significant contribution to our projects, and that we will continue working and relying on this technology. The aim is to improve our processes, speed up timing, and continue driving this disruptive technology in the engineering and construction industry.
Clarification: this Newsletter was not written with artificial intelligence (for now!), but we did use it to design the cover image.
Until next time!
The Communications team at Techint Engineering & Construction