Home > Latest

Key Bridge Reconstruction: Engineers Look to AI, Machine Learning to Accelerate Rebuild Timeline

Key Bridge Reconstruction: Engineers Look to AI, Machine Learning to Accelerate Rebuild Timeline

By Arnab Ghosh, Global Sales Engineering Director at Accuris

The collapse of the historic Francis Scott Key Bridge in Baltimore in March has left the city and nation grappling with the daunting task of reconstruction. More than a symbolic loss, the bridge’s destruction has disrupted a critical link in the eastern corridor’s supply chain, guaranteed to cause havoc on both business operations and daily life within the region. As recovery efforts continue amidst heaps of debris, engineering experts estimate that a complete rebuild adhering to stringent safety standards could take a staggering 10-15 years using existing methodologies and design approaches.

Traditionally, architectural engineering and construction (AEC) projects, specifically for large infrastructure such as the building of bridges, have extensive and rigid design and planning phases. For over 200 years, the engineering discipline has formalized their approach to consider the hundreds of factors at play during such a process. This includes impacts on traffic flow, determining how precast concrete segments are manufactured and delivered, site-specific factors such as soil conditions, outlining necessary land purchases, and considering environmental elements, such as climate change and effects on wildlife. All while ensuring that projects adhere to both local and national standards and regulations.

Therefore, it comes as no surprise that it could be more than a decade before Key Bridge is back in operation. Daily, engineers face a painstaking process in trying to locate the right tools, materials, and suppliers that fit their complex designs while ensuring compliance with ever-evolving regulations—a needle-in-a-haystack endeavor for even the brightest minds. The longevity of critical infrastructure construction is often the topic of public conversations but the rebuilding of a widely utilized transport channel adds serious urgency and complexity. So how can engineers eliminate some of this strain and complete projects on significantly reduced timelines?

AI Enters the Chat

Artificial intelligence and generative AI tools such as ChatGPT and Jasper.ai have captivated the attention of organizations across all industries, especially over the past year. And they have earned the right to do so, boasting the potential to improve efficiencies and processes to save valuable time and resources. While construction and engineering sectors have been somewhat slower to modernize their digital infrastructure, leveraging AI and machine learning (ML) can significantly streamline design and construction processes, particularly for high-priority, large-scale, and intricate projects, such as the reconstruction of the Key Bridge.

AI and ML are strongest in the presence of large data sets, where the technology can evaluate existing information and work to synthesize such data and provide recommendations. Construction, manufacturing and engineering are sectors rich with troves of regulations, historical data, project execution documents, best practices, lessons learnt, and design blueprints—making them ideal candidates for the effective application of AI and ML.

Engineers spend more than 40 percent of their time searching for and processing information when they don’t have credible sources rightly available, according to recent market research by Accuris. Obtaining accurate information in a timely, concise manner is an age-old pain point for many engineering professionals.

With the Key Bridge, and most other critical infrastructure projects, dissecting endless local, state and national regulations and standards is perhaps the most challenging aspect of the design and build process. Deploying AI and ML algorithms and technology can help engineers sort through millions of pages of standards, codes, and regulations, and compare and share versions of standards as they change. Information-gathering tasks that could have taken days in the past can now be completed in a matter of minutes. New natural language processing tools can help engineers turn their questions (for example, “Are there alternative materials I can use for this pipe that would still meet the pressure requirements of the current standard?”) into quick and accurate answers without having to write a complicated prompt.

AI and ML can also be used to optimize the design process to create new, higher quality concepts faster. By uploading previous blueprints and plans to an AI algorithm and embedding the specific parameters of new projects (weight capacities, temperature changes or earthquake protection, etc.), engineers can quickly pinpoint possible development approaches based on the analysis of past data. In addition, it can enable engineers to predict the future performance of such infrastructure. Training machine learning data models on historic data can support long-term inspection and maintenance costs for structures and other assets so that regions can allocate specific materials and resources to these causes.

In terms of planning and scheduling construction activities, AI technology also presents significant advantages to bolster material selection practices. For highly specialized, regulated projects, engineers spend months in search of appropriate suppliers and materials, oftentimes causing expensive delays to construction. AI can assist greatly in this realm to quickly catalog where to obtain necessary parts and develop best approaches to deliver such goods. AI enabled project planning and scheduling algorithms can leverage past project plans, budgets, and schedules to optimize construction sequences, resource allocation, and logistics, leading to more efficient project timelines and reduced costs. AI models trained on historical project data can also be used to identify potential risks, bottlenecks, and areas for improvement, enabling proactive risk mitigation strategies.

Beyond design planning and construction, AI can improve the safety of critical infrastructure. In terms of Key Bridge, having cameras on the bridge, where data is ingested by a computer vision AI algorithm that can automatically calculate deviations in normal surrounding activity and alert necessary services or close the approaching traffic to the bridge, could prevent similar instances and save lives. Airports have already deployed such technology to identify suspicious packages for removal.

Best Practices and Limitations 

The success of AI and ML algorithms in AEC projects largely depends on the information that is provided to it. To avoid design errors and regulatory inaccuracies, data input must be of high quality as well as up-to-date and accurate.  Data governance is an essential component of this process, and implementing policies and guardrails to ensure proper data quality throughout this process is essential for avoiding misconceptions. Education amongst engineers and project managers is also pivotal to the concept’s success and teams must be aware of its strengths and limitations.

Cybersecurity is another important factor, especially when leveraging AI technology for designing national critical infrastructure like bridges and airports. Cyber criminals may penetrate AI software to gather intel about foreign nation states to execute attacks. Organizations must proceed with caution about the data that is provided to such tools as AI remains a largely unregulated industry. Creating strict federal regulations is critical to the AEC industry, especially as more organizations start to leverage AI technology.

Explainability and interpretability of generative AI models is a topic of great discussion and must be prioritized especially in the AEC industry. The use of explainable AI techniques and interpretable machine learning models, which can provide insights into engineer decision-making processes and enable better human understanding is critical to ensure project deliverables are trustworthy.

Stakeholder engagement and continuous education of engineers, contractors, regulators, and the public, to promote understanding, trust, and acceptance of AI technologies in infrastructure projects is also a must for the AEC industry to successfully benefit from this AI boom.

AI will not replace humans—rather, it will act as a companion to streamline workflows and spark creative recommendations. It certainly boasts the potential to completely transform AEC projects like the Key Bridge rebuild, helping engineers to design and construct critical infrastructure quicker and more efficiently than ever before. The technology will ultimately help engineers accelerate material selection and keep pace with evolving safety standards, to deliver smarter, compliant designs in a fraction of the time.