In a Gemini Call presentation, Professor Abdel-Rahman Tawil and Ahmad Wahab from Birmingham City University showcased their work on the DIATOMIC project—a sophisticated digital twin platform that’s revolutionising how cities monitor and manage traffic and air quality through AI-driven insights.
From Data Chaos to Urban Intelligence
Birmingham faces a challenge familiar to cities worldwide: hundreds of sensors generating continuous streams of traffic and air quality data, yet struggling to transform this information into actionable insights. The Traffic and Air Quality Digital Twin (TACDT), developed as part of the DIATOMIC initiative, represents a transformative solution to this urban data deluge.
Professor Tawil, leading the BCU team, introduced TACDT as more than just another monitoring system. “At its core, TACDT is a platform that ingests real-time data from hundreds of sensors deployed across the city, covering everything from vehicle flow to air quality levels,” he explained. The platform creates a virtual replica of Birmingham’s Clean Air Zone, enabling city planners to simulate, predict, and optimise urban systems in ways previously impossible.
The DIATOMIC Foundation: Three Years of Innovation
The DIATOMIC project—standing for Data Innovation Accelerating Transformative Organisational Metasystems for Impactful Change—represents a collaboration funded by Innovate UK and the West Midlands Combined Authority. Working alongside Connected Places Catapult, three Birmingham universities, and industry partners Siemens, the team has built a composable digital twin platform that’s been operational since September 2024.
Now in its third year, DIATOMIC focuses on leveraging the platform within Birmingham’s Knowledge Quarter and Wolverhampton’s Green Innovation Corridor investment zones, fostering cross-sector collaboration and data-driven policymaking to drive economic growth.
Architectural Excellence: Building for Scale and Flexibility
The platform’s layered reference architecture demonstrates sophisticated engineering thinking. Starting from the city layer (physical sensors and IoT devices) and building through data management, information modelling, AI/ML analytics, and application layers, the system culminates in community engagement interfaces.
This modular approach ensures the platform can adapt to new requirements without fundamental restructuring. As Professor Tawil noted, “This layered design allows us to manage complexity whilst enabling a wide range of urban applications.”
At the heart of the system lies a hexagonal architecture pattern for the digital twin engine—a design choice that ensures flexibility and maintainability. “This architecture style ensures that we can flexibly plug in new data sources, use cases, or AI modules without changing the core logic,” Tawil explained.
AI Canvas: Democratising Urban Analytics
Ahmad Wahab’s demonstration revealed the platform’s most innovative feature: an AI Canvas that transforms how cities approach predictive modelling. Rather than requiring data scientists for every analysis, the platform enables users to create localised AI models by simply selecting relevant sensors on an interactive map.
“One of the greatest challenges we’ve encountered when working with traffic and air quality is how do we model the data from individual sensors to holistic models that can include all sensors,” Wahab explained. The solution allows users to drag and drop sensors—red markers for traffic counters measuring vehicle numbers and speeds, green for air quality monitors tracking NO₂, PM2.5, and other pollutants—creating custom prediction models in under a minute.
The LLM Revolution: Natural Language Meets Urban Data
Perhaps the most striking innovation is the integration of Large Language Models (LLMs) for data exploration. Rather than navigating complex databases or writing SQL queries, users can simply type natural language questions like “Show me the top 10 correlations with traffic sensor RTM when there are around 17 buses.”
The system responds with not just data, but contextualised insights including:
- Summary of results with key findings
- Alignment verification confirming the response matches the original query
- Limitations acknowledgement highlighting any constraints in the analysis
- Follow-up suggestions recommending next analytical steps
This approach addresses a critical challenge in urban analytics: with over 200 sensors generating continuous data, finding meaningful correlations manually would be virtually impossible. The LLM-powered interface democratises access to these insights, enabling policymakers without technical backgrounds to explore complex urban dynamics.
Real-World Impact in Birmingham’s Clean Air Zone
The platform’s deployment in Birmingham’s Clean Air Zone demonstrates tangible benefits for urban management:
Predictive Power
By combining historical data with real-time feeds, TACDT can forecast traffic patterns, pollution spikes, and the impact of interventions like school holidays on air quality. This predictive capability enables proactive rather than reactive city management.
Scenario Testing
City planners can simulate various scenarios—from changes in traffic flow patterns to the introduction of new vehicle types—understanding impacts before implementing potentially costly changes.
Cross-Domain Insights
The platform reveals hidden connections between seemingly unrelated factors. For instance, the correlation analysis might show how traffic patterns at one junction influence air quality at a school several streets away, insights that would be impossible to detect through traditional monitoring.
Experimental Frontiers: Augmented ETL and Automated Reporting
The team is pushing boundaries with experimental features that could reshape urban data management:
Augmented ETL Processes: Using generative AI to automate data pipeline creation, reducing the technical burden of integrating new data sources.
Dynamic Dashboard Generation: LLMs that can create custom visualisations on demand, adapting to specific stakeholder needs without manual configuration.
Automated Reporting: AI-generated reports that translate complex technical data into accessible narratives for different audiences, from technical teams to public consultations.
Lessons for Digital Twin Practitioners
The DIATOMIC project offers valuable insights for organisations embarking on digital twin initiatives:
- Start with Clear Use Cases: Focus on specific challenges (traffic congestion, air quality) rather than attempting to model everything simultaneously.
- Embrace Composable Architecture: Build flexibility into your platform from the start—requirements will evolve.
- Democratise Access: Complex analytics become valuable only when stakeholders can actually use them. Natural language interfaces remove barriers to adoption.
- Validate Continuously: The platform’s emphasis on alignment verification ensures AI-generated insights remain trustworthy and relevant.
- Think Ecosystem, Not Silo: Data becomes more valuable when shared. DIATOMIC’s platform approach ensures any application plugged into the system can access cleaned, validated, structured data.
The Road Ahead: Scaling Urban Intelligence
As cities worldwide grapple with urbanisation, climate change, and the need for sustainable growth, Birmingham’s TACDT offers a blueprint for transformation. The platform demonstrates that the challenge isn’t collecting data—modern cities are awash with it—but rather transforming that data into intelligence that drives better decisions.
The integration of LLMs represents a paradigm shift in how cities can interact with their digital twins. No longer confined to technical specialists, these powerful analytical tools become accessible to anyone who can formulate a question in plain English.
For Birmingham, this means cleaner air, smoother traffic, and evidence-based policies. For the broader digital twin community, it represents proof that sophisticated urban analytics can be both powerful and accessible, setting a new standard for smart city initiatives worldwide.
The Digital Twin Hub’s weekly Gemini Calls continue every Tuesday at 10:30am BST, showcasing innovations that are shaping the future of connected digital twins. All sessions are recorded and available to community members.
Ready to explore how AI-powered digital twins can transform your city’s data into actionable intelligence? Join the Digital Twin Hub community to connect with innovators leading this urban revolution.
Leave a comment