Case study: Digital twins, simulation & synthetic data — a city or utility pilot and the results so far

Cities and utilities across the United States are deploying digital twins to manage aging infrastructure, optimize energy distribution, and plan for climate resilience. Unlike traditional modeling approaches that rely on static snapshots of system behavior, digital twins integrate real-time sensor data with physics-based simulations to create continuously updated virtual replicas of physical assets. The results from early municipal and utility pilots reveal both the transformative potential of this technology and the practical challenges that procurement teams must navigate. This case study examines three pioneering deployments, synthesizes their measured outcomes, and distills transferable lessons for jurisdictions evaluating digital twin investments.

Why It Matters

The American Society of Civil Engineers estimates that the US faces a $2.6 trillion infrastructure investment gap through 2030, with water systems, electrical grids, and transportation networks all requiring significant upgrades. Digital twins offer a pathway to extend the useful life of existing assets while improving operational efficiency and reducing maintenance costs. According to a 2025 report from McKinsey, utilities that have deployed digital twins for grid management report 15 to 25 percent reductions in unplanned outages and 10 to 18 percent decreases in maintenance expenditures.

For procurement professionals, the relevance extends beyond operational efficiency. The Infrastructure Investment and Jobs Act (IIJA) allocated $65 billion for power grid modernization and $55 billion for water infrastructure. Federal grant programs increasingly favor proposals that include digital monitoring and predictive analytics capabilities. The Department of Energy's Grid Resilience Innovation Program specifically identifies digital twin technology as a qualifying investment for grant funding. Understanding what these pilots have actually delivered, rather than what vendors promise, is essential for structuring competitive proposals and negotiating realistic contracts.

The synthetic data dimension adds another layer of strategic importance. Many municipal systems lack the historical operational data needed to train machine learning models effectively. Synthetic data generation, using physics-based simulations to create realistic training datasets, enables AI deployment even in data-sparse environments. This capability is particularly relevant for climate adaptation planning, where historical data may not reflect future conditions.

Background and Context

Digital twin technology originated in aerospace and manufacturing, where companies like General Electric and Siemens built virtual replicas of jet engines and factory equipment to predict maintenance needs and optimize performance. The migration to municipal infrastructure began in earnest around 2020, driven by three converging factors: the declining cost of IoT sensors, the maturation of cloud computing platforms capable of running complex simulations, and growing regulatory pressure to improve infrastructure resilience.

A municipal digital twin typically consists of four layers. The physical layer includes sensors, meters, and monitoring devices deployed across the infrastructure. The data layer aggregates, cleans, and stores information from these sensors alongside historical records and external data sources such as weather feeds. The model layer contains physics-based simulations, statistical models, and machine learning algorithms that replicate system behavior. The application layer provides dashboards, alerts, and decision-support tools for operators and planners.

The US market for infrastructure digital twins reached $4.2 billion in 2025, according to MarketsandMarkets, with utilities accounting for approximately 35 percent of total spending. Water utilities represent the fastest-growing segment, driven by EPA consent decrees requiring improved monitoring of combined sewer overflows and lead service line replacements.

Pilot 1: New York City Department of Environmental Protection Water System

The New York City Department of Environmental Protection (NYC DEP) launched its digital twin pilot in 2023, creating a virtual replica of a 42-square-mile section of the city's water distribution network covering portions of Queens and Brooklyn. The pilot integrated data from 1,200 pressure sensors, 380 flow meters, and 85 water quality monitoring stations with a hydraulic model of 3,400 miles of water mains.

The primary objectives were threefold: reduce water main breaks through predictive maintenance, improve water quality monitoring response times, and optimize pumping operations to reduce energy consumption. NYC DEP partnered with Bentley Systems' OpenFlows platform and IBM's Maximo for asset performance management.

Measured outcomes after 18 months of operation:

The digital twin reduced unplanned water main breaks in the pilot area by 22 percent compared to the five-year historical average. The system identified 47 locations at elevated risk of failure based on pipe age, material, soil conditions, pressure transients, and historical break patterns. Of these flagged locations, 31 were proactively repaired or replaced before failure occurred. The predictive model achieved a positive prediction rate of 66 percent, meaning roughly two-thirds of flagged locations would have experienced failures within the subsequent 12 months based on statistical analysis.

Energy consumption for pumping operations decreased by 14 percent through optimized pump scheduling that accounted for real-time demand patterns, electricity pricing, and system pressure requirements. This translated to approximately $3.8 million in annual energy savings across the pilot zone. Water quality incident response times improved from an average of 4.2 hours to 1.7 hours, as the digital twin enabled operators to trace contamination events upstream and isolate affected zones more rapidly.

Total pilot investment was $28 million, including sensor deployment, platform licensing, systems integration, and staff training. NYC DEP estimates a payback period of 5.2 years based on avoided emergency repairs, energy savings, and reduced water losses.

Pilot 2: Austin Energy Grid Digital Twin

Austin Energy, the eighth-largest publicly owned electric utility in the US, deployed a grid digital twin in 2022 to manage the integration of distributed energy resources (DERs) across its service territory. With over 45,000 rooftop solar installations and growing battery storage adoption, the utility faced increasing challenges in maintaining grid stability and managing bidirectional power flows that traditional distribution management systems were not designed to handle.

The utility partnered with GE Digital's GridOS platform, integrating data from 520,000 smart meters, 3,200 distribution transformer monitors, and weather stations across the service territory. The digital twin simulates power flows across all 43 distribution feeders in near real-time, updating every 15 seconds.

Measured outcomes after 24 months:

Voltage violations on feeders with high solar penetration (exceeding 40 percent of transformer capacity) decreased by 38 percent. The digital twin enabled operators to identify impending voltage issues 2 to 4 hours before they would have been detected by traditional SCADA systems, allowing preventive action through capacitor bank switching, transformer tap changes, and targeted DER curtailment.

Transformer overloading events decreased by 27 percent through improved load forecasting that accounts for behind-the-meter solar generation and EV charging patterns. The digital twin's synthetic data capabilities proved particularly valuable here: by generating thousands of simulated scenarios combining extreme heat events with varying levels of solar generation and EV charging demand, Austin Energy identified 12 distribution transformers at risk of failure that traditional planning methods had not flagged.

Renewable energy curtailment decreased by 31 percent as the digital twin optimized the dispatch of utility-controlled battery storage systems to absorb excess solar generation rather than curtailing rooftop installations. This resulted in an additional 18 GWh of renewable energy delivered to the grid annually.

The total investment was $19.5 million over three years, with Austin Energy reporting that avoided infrastructure upgrades alone (deferred transformer replacements and feeder reconductoring) saved $14.2 million in the first two years.

Pilot 3: Miami-Dade Water and Sewer Department Resilience Planning

Miami-Dade County's Water and Sewer Department (WASD) initiated a digital twin project in 2023 focused on climate resilience planning for its wastewater collection and treatment infrastructure. With sea level rise projections of 10 to 17 inches by 2040 and increasing intensity of rainfall events, the department needed to evaluate the performance of its 8,800-mile sewer network under future climate scenarios that have no historical precedent.

WASD partnered with Autodesk's InfoWorks ICM platform and used synthetic data generation extensively to model conditions that the existing system has never experienced. The digital twin incorporates LiDAR-derived terrain models, groundwater elevation data from 340 monitoring wells, tidal gauge measurements, and rainfall intensity projections from downscaled CMIP6 climate models.

Measured outcomes after 15 months:

The digital twin identified 23 pump stations at risk of inundation under 2040 sea level scenarios, of which only 9 had been flagged in the department's previous capital improvement plan. The synthetic scenario analysis revealed that the compound effect of king tides, heavy rainfall, and elevated groundwater during wet season creates failure modes that traditional steady-state modeling cannot capture.

Using the digital twin's scenario analysis capabilities, WASD redesigned its $4.5 billion capital improvement program, reprioritizing $780 million in investments toward the most vulnerable assets. The department estimates this reprioritization will avoid $1.2 billion in reactive emergency repairs over the next 15 years compared to the original plan.

Operational improvements included a 19 percent reduction in sanitary sewer overflows (SSOs) during the 2024 hurricane season compared to the three-year average, achieved through proactive pump station staging and temporary flow diversion guided by digital twin predictions. Response time for SSO events decreased from 3.1 hours to 1.4 hours.

Total investment was $22 million, with additional annual operating costs of $3.2 million for data management, model updates, and platform licensing.

Key Lessons and Transferable Insights

Data Infrastructure Is the Critical Path

Across all three pilots, sensor deployment and data integration consumed 40 to 55 percent of total project budgets and represented the primary schedule risk. NYC DEP spent 14 months on sensor installation and data validation before the digital twin became operational. Austin Energy benefited from its existing smart meter infrastructure, reducing this phase to 8 months. Procurement teams should plan for data infrastructure as the dominant cost driver, not software licensing.

Organizational Change Management Cannot Be Underestimated

All three utilities reported significant resistance from operations staff accustomed to experience-based decision-making. Austin Energy invested $1.8 million in training programs and embedded digital twin specialists in control rooms for the first 12 months. Miami-Dade WASD created a dedicated Digital Infrastructure team of 14 staff members. Organizations that underinvest in change management consistently report lower utilization rates and diminished returns.

Synthetic Data Unlocks Climate Adaptation Use Cases

Miami-Dade's experience demonstrates that synthetic data generation is not merely a workaround for missing historical data but a strategic capability for climate adaptation planning. The ability to simulate conditions that have never occurred, and may not occur for decades, transforms digital twins from operational tools into strategic planning platforms. This capability directly addresses the limitations of traditional infrastructure planning methods that extrapolate from historical patterns.

Vendor Lock-in Risks Require Active Mitigation

Each pilot involved proprietary platforms with varying levels of data portability and interoperability. NYC DEP negotiated contractual provisions requiring data export in open formats (CityGML and IFC) and API access for third-party analytics tools. Procurement professionals should require open standards compliance and data portability guarantees as non-negotiable contract terms.

Start Focused, Scale Incrementally

All three pilots began with geographically bounded service areas and specific operational objectives rather than attempting enterprise-wide deployment. This approach allowed teams to validate data quality, refine models, and demonstrate value before requesting expanded budgets. Austin Energy's phased approach, starting with 5 of its 43 feeders before expanding, reduced risk and built internal confidence.

Action Checklist

• Conduct a comprehensive sensor and data infrastructure audit before issuing RFPs for digital twin platforms
• Require vendor proposals to separate data infrastructure costs from software licensing and professional services
• Negotiate data portability provisions and open standards compliance in all platform contracts
• Allocate 15 to 20 percent of total project budget for staff training and organizational change management
• Define 3 to 5 specific operational KPIs that the digital twin must improve, with measurable targets and verification methods
• Plan for a phased deployment starting with a bounded pilot area and expanding based on demonstrated results
• Evaluate synthetic data generation capabilities for climate resilience planning, particularly in coastal and flood-prone jurisdictions
• Establish data governance policies covering sensor data ownership, retention, and sharing agreements

FAQ

Q: What is a realistic budget range for a municipal digital twin pilot? A: Based on the three pilots examined, expect $15 to 30 million for a meaningful pilot covering a bounded service area with 1,000 to 3,000 sensor points. This includes sensor deployment (40 to 55 percent), platform licensing and integration (25 to 30 percent), and training and change management (15 to 20 percent). Annual operating costs typically run 12 to 15 percent of initial capital investment.

Q: How long before a digital twin delivers measurable operational improvements? A: Plan for 18 to 24 months from project initiation to verified operational improvements. This includes 6 to 14 months for sensor deployment and data validation, 3 to 6 months for model calibration and testing, and 3 to 6 months of parallel operation before the digital twin informs real-time decisions.

Q: Can digital twins integrate with existing SCADA and asset management systems? A: Yes, but integration complexity and cost vary significantly. Modern platforms from Bentley, GE Digital, and Autodesk provide standard connectors for common SCADA protocols (DNP3, Modbus, IEC 61850) and asset management systems (IBM Maximo, SAP PM). Legacy systems with proprietary protocols may require custom middleware costing $200,000 to $500,000.

Q: What staffing is required to operate a municipal digital twin? A: The pilots examined required 8 to 14 dedicated staff for ongoing operations, including data engineers, model specialists, and application administrators. Many municipalities partner with system integrators for the first 2 to 3 years while building internal capabilities. Budget $800,000 to $1.5 million annually for dedicated digital twin operations staff.

Q: How does synthetic data generation work for infrastructure planning? A: Physics-based simulation engines create virtual datasets representing conditions the real system has not experienced, such as extreme weather events, future demand growth, or sea level rise scenarios. These synthetic datasets train machine learning models and stress-test infrastructure designs without waiting for actual extreme events. The approach is particularly valuable for climate adaptation, where historical data is an unreliable predictor of future conditions.

Sources

• McKinsey & Company. (2025). Digital Twins in Infrastructure: From Pilots to Scale. New York: McKinsey Global Institute.
• American Society of Civil Engineers. (2025). Report Card for America's Infrastructure: 2025 Update. Reston, VA: ASCE.
• MarketsandMarkets. (2025). Digital Twin Market for Infrastructure: Global Forecast to 2030. Pune, India: MarketsandMarkets.
• US Department of Energy. (2025). Grid Resilience Innovation Program: Technology Eligibility and Selection Criteria. Washington, DC: DOE.
• New York City Department of Environmental Protection. (2025). Water Distribution Network Digital Twin: 18-Month Performance Report. New York: NYC DEP.
• Austin Energy. (2025). Grid Digital Twin for Distributed Energy Resource Management: Two-Year Evaluation. Austin, TX: Austin Energy.
• Miami-Dade Water and Sewer Department. (2025). Climate Resilience Digital Twin: Phase 1 Results and Capital Program Optimization. Miami, FL: Miami-Dade County.
• Bentley Systems. (2025). OpenFlows Municipal Water Digital Twins: Implementation Guide and Performance Benchmarks. Exton, PA: Bentley Systems.