Introduction

Cities are undergoing a fundamental shift in how energy is produced, distributed, and consumed. Electric vehicles (EVs), rooftop solar, and smart devices are no longer niche technologies—they are becoming part of everyday urban life.

Yet the way we model energy systems has not kept pace.

For decades, planners and utility operators treated buildings as “black-box” loads. In simple terms, this means we only cared about how much electricity a building used—not when, why, or how that usage might change. To estimate demand, models relied on averaged assumptions like peak demand factors, which smooth out real-world variability.

That approach worked when energy consumption was relatively stable.

It no longer does.

Modern energy use is:

  • Dynamic — constantly changing throughout the day
  • Stochastic — partly unpredictable due to human behavior
  • Correlated — many users acting at the same time (for example, plugging in EVs after work)

As cities commit to Net-Zero emissions, these simplifications become a liability. We need models that reflect reality more closely—models that understand not just infrastructure, but also behaviour and policy.

This is where Digital Twins come in.

A digital twin is a virtual representation of a real system that evolves over time. But for urban energy systems, it must go beyond a 3D model. It must connect three critical dimensions:

  • The physical world (buildings, grids, devices)
  • The human layer (occupancy, mobility, behaviour)
  • The governance layer (zoning, policies, regulations)

This idea aligns with research on Cyber-Physical-Social Systems (CPSS), which emphasizes that human behaviour—such as travel patterns and occupancy—is essential for understanding energy demand, especially with EV adoption (Zhou et al., 2020).

In this post, we explore a modelling approach called a relational digital twin. The goal is not just to predict energy demand, but to create a system that can anticipate, adapt, and actively manage it.

The Physical and Governance Layer — Building the Urban Skeleton

Every digital twin starts with structure.

Before we can simulate behavior or predict demand, we need a clear representation of what exists in the real world. This foundational layer can be thought of as the urban skeleton—a structured map of the district and the rules that govern it.

The model begins at a broad level and gradually becomes more detailed. At the top is the district, representing the overall area of interest. This is divided into parcels, which correspond to individual pieces of land with defined ownership and boundaries. Each parcel may contain one or more buildings, and within those buildings are the actual energy-consuming or energy-producing devices—such as heating systems, EV chargers, or solar panels.

To keep the model practical, buildings are not described in exhaustive detail. Instead, they are assigned archetypes, such as “post-1990 residential” or “commercial office.” These archetypes carry typical energy characteristics, allowing the model to remain realistic without requiring detailed data for every structure (Abbasabadi & Ashayeri, 2019).

At the lowest level, meters connect devices to the grid. This creates a continuous chain that links individual pieces of equipment to the broader energy system. In effect, the model can trace electricity use from a single device all the way up to the district level.

However, cities are not defined by physical structures alone—they are shaped by rules.

This is where the governance layer comes in. Zoning regulations determine what can be built and where. Land use policies define whether an area is residential, commercial, or mixed-use. Development phases introduce a time dimension, allowing the model to simulate how a district evolves over years or decades.

These constraints ensure that the digital twin remains grounded in reality. For example, a high-density commercial building cannot suddenly appear in a low-density residential zone. Similarly, infrastructure growth must align with development timelines.

By linking physical assets with regulatory constraints, the model ensures that every simulated outcome is both technically feasible and legally plausible. This is what distinguishes a true digital twin from a simple visualization.

The Mobility–Energy Nexus — Bringing the System to Life

With the physical structure in place, the digital twin has a body—but it is still static.

What brings it to life is movement.

In modern urban energy systems, the most important driver of change is the Electric Vehicle (EV). Unlike traditional loads, EVs are mobile. They connect transportation networks with the electricity grid, creating a new and complex relationship between how people move and how energy is consumed.

To understand this, consider a simple scenario. A vehicle arrives at a residential building in the evening. In a traditional model, this might immediately translate into an increase in electricity demand. But in reality, several conditions must align before charging occurs.

The person must be home. The vehicle must need charging. The user must decide to plug in—or allow the system to do it automatically.

This is where the model introduces a critical concept: behaviour.

Instead of treating devices as the sole drivers of demand, the digital twin separates the physical asset (the charger) from the user’s intent (how and when it is used). This distinction allows the system to represent real-world variability much more accurately.

Traffic data provides information about when vehicles arrive in different parts of the district. Occupancy data indicates whether people are present in buildings. When these two are combined, the model can determine when charging is likely to occur—and when it is not.

This alignment is essential for capturing what engineers call the diversity factor: the fact that not everyone uses electricity at the same time (He et al., 2022). Without this, models tend to overestimate peak demand.

The real power of this approach, however, lies in flexibility.

Each user or group of users can be assigned a behaviour profile, which defines how they interact with energy systems. Some users may prefer immediate charging, while others may allow their charging to be delayed. This willingness to adapt is captured as a flexibility factor, ranging from no flexibility to full participation in managed programs.

When the grid is under stress, the system can identify which loads are flexible and adjust them accordingly. Charging may be delayed, reduced, or shifted to a different time—all while ensuring that user needs are still met.

In this way, EVs become more than just a source of demand. They become a tool for balancing the grid.

This is the essence of the mobility–energy nexus: a dynamic interaction where transportation behavior directly influences energy systems, and vice versa.

The Intelligence Layer — From Prediction to Intervention

Up to this point, the digital twin has a body and a pulse. It knows what exists in the district and how people move and behave within it. But insight alone is not enough.

A truly useful digital twin must answer a harder question:

What is going to happen next—and what should we do about it?

This is where the Intelligence Layer comes in.

At the center of this layer is the forecasting engine, which brings together multiple streams of data to predict how the grid will behave over time. Weather conditions influence heating and cooling demand. Occupancy patterns determine baseline energy use. EV charging behavior introduces variability tied to mobility.

By combining these inputs, the model produces a time-based view of electricity demand across the district.

What makes this powerful is its level of detail. Instead of providing a single aggregate number, the model identifies where stress will occur within the network. It can reveal hidden risks—future overloads that emerge from the interaction of growth, behavior, and infrastructure. These “ghost peaks” are often invisible in traditional models until it is too late.

But prediction alone does not solve the problem.

When the model detects that a constraint is about to be violated, it activates a Distributed Energy Resource (DER) program. This is the mechanism through which the system intervenes.

Using its relational structure, the twin identifies which assets are connected to the affected part of the grid and determines which of them are flexible. EV chargers, batteries, and other controllable devices can then be adjusted—shifting demand in time rather than increasing supply.

These actions are governed by predefined policy constraints, such as maximum grid capacity or emission limits. The system continuously evaluates whether it is operating within acceptable boundaries and adjusts accordingly.

The result is a closed-loop system that does not just observe the grid, but actively manages it.

The Economic Layer — Turning Grid Events into Decisions

Technical solutions alone are not enough. For any system to work in practice, it must align with economic reality.

The Economic Layer ensures that the behaviours modelled in the digital twin are not only possible, but also desirable from a financial perspective.

In traditional systems, electricity pricing is often static. But in reality, the cost of supplying electricity varies depending on demand. During peak periods, the grid is under stress, and the cost of meeting that demand increases.

To reflect this, the model introduces dynamic pricing. When demand is high, prices rise. When demand is low, prices fall. These signals are fed back into user behaviour.

For example, a user who allows their EV charging to be managed may benefit from lower energy costs. In contrast, charging during peak periods may become more expensive. Over time, these signals encourage users to shift their behaviour in ways that support grid stability.

Beyond pricing, the model can also account for avoided infrastructure costs. If a coordinated demand response program prevents the need for a costly grid upgrade, the savings can be quantified and redistributed as incentives. This creates a direct link between system-level benefits and individual rewards.

The model also incorporates carbon economics, assigning a cost to emissions when carbon limits are exceeded. This influences both operational decisions and long-term planning, encouraging cleaner energy use and investment in low-carbon technologies.

By integrating these economic signals directly into the simulation, the digital twin becomes more than a technical tool—it becomes a decision-making framework that reflects how real-world systems operate.

Bringing It All Together — From Simulation to Strategy

Each layer of the digital twin serves a purpose, but its true value lies in their integration.

When physical infrastructure, human behaviour, predictive intelligence, and economic signals are combined, the model becomes a powerful tool for scenario analysis.

Consider a proposed urban development. On its own, it may appear feasible. But when placed within the digital twin, its interactions with the broader system become clear. The model can simulate how it affects traffic patterns, how it changes energy demand, and how it interacts with existing grid infrastructure.

Now extend this further.

What happens during a future heatwave, when cooling demand surges and EV charging peaks at the same time? Can the system handle the load? If not, can flexibility programs mitigate the impact? And if those programs are implemented, will users participate given the economic incentives?

These are complex, interconnected questions—and they cannot be answered in isolation.

The digital twin addresses them by running integrated scenarios and evaluating outcomes across multiple dimensions. It can assess technical performance, economic viability, and policy compliance simultaneously. In doing so, it provides a structured way to test whether a strategy is not just desirable, but achievable.

This ability to test policy against physical reality is what sets relational digital twins apart.

They transform models from passive representations into active decision-support systems, enabling cities to plan with confidence in an increasingly uncertain future.

Conclusion

Urban energy systems are becoming more complex, driven by electrification, decentralization, and human behaviour.

Static models, built on simplified assumptions, are no longer sufficient.

Relational digital twins offer a new approach—one that connects infrastructure, behaviour, policy, and economics into a unified system. By doing so, they enable cities to anticipate challenges, optimize operations, and make informed decisions about the future.

In the transition to Net-Zero, this is not just an advantage.

It is a necessity.

References and Further Reading