Modeling Urban Energy Demand as a Continuous Fluid Flow

17 concepts7 papers1 proposal

Traditional urban energy forecasting treats cities as collections of isolated, discrete nodes. Whether analyzing individual buildings or aggregating regional smart-meter data into isolated time-series vectors, conventional models often ignore the underlying spatial geography. As noted by Nan Li et al., these discrete approaches frequently lack the spatial and temporal resolution required for highly detailed grid management.

The core limitation of a purely time-series perspective is that it overlooks how spatial characteristics—such as building layouts and urban geography—interact with human behavior (Abrar Ahmed et al.). Flattening a city's complex, multi-dimensional spatial relationships into one-dimensional representations obscures the dynamic reality of how energy use moves (Shenli Jia et al.). In reality, urban energy consumption is not static; it migrates across a city's landscape as populations move throughout the day.

To capture these dynamics, a recent research direction reimagines urban energy tracking. Instead of viewing energy demand as isolated data points on a map, this framework treats electricity consumption as a continuous, flowing fluid.

The City as a Continuous Demand Field

To bridge the gap between spatial layout and temporal shifts, the authors propose transforming discrete smart-meter data into continuous 2D "density maps" (Shenli Jia et al.). By preserving the two-dimensional spatial structure of a city throughout the analytical process, researchers can observe how energy hotspots emerge, grow, and dissolve over time.

However, simply observing static snapshots of energy density is not enough to understand urban dynamics. The true value lies in modeling the migration between these hotspots. As crowds of citizens move from commercial centers during the day to residential suburbs in the evening, the load on the electrical grid shifts accordingly. To map this migration, researchers have adapted potential flow theory—a core framework borrowed from fluid mechanics—to model the spatio-temporal dynamics of urban energy consumption (Xiufeng Liu 0001 et al.).

The Intuition of Potential Flow Mapping

How do you transform discrete electricity readings into a fluid velocity field? The mechanism relies on treating changes in consumption density as a potential function.

[Time T1: Density Map]  \
                         -->  [Calculate Difference Map]  -->  [Compute Gradient]  -->  [Vector Flow Field]
[Time T2: Density Map]  /

As outlined by Xiufeng Liu 0001 et al. and Xiufeng Liu 0001 et al., the process follows three main steps:

  1. Continuous Density Estimation: Discrete consumption points are converted into continuous surfaces using techniques like Kernel Density Estimation (KDE). This creates a smooth landscape of energy demand across the city for any given time interval.
  2. Differential Mapping: The framework calculates the difference between two sequential density maps (e.g., comparing 2:00 PM to 6:00 PM). This highlights areas where energy demand has accumulated (sinks) and areas where it has depleted (sources).
  3. Gradient Vector Generation: By treating this differential map as a potential field, the model calculates its spatial gradient. The gradient vectors point in the direction of the greatest change, effectively generating a velocity field that characterizes the "flow" of demand across the region.

The resulting vector field provides a clear spatial-temporal signature of urban mobility. Instead of merely showing that a commercial zone's usage dropped while a residential zone's usage rose, the flow map visualizes the path and velocity of the demand migration between them.

High-Resolution Forecasting and Grid Management

Shifting from discrete time-series to continuous potential flow models fundamentally changes how we predict future energy needs. Traditional forecasting models struggle with fine-grained spatial variations because they treat each location independently. By modeling energy demand as a continuous field governed by fluid dynamics, frameworks can naturally capture the complex spatial dynamics that time-series methods overlook (Jieyang Peng et al.).

This fluid analogy makes it possible to build highly interpretable, high-resolution forecasting tools. As demonstrated by Zhibin Niu et al., visualizing demand shifts as vector flow fields allows grid operators to interactively explore how human mobility and seasonal changes influence energy distribution. By understanding exactly where and when energy demand is migrating, utility providers and urban planners can optimize energy dispatch, prevent localized grid overload, and more effectively integrate intermittent renewable energy sources into the smart grid.

Key Foundational Work

VAP: A Visual Analysis Tool for Energy Consumption Spatio-temporal Pattern Discovery (2020) — contributed 3 concept elements to this direction.
Predicting HVAC Energy Consumption in Commercial Buildings Using Multiagent Systems (2013) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_spatio_temporal_demand_dynamics_6567dbaf,
  title = {spatio-temporal demand dynamics},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/spatio-temporal-demand-dynamics/rdr_6567dbaf969841f3824128c6413b77c2},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_spatio_temporal_demand_dynamics_6567dbaf_v1,
  title = {spatio-temporal demand dynamics},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/spatio-temporal-demand-dynamics/rdr_6567dbaf969841f3824128c6413b77c2/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake