A study towards applying thermal inertia for energy conservation in rooms

Yi Yuan, Dawei Pan, Dan Wang, XiaoHua Xu, Yu Peng, Xiyuan Peng, Peng-Jun WanACM Trans. Sens. Networks 2013 · DOI

TL;DR

This paper introduces a novel approach to energy conservation in buildings by leveraging "thermal inertia" – the residual heat or coolness in a room after it's no longer actively heated or cooled. They developed a system that uses sensor networks to build energy-temperature correlation models and then applies these models to optimize room scheduling, leading to significant energy savings.

What Problem Does This Paper Address?

Buildings, particularly commercial ones, are major energy consumers, with HVAC systems being a significant contributor. While existing solutions often focus on simply turning off systems when rooms are unoccupied, this paper highlights a missed opportunity: the lingering thermal state of a room. After a meeting or class ends, a room doesn't immediately revert to ambient temperature. This residual "thermal inertia" represents stored energy that can be exploited. The challenge is to quantify this thermal inertia and integrate it into practical building management systems, specifically room scheduling, to achieve greater energy efficiency beyond simple on/off controls.

Methods

The core of the proposed solution is a two-component framework:

  1. Energy-Temperature Correlation Model: This component aims to predict the energy required to re-cool a room to a target temperature based on its current state and external factors. The authors employ a method that combines rudimental thermodynamics (Fourier's law of heat conduction) with sensor data. They build a "virtual perfect room" model and then calibrate its parameters (invariants) using data collected from a sensor network. This approach allows for individual room modeling without needing extensive building specifications.
  2. Energy-Aware Scheduling Algorithms: These algorithms use the predictions from the correlation model to schedule meetings. They consider not only standard requirements like time, capacity, and facilities but also the predicted re-cooling energy. The authors developed an optimal algorithm for uniform rooms and two efficient heuristics for scenarios with non-uniform room capacities.

A key technical contribution is the design of a two-tiered sensor network. The first tier consists of enhanced Imote2-based electricity meters to monitor AC usage, and the second tier uses TelosB sensors for temperature readings. They also developed a long-range data communication module for the Imote2 nodes.

Datasets

The authors evaluated their system in two settings:

  • Real-world data from The Hong Kong Polytechnic University (PolyU): They used classroom configurations and schedules from Spring 2011. For simulation purposes, they constructed models for the 155 classrooms based on their size and assumed material properties, rather than building models from real-time data for each classroom.
  • Synthetic room arrangements: These were generated semi-randomly to test various scenarios, including different room capacities and meeting lengths.

Key Insights and Ideas

  • Thermal inertia is a valuable resource: The paper convincingly argues that the residual thermal state of a room is not just a passive phenomenon but an exploitable resource for energy conservation.
  • Individual room modeling is crucial: Recognizing that each room has unique thermal properties (materials, size, shape) is key. Their approach of building individual energy-temperature correlation models for each room is a significant step towards accurate prediction.
  • Thermodynamics meets sensing: The integration of physical principles (heat transfer) with real-time sensor data provides a robust way to create practical energy models for buildings.
  • Scheduling can be optimized for energy: By incorporating energy predictions into the scheduling process, significant energy savings can be achieved beyond traditional scheduling methods.
  • Target temperature is a powerful lever: The results show a strong correlation between target temperature and energy consumption, suggesting that setting slightly higher target temperatures can yield substantial savings.

Key Results

  • 20% energy savings: In simulations using real PolyU class schedules, the proposed Energy-RS algorithm achieved approximately 20% energy savings compared to existing schedules.
  • Seasonal variations: Energy savings were more pronounced in summer (up to 26% in July) than in winter (around 15%).
  • Significant reduction in re-cooling energy: In synthetic scenarios, their algorithms (Energy-RS and TimeUr-RS) reduced re-cooling energy by more than half compared to a basic ad-hoc scheduling algorithm (RS).
  • Accuracy of the model: Experimental validation showed that the energy-temperature correlation model provided reasonable predictions for short-term re-cooling energy, with daily errors around 10% in one experiment.

Limitations and Future Work

The authors acknowledge several limitations:

  • Pre-determined meeting schedules: The current framework assumes meetings are scheduled in advance, which might not apply to all scenarios (e.g., online room booking for companies).
  • Simplified thermal model: The current model doesn't account for other heat sources like human activity or electronic devices.
  • Fixed target temperature: The simulations often used a fixed target temperature, and future work could explore dynamic target temperature adjustments.
  • General air conditioners: The electricity meter used is for general AC units; advanced meters for central HVAC systems are needed.
  • Single objective: The scheduling algorithms primarily focus on energy conservation, neglecting other important factors like user comfort or travel time between rooms.

Future work includes developing more detailed thermal models, exploring online room scheduling, incorporating comfort as an objective, and developing advanced metering solutions. They have also released an open-source MATLAB package for their energy-temperature correlation model.

Summary updated

BibTeX
@article{yuan2013study,
  author = {Yi Yuan and Dawei Pan and Dan Wang and XiaoHua Xu and Yu Peng and Xiyuan Peng and Peng-Jun Wan},
  title = {{A study towards applying thermal inertia for energy conservation in rooms}},
  journal = {ACM Trans. Sens. Networks},
  year = {2013},
  doi = {10.1145/2529050},
}
Figure