Modelling Office Energy Consumption: An Agent Based Approach

Tao Zhang, Peer-Olaf Siebers, Uwe AickelinSsrn 2010 · DOI · OpenReview

TL;DR

This paper introduces an agent-based model to simulate office energy consumption by integrating organizational policies, technology, equipment, and human behavior. The model, validated against real-world data from a university building, offers a novel approach to understanding and optimizing energy usage in complex office environments.

What Problem Does This Paper Address?

Offices are significant energy consumers, contributing to climate change concerns. However, understanding and managing office energy consumption is a complex challenge due to the interplay of multiple factors: organizational energy management policies, the technologies installed (like metering and automation), the types and usage of electric equipment, and crucially, the behavior of the people working in these spaces. Existing research often focuses on isolated aspects like energy prediction or technology, neglecting the integrated, human-centric nature of the problem. This paper aims to bridge this gap by developing a comprehensive model that captures these four key elements and their interactions.

Methods

The core of this research is an agent-based model (ABM) developed using AnyLogic 6.5.0. The model simulates the energy consumption of an office floor by representing different entities as agents:

  • Energy User Agents: These agents embody the behavior of staff and students. Their actions are based on an extensive empirical survey, categorizing them into stereotypes like "Early Birds," "Timetable Compliers," and "Flexible Workers," each with defined arrival/leave times and energy-saving awareness levels. Their states include being out of school, in corridors, in their own office, or in other rooms, influencing their interaction with appliances.
  • Light Agents: These are passive agents that react to the presence of energy user agents (via sensors) or manual switch actions.
  • Computer Agents: Similar to light agents, these are passive, with states of "off," "on," and "standby," influenced by user activity.
  • Base Electric Appliances: These represent continuously running equipment, contributing to a fixed base energy consumption.

The model integrates four key elements:

  1. Organizational Energy Management Policies/Regulations: Implicitly represented through the automated lighting strategy and the potential for staff control.
  2. Energy Management Technologies: Modeled through light sensors and the concept of automated vs. staff-controlled strategies.
  3. Electric Appliances and Equipment: Differentiated into base (always on) and flexible (lights, computers) appliances.
  4. Human Behavior: Captured through the detailed state charts and stereotypes of energy user agents, including their energy-saving awareness.

A mathematical model is also presented to explain the relationship between base and flexible consumption, influenced by user behavior parameters.

Datasets

The study utilizes data from a case study of the School of Computer Science at the University of Nottingham. This includes:

  • Building Plan: Used to define the spatial layout and agent placement.
  • Details of Rooms and Electric Equipment: Provided by the Estate Office.
  • Energy Consumption Data: Real-world electricity consumption data from the school building, provided by the Estate Office, used for model validation.
  • Empirical Survey Data: Extensive questionnaire and observation data collected from staff and PhD students regarding their energy use behavior, which formed the basis for developing the energy user agent stereotypes and behaviors.

No new benchmark datasets were introduced; the study leverages existing building data and empirical observations.

Key Insights and Ideas

  • Holistic Integration is Crucial: The paper strongly advocates for integrating organizational policies, technology, equipment, and human behavior for a realistic simulation of office energy consumption. Isolating these factors leads to incomplete understanding.
  • Human Behavior is a Dominant Factor: The detailed modeling of energy user agents, based on empirical surveys, highlights how individual actions and awareness levels significantly impact overall energy use.
  • Automated vs. Staff-Controlled Lighting Debate: The simulation reveals that, in the studied context, an automated lighting strategy is more energy-efficient than a staff-controlled one, challenging anecdotal beliefs. This is attributed to the 20-minute delay in automated systems, which prevents immediate shut-off, but ultimately leads to lower overall consumption compared to potentially less consistent manual control.
  • Social Interaction Amplifies Energy Saving: Increasing the "contact rate" between energy user agents (simulating enhanced communication about energy issues) demonstrably reduced overall electricity consumption, suggesting the power of social influence and awareness campaigns.
  • Disaggregating Consumption is Possible: The model can effectively break down energy consumption into components like base load, lights, and computers, providing insights that are difficult to obtain through simple metering.

Key Results

  • Model Validation: The simulation results for reproducing the current energy management strategy (automated lighting) closely matched the real school electricity consumption patterns, validating the model's accuracy.
  • Automated Lighting Efficiency: The simulation showed that the automated lighting strategy resulted in substantially lower electricity consumption compared to a staff-controlled strategy, despite similar peak consumption.
  • Impact of Social Interaction: Enhancing interactions between energy user agents led to a noticeable decrease in overall electricity consumption.
  • Consumption Breakdown: In the daytime (Monday-Friday), lights accounted for approximately 55% of flexible electricity consumption, computers around 7%, and base consumption was 92% in evenings and weekends.

Limitations and Future Work

Limitations: * Fixed Agent Stereotypes: Energy user agents are assigned a fixed stereotype (e.g., "Early Bird") and cannot switch between them during a simulation. In reality, people's behaviors can evolve. * Assumption on Interaction Impact: The assumption that enhanced interactions increase energy-saving awareness is based on research where users bear energy costs. This might not directly translate to office environments where staff don't pay their own energy bills. * Simplified Appliance Models: The model focuses on lights and computers, with potential for incorporating more complex appliance types and interactions.

Future Work: * More Flexible Appliances and Interactions: Incorporate a wider range of electric appliances and more intricate human-appliance interactions. * Large-Scale Organization Modeling: Extend the model to simulate larger organizations with complex energy management strategies. * Gas-Fired Heating Systems: Adapt the model to include gas consumption by drawing analogies with electricity consumption. * Psychological Factors and Satisfaction: Integrate more psychological factors into energy user agents to study how to optimize energy consumption while maintaining staff satisfaction. * Dynamic Stereotype Switching: Allow energy user agents to change their stereotypes over time.

Summary updated

Abstract

In this paper, we develop an agent-based model which integrates four important elements, i.e. organisational energy management policies/regulations, energy management technologies, electric appliances and equipment, and human behaviour, based on a case study, to simulate the energy consumption in office buildings. With the model, we test the effectiveness of different energy management strategies, and solve practical office energy consumption problems. This paper theoretically contributes to an integration of four elements involved in the complex organisational issue of office energy consumption, and practically contributes to an application of agent-based approach for office building energy consumption study.
BibTeX
@misc{zhang2010modelling,
  author = {Tao Zhang and Peer-Olaf Siebers and Uwe Aickelin},
  title = {{Modelling Office Energy Consumption: An Agent Based Approach}},
  year = {2010},
  eprint = {1607.06332},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1607.06332},
  howpublished = {Ssrn},
}
Figure