Modelling Office Energy

Consumption: An Agent Based

Approach

Tao Zhang, Peer-Olaf Siebers, Uwe Aickelin

Intelligent Modelling & Analysis Group, School of

Computer Science

University of Nottingham

Agenda

• Overall Project Background

• Office Energy Consumption

• Case Study• Case Study

• Simulation Experiments

• Conclusions

Overall Project Background

• EPSRC Energy & Complexity Science Call

(2008)

– Using complexity science approaches (e.g. agent-based

simulation, system dynamics, dynamic network models

and control theories) to tackle energy challenges (e.g. and control theories) to tackle energy challenges (e.g.

policy/regulation/intervention, social-technical aspects of

innovative energy technology adoption)

Overall Project Background• Four Projects Funded

Project Universities

Preventing wide-area blackouts through

adaptive islanding of transmission

networks

Edinburgh, Durham, Southampton

Complex Adaptive Systems, Cognitive

Agents and Distributed Energy (CASCADE): De Montfort University

Agents and Distributed Energy (CASCADE):

a Complexity Science-Based Investigation

into the Smart Grid Concept

Future Energy Decision Making for Cities -

Can Complexity Science Rise to the

Challenge?

Nottingham, Leeds

SCALE (SMALL CHANGES LEAD TO LARGE

EFFECTS): Changing Energy Costs in

Transport and Location Policy

UCL

Overall Project Background

• Future Energy Decision Making for Cities - Can

Complexity Science Rise to the Challenge?

– Cities have a vital role in future UK energy

sustainabilitysustainability

– Cities regard energy as someone else’s problem

– Cities lack knowledge, experience tools for local

energy decision-making

Overall Project Background

• Future Energy Decision Making for Cities - Can

Complexity Science Rise to the Challenge?

– Using complexity science tools to deliver models

that enable cities define their current energy that enable cities define their current energy

situation and then reach a balanced decision in

future energy planning and implementing

sustainability targets

– Developing decision support frameworks that are

applicable to all cities in the UK

Overall Project Background

generation transmission distribution

End-use

Intervention

Distributed

generation

End-use End-use

technology

Energy

service

City

boundary

End-use

efficiency

Behavioural

change

Overall Project Background

Overall Project Background

• Future Energy Decision Making for Cities - Can

Complexity Science Rise to the Challenge?

– A complexity science project involving economics,

engineering, mathematics, organisational engineering, mathematics, organisational

behaviour and social psychology

– Two teams

• Leeds team: engineers, mathematicians, energy

economists, social psychologists

• Nottingham team: simulation scientists

Office Energy Consumption

• A sub-project under the City Energy Future

Project

• Target the organisational behaviour of using

energyenergy

– Reasons: UK government’s 2020 target of cutting

emission (by 34% of 1990 levels)

– 14% of overall energy consumption is in the

service sector (e.g. heating, lighting, computing)

Office Energy Consumption

• An integration of four elements

Energy Management Policies Made

by the Energy Management Division

Energy Management

Technologies

Office Electric Equipment and

Appliances

Staff’s behaviour of using

energy

Office Energy Consumption

• Previous literature primarily focuses on building energy consumption prediction, energy management technology development and building energy consumption benchmarks, and ignores human factorsbenchmarks, and ignores human factors

• We aim to develop a simulation model integrating the four elements and provide decision support for energy management divisions

Office Energy Consumption

• Specifically focus on electricity consumption

• Electricity are consumed by electric appliances

and equipment

• Two kinds of office building electric • Two kinds of office building electric

appliances: base appliances and flexible

appliances

Office Energy Consumption

Case Study• Case: First Floor, School of Computer Science, Jubilee Campus, University of Nottingham

Case Study

Item Number

Rooms 47

Lights 239

Computers 180

Details of Rooms and Electric Equipment and Appliances on the First Floor

Computers 180

Printers 24

Information Displays 4

Energy Users 213

Case Study

Flexible appliances Base Appliances

Computers Printers

Lights Information Displays

Other small appliances Servers

Details of Rooms and Electric Equipment and Appliances on the First Floor

Other small appliances Servers

Network Device

Case Study

• Two Research Questions

– Is automated lighting strategy always energy-efficient

than staff-controlled lighting management strategy?

– What are the proportions of electricity consumed by –

lights and computers respectively?

Case Study

• Agents

– Energy User, i.e. Staff and Students (proactive agents)

– Computers (passive agents)

– Lights (passive agents)– Lights (passive agents)

Case Study

• Behaviour of Energy User Agents

Case Study

• Archetype of Energy User Agents (work time)

Agent Archetype Percentage Arrival Time Leave Time

Early Birds 8% Monday to Friday, between 5am and

9am, random uniform distribution

Monday to Friday, between 5pm

and 6pm, random uniform

distributiondistribution

Timetable

Compliers

53% Monday to Friday, between 9 am and

10 am, random uniform distribution

Monday to Friday, between 5pm

and 6pm, random uniform

distribution

Flexible Workers 39% Monday to Friday, between 10 am

and 1 pm, random uniform

distribution

Monday to Friday, between arrival

time and 23pm, random uniform

distribution

Case Study

• Archetype of Energy User Agents (Energy Saving)

Archetype of

Agent

Percentage energySavingAwareness Probability of Switching

Off Unnecessary Electric

Appliances

Probability of Sending

Email about Energy

Issues to Others

Environment

Champion

1% Between 95 and 100,

random uniform

distribution

0.95 0.9

distribution

Energy Saver 8% Between 70 and 94,

random uniform

distribution

0.7 0.6

Regular User 31% Between 30 and 69,

random uniform

distribution

0.4 0.2

Big User 60% Between 0 and 29,

random uniform

distribution

0.2 0.05

Case Study

• Behaviour of Computer Agents

Case Study

• Behaviour of Light Agents

Case Study

• Model Implementation

Computer agent 1

Base electric appliances

Base Electricity Consumption

Computer agent 1

System Level

Electricity

Consumption of

the School

Computer agent 2

Computer agent n

Light agent 1

Light agent 2

Light agent n

Energy user agent 1

Energy user agent 2

Energy user agent n

Flexible Electricity Consumption

Simulation Experiments

• Experiment 1: Replicate current policy

Simulation Experiments

• Experiment 2: Automated Strategy vs. Staff-Controlled Strategy

Simulation Experiments

Experiment 3: Understand the proportions of electricity consumed by lights and computers

Conclusions

• An computational simulation model integrates

four elements (i.e. energy technology,

management strategy, appliances, and users

behaviour) in office building energy behaviour) in office building energy

consumption

• Agent-based simulation: an effective decision

support tools for office building energy

management

Thank you for your attention

Hello, welcome back to

the world of complexity. I

am an agent living in a

computer

30