Transcript www.otago.ac.nz

Encouraging Electricity Savings in a
University Hall of Residence
Through a Combination of Feedback,
Visual Prompts, and Incentives.
Presented by: Marthinus Bekker
Contributors: Tania Cumming, Dr Louis Leland Jr,
Julia McClean, Niki Osborne, & Angie Bruining
Department of
Psychology
University of Otago
How we went about it
• Control study with a preceding baseline
Week 1
Baseline data
recording starts at
both Colleges
Week 2
Week 3
Week 4
Week 5
Week 6
Week 7
Intervention starts at
Salmond College, normal
readings continue at
Cumberland College
• Control: Cumberland College
• Intervention: Salmond College
Finish
Electricity
Recordings stop at
both Colleges
The Settings
Control
Intervention
Cumberland College
Salmond College
• 326 residents
• Mostly 1st year University
students
• Aged 18 to 20 years old
• Gender ratio of 59% female
and 41% male
• Cumberland College is
located in Central Dunedin in
an old nurses residence
• A steam plant situated
further along the street
drives the majority of
heating, and hot water
• 190 residents
• Mostly of 1st Year University
students
• Aged 18 years or older
• Gender ratio of 63% female
and 37% male
• Salmond College is located in
North Dunedin, in a purpose
built building.
• An on-site steam plant
generates heating and hot
water
The Results
Day
Night
College
Mean %
Electricity saved
Control
5.90%
Intervention
16.19 %
Control
6.44 %
Intervention
10.61 %
 Substantial differences between control and
intervention savings
Please note differences from abstract due to revision of data since
Summary
• Use of Feedback, Incentives & Visual
prompts
• Significant difference in savings
• Opportunity for substantial savings
across many colleges
• Significant savings can be achieved with
little effort & investment
Recommendations
• Full year study, with first semester as
baseline
• Long term follow-up to assess spill over
• Daily readings (or could do weekly)
• Either:
- Use control college
OR
- Regression equation that predicts expected
usage from baseline period and
temperature, humidity, population, etc..
Predicting electricity usage in
University Colleges of Residence
By: Marthinus Bekker & Dr Louis Leland Jr,
The Idea
Phantom
Control
Control
Temperature
• 326 residents
Humidity
• Mostly 1st year University
Previous years usage
students
Day
theyears
Week
• Aged
18 of
to 20
old
Light,
UVA,
• Gender
ratio of
59% UVB
female and
41% male Rain
• CumberlandETC….
College is located
in Central Dunedin in an old
nurses residence
• A steam plant situated further
along the street drives the
majority of heating, and hot
water
Intervention
• 190 residents
• Mostly of 1st Year University
students
• Aged 18 years or older
• Gender ratio of 63% female and
37% male
• Salmond College is located in
North Dunedin, in a purpose
built building.
• An on-site steam plant
generates heating and hot water
How we are doing it
1. Obtain archival electricity with the help of
the University’s Energy Manager
2. Obtain archival weather data through the
physics department weather station and
NIWA
3. Obtain other variables such as semester times
4. Go mining with the various variables using
multiple regression analysis
Multiple regression analysis
In multiple linear regression, the relationship between several
independent variables and a dependent variable is modeled by
a least squares function, called the linear regression equation.
This function is a linear combination of the various model
parameters, called regression coefficients.
A linear regression equation with one independent variable would
represent a straight line.
The results are then statistically analysed for significance and
predictive value.
Different versions of these equations can then be
compared to find the best one.
Multiple regression analysis
Temperature
Day of the Week
Day of the Year
Wind direction
Humidity
Previous years usage
Global radiance
Period of the Day
Wind speed
UVA
Year
Rain
Pressure
Wind direction
Wind speed
Equation
Rain
UVB
Multiple regression analysis
Equation
Electricity usage = Last years electricity usage, Hour of day (Dummy), Year, Day of
the year, Day x Year2, Temperature, UVA, UVB, Day of the week (Dummy), Global
radiance
The above variables strongly predicts electricity usage, R2Adjusted
(variance explained)=0.929(19,2632), p <0.000
Standard Error=26.880
Graph
Electricity usage per 4hr period (kWh)
600
600
500
500
400
400
300
300
200
200
100
100
00
17/02/2005
17/02/2005
5/09/2005
5/09/2005
24/03/2006
24/03/2006
10/10/2006
10/10/2006
28/04/2007
28/04/2007
Date
14/11/2007
14/11/2007
1/06/2008
1/06/2008
18/12/2008
18/12/2008
Questions, comments &
Suggestions
Thank You
•
•
•
•
•
OERC for funding this summer bursary
project
Hans Pietsch for the electricity data
Brian Niven for help with the non linear
transformations
Dr Louis Leland for his guidance and
support
Psychology department for the facilities to
do all this