Established Base Important Assumptions on PC Sector

Sector
Established Base
Energy Consumption
PCs
Regression Model +
Intuition (w/o ref. but
with middle checkpoints)
Regression Model +
Intuition (w/o ref.)
Data Centers
Regression Model
Regression Model
Mobiles
Regression Model +
Intuition (w/o ref.)
Intuition (w/o ref.)
Gaming Consoles + TV
Regression Model +
Intuition (w/o ref.)
Regression Model +
Intuition (w/o ref.)
1-3 Historial Data
Points
3-5 Historial Data
Points
5+ Historial Data
Points
Types of Regression Models
• Following types of regression models were considered
–
–
–
–
–
–
–
Logarithmic – ln(x)
Linear – ax+b
Quadratic – ax2+bx+c
Polynomial – a1xn + a2xn-1 + .. + anx + c
Power - abx
Exponential - aex
Combination of above families
Example of Regression Models
Exponential
Polynomial
Linear
Power
Logarithmic
Important Assumptions on PC Sector
Established Base
 The PCs sector was divided into two main sectors:
 Desktop PCs
 Laptop PCs
 Percentage share of each sector was in accordance with the Smart2020 report
 In 2002,
─
─

By 2020,
─
─

desktop had 84% of market share
Laptops had 16% of market share
laptops taking 74% of market share
Desktops have 26% of market share
linear increase in use of laptops was assumed
4
Important Assumptions on PC Sector
Established Base Computation – Case I
 PC sales numbers from Gartner are used to estimate the established base
 Needs an accurate estimate of PC life cycle to get the established base
 Three scenarios showing the life cycle of PCs are obtained
 3-year “typical” life cycle as mentioned in Gartner news release
 4-year life cycle derived from Gartner’s “1.1 Billion PCs in Use in 2008” statement
 5-year life cycle for laptops and 6-year life cycle for desktops as mentioned in “European
Commission DG TREN : Personal Computers (desktops and laptops) and Computer Monitors
─
Can be obtained at www.ecocomputer.org
 Plugging in the recent data (2009 and 2010(est.)) from Gartner Inc. gave different
estimates than before

Estimating PC life cycle as 3 or 4 years gave surprisingly low estimates even with the most
aggressive curve fit (exponential growth)
─
─

2.75 Billion PCs in 2020 for 3 year life cycle
3.6 Billion PCs in 2020 for 4 year life cycle
Estimating PC life cycle of 5 years gave the most reasonable estimate of PC established base
─
Closely matches Smart2020 established base prediction of 4.067 Billion PCs by 2020
•
A reasonable 50% PC penetration by 2020

4.3 Billion PCs in 2020
5
Important Assumptions on PC Sector
3-Year Life Cycle
4-Year Life Cycle
6
Important Assumptions on PC Sector
5-Year Life Cycle
7
Important Assumptions on PC Sector
Established Base Computation – Case II
 Internet Users numbers from CIA World Factbook to derive PC established base
 Assume number of internet users to reflect PC base closely
─
─
Non-internet connect PCs and sharing of PCs cancel each other out maybe?
Only 3 world data points but gives very good curve fit as the data points are spread over a long period
•

2002, 2005 & 2008 years estimates
Gives an estimate of 5.75 Billion Pcs by 2020
8
Important Assumptions on PC Sector
Energy Consumption
 To estimate the energy consumption of a PC, its operation was divided into 3 different modes:



Active mode – when PC is on and CPU is functioning
Sleep mode – when CPU is on standby
Off mode – when CPU is off but is plugged into the socket
 Usage patterns for different modes of operation are obtained from TIAX from the report submitted to
the US Dept. of Energy.

The usage pattern values was based on data of year 2005
─
Lower bound estimate that usage in future years will be at least as much as 2005 was made
 Sleep & Off modes power consumption from the ENERGY STAR ratings
 The latest energy star ratings was used to define the lower bound on sleep and off mode power consumption
─
All PCs are assumed to satisfy Energy Star ratings
 As the top 5 models PC power numbers weren’t available, the active PC power consumption of
some popular models listed in following websites was used




Univ. of Pennsylvania, Information Systems and Computing
Apple Energy and Environment website
White paper “Review of Computer Energy Consumption and Potential Savings” sponsored by Dragon Systems
Software Limited.
Note: Lowest values of power consumption were taken
─
for having lower bound estimates for the energy consumption
9
Important Assumptions on PC Sector
Energy Consumption
 All the power consumption analysis was done independently for desktops and laptop
PCs

decrease in power consumption of desktops – polynomial fit
─

in accordance with the historical trends observed
Two curves showing both increase and decrease in power consumption of laptops – polynomial
fit
─
Ambiguous historical data in which some models have increased energy consumption while some show a
slight decrease
10
Important Assumptions on Data Center Sector
Established Base
 Established base numbers till 2020 were extrapolated based on Gartner and Koomey’s
reports

Matched Smart 2020 report’s extrapolated establised base numbers
Energy Consumption
 A linear increase in energy consumption of data centers was assumed
 As opposed to Smart2020’s view of no increase in energy consumption
 Equal energy consumption for running and cooling the data center was assumed for all
extrapolations
─
In accordance with assumption by Koomey et al.
 Half the energy consumption of data centers is assumed to be from electronics usage
 The other half is by cooling
11
Important Assumptions on Mobile Sector
 Only mobile devices have been considered in this study
 Other telecom devices not considered
─

Like IPTV boxes, broadband modems
Telecom infrastructure was not considered
Established Base Computation – Case I
 Extrapolations using curve fitting were done based on the historical data of Gartner Inc.
 Life cycle of a mobile phone was assumed to be 3 years
 2009 data from Gartner was used as well
 Best regression model gives an estimate of 6.5 Billion established mobile base by 2020
12
Important Assumptions on Mobile Sector
Established Base Computation – Case II
 Extrapolations using curve fitting were done based on the data from CIA World
Factbook.
Best regression models give an estimate of 20-50 Billion established mobile base by 2020
 Two reasonable fits were found

─
─

Linear model gave 11 Billion established base
Power model gave 7.3 Billion established base
To decide on the growth trends, the basket approach was used
─
Was not conclusive
Linear
Regression
13
Important Assumptions on Mobile Sector
Established Base Computation – Case II
Power Regression
14
Important Assumptions on Mobile Sector
“Basket” Approach
OECD
Developing
Under Developed
USA
UK
Japan
France
Australia
Brazil
Russia
India
China
Zimbabwe
Congo
Afghanistan
Somalia
Nigeria
Best Fit - Polynomial
Total established
base for these
countries is 12
Billion by 2020!!
15
Important Assumptions on Mobile Sector
“Basket” Approach
Linear Fit – 97.2% accurate
Total established
base for these
countries is 5.5
Billion by 2020
Power Fit – 91.2% accurate
Not
Conclusive!
16
Important Assumptions on Mobile Sector
Energy Consumption
 For the mobile phone power consumption
 Energy consumption of a device is equal to the battery energy storage capacity
─
Different values based on charging frequency were given
•
•
•
─

lower bound estimate: charging once in two days
Average estimate: charging once a day
Higher bound estimate: charging twice a day
Doesn’t impact the total values by much as Mobile sector energy consumption is negligible compared to
other sectors
Popular Nokia models were only considered for obtaining the energy consumption per device
─
As most of the similar cellphones have the same battery capacity
17
Important Assumptions on Gaming Sector
Established Base
 Extrapolations using curve fitting were done based on the historical data of NPD group and
VGChartz

A quadratic increase in established base is assumed
─
Gives a reasonable 700,000 game consoles by 2020
 Number of televisions is same as the video game consoles in that year
─
As televisions are assumed to be used with gaming consoles in our analysis
Energy Consumption
 Power consumption values were obtained from:
 The Energy Saving Trust, “The ampere strikes back”
─
─

Detailed power consumption of various popular gaming consoles
Different modes of power consumption
Projections to power consumption were assumed to increase logarithmically
─
─
Gave a reasonable 250 watts per device by 2020
Linear increase gave unreasonably high 800Watts per device by 2020
 Usage statistics of gaming device was obtained from Consumer Electronic Association (CEA)
 Alternate scenarios with higher gaming device usage will be plotted as well
18
Important Assumptions on Gaming Sector
Energy Consumption
 Weighted average of the different gaming consoles based on their market share was done
 to obtain the energy consumption per device
─
Different power modes were considered
 Television power consumption was based on ”Efficiency standards for televisions”, California Energy
Commission, 2008.

Television power consumption is assumed to be decreasing over the years
─
An constant average TV screen size 26” is taken for all extrapolations
19
Our Projections on Total Energy
Consumption of ICT Sector
Total energy consumption in (billion KWhrs/yr)
2008
2012
Will be
updated
based on case
choice
2020
Data Centers
192.676
326.7
695
PCs
354.11
493.34
1015.28
Mobile
6.215
11.096
30.55
Gaming Consoles & TV
52.6
117.8
418.6
605.601
948.936
2159.43
Total
EIA
2008
2012
2020
Total Electricity
consumption
19,460
22,010
27,500
IT % of total
3.11%
4.31%
20
7.85%
Our Projections on Carbon Footprint of ICT
Sector
Will be
updated
based on case
choice
Total carbon footprint in (Mega tonnes of CO2)
2008
Conversion Factors:
(1KWhr of electricity = ? lbs of
CO2)
1.306
2012
1.27
2020
(Our)
1.238
2020
(Smart2020)*
1.44
Data Centers
114.3795
188.595
PCs
210.2126
284.7917
Mobile
3.68945
6.405418
Gaming Consoles
31.22527
68.00273
17.19132 10
235.5576 -----
Total
359.5068
547.7949
1215.17
2.48%
3.78%
8.38%
IT % of total Carbon
Cap (14.5 Gt of CO2
per year)
391.0955 259
571.3257 500
*Embodied carbon & Telecommunications Infrastructure has been ignored in this analysis
21
769
5.3%