165
energy-efficient equipment could result in significant energy savings. This includes electric induction hobs which are up to 50% more efficient than a traditional electric hob, open-top gas ranges with individual burners as they are more efficient than a large single burner, combination ovens that offer convection, steam and combination cooking, as they can reduce energy costs by around 50% because they offer faster cooking times,
■ Innovative insulation and intelligent control systems: Most equipment is under-insulated to keep costs down or to minimise the space it takes up, but new materials mean more efficient and thinner insulation. Together, insulation and advanced controls can dramatically reduce energy costs.
166 Figure 4.8 Figure 3 European Co-location Market 2013
Source: CBRE (2013) European Data Centres MarketView
In 2013, total co-location in Frankfurt was the highest recorded for 4 years with an annual increase of 20% compared to 2012. Additionally, in 2013, London data centre uptake increased by 26% due to the award of several large contracts and an upturn in retail leasing activity.173 This is further illustrated by the power requirement of the UK data centre industry growing by 8.8% to 3.10 GW in 2013, compared to 2012 levels. The manufacturing sectors;
banking and financial services; healthcare; telecoms and media are most responsible for driving growth.174
If no energy efficiency improvements occur, energy consumption is assumed to grow inline with market growth; i.e., 2.5% per year through 2050. This reflects the increase in data volumes due to the growth of mobile computing, social networks, and the spread ICT in all aspects of private and work life. This has resulted in the continuous increase in both energy densities within the typical data centre and increased cooling requirements. Table 4.7 presents anticipated growth in energy consumption from 616PJ (14.7 Mtoe) at 2012 to 1,577 PJ (37.7Mtoe) in 2050, assuming no energy efficiency improvements are implemented. However, there is a growing trend towards green data centres globally. It is estimated that the green data centre175 market will grow from US$17.1 billion in 2012 to US$45.4 billion by 2016 – at a compound annual growth rate of nearly 28%.176. This trend is reflected in the improvement in Power Usage Effectiveness (PUE)177 over time. The average PUE has improved from 2.50 in 2007 to 1.89 in 2011.178 In 2014, the PUE has improved to 1.7, reflecting the fact that the biggest infrastructure efficiency gains have already happened, and further improvements will require significant investment and effort, with increasingly diminishing returns.179 Table 4.7 presents estimated improvements in PUE through 2050. Between 2012 and 2014, PUE improved at a rate of 2.8% per year. It is assumed that this rate will halve through 2020 (1.4%
173 Ibid
174 Datacentre Dynamics (20 November 2013) The DCDi 2013 Census – UK figures Available at http://www.datacenterdynamics.com/focus/archive/2013/11/dcdi-2013-census-%E2%80%93-uk-figures
175 A Green Data Centre has mechanical, lighting, electrical and computer systems designed for maximum energy efficiency and minimum environmental impact
176 Pike Research (2012) Green Data Centres in Navigant Research (2012) The Green Data Center Market Will Surpass $45 Billion by 2016 available at http://www.navigantresearch.com/newsroom/the-green-data-center- market-will-surpass-45-billion-by-2016
177 PUE is a ratio of the amount of power entering a data centre by the power used to run the computer infrastructure within it. An ideal PUE is 1.0.
178 2014 Data Centre Industry Survey; Uptime Institute; http://journal.uptimeinstitute.com/2014-data-center- industry-survey/
179 Ibid
167
per year), and will halve with each following decade. By 2040-2050, the improvement in PUE is assumed to be 0.2% per year.
These assumptions have been overlain to develop an estimate for business-as-usual (BAU) energy consumption for the information and communications sector through 2050 (Table 4.7
; Figure 4.9).
Table 4.7 BAU energy consumption trend for the information and communications sector based on projected market energy consumption assuming no energy efficiency (EE) improvements and PUE improvement through 2050.
2012 2020 2030 2040 2050
Energy consumption assuming no EE improvements (PJ)
616 751 962 1231 1577
PUE 1.8 1.58 1.48 1.42 1.40
BAU energy consumption (PJ) 616 661 788 974 1226
Figure 4.9 Projected BAU energy consumption trend for Information and communications sector
Considering that the ideal PUE is 1.0; if it assumed that BAU PUE in 2030 and 2050 is 1.48 and 1.40, respectively, the technical potential for the sector is assumed to be 255 PJ (6.1 Mtoe) in 2030, and 350 PJ (8.4 Mtoe) in 2050.
4.3.2 Energy usage profile
There are four key processes that consume the most electricity in the data centre environment, these are: the ICT load; cooling and ventilation; Uninterruptable Power Supplies (UPS) and power distribution overheads; lighting and other building overheads. A typical distribution of the energy consumption of these in a data centre is outlined in Table 4.8.
Table 4.8 Distribution of energy consumption in data centres
Data Centers Approximate % of total
consumption
The IT load 40%
The cooling system 35%
168
Data Centers Approximate % of total
consumption
UPS and power distribution 20%
Lighting and other overheads 5%
Source: Falcon Electronics PTY (n.d) Measurement of data centre power consumption
4.3.3 Opportunities
Table 4.9 highlights some of the equipment and infrastructure approaches to improving energy efficiency in data centres.
Table 4.9 Energy efficiency improvement in data centres
Best Practices Opportunities
Typical Investment
180
Typical Cost Savings Efficient Equipment Adopt Minimum Energy
Performance Requirements (EU Energy Star or
SPECPower)
€€€ For IT legacy equipment, savings could be between 30% to 50%
Consolidate and Optimize
Use Virtualization to
Consolidate and Optimize the exploitation of Physical Devices
€€€ 63% reduction in total cost of ownership (i.e., financial impact over the product life cycle)181
Implement Cloud Computing to shave peaks, by balancing data center workloads
€€ 29% compared to an internal IT system182 Power Generation &
Distribution
High Voltage DC (HVDC) to reduce to step down loss/conversion
€€€ Between 5% to 10% of facility power
Install PV panels where cost- effective and maximize auto consumption of DC power, avoiding DC-AC-DC losses
€€ DC-AC-DC can involve 30- 40% of losses (PV produce DC, converted to AC for the grid, then reconverted to DC for computing use
Use efficient Uninterruptable Power Supplies (UPS) to match load to demand
€€€ Capital and operational savings of 20%
Heating & Cooling Increase temperature ranges to decrease the workload on the HVAC systems
€ Up to 4% of legacy costs per degree change
Increase humidity set points to decrease workload on humidification and cooling systems
€ Up to 50% of legacy costs183
Hot/cool aisle thermal containment
€€ 30% of legacy costs
180 Very rough estimate only (for 10,000ft data centre): € = <€50,000, €€ = €50,000 - €500,000, €€€ = €500,000+
181 Storage Networking Industry Association (2008), Building the Green Data Centre; Towards Best Practices and Technical Consideration
182 http://broadcast.rackspace.com/hosting_knowledge/whitepapers/Cloudonomics- The_Economics_of_Cloud_Computing.pdf
183 http://advice.cio.com/michael_bullock/cool_ways_to_save_money_in_the_data_center?page=0%2C
169
Best Practices Opportunities
Typical Investment
180
Typical Cost Savings Replace air conditioners with
air handlers and economizers
€€ Up to 50% of legacy costs Variable speed fan drives €€€ 4 to 10 months return on
investment (ROI)184
Lighting Advanced Lighting Technology €€ Up to 70% of legacy lighting consumption depending on lighting technology Occupancy Control € Up to 80% of lighting load
depending on occupancy