Print this page

Published: 9 March 2011

Curbing IT’s appetite for energy

Graeme O’Neill

The information technology (IT) sector’s phenomenal growth and appetite for energy have produced a bulge in its carbon waistline. It’s not a good look when the world is warming at a rate unprecedented in recorded history. Just how big is the IT sector’s footprint in Australia?

Cloud computing and virtualisation can optimise spare capacity in a system or network, reducing hardware requirements and cutting energy use.
Credit: istockphoto

The ICT (information and communications technology) industry continues to make quantum gains in processing power while improving its energy efficiency – measured in Floating Point Operations Per Second (FLOPS)1 per watt. But the sector’s burgeoning growth more than negates the savings achieved through increased efficiency.

SMART 2020, a 2008 report produced by The Climate Group on behalf of the Global e-Sustainability Initiative (GeSI), claims the ICT sector could increase its emissions offsets by a factor of five, through helping facilitate emissions reductions in other industries.

However, British environmentalist and author Mr Jonathon Porritt has said that this claim is ‘quite something’ given that emissions from IT are set to balloon from three per cent of total global emissions in 2009 to six per cent by 2020. Mr Porritt put his views in a foreword to another report, Green IT: The Global Benchmark, commissioned by Japanese ICT giant Fujitsu and published in 2010. He noted that this later Fujitsu report ‘provides an incredibly helpful, euphoria-dispelling reality check.’

The report was authored by Australian green ICT expert, Mr Graeme Philipson, Research Director of Sydney-based sustainability consultancy Connection Research. Mr Philipson says that while the application of green ICT to other industry sectors will undoubtedly yield substantial offsets in emissions, the SMART 2020 report’s best-case scenario of a 5-fold offset is unrealistic given the projected increase in the ICT industry’s share of global carbon emissions.

The Fujitsu report was one of the first to employ Connection Research’s Green IT Maturity Index, originally developed in collaboration with RMIT University in 2009. Fujitsu has since adopted the improved 2010 version across its global operations. Mr Philipson hopes the index will become the global benchmark for assessing green IT progress.

‘We surveyed 500 organisations in four countries – the UK, the US, India and Australia – on the basis that the more we surveyed, the better the data,’ he explains. ‘We asked each organisation 100 questions to rank its progress in five defined areas of green IT, across the range of its operations. We then calculated the average for the 500 companies, which yielded the index.’

The five areas in the Green IT Maturity Index are:

  1. green IT lifecycle (procurement and disposal)

  2. end-user IT efficiencies

  3. enterprise and data centre IT efficiencies

  4. use of IT as a low-carbon enabler

  5. effectiveness of Green IT measurement and monitoring.

The average Green IT Maturity level for the 500 companies surveyed in the Fujitsu report is low – just 56.4 on a 100-point scale. The best-performing country was the UK, with an overall Green IT Readiness Index of 61. The US was next with 58.6, then Australia (53.9), which was penalised for its low level of measurement. India was fourth, with a score of 52.

The report concluded that ‘there is consistently very low performance in the metrics that enable green IT to be properly measured and monitored, and environmentally unsound IT procurement and disposal practices remain widespread’.

Greenpeace International recently ranked Fujitsu third overall – behind Cisco and Ericsson, but ahead of Google – among international ICT companies on its Cool IT Leaderboard. Greenpeace’s Cool IT Challenge, launched in 2009, calls on ICT companies to ‘innovate, mitigate their carbon footprints, and advocate for significant policy changes in the mutual interest of business and the climate’.

Mr Philipson also wrote a recent Connection Research report for the Australian Computer Society entitled Carbon and Computers: The Energy Consumption and Carbon Footprint of ICT Usage in Australia in 2010. He says it is the first time anyone has attempted to calculate a definitive carbon footprint for the industry in Australia – which his best estimate puts at 2.7 per cent of national emissions. This figure reflects energy consumption by computers, air-conditioning and ancillary equipment, but excludes embodied energy – the energy consumed in manufacturing and transporting IT equipment.

‘I’ve seen a study that suggests that more than half the carbon cost of a laptop is accounted for by its manufacture, but it’s not included in any carbon targets in any jurisdiction, because it’s impossible to measure accurately,’ Mr Philipson explains.

Apart from the quick energy-use cuts that can be made by reducing standby power, current trends towards cloud computing and virtualisation offer further savings.

Cloud computing is a concept for IT architecture, in which all IT applications and data are kept on ‘virtual servers’ on the internet. Virtualisation consolidates multiple specialised applications – formerly run on dedicated servers – on a single server and storage system. The Commonwealth Bank, Westpac, Visy and Komatsu are among large Australian companies that have already committed to cloud computing, and The Royal Australian College of General Practice plans to provide GPs with cloud-based e-health applications.

CSIRO is working with National Information and Communication Technologies Australia and Microsoft to implement Microsoft’s Azure cloud-computing facilities through the National Computational Infrastructure at the Australian National University. The aim is to develop systems to help the Australian ICT industry make wider use of cloud computing.

CSIRO is using virtualisation as part of its energy-saving consolidation of 60 computer rooms spread around Australia into a single ‘green’ data centre in Canberra. This makes use of the 95–98 per cent spare capacity that most corporate servers have available at any one time.

Interviewed by Computerworld magazine, CSIRO’s Director of Strategic Planning for IT Infrastructure and Science Support, Mr Peter Czeti, commented: Applications don’t like being away from their data by 2000 km or so, so that’s where the virtualisation is creeping in’.

Virtualisation requires less equipment, because centralised servers are run at 70–80 per cent of capacity; individual servers typically run at only 20–30 per cent of capacity.

‘The disadvantage is that [virtualised IT solutions] don’t provide the power savings you might expect,’ notes Mr Philipson. ‘Organisations usually use virtualisation to reduce their hardware requirements, rather than to save energy.’ He believes more work is required to measure and monitor the energy savings brought by virtualisation and cloud computing.

Ancillary equipment – particularly air-conditioning – accounts for around half the energy consumed by major data centres. Mr Philipson says that in cool areas close to renewable energy sources such as hydro or wind power, say in Tasmania or Jindabyne in NSW, clustering data centres would slash their emissions.

GPU = more energy-efficient supercomputing

The CSIRO GPU research cluster operates Australia’s ‘greenest’ supercomputer, which is powered by Graphics Processing Units (GPUs). The supercomputer ranks 13th on an internationally recognised list of the world’s 500 fastest and most energy efficient supercomputers – the Green500 List.2

GPUs speed up data processing by allowing a computer to massively multi-task through parallel processing. GPU-based supercomputers are 2–10 times as energy efficient as regular CPU (central processing unit) supercomputers, completing calculations around 10–100 times faster. GPUs are also much cheaper to purchase and occupy a fraction of the rack space, which also reduces cooling and data centre costs.

Recently upgraded by Xenon Systems, the CSIRO GPU cluster, located in a data centre in Canberra, achieved a speed of 52.55 TeraFLOPS (trillion Floating Point Operations Per Second) at an energy efficiency of 555.5 MegaFLOPS per watt. This means it can perform 55 550 000 000 calculations for the equivalent power consumption of a 100 W incandescent light.

Dr John Taylor, CSIRO’s Computational and Simulation Science leader, says energy efficiency and value for money were big draw cards for the organisation.

‘The high-performance computing industry has reached an energy wall, where the cost of operating supercomputers is about to exceed the purchase cost: primarily due to energy costs. The current focus in the high-performance computing industry is now very much on reducing the energy used to perform computations,’ says Dr Taylor.

‘We’re solving big national research problems at CSIRO, particularly in climate and environmental science, so it’s important that we focus on energy efficiency for CSIRO computers used to support computational and simulation science.’

Scientists using the GPU cluster have already seen a dramatic improvement in IT performance in a range of research projects, from 3D image reconstruction and analysing genetic data in wheat breeding experiments, to modelling interactions between nutrients and plankton in the ocean, and investigating materials science and physics problems.

GPU supercomputers enable researchers to carry out data-intensive modelling work in areas such as materials science.

1 The FLOPS unit is a measure of the performance of a computer per watt of energy consumed. FLOPS indicate how many mathematical operations involving decimal fractions a computer can handle per second.
PCs are measured in millions of flops (megaflops), mainframe computers in billions of flops (gigaflops), and super computers in trillions of flops (teraflops).

Published: 9 March 2011

Zero Carbon Australia plan, revisited

Matthew Wright and Patrick Hearps

In 2010 the Beyond Zero Emissions group released a report with the University of Melbourne’s Energy Research Institute claiming that Australia could be powered by renewable energy sources by 2020. Here its lead authors reply to some of the points raised by Dr Mark Diesendorf’s review of the report in ECOS 157.

This Gemasolar CST plant in Seville, Spain, is despatching electricity to the Spanish grid.
This Gemasolar CST plant in Seville, Spain, is despatching electricity to the Spanish grid.
Credit: Torresol Energy/SENER

The Zero Carbon Australia (ZCA) Stationary Energy Plan sets out strategies for powering Australia with 100 per cent renewable energy by 2020. While the plan stands alone as the only technical blueprint for completely decarbonising the domestic energy sector, it is a work in progress. There are areas to improve and some clarifications we would like to make about some of the recommendations.

Our research was undertaken with two explicit parameters: energy technologies selected had to be both commercially available and from carbon-free renewable energy sources. This explains why the ZCA Plan identifies a 60/40 mix of concentrated solar thermal (CST) power and large-scale wind developments as the backbone of a decarbonised energy system. Together with existing hydropower, investment in CST with molten salt storage, backup from a small percentage of biomass power, an upgraded electricity grid, and comprehensive energy efficiency measures, Australia can reliably meet its energy needs from renewable electricity generation. The technologies selected were not preordained; rather they were chosen on the basis that they worked within ZCA’s parameters.

The ZCA scenario also includes natural gas. Under the plan, Australia would use existing gas infrastructure in a staged scale-back, until the last gas power plants are mothballed in 2020. The most carbon-intensive coal power plants must be first to be decommissioned as large-scale renewables come online, made possible by the deployment of CST power towers with molten salt storage for 24-h operation.

CST is a nascent, commercially available energy technology. At November 2010, there were 632.4 electrical megawatts (MWe) of CST operating in Spain, including 250 MWe with storage, and a further 422 MWe in the US. Another 2000 MWe are in advanced stages of construction and development in Spain. This project pipeline amounts to over a US$20 billion investment. Meanwhile, in the US, federal loan guarantees and cash grants have fostered the approval of over 4 000 MW of CST, many of which have begun construction.

The CST plants in the ZCA Plan are modelled on the Spanish Gemasolar plant, which is now dispatching electricity to the Spanish grid. Our cost projections are based on those from existing projects in the US and Spain, with provisions for significant cost reductions following the first 1000 MWe installed.

The infrastructure rollout proposed under the ZCA plan, including these CST plants, is well within Australia’s industrial capability. Dr Diesendorf presents a global shortage of electrical engineers as a constraining factor. However, CST plants constructed under the ZCA plan would be replicated with a standardised series of plants, reducing the need for electrical engineers who are mostly required during the design phase.

As to the value of an east–west transmission link, more detailed modelling will be conducted for version 2.0 of the ZCA plan. Even without this data, it is premature to rule out the cost effectiveness of a transcontinental grid. Siemens proposes an east–west link in its 2010 report Picture the Future: Australia – Energy and Water. High-voltage direct current (HDVC) infrastructure is already in widespread use in the US, Canada, Europe and South America, and China has now commissioned the 2071 km Xiangjiaba-Shanghai 800 kV Ultra HVDC link.

The ZCA plan puts forward a single scenario largely in order to identify the specific challenges around implementation. We do not claim that the current iteration of the ZCA plan is the optimal solution. We would like to invite engineers and scientists from around Australia to provide their services as pro bono researchers with the Zero Carbon Australia project and make version 2.0 an even stronger document than the first.

We don’t think the Zero Carbon Australia initiative is brave. We think it’s necessary.

Matthew Wright and Patrick Hearps are lead authors of the Zero Carbon Australia Stationary Energy Plan. Matthew Wright is Executive Director of Beyond Zero Emissions and the 2010 Environment Minister’s Young Environmentalist of the Year. Patrick Hearps is a research fellow at the University of Melbourne’s Energy Research Institute.

More information

Mark Diesendorf’s review of the ZCA plan (‘Ambitious target does not quite measure up’):
ZCA plan:
Basis for cost projections for CST plants:
US National Energy Renewable Laboratory –

ECOS Archive

Welcome to the ECOS Archive site which brings together 40 years of sustainability articles from 1974-2014.

For more recent ECOS articles visit the blog. You can also sign up to the email alert or RSS feed