David Clarke has a ?23m plan for a fantasy power station. From the outside, it would look like any other coal or gas power plant. There would be one big difference, however - it would not emit greenhouse gases.  "The project will see a plant capable of capturing up to 95% of carbon dioxide emissions built by the middle of 2015," said Clarke, boss of the Energy Technologies Institute (ETI) - a research lab in Loughborough backed by government and companies such as BP, Shell, Rolls-Royce, Eon and EDF.  Clarke reckons ETI has mastered a technique called carbon capture and storage (CCS). He believes it can remove the carbon from power station emissions so we can go on burning coal and gas to create electricity. The carbon dioxide would be extracted, piped away and then injected into the ground instead of being released into the atmosphere.  The idea has been championed for years by a number of academics and energy experts. although dozens of projects are under development around the world, critics have insisted the technology is too expensive - and potentially dangerous.  Residents near the world's largest test site in Saskatchewan, Canada, have been evacuated from their homes after claims that carbon dioxide may have been leaking from an old mine where the gas is stored.  Clarke thinks he and his colleagues at the ETI have a "next generation" technology that can answer all the questions. He is now looking for sites to build a power plant to prove it is viable.  If he is right, it may be possible to fit the system in power stations all over the world - radically cutting emissions. a new generation of power stations could be built to run on fossil fuels but without the usual emissions. Clarke said this was an important step.  "By the time we get to 2020 all coal power stations will be close to being retired so then we are going to need new ones to be built," he said.  First, Clarke, a former Rolls-Royce research scientist, and his team need to prove that the technology works. Critics claim rolling out CCS would be so costly it would add as much as 25% to consumers' power bills.  Clarke, however, said the technology ETI is working on is so efficient that consumers' pockets are unlikely to be hit. "With the technology today you are looking at electricity prices rising more than 10%, but that's today. CCS is quite new but we are looking to demonstrate technology that will avoid that price rise."  CCS is a three-step process.  It begins with the removal of carbon dioxide from the coal to be burnt in the power plant. This is done in a "gasifier", where the fuel is blasted with oxygen and steam to produce carbon monoxide and water. The gas is converted into carbon dioxide and hydrogen.  Requiring temperatures of up to 700C, it is a highly energyintensive process that gobbles up a large chunk of the power the plant produces, making it by far the most expensive part of the process.  "We think the capturing [equipment] accounts for about 70% of the overall cost of CCS and adds 25% to 90% to the cost of building a power station," Clarke said.  The second stage is transportation.  The carbon dioxide is cooled to a liquid and pumped through pipelines. This also requires additional power, bumping up the costs. ETI says these can be kept to a minimum by locating the plant close to the storage site.  ETI hopes it can use some of Britain's existing oil and gas pipelines but it expects new ones will also have to be built.  "In theory what we are doing is just reversing the process that extracts oil and gas from the North Sea. Our ability to use current infrastructure will be important to minimising the costs for any site," Clarke said.  Storing the carbon dioxide is the final step and the one some experts say there is the most concern about. ETI plans to inject it thousands of feet underground into old, disused oil fields and saline aquifers, underground rocks saturated with salt water.  The scale of available storage is unclear - some estimates suggest there is only enough space to store carbon equivalent to 10 to 20 years of power station output,