Frozen future

The world's largest medical experiment will be launched in Manchester next summer with the aim of improving the prevention, diagnosis and treatment of cancer, heart disease, Alzheimer's, diabetes, arthritis and other illnesses that cause premature death.

The

will gather, store and protect a bank of anonymous medical data from 500,000 participants, and keep blood and urine samples in fridges between -20°C and -80°C for decades, so scientists around the world can study how genes, lifestyle and environment affect their risk of disease.

The samples will be housed in a fridge the size of a large lorry container, and be separated, stored, retrieved and tracked with specially designed robots.

The engineers, who have spent more than two years designing the repository, say it was challenging to design an automated system that works in sub-zero temperatures. 'We were scared the whole thing would turn into a big ice cube,' said Justin Owen from

(TAP).

Owen is the systems architect responsible for the technical delivery of the Biobank project. 'Unless you can control the rate of moisture going into the box and have something to get rid of it, over a 20- to 25-year period it is just going to be unworkable,' he said.

After extensive testing on a smaller model of the bank, Owen said the researchers found that if the fridge was purged of all its air, which is kept at a few parts per million of moisture, any accumulated ice would disappear. 'It was subliming into the very dry air around it and being carried out of the box,' he said. 'That was something we expected to happen but not at such a fast rate.'

Another challenge was designing installation for the refrigerated store. One conventional technology that industries use for refrigeration is the heater mat, designed to prevent a freezer floor and subsequently sub-floor, concrete and subsoil from freezing over and causing massive fractures to a structure.

'We weren't confident that the heater mat would be the right technology because no-one would guarantee it for 25 years — the length of the Biobank project,' said Owen.

After researching with other manufacturers, the TAP team decided on an ancient technology. 'We decided on a ventilated floor that looked like a Roman hypocaust, where they used to pump hot air underneath their floors to heat them,' he said. 'So the floor of the refrigerated box will stand on ventilated, load-bearing plastic grilles so air can percolate beneath the box. This provides enough of a thermal break between its underside and the concrete so we don't get that problem with fracture. It's intrinsically very reliable because there is really nothing to break.'

Once the researchers were sure the Biobank would not turn into an ice sculpture, they were able to design the special robots to do the work inside.

The TAP team designed a 1.5-tonne robot to access samples contained in two storage compartments cooled by liquid nitrogen and constantly held at -80ºC.

'One of the key pieces of technology we developed here was the means for maintaining the robotics at -20ºC, which is about as cold as you can comfortably go without spending a tremendous amount of money,' said Owen. 'We designed the robots to work isolated from the -80ºC compartments.

'Also, these compartments have no moving parts or cables so if you look inside them you find a very simple design with very careful choice of materials.'

While the -80ºC compartments are mostly isolated from the -20ºC area, small problems do occur when the drawers have to be briefly opened, said Frank Tully, the lead robotics engineer.

'There are many problems like differential expansion,' said Tully. 'When a robot pulls out a drawer from the -80ºC compartment, it expands by 3mm because it's going from -80ºC to -20ºC. So the robot is never sure where exactly it is.'

Owen said early on in the project the team had to decide whether to develop a rigid design for the compartments that the robots would be pre-programmed to understand. It was a big challenge because the robots are needed to operate compartments that weigh 3.5 to four tonnes when empty and almost 30 tonnes when stored at full capacity.

'We had this choice to make everything super-rigid and perfectly well aligned, which is not an easy thing to do because we're moving huge lots of metal around here,' he said. 'What we instead decided was to design the robotics so they're a little bit more intelligent and sophisticated.'

Tully said the engineers began with a simple idea to design drawers that are easier to find. 'We made the handle as big as possible so the robot is very unlikely to miss,' he said. 'Once it finds the drawer it can open it, and then it will use a sensor to detect the edge of the drawer.'

Owen added: 'The robot is able to make an active decision sometimes in real-time. It can detect the end of the drawer and say, "OK, I know where the rack should be." It does that quick calculation then moves to that position.'

The process is accurate but not always graceful. 'Instead of having it try to find the front of the drawer, we run the robot into it, albeit under control,' said Owen. 'So it bumps into it and says, "Ah, I've hit the drawer now, I must be at the right place." Then it moves on to do the next thing.'

Rob Meaker, the TAP team project manager, notes the skill it takes for the robot to find one specific 1.4mm tube among so many others. 'It must go over a 27m length to find the exact location of one tube and pick it out of 10 million other tubes. It is quite a challenge.'

Owen added that each tube has a 2D barcode at the bottom. 'So our tracking software knows the position of every tube in the store,' he said.

Part of the reason the team is so confident in its system is because of the amount of time it spent reviewing all the risks for failure. 'We are more than 90 per cent prepared for any risk of failure,' said Meaker.

Owen said the team believes an automated system works much more reliably than a manual one. 'We don't lose samples where with a manual system they are easier to lose or misplace,' he said. 'Also, with a manual system it's very easy to drop a load of samples.'

An automated system also allows greater storage density. 'We can get 10 million samples into something that is the size of a large lorry container,' he said. 'To do that manually, with liquid nitrogen, you might need something half the size of a football field.'

Owen added that the confidential nature of the project makes automation a better choice. 'With manual systems, samples can get mixed up and this study is bound by strict ethics,' he said. 'People can withdraw their samples at any time. It would be very embarrassing for Biobank or any organisation like this to say we can't find them.'

After more than two years of hard work, the team thinks this is the kind of project they can look back on and be proud of. 'It's potentially something that is going to benefit a lot of people, and looking to the future our work might help improve the quality of millions of people's lives,' said Meaker. 'It's a satisfying feeling.'