Twenty-nine of the 220 laboratories under construction are "close control" rooms, in which a given temperature has to be maintained with a tolerance of one-tenth of a degree centigrade – any variation, and the measured size of the biscuit will change.
The project team at the NPL struggled to meet that extremely rigorous requirement in the time allowed, and suffered accordingly. As you will know from the news, Laing has declared losses of £40m on the project. The full story of this has yet to be revealed, and its legal consequences have yet to be resolved. The team refused to discuss the details of the project, so Building spoke to some specialists to find out how such demanding specifications can be met.
Gary Tucker, senior partner at consulting engineer Hoare Lea, and an expert in the design of laboratory air-conditioning systems, likens the technology needed to the control of a fighter aircraft, which has to respond instantly to changes in its environment. This requires a computer-controlled system that can monitor and control all aspects of the labs' environment. External influences, such as vibration or dust, must be entirely absent and a constant humidity maintained.
Conventional air-conditioning systems are inadequate for producing these conditions because they work by producing air several degrees below the desired temperature. This mixes with the warmer air to achieve the correct temperature, but there is no guarantee that it will be uniform across the room.
For a measurement to be accurate, the temperature across the experiment must be uniform. Achieving this is made more difficult by the introduction of randomly moving heat sources (that is, humans). Tucker explains: "If there are no temperature introductions, it is easy. Once you've got people moving around, putting heat and moisture into the space, it becomes very difficult." The lights and electric motors that power laboratory equipment are also heat sources that must be factored into the equation.
The easy way to get the degree of control required is simply to buy controlled rooms, designed by a specialist manufacturer, and put these in the middle of a lab. The advantage of this approach is that rooms are designed and tested as a complete system.
Tucker says he prefers this approach because, otherwise, "you are putting together components that haven't worked together before. It's the difference between a product and a prototype".
So what do you do when you cannot use an off-the-peg solution (as was the case at the NPL)? The answer is that the rooms must be designed as bespoke systems. Ideally, they should be in the centre of a building, with no external windows, and a high level of insulation, sealed, and fitted with conditioned airlocks. A "buffer zone" of rooms, with progressively reducing temperature tolerances around the critical area helps to maintain the temperature.
Pumping very high volumes of air into the room is the only way to offset heat gain from equipment and people. This air has to be supplied at the lowest temperature allowed – so if the target temperature was 20°C with a tolerance of 0.1°C, the air would be supplied at 19.9°C. And it has to be delivered at a low pressure over a large area – usually entering through a perforated ceiling and exiting through the floor. It is directed through the room in a laminar manner – that is, it goes from ceiling to floor in a straight line to avoid turbulence, which could blow dust into the critical area. Parker says 600 air changes an hour is not unusual.
At the NPL, only the areas where experiments are carried out have close control. These are not isolated from the rest of the laboratory, so large volumes of air are directed from a wall plenum horizontally across the experimental benches to cope with heat gain. Tucker and Andy Parker, who looks after the business side of Amec's pharmaceutical and chemical industries, agree that this approach sounds appropriate.
The rest of the system has to be designed, and must function, with extraordinary precision if it is to work within its targets. Very accurate temperature sensors are needed, and must be calibrated correctly. This is done by linking special control systems to sensors and valves, as standard building management systems respond too slowly. The air is cooled below the target temperature using a water-cooled coil and reheated to the correct temperature using an electric coil, as this responds instantly to change and offers a fine degree of control.
Large amounts of plant are needed to move the huge volumes of air required. Tucker believes that the more plant is needed in an application, the greater the scope for problems, as the plant has to offset its own heat gains – rather like a dog chasing its tail.
Tucker suggests that a tolerance of 0.1°C, as required in several of the NPL labs, would push the close-control computer modelling program to its limit: every last detail would have to be modelled exactly.
Tucker says: "Everything has to be perfect and thought through – if there is a problem, you are snookered."
One reason for the problems faced at NPL could stem from the exacting requirement of the individual laboratories and the bespoke nature of the project: the 29 high-tech laboratories that feature close-control systems were, in themselves, one large experiment in air-conditioning. Bearing that in mind, the problems that Laing experienced are more than understandable.
Tucker certainly appreciates the enormous difficulties the team faced. "I have had personal experience of getting these rooms to work," he says. "It's very difficult and I have every sympathy with the teams trying to sort it out."
