Synopsis

Keywords: Physics & Engineering: Hardware

Superconducting MRI scanners have a high baseline power consumption, This is caused by the cryocooler which is on all the time. There is potential to reduce this power by operating the cryoycooler such that does not generate more cooling than the magnet needs.

It's about cryocoolers and gradients

The power consumption of an MRI scanner is about an order of magnitude higher than of a CT scanner or other medical imaging systems. The two most power-hungry subsystems are the main field magnet (assuming that it is superconducting) and the gradient system. Useful practical data on the energy consumption of such systems were recently reported by Heye et al. [1] and Woolen et al. [2].
Although a superconducting magnet at field does not generate any power of its own (by virtue of the superconducting state), it needs refrigeration to keep it at its operating temperature. Today’s cryocoolers, used to keep the magnet cold, typically draw 7 kW from the mains, to produce a mere 1-1.5 W of refrigeration at the superconducting coil, at 4.2K. In addition, the cryocooler cools away 40-60W at an intermediate temperature (typically 40K), which is the radiation heat received by the shield between the room temperature wall of the magnet and the cold mass with the superconducting coil. Since the magnet is always cold and at field, the refrigeration power is basically constant over time, independent of whether the scanner is in use or not.
The power drawn from the mains to drive the gradient system is, to first approximation, equal to the ohmic dissipation in the coil and other conducting parts of the circuit when the scanner is in use. The enormous peak power delivered by the amplifier to excite the coils to a certain amplitude is recuperated to a large extent in a capacitor bank when the gradients come back to zero. The dissipation is very much dependent on the MRI sequence, typically of the order of 20 kW, but much smaller when the scanner is idle. So, although the gradient power may have high peaks, the time averaged power used by the gradients is generally much lower than the refrigeration power. Switching off the power amplifiers when they are not needed certainly helps to save power. Using less gradient-intensive sequences also helps. The observation by Woolen that 3T scanners use significantly more power than 1.5T scanners is probably due to the fact that high field systems tend to have higher performance gradients.

Cryocooling for zero boil-off magnets, Eco mode

Because all modern superconducting MRI magnets have basically the same size (wall area) and similar thermal insulation quality, the amount of heat to be taken up by the cryocooler is more or less independent of manufacturer and also largely independent of the field strength of the magnet (in the range between 0.5 and 3 T). The heat leak into the magnet’s cold mass is usually significantly less than what the cryocooler extracts when it is running normally. This excess cooling capacity would cool the magnet to a temperature well below 4.2 K. For a bath cooled magnet this would lead to a reduction of the pressure in the tank below atmospheric pressure. This is undesirable because outside air can then be sucked in if there are small leaks, leading to icing of the cryostat’s neck tube. To prevent this from happening, the cooling power is regulated. The classical way this regulation has been implemented by all MRI vendors is to energize a heater in the helium bath as soon as the pressure in the tank drops below the set point, so that the total heat arriving at the refrigerator is equal to the cooling power at the desired temperature/helium pressure. In this mode, the refrigerator is always running at full power. A more economical way to keep the temperature/pressure at the desired value is to use the measured temperature to periodically switch the cryocooler off and on, in the same way as every household refrigerator is kept at a constant temperature by means of a thermostat. This so-called eco-mode is now gradually being implemented by at least one manufacturer. There is no plausible reason why this cannot be done on all zero boil-off magnets installed worldwide. The margin between actual heat load on the coil and the cryocooler power varies from system to system and it will also gradually decrease over time as the performance of the cryocooler degrades over a service interval. Test results suggest a potential power saving of 30% by cycling the cryocooler during the idle time of the system. In principle, the eco-mode can also be used when the scanner is in use; most scans will hardly cause any additional heat load on the cold mass. However, this must be implemented with care, it is not a good idea to switch the cryocooler on or off while a scan is in progress because the transient could have some impact on image quality.

Helium-free (dry) magnets

If the magnet is a so-called dry magnet without a helium bath at atmospheric pressure, there is no simple recipe for energy saving. If the cryocooler stops working on such a magnet, it warms up relatively quickly to a temperature where it must be ramped down to prevent a quench. Depending on the design of the magnet this ride-through time can be up to a few hours. If it is likely that within this time a malfunction can be corrected, it makes sense to let the cryocooler cool the magnet to the lowest achievable temperature, to maximize this ride-through time. In that case there is little opportunity for any eco-mode operation. But if the possibility to correct a problem before the magnet gets too warm is limited in the first place, or if there is enough redundancy in the cooling system that the risk of malfunction is very small, then it makes sense to set the magnet temperature to the highest possible value, which may even be above 4.2K and make full use of the possible energy saving by cycling the cryocooler.

Future options

The poor efficiency of today’s MRI cryocoolers is partly due to laws of thermodynamics (Carnot), partly because the refrigerators used in MRI are of a very simple design with only one moving part (Gifford-Mc Mahon). This makes them robust, cheap and MRI compatible, but their efficiency is lower than of some other types of refrigerator. There is certainly potential to improve the efficiency of these machines. But the largest potential improvement in cryocooler efficiency would be to increase the operating temperature of the magnet. A typical cryocooler with a power of 5 W at 20K requires only about 3.5 kW primary power. The high temperature superconductors that could enable such systems are still rare and expensive, but this may change in the foreseeable future.

Acknowledgements

No acknowledgement found.