The CSIRO Groundwater Cooling Project - Cooling Australia's Latest Supercomputer with Groundwater

The CSIRO has designed and implemented a geothermal solution using ambient temperature groundwater to cool the recently built Pawsey Supercomputer Centre in Kensington, Western Australia. The corresponding groundwater cooling (GWC) project has developed the system which runs by pumping cool water (~21°C) from the Mullaloo aquifer located approximately 35 to 120 m depth, through an above-ground heat exchanger to cool the supercomputer. The warmed water (up to a maximum of ~31°C) is then reinjected back into the same aquifer, slightly downstream, resulting in no net consumption of water. Two warm water injection boreholes are separated from two cold extraction wells by approximately 340 m. An additional two boreholes are placed in between the injector pair and the extractor pair to potentially serve as a shield by reinjecting cold water to delay thermal breakthrough. The requirements and optimal usage of this shielding functionality is one of the research questions the GWC project will be investigating in the coming years. Nine monitoring wells located in close proximity to the site have been equipped with sondes to collect temperature, pH and other water quality data. These data are in part used to calibrate numerical simulations to help quantify uncertainty, calibrated with various measurements, including real time data as well as regular manual sample analyses. The repository for these data is maintained by CSIRO and partly available to the research community through web portals. Numerical simulations are performed using various codes, including Feflow (Diersch, 2014; Trefry and Muffels, 2007) and MOOSE (Gaston et al., 2009), to model the performance of the GWC system. The GWC system generates a thermal plume in the aquifer which diffuses away from the injection wells and moves westwards with the groundwater flow. Simulation results show that the effects of the GWC system on the environment were very low in the configuration used for the first year of operation. Various scenarios were investigated to understand the impact of the system on water temperature and drawdown around the site under different configurations. Results suggest a minimal impact with drawdown levels off-site less than 0.8 m and recovering to less than 0.1 m within a year after the cessation of pumping. For a 2.5 MWth system, the time to thermal breakthrough ranges from 3.7 to 4.7 years for a 0.1°C temperature increase at the production wells, and from 6.1 to 8.0 years for a 1°C temperature increase at the production wells. The shielding strategy using the cold reinjection wells was found to be effective at delaying thermal