The Shire of Murchison in western Australia is the size of the Netherlands but has a population of 110. Viewed from above, a spattering of white, pointy-nosed dishes stand out against the landscape’s red sands and hunched green shrubs. There are 36 dishes in total, which together form the Australian Square Kilometre Array Pathfinder (ASKAP), the world’s fastest survey radio telescope. The area is also home to a collection of spiderlike antennas that observe the universe as part of the Murchison Widefield Array (MWA).
Beginning next year, those lonely instruments in Murchison will be joined by a set of tall antennas resembling Christmas trees. They will combine to form an even grander telescope: the Square Kilometre Array. When completed in 2030, the SKA will be the world’s largest telescope, with thousands of dishes and a million antennas in Australia and Africa canvasing a spectrum of frequencies.
The SKA Organisation’s selection of remote areas including Murchison and the South African desert ensures a radio-quiet environment for obtaining clean data. But it also introduces a daunting challenge. Each of the dishes and antennas will need electricity. Processing all of the signals they collect and correlate will require even more.
Unlike other large science projects, including the Large Hadron Collider near Geneva, the Very Large Array in New Mexico, and the Laser Interferometer Gravitational-Wave Observatory in Louisiana and Washington State, the SKA telescope array cannot simply plug into existing energy infrastructure. The proposed sites will need either new transmission lines spanning at least 100 kilometers or their own power generation capacity.
“We’re unique in that we have a lot of very sensitive sensors in remote areas,” says Adriaan Schutte, the SKA project’s power engineer. In addition, the data from the sensors have to be processed very quickly, since the volume of raw data is too large to store.
Power and the infrastructure to supply it are not cheap. SKA Organisation documents seen by Physics Today estimate that currently up to a quarter of the capital and operational budget for the project’s phase 1, which constitutes about 10% of the final envisioned array, will go toward power. With phase 1 construction expected to begin next year, SKA officials are trying to strike the right balance between transporting electricity from distant utilities and installing solar generation facilities on site—without breaking the bank. Millions of euros are riding on these decisions, which are sure to influence the strategies of future high-profile research facilities that are built in remote areas.
A changing energy landscape
At the SKA meeting in Banff, Canada, in July 2011, all eyes were on the politics surrounding the site choice for the vast project. South Africa and Australia were each vying to host the full array, and the meeting was to be the last international SKA forum before the decision.
But even then, the electricity issue was simmering under the surface. In a breakaway session, a telescope engineer told colleagues that power consumption alone could cost tens of millions of euros each year, a sizable chunk of the array’s projected €100 million annual operating cost. If costs did not come down, the energy requirements had the potential to hobble or even sink the project.
At this stage, prior to the site decision, the SKA’s design was still up in the air. The extreme upper power estimates for the entire telescope were in the region of 1 GW. One option for supplying that electricity was plugging into the local grid of whichever country was selected and running shielded transmission cables to the dishes and antennas. Another possibility was adding renewable-energy-generation outposts to power distant stations. Both options carried potentially untenable price tags. SKA officials were betting that energy capabilities would follow a similar trajectory as computing has under Moore’s law, giving the team lower-cost options for powering its vast telescope array.
Indeed, the energy landscape has changed dramatically over the past eight years. The cost of electricity from renewable energy sources has rapidly dropped, particularly for solar photovoltaics, which fell by nearly three-quarters from 2010 to 2017, according to the International Renewable Energy Agency. The SKA Organisation had always voiced a commitment to green energy from an environmental perspective; now the financial case for installing small, independent solar sources at remote dishes and antennas is sound as well. Since 2014, the SKA’s power cost estimates have been cut in half, due mainly to the drop in renewable energy costs and to greater energy efficiency from the computing facilities that will crunch data from the array.
Lighting up the Christmas trees
After years of planning, construction on the SKA is slated to begin in 2020. When completed, the midfrequency array in Africa will comprise about 2000 dishes, with its core in South Africa and additional dishes in Botswana, Ghana, Kenya, Madagascar, Mauritius, Mozambique, Namibia, and Zambia. Australia is set to host the low-frequency array’s million antennas. In 2017 the SKA board finalized the plans for phase 1 of the project: 197 dishes in South Africa and around 130 000 antennas in Australia.
| SKA Phase 1* by the numbers | ||
|---|---|---|
| Africa | Australia | |
| 197 dishes | Number of instruments | 130 000 antennas |
| 350 MHz–14 GHz | Frequency range | 50–350 MHz |
| 150 km | Max distance between instruments | 65 km |
| 5–6 MW | Onsite power needs | 2–3 MW |
| 2 TB/s | Raw data output | 157 TB/s |
In Murchison, the SKA’s array of 2-m-high antennas will, among other things, investigate ancient hydrogen abundance and probe the primordial post–Big Bang universe. The closest energy infrastructure to the Australian array is a gas line 150 km distant; the closest electricity grid is 300 km away. This means the site requires its own power station.
Fortunately, the scientists working on the ASKAP, a precursor to the SKA, have already demonstrated the feasibility of a stand-alone station with a significant renewables component. Last year the Commonwealth Scientific and Industrial Research Organisation, which manages the ASKAP and MWA, completed a permanent hybrid solar photovoltaic and diesel power station. Solar energy carries the site through the day, and the generators kick in at night. The station is shielded to stop its components from creating their own radio emissions and interfering with the telescopes, says Antony Schinckel, ASKAP director and the SKA Australian infrastructure consortium lead.
The station is designed to produce 1.1 MW, the maximum load expected in summer with the on-site telescopes complete and fully operational. A large percentage of that load goes toward the digital signal processing systems, and that is even before data are sent to the Pawsey Supercomputing Centre in Perth for further analysis.
SKA phase 1 will require more than twice that much power at the telescope site, says Schutte, and its sprawling configuration comes with unique challenges. The Christmas-tree antennas will have a geographic spread of about 65 km; it is expensive to power remote antennas via shielded and buried cables linked to a central power station. The upper estimates in the SKA’s documents put the total cost at €100 000 per kilometer for 300 km, a €30 million hit that doesn’t include the cost of power generation.
But at less than 15 kW apiece, the antenna stations themselves do not, in the grand scheme of the SKA, consume a lot of power, which opens up the possibility of powering them individually. “It is likely that the outermost antenna stations will be powered by stand-alone renewable power systems,” Schinckel says.
As for the computer processing component of the energy equation, engineers are biding their time as computing and energy production continue to become more efficient. “SKA phase 1 is a lot more energy efficient than if we’d built it 10 years ago,” says Schutte. “In fact, the low-frequency array would not have been possible 10 years ago.” He is expecting those trends to continue.
Protecting delicate dishes
In South Africa, almost 200 three-story midfrequency dishes will look at frequencies of 400 MHz and up to investigate the effects of dark energy, hunt for transient radio events, and survey pulsars. There, the challenges are the dish hardware and cooling systems. The site is about 80 km from the nearest town of Carnarvon and currently is home to the 64-dish MeerKAT, a precursor to the SKA that will eventually be folded into phase 1, as well as the KAT-7 telescope and the HERA array. Data from the dishes are correlated on site before being transported to Cape Town via fiber-optic cable for further analysis.
Next year another 133 dishes will be added to MeerKAT. SKA phase 1’s midfrequency array will look like a giant star, with a dense cluster of dishes in the core and three spiral arms of dishes extending up to 120 km.
Unlike the ASKAP, MeerKAT gets its power directly from a local power utility via a 40-year-old transmission line. That approach is unlikely to be feasible once the SKA dishes start going up. MeerKAT has a peak load of about 1.9 MW; an estimate for the SKA phase 1 load is about 5 MW. The current system will struggle to reliably meet that level of demand.
Other factors imperil the current energy delivery system. South Africa often experiences electricity blackouts due to supply problems at its utility. Lightning also poses a threat. So MeerKAT has a significant backup power system, with three diesel generators capable of supplying power to the entire site for a number of days. But that is not enough to satisfy the consumption for the SKA. Each SKA dish contains multiple receivers that need to be cryogenically cooled. If the power goes down for more than a few minutes, it can take days to restore the temperature and vacuum around a receiver, says Schutte.
As a result, SKA officials plan to make stand-alone solar a major part of the final energy configuration. “The distances between these dishes increase with distance from the core, which means that outer dishes are significant distances from each other,” says Craig Smith of the South African Radio Astronomy Observatory. “The reticulation of power to these dishes will be a significant capital expense.” For example, installing underground power cables at the remote sites could cost up to €60 000 per kilometer. In South Africa, the biggest gap between dishes will be 30 km, and there will be many gaps.
Current thinking has the site divided into zones, with a mixture of underground cabling, above-ground power lines, and stand-alone solar. The SKA organizers have already put out two calls for quotes on power systems for the South African site.
Decision time
Even as construction of its phase 1 array approaches, the SKA Organisation cannot make any final decision on its energy strategy. It is in the process of setting up an intergovernmental organization similar to CERN, and it is that organization that will decide what the power configuration will look like. Fortunately, it will cost substantially less than feared back in 2011.
The decision could have ramifications greater than how it affects the SKA’s budget. Telescope designers learn from the mistakes and successes of the telescopes that have gone before them, and future science facilities could benefit from the SKA’s pathfinding.
For example, a number of countries are eyeing the Moon as the site of an ambitious future radio telescope. Blocked from humanity’s radio emissions, the far side of the Moon is ideal for low-frequency radio astronomy. As is the case for the SKA, power is a concern. “The biggest problem for all of these missions is the power supply,” says radio astronomer Heino Falcke of Radboud University Nijmegen. The Moon’s far side alternates two weeks of darkness with two weeks of sunshine. A solar-based system would need to be able to store energy for each two-week period of darkness.
“Power is almost the one and only issue for operations, but it is the same technology we use on Earth,” Falcke says. For a lunar mission, as for the SKA, power is a problem that needs to be solved, without breaking the bank.
