A paper examining this potential was written by myself and my research team. The paper selected the United States for modeling because it is a large-sized system with many generation sites, has a great range of weather regions, has very good load data from past years and is a major emitter of greenhouse gases. The examination demonstrated that while more than 90 per cent of the electricity sector could be powered by solar energy with sufficient thermal storage, the coverage left extensive times for when blackouts could occur unless large auxiliary, and probably polluting generating systems, were present.

It was speculated, however, that there might be significant synergies between solar and wind, not least because the wind blows at night when solar is absent, but also because there was an idea that winter winds were strong when solar was diminished. This was a proposition that needed some confirmation.

Setting study parameters

The 2006 US electricity supply data used for the paper was calculated on an hour-by-hour basis, using US Government energy load data for that year and 2006 National Renewable Energy Laboratory (NREL) solar and wind weather data for load match modelling. These datasets were assembled into national output data.

The paper is directed at technical feasibility only. My colleagues and I adopted the premise that the technologies assumed for the modelled systems should be already available, and that large technical advances should not be required to produce a functioning national energy system without blackouts.

The general solar system design was adapted to peaking using components available today. Thermal storage is now being widely installed on new solar plants and is regarded as a proven entity.

The grid is assumed not to be a blockage, and we assumed it to have a high voltage direct current backbone and high voltage alternating current ‘ribs’. The grids in the US and Australia are both in need of replacement and such technology should be used where appropriate to allow energy transfer over as large an area as possible.

Defining the study data

Our wind dataset used the Eastern Wind Integration and Transmission Study, and a corresponding Western Wind Integration and Solar Transmission study was developed by NREL. The results of our integration of these are given in Figure 1, which shows a wind resource peaking in winter, as we had hoped, with a peak of 1.02 terawatts (TW) of coincident generator output. In summer, the entire wind fleet can drop as low as 67 gigawatts (GW), so the output is highly variable.

Nevertheless, this was anticipated to be the replacement for low-cost base load in the system, with wind being as cheap as, or cheaper, than coal and nuclear generation.

The 2006 electricity load was developed from data in the Federal Energy Regulatory Commission’s (FERC) online database. In the end, the separate Energy Information Administration estimate of 4,107 terawatt hours (TWh) of national electricity used for that year was adopted but the hourly load variations in the FERC data were used to allocate the electricity hourly throughout the year. The result is shown in Figure 2 and reveals a highly variable summer peaking load, with air-conditioning a significant cause of the summer peak.

Solar plant output data was calculated for NREL data using a solar simulation program developed by my team and I, but this gives outputs very similar to the NREL Solar Advisor Program for a parabolic trough, which can be used for similar calculations by others who may want to perform such calculations. The characteristic is not shown because solar and thermal storage is simply used to bridge the gap between the wind resource and the load, and is the difference between these two, with any excess energy being discarded or dumped.

It was found necessary to overbuild the solar/wind fleet to ensure full coverage at low resource times, which mainly occurred in December and January. The overbuild was based on 25–35 per cent in excess of the annual production potential of 4,107 TWh and is called redundancy by the authors, but this redundancy also useful in ensuring that sufficient electricity is available if there are breakdowns. The conventional US generation also had an overbuild of 39 per cent, used to ensure coverage from breakdown and periodic maintenance.

Study results

With a 25 per cent redundancy, the fleet achieved 99.99 per cent coverage of load using 18 hours of thermal storage, and the annual distribution of solar and wind was 63 per cent and 37 per cent respectively. With 35 per cent redundancy (a fleet size increase of 8 per cent), 100 per cent coverage was achieved with ten hours of thermal storage, and the annual distribution of solar and wind was 62 per cent and 38 per cent respectively. The higher fraction of solar was due to a better match to the higher peak load in summer. Full coverage of the load is possible without base load technology.

The final few percent of matching, however, is very expensive and 100 per cent can also be achieved by using a 25 per cent redundancy system with only three or four hours of storage and an added 1–2 per cent of combustible backup fuel. Solar thermal plants can run from direct sun, thermal storage, or fuel. This seems a substantially cheaper approach because capital costs are lowered by 8 per cent and storage costs are cut significantly.

There are several backup options: natural gas, biomass fuel, hydrogen created from dumped electricity, and methane created from dumped electricity and hydrogen, combined chemically with carbon dioxide (methanation). The use of dumped energy is highly attractive in that the solar fleet dumped energy can be fully utilised for additional income, but hydrogen can require expensive storage while methanation requires purchase of expensive carbon dioxide from carbon capture storage (CCS) or using advanced technology and more energy to gather it from the air. Hydrogen can be used both for direct reduction of iron ore and aircraft fuel however as CCS is not yet proven for large scale use with coal, it was not assumed in this paper.

By calculating the potential of wind and solar, it is not suggested that technologies like geothermal and hydroelectricity cannot be accommodated. These non-storage technologies, however, would be used to reduce wind usage and not solar thermal. Also, should any new technologies such as very low-cost electrical storage appear, these would only increase the likelihood of a lower cost system.

We preferred not to speculate if or when they might arrive as this exercise was about existing technology being used.

Study conclusion

In conclusion, my colleagues and I found that it does appear that in 2006 the US electrical grid could have been powered by wind and solar thermal electricity alone, using already commercially proven technology.

The common notion that we need base load sources like nuclear and coal CCS is not supported by this work. In the future, low-cost batteries might improve the picture, but that part of the future is unclear; we cannot yet assume batteries will be cheaper than thermal storage.

A second paper showing the powering of an electrified US, totally supplied by wind and solar, is being prepared.

Dr Mills is Executive Vice President, Chief Scientific Officer and Founder of Ausra. He is a former President of the International Solar Energy Society, and served as inaugural Chair of the International Solar Cities Initiative. Dr Mills is known for pioneering Compact Linear Fresnel Reflector technology and is known for his work in non-imaging optics, solar thermal energy and photovoltaic systems conducted over 32 years.