The Importance of Energy Payback Time in Photovoltaics

Craig Peters
October 23, 2010

Submitted as coursework for Physics 240, Stanford University, Fall 2010

Fig. 1: Annual PV capacity added to grid (dark blue), total installed PV capacity on grid (red) and total net PV capacity on grid (yellow).

The world is experiencing a shift into clean sources of energy that is unparalleled historically. Fears of global climate change, coupled with concern over energy independence and the depletion of traditional fossil fuel sources, is driving this shift. A number of "clean" viable technologies have emerged to replace fossil fuels including wind turbines, solar thermal and solar photovoltaics (PV). Many compelling arguments, which will not be explored here, have been made for PV to become the dominant source of renewable power generation. PV has made, on a year-on-year growth basis, significant headway in the marketplace. The global PV market has grown at a compound annual growth rate of about 40% for the past 10 years. [1] Underpinning this growth have been generous government subsidies in countries like Germany, Spain, Italy and the US. However, for PV to satisfy a significant percentage of the world's energy demand it must ultimately be scalable, cost competitive with traditional fuel sources and have a short enough energy payback time (EPBT). EPBT is the point in time at which PV device has produced as much power as it took to produce the PV device itself. It is this last point that is generally ignored by the policy makers and industry players.

In order to illustrate the magnitude of the energy problem, global energy usage is currently 15 TW and expected to increase by over 50% by 2035 based on the US Energy Information Administration's International Energy Outlook 2010. The key questions are (1) can we satisfy much of this new demand through PV and (2) can we further reduce our current dependence by replacing existing fossil fuel burning plants with PV. Though important, I will ignore the issues of the cost of PV modules and the limited supply of certain key elements, such as indium and tellurium. I will instead focus on the growth rate required in the PV industry and what this means as it relates to the EPBT for PV modules.

Current global PV production capacity is approximately 13 GW. Let's assume that we want to satisfy 100% of the expected growth in demand by 2035 using PV. This implies a need to have an installed PV capacity of 7.5 TW by 2035. In order to achieve this capacity the PV industry would need to grow at a compound annualized growth rate (CAGR) of ~20%. Let's also assume, for the sake of simplicity, that this CAGR also coincides perfectly with the projected growth in energy demand. This is not necessarily a reasonable assumption but one that will not invalidate the conclusions in this paper. Importantly, we need to consider the amount of energy that is consumed in the process of manufacturing the PV modules. Fig. 1 shows the growth in installed capacity assuming a 20% CAGR. The red line shows the total installed PV capacity at the end of each year while the blue line shows the PV capacity added in any given year.

Fig. 2: Annual PV capacity added to grid (dark blue), total installed PV capacity on grid (red) and total net capacity on grid (yellow). This assumes a 12% CAGR for the PV industry with a one-year energy payback time for the PV modules.

The yellow line, which shows the total net capacity on the grid, is the critical one. In order to generate this I have assumed a one year EPBT for the modules produced. This means that the energy needed to manufacture a solar panel is recapture by the solar panel after one full year of use. So if I manufacture a 100-Watt solar panel, it takes another 100-Watt solar panel one full year of operation to product the energy to manufacture it. This effectively reduces the amount of PV generation capacity on the grid while the industry is still growing and producing new panels. For the plot I have assumed that the energy used to produce new PV modules comes exclusively from previously installed PV modules. Clearly this is not possible for the first two years but we can make this allowance without diluting the main point. We see that by 2035 we approach 7.5 TW of installed capacity (red line), which was our expected growth in energy demand. However, due to the fact that it takes one year of operation of the PV module to generate the energy it took to produce that module, we know intuitively that the net capacity on the grid should be lower (yellow line). In fact, at the end of 2034 we would have 6.2 TW of installed capacity but given the need to add over 1.25 TW of new capacity that year the net capacity on the grid would only be ~5 TW. This implies that we would need to fulfill the excess 2.5 TW of energy demand through some other means.

Now if we take a more modest approach and assume that PV will satisfy 25% of the expected growth in demand, or about 1.9 TW of capacity, then we arrive at a CAGR of 12%. Fig. 2 shows a similar plot to Fig. 1 but with a 12% CAGR. We see that in 2035 we would have about 2 TW of installed PV capacity but for that year we would only have 1.5 TW of net capacity on the grid. This would mean that we would have to satisfy 500 GW of excess demand with other fuel sources.

Currently there is a wide range of EPBT for PV modules. One detailed analysis of fixed versus active tracking systems put the range of time between 2-5 years, which is significantly greater than the one year I assumed. [2] This would further lower the net capacity added to the grid and increase the need to alternative fuel sources, including traditional fossil fuels. Though it's challenging to get the exact EPBT for thin film module producers, First Solar has claimed an EPBT at or below one year. The key point is that we are in a difficult position. In order to fulfill a significant fraction of our energy demands through PV we need a substantial CAGR in capacity. However, this new capacity requires energy in order to produce it, which reduces the net capacity added to the grid during this growth period. This reduced capacity then requires substantial capacity from other fuels sources, including fossil fuels, to be used to satisfy the expected demand.

On a final note, one of the key reasons for the long EPBT is the fact that PV modules are delivering power for on average 5.5 hours of one-sun intensity per day. If they were able to produce power throughout the day, like a coal, natural gas or nuclear power plant, the EPBT would be reduced by a factor of 4, which is significant.


[1] W. Hoffman, "PV Solar Electricity: From a Niche Market to One of the Most Important Mainstream Markets for Electricity," Optical Sciences 140, 29 (2009).

[2] O. Perpinan et al., "Energy Payback Time of Grid Connected PV Systems: Comparison Between Tracking and Fixed Systems," Prog. Photovolt: Res. Appl. 17, 137 (2009).