Published on January 31st, 2019 | by Jake Richardson0
Solar Power Research Study Investigates Rapidly Declining Costs
MIT researchers studied the steady decrease in solar PV module costs over the last several decades and had their results published in the December 2018 issue of the journal Energy Policy. In their study paper, the authors wrote that of the energy technologies, solar PV has undergone the fastest cost decline. More specifically, they referenced this striking fact, “PV module costs fell by about 20% with every doubling of cumulative capacity since the 1970s.”
Past studies are mentioned, along with their approaches to evaluating solar PV costs, but they state they want to use a new framework, and model which includes a cost equation and low and high-level mechanisms. The cost equation includes module efficiency, wafer area, and manufacturing plant size. While some of these terms and concepts might seem challenging to relate to for lay people, meaning non-scientists, one of the researchers, Dr. Göksin Kavlak, answered some questions about the research for CleanTechnica.
1. From 1980–2012, government-funded and private R&D was the most important high-level contributor to decreasing PV module costs. How did you determine this is true?
R&D efforts targeting PV technologies have existed at private companies, national laboratories, and universities, and PV has benefited from these efforts. However, how much R&D or other high-level drivers such as scale economies contributed to cost reduction was not known.
To estimate the contribution of high-level drivers like R&D, we developed a method that starts with estimating the low-level changes first. In other words, we first calculate the cost change at the device level due to changes in module efficiency, material use, input prices, and so on. We then ask which high-level drivers were most responsible for each device-level change. This way we are able to connect cost to high-level drivers.
R&D drove several device-level changes including improvements to module efficiency, wafer area, and silicon usage. Adding up the cost change contributions of these device-level changes, we found out that the contribution of R&D to cost change was significant.
2. Were there specific aspects of government-funded and private R&D that contributed to these lower costs?
In general, government-funded R&D tends to focus more on developing a fundamental understanding of the technology, while private R&D tends to be more applied and have specific commercial objectives. In the case of PV, both types of R&D played a significant role in improving the technology and reducing costs. They are not necessarily isolated from each other though. Government-funded and private R&D often went hand in hand through public-private partnerships. For instance, the US government supported R&D in emerging PV companies through its PV Manufacturing Technology (PVMaT) program in the 1990s. This helped companies access financial resources as well as technical expertise developed at government facilities like NREL, and accelerated decreases in production costs.
In our ongoing work, we are delving into the specific innovations that originated from private and public R&D to investigate the differences between innovation types in more detail. Stay tuned!
3. Were there specific government programs that were effective in reducing PV module costs?
The PV Manufacturing Technology (PVMaT) program that I mentioned previously was one such program. In the US, there were also other programs, such as the ‘Flat-Plate Solar Array’ that took place from the mid-1970s to mid-1980s and helped develop silicon refining methods, module materials that are critical for durability of the modules, and manufacturing techniques. Japan’s ‘Sunshine Project’ was a similar program.
Besides programs targeting cost reduction through technology development, governments also implemented other programs and policies to encourage the growth of a PV market. Germany and Japan’s residential roofs projects in the 1990s, and later on tax credits in the US, and feed-in-tariffs in Germany and other countries are examples of programs which indirectly affected costs by signaling manufacturers to increase capacities and other stakeholders in the value chain to respond to market growth. In our model, the effects of market expansion on module cost are represented by several high-level drivers such as economies of scale and learning-by-doing,
4. In the abstract of your study, it says “Our method begins with a cost model that breaks down cost into variables that changed over time.” What are those variables?
The variables represent various determinants of module costs. We have eight variables in our cost model: module efficiency, silicon price, silicon usage, non-silicon materials costs, wafer area, manufacturing yield, manufacturing plant size, and an aggregate variable indicating labor, operation & maintenance, electricity, and depreciation of the plant and equipment.
These variables describe the state of the technology in a given year and together make up the total cost. By tracking how the variables changed over time, we estimate their contribution to total cost reduction.
5. The study paper also references scale economies, for the unaware, what does that term mean in the context of the solar power industry?
In our paper, scale economies means that the unit cost of a product, e.g. a PV module, decreases as the scale of the manufacturing plant increases. There can be various reasons for this. For example, as the plant size increases, fixed costs are spread over a greater output, which reduces per-unit costs. In many manufacturing settings, capital and operating costs tend to rise less than proportionately with plant output. Another reason is that a larger plant can allow more specialized tasks, leading to production efficiencies.
Another instance where we observe scale economies is discounts in raw material prices when materials are purchased in large quantities. In the context of PV manufacturing, this refers to unit prices of input materials like polysilicon and other materials going down when they are purchased in large quantities.
6. Based on your research, do you expect the cost of PV modules to continue decreasing?
They will most likely continue decreasing. In fact, since 2012, which was the last year of our study period, module costs have already come down from about 1 $/W to 0.30 $/W. This was mainly due to changes in module efficiency, materials prices and usage, and capital expenses. Considering the ongoing advances in manufacturing, all of the variables of our cost model show improvement potentials to varying degrees. Going forward, the downward trends of some kind will likely continue.
7. If so, can you estimate how much over the next 10 years?
We provide a tool to estimate future costs, too, as well as analyzing historical costs. By populating the model with the expected values of module efficiency, yield, silicon price, and other variables over the next 10 years, we could come up with scenarios for future cost. The expected values could be obtained from detailed technology roadmaps out there which put forward what the values of these variables may be in the future. Rather than coming up with specific numeric estimates, our goal is to provide a tool for this estimation to be made.
8. What might be some key implications of your study from the perspective of the future of PV module manufacturing?
Developing a cost model allowed us to identify the determinants of cost and to track progress in each of them. As I mentioned earlier, we observed that there is a potential for improvement in all of the variables, so we expect module costs to come down further.
High-level drivers like R&D and economies of scale have played important roles in the past and will likely continue to do so. The conventional silicon technology can still benefit from R&D since there are several challenges to be addressed to increase module efficiency while using materials and equipment more productively. R&D has a crucial role in helping develop emerging technologies like perovskites. Gains from scale economies that we observed for silicon technology will likely play a role in these technologies if/when they move on to large-scale manufacturing.
Image Credit: Dr. Göksin Kavlak