SOLAR PV PANEL WASTE PROJECTIONS

Analysis
Typography

PV panel waste streams will increase alongside worldwide PV deployment. This publication is the first to quantify potential PV panel waste streams in the period until 2050. 

As outlined in Figure 1, a three-step approach is used to quantify PV panel waste over time. First, this chapter analyses trends and future global solar PV growth rates from 2010 to 2050, which is a main input to waste volume estimation. Next, the PV panel waste model and main methodology used in this report are explained. The last section summarises the findings and provides PV panel waste predictions globally and by country.

1. GLOBAL SOLAR PV GROWTH 

In 2015 capacity to generate renewable energy increased by 8.3% or 152 GW, the highest annual growth rate on record (IRENA, 2016b). Global solar PV capacity added in 2015 made up 47 GW of this increase, cumulatively reaching 222 GW at the end of 2015, up from 175 GW in 2014 (IRENA, 2016b). The bulk of these new installations was in non-traditional PV markets, consolidating the shift in major PV players. Traditional PV markets such as Europe and North America grew 5.2% and 6.3% in 2015 respectively. By contrast, Latin America and the Caribbean grew at a rate of 14.5%, and Asia at a rate of 12.4%. Asia alone thereby witnessed a 50% increase in solar PV capacity in 2015, with 15 GW of new PV capacity installed in China and another 10 GW in Japan. Main global PV leaders today include China (43 GW of cumulative installed capacity), Germany (40 GW), Japan (33 GW) and the US (25 GW).

To account for current and future waste streams for solar PV, global PV growth rates were projected until 2050. These rely on results from previous work on PV forecasts by both IRENA and the IEA. For projections to 2030, REmap (see Box 1), IRENA’s roadmap for doubling the global share of renewables, was used (IRENA, 2016a). For 2030-2050, the projections are based on IEA’s Technology Roadmap on Solar Photovoltaic Energy (see Box 2) (IEA, 2014).

 

As shown in Figure 2, global cumulative PV deployment accelerated after 2010 and is expected to grow exponentially, reaching 1,632 GW in 2030 and about 4,512 GW in 2050.

To develop annual estimates of PV capacity between 2016 and 2030, an interpolation was made between IRENA’s REmap estimates for 2015, 2020 and 2030. To achieve this, an average annual growth rate was calculated between each five-year period, amounting to 8.92%. In some selected countries, the individual growth rates may be adjusted higher or lower due to political and economic uncertainties foreseen. To extend the model projection to 2050, more conservative growth projections were assumed for 2030-2050 with annual growth rate of about 2.5%. This extrapolation was matched with the forecast of the IEA’s PV Technology Roadmap. 

The final projections of global PV growth to 2050 are shown in Table 1 and were used to model global waste streams in the next chapter.

2. PV PANEL WASTE MODEL 

The objective of this report is to quantify future PV panel waste streams. Most waste is typically generated during four primary life cycle phases of any given PV panel. These are 1) panel production 2) panel transportation 3) panel installation and use, and 4) end-of-life disposal of the panel. The following waste forecast model covers all life cycle stages except production. This is because it is assumed that production waste is easily managed, collected and treated by waste treatment contractors or manufacturers themselves and thus not a societal waste management issue. 

Future PV panel waste streams can be quantified according to the model described in Figure 3. The two main input factors are the conversion and probability of losses during the PV panel life cycle (step 1a and 1b). They are employed to model two waste stream scenarios using the Weibull function, the regular-loss and the early- loss scenario (step 2).

The next section provides a step-by-step guide showing details of the methodology and underlying assumptions.

To estimate PV panel waste volumes, installed and projected future PV capacity (megawatts or gigawatts-MW or GW) was converted to mass (metric tonnes-t), as illustrated in Table 2. An average ratio of mass of PV per unit capacity (t/MW) was calculated by averaging available data on panel weight and nominal power. For past PV panel production, the nominal power and weight of representative standard PV panel types was averaged from leading producers over five-year intervals (Photon, 2015). The panel data sheets of Arco, Siemens, BP, Solarex, Shell, Kyocera, Sharp, Solarworld and Trina were considered. 

For future PV panel production, the data are based on recent publications (Berry, 2014; IEA, 2014; IRENA, 2014; Marini, 2014; Raithel, 2014; Lux Research, 2013 and Schubert, Beaucarne and Hoornstra, 2013). 

This report’s model includes a correction factor to account for panels becoming more powerful and lighter over time. This is due to optimisation of cell and panel designs as well as weight reductions from thinner frames, glass layers and wafers. The correction factor is based on an exponential least-square fit of weight-to-power ratio for historic and projected future panels. Figure 4 shows how the weight-to-power ratio is continuously reduced over time due to further developments in PV technologies such as material savings and improved solar cell efficiencies.

The potential origin of failures for rooftop and ground-mounted PV panels was analysed independently from PV technology and application field to estimate the probability of PV panels becoming waste before reaching their estimated end-of-life targets. The three main panel failure phases detected are shown in Table 3 (IEA-PVPS, 2014a):

• Infant failures defined as occurring up to four years after installation (average two years);

• Midlife failures defined as occurring about five to eleven years after installation; 

• Wear-out failures defined as occurring about 12 years after installation until the assumed end-of-life at 30 years. 

Empirical data on causes and frequency of failures during each of the phases defined above were obtained from different literature (IEA-PVPS, 2014a; Padlewski, 2014; Vodermayer, 2013 and DeGraaff, 2011). Independent of those phases, Figure 5 provides an overview of the main causes of PV panel failure.

The main infant failure causes include light-induced degradation (observed in 0.5%-5% of cases), poor planning, incompetent mounting work and bad support constructions. Many infant failures have been reported within the electrical systems such as junction boxes, string boxes, charge controllers, cabling and grounding. 

Causes of midlife failures are mostly related to the degradation of the anti-reflective coating of the glass, discoloration of the ethylene vinyl acetate, delamination and cracked cell isolation. 

Causes of frequently observed failures within all phases in the first 12 years - after exposure to mechanical load cycles (e.g. wind and snow loads) and temperatures changes - include potential induced degradation, contact failures in the junction box, glass breakage, loose frames, cell interconnect breakages and diode defects. 

In the wear-out phase, failures like those reported in the midlife phase increase exponentially in addition to the severe corrosion of cells and interconnectors. Previous studies with statistical data on PV panel failures additionally observe that 40% of PV panels inspected suffered from at least one cell with microcracks. This defect is more commonly reported with newer panels manufactured after 2008 due to the thinner cells used in production. 

These failures and probability of loss findings, alongside data from step 1a (conversion factors) are used to estimate PV panel waste streams (step 2). 

On the basis of step 1a and 1b, two PV waste scenarios were defined (see Table 4) – the regular-loss scenario and early-loss scenario.

Both scenarios are modelled using the Weibull function as indicated in the formula below. The probability of losses during the PV panel life cycle is thereby determined by the shape factor α that differs for the regular-loss and early-loss scenario.

Both scenarios assume a 30-year average panel lifetime and a 99.99% probability of loss after 40 years. A 30-year panel lifetime is a common assumption in PV lifetime environmental impact analysis (e.g. in life cycle assessments) and is recommended by the IEAPVPS (Frischknecht et al., 2016). The model assumes that at 40 years at the latest PV panels are dismantled for refurbishment and modernisation. The durability of PV panels is thus assumed to be in line with average building and construction product experiences such as façade elements or roof tiles. These also traditionally have a lifetime of 30-40 years. 

Neither initial losses nor early losses were included in the regular-loss scenario. The results from Kuitsche (2010) are used directly, assuming an alpha shape factor in this scenario of 5.3759 (see Table 5).

In the early-loss scenario, the following loss assumptions are made based on an analysis of the literature and expert judgement (IEA-PVPS, 2014a; Padlewski, 2014; Vodermayer, 2013 and DeGraaff, 2011): 

• 0.5% of PV panels (by installed PV capacity in MW) is assumed to reach end-of-life because of damage during transport and installation phases ; 

• 0.5% of PV panels will become waste within two years due to bad installation; 

• 2% will become waste after ten years;

• 4% will become waste after 15 years due to technical failures. 

The early-loss scenario includes failures requiring panel replacement such as broken glass, broken cells or ribbons and cracked backsheet with isolation defects. However, only panels with serious functional or safety defects requiring entire replacement are included, while other defects that, for example, reduce power output or create panel discoloration are ignored. 

In the early-loss scenario, the shape factor was calculated by a regression analysis between data points from literature and also considered early failures (see Table 5). The resulting alpha shape factor of 2.4928 for the early-loss scenario is lower than literature values presented. This is because it includes early defects that yield higher losses in the first 30 years and lower losses in later life should a panel last longer. 

For each scenario (regular-loss and early-loss), the probability of failure value (alpha) is multiplied according to the Weibull function by the weight of panels installed in a given year. Since a bigger alpha value is used in the regular-loss scenario, the curve ascends smoothly and intersects with the early-loss scenario curve at the nominal lifetime point of 30 years. In line with the Weibull function and due to the different assigned alpha parameters, regular-loss and early-loss scenarios have the opposite effect after 30 years. Hence, the regularloss scenario indicates a higher probability of loss from 30 years on (see Figure 6).

The above modelling produces PV panel waste projections by country up to 2050. The next section summarises the findings of the model. 

3. PV PANEL WASTE PROJECTIONS 

Global PV panel waste outlook

Total annual e-waste in the world today accounts for 41.8 million t (Baldé, 2015). By comparison, cummulative PV panel waste will account for no more than 250,000 t by the end of 2016 according to the early-loss scenario modelled in this report. This represents only 0.6% of total e-waste today but the amount of global waste from PV panels will rise significantly over the next years. 

Figure 7 displays cumulative PV panel waste results up to 2050.

• In the regular-loss scenario, the PV panel waste accounts for 43,500 t by end 2016 with an increase projected to 1.7 million t in 2030. An even more drastic rise to approximately 60 million t could be expected by 2050. 

• The early-loss scenario projection estimates much higher total PV waste streams, with 250,000 t alone by the end of 2016. This estimate would rise to 8 million t in 2030 and total 78 million t in 2050. This is because the early-loss scenario assumes a higher percentage of early PV panel failure than the regular-loss scenario. 

Based on the best available information today, this report suggests the actual future PV panel waste volumes will most likely fall somewhere between the regular-loss and early-loss values.

Annual PV panel waste up to 2050 is modelled in Figure 8 by illustrating the evolution of PV panel end-of-life and new PV panel installations as a ratio of the two estimates. This ratio starts out low at 5% at the end of 2020, for instance (i.e. in the early-loss scenario, annual waste of 220,000 t compared to 5 million t in new installations). However, it increases over time to 4%-14% in 2030 and 80%-89% in 2050. At that point, 5.5-6 million t of PV panel waste (depending on scenario) is predicted in comparison to 7 million t in new PV panel installations. 

A feature of the Weibull curve shape factors for the two modelled scenarios is that the estimated waste of both scenarios intersects. The scenario predicting greater waste panels in a given year then switches. The intersection is projected to take place in 2046. This modelling feature can be observed in Figure 8 which shows the volume of PV panel waste amounting to over 80% of the volume of new installations as a result of the early-loss scenario in 2050. The comparable figure for the regular-loss scenario exceeds 88% in the same year.

Waste projections by country 

Detailed PV panel waste estimates by selected countries are displayed in Table 6 from 2016 up to 2050. The countries were chosen according to their regional leadership when it comes to PV deployment and expected growth. 

The projections are modelled using the same Weibull function parameters as the global estimates of the previous section. Projected waste volumes of PV panels in individual countries are based on existing and future annual installations and rely on input data available for each country. The historic cumulative installed PV capacity was used as benchmark in each country alongside future projections to 2030 using IRENA’s REmap and for 2030 to 2050 IEA's PV Technology Roadmap, with a simple interpolation.

PV panel waste projections until 2030

The results modelled indicate that the highest expected PV panel waste streams by 2030 are in Asia with up to 3.5 million t accumulated, depending on the scenario. Regional Asian champions in renewable energy deployment will therefore also experience the highest waste streams. For example, China will have an estimated installed PV capacity of 420 GW in 2030 and could accumulate between 200,000 t and 1.5 million t in waste by the same year. Japan and India follow, with projections of between 200,000 t and 1 million t, and 50,000-325,000 t in cumulative PV-waste by 2030 respectively. 

Europe is predicted to present the second largest PV waste market with projected waste of up to 3 million t by 2030. Germany, with an anticipated 75 GW of PV capacity, is forecasted to face between 400,000 and 1 million t of PV panel waste by 2030. Other future significant PV waste markets are projected to include Italy and France. 

With an expected cumulative 240 GW in deployed PV by 2030, the US will lead in terms of total installed PV capacity in North America. It is projected to generate waste between 170,000 and 1 million t by then. Countries such as Canada (up to 80,000 t) and Mexico (up to 30,000 t) will also experience rising PV waste streams by 2030.

By 2030 Africa and Latin America are predicted to also see expanding PV-waste volumes. South Africa (8,500-80,000 t by 2030) and Brazil (2,500-8,500 t by 2030) will be regional leaders in this respect. Other significant PV-waste markets by 2030 will include the Republic of Korea with cumulative waste of 25,000-150,000 t and Australia with 30.000-145,000 t. 

Waste volume surge in 2030-2050 

Given the worldwide surge in PV deployment since 2010 and average lifetime and failure rates for panels, waste volumes are certain to increase more rapidly after 2030. Whereas in 2030 the top three PV panel waste countries are expected to include China, Germany and Japan, the picture slightly changes by 2050. By then, China is still predicted to have accumulated the greatest amount of waste (13.5-20 million t). However, Germany is overtaken by the US (7.5-10 million t), Japan is next (6.5-7.5 million t) and India follows (4.4-7.5 million t). The regular-loss and early-loss waste estimates by top five countries in 2030 and 2050 are displayed in Figure 9. 

The analysis presented in this chapter develops quantitative estimates for PV panel waste streams until 2050 by country and region as well as on a global scale. At the same time, PV panels and consequently their waste differ in composition and regulatory classification.

 

Source: IRENA (© IRENA 2016)