Contents - Previous - Next
This is the old United Nations University website. Visit the new site at http://unu.edu
The major obstacle to the widespread use of PV is currently the high capital and installation costs of the system. Owing to the absence of moving parts and to the simplicity and reliability of PV systems, operating and maintenance costs can be very low if the lifetime is long enough. "Fuel" costs are obviously zero.
At present, installation costs are more or less equally shared between module costs and "Balance-of-System" (BOS) costs. PV modules4 costs have fallen sharply during the past two decades as the global market grew to 61 MW peak power worldwide shipments in 1994. The average selling price of complete PV modules fell from around US$50 per peak watt production in 1976 to less than US$5.5 in 1994 (Strategies Unlimited 1995) (see fig. 7.3).5 In terms of current US$/Wp prices, this corresponds to a general trend of price decline (about 10 per cent per year) over the past decade. In 1988-1990, however, supply shortages of crystalline silicon wafer scraps led to briefly higher prices of PV modules (actually a slower price decline, in real terms). After 1991, the previous rate of price decline resumed, owing to additional capacity coming on-line and a new surplus of electronics-grade silicon feedstock. The present (1995) PV market situation is characterized by over capacity combined with increased price competition in the power module market (Vigotti 1994a). In the near term, price declines can be expected to continue.
But what can be said about long-term forecasts? Will PV module prices continue their historical downward path? At what rate of decline? Will there be a saturation point?
According to optimists, there is enormous potential for further PV cost reduction through technological innovation and economies of scale. This situation is depicted in the "PV learning curve" shown in figure 7.4. Based on the historical average price pattern, possible PV specific module prices down to US$1/Wp for a world PV market of 1-10 GWp are forecast. Pessimists, on the other hand, point out that this learning curve does not reflect the time variable. So, although the cost decline will almost certainly happen eventually, in fact it could be delayed for a considerable time.
Fig. 7.3 PV mean module prices and PV market
However, there are at least three good reasons to be optimistic about specific cost reductions (US$/Wp) of PV systems. The first reason is that, until now, crystalline silicon modules using electronics grade silicon scraps as a feedstock have dominated the world PV market. As already mentioned, their manufacturing processes derive from the electronics industry and are not optimized for PV cell production. The annual production capacity is fairly low (max. 5 MWp/ year for a single plant). Electronics-grade silicon scraps are becoming scarce and expensive. Significant cost reductions will be achieved as soon as dedicated production facilities for solar-grade silicon are justified by demand. As a matter of fact, Siemens Solar recently announced a target price of US$2/Wp for Czochralsky-based silicon crystal modules to be achieved through wafer geometry, process changes, and materials handling measures within a typical production capacity of 25-100 MWp per year (Strategies Unlimited 1994). On the same occasion, Solarex Corp. presented its module production cost goal of US$1.20/Wp within the framework of the US Department of Energy PVMat programme, to be reached by improved cutting, automated module assembly, and a threefold increase in production capacity.
Fig. 7.4 The PV "learning curve"
Even sharper module cost reductions can be expected in the case of thin film PV cells, irrespective of the basic semiconductor employed (amorphous silicon, CdTe, CIS, or others). First, this is due to the use of a much smaller amount of semiconductor material and to much lower energy consumption rates. Secondly, thin-film manufacturing techniques (direct deposition) allow the direct manufacturing of 1,000 cm2 integrated solar modules (i.e. a-St) and are particularly well suited for mass production.
The cost goals for thin-film modules are summarized in table 7.2 (Kelly 1993). In accordance with other authors (Coiante and Barra 1992), these considerations have been incorporated in figure 7.4 by assuming that another learning curve, with a faster-declining slope, will apply in the case of thin films. Such a two-slope learning curve is nothing new in the history of technology. Several products, and particularly semiconductor devices, have actually followed this type of historical pattern (Ayres and Martinas 1992). The present market situation, on the verge of a change in the slope of the learning curve, just reflects a maturing PV industry, emerging from the laboratory and entering the real energy market.
Table 7.2 Cost goals for thin-film PV modules
Module efficiency (%)
Annual production capacity (MWp/year)
Expected module cost (US$/Wp)
|CdTe and CIS||10||10||1.19-1.86|
Source: adapted from Kelly (1993).
In fact, in late 1994, a US natural gas company (Enron Corporation) entered into a joint venture with an oil company (Amoco Corporation) to form Amoco/Enron Solar and acquire 50 per cent ownership of Solarex (Strategies Unlimited 1994). Under the venture, it will participate in the US Department of Energy (DOE) Solar Enterprise Zone project, building up a 100 MWp amorphous silicon solar power plant to be completed by 2003. Beginning in 1997, the PV electricity produced will be sold at the remarkable price of US$0.055/kWh, which is fully competitive with conventional electric power sources. Of course, this is the result of the DOE subsidy. However, most important, and for the first time in history, the development of a sustained long-term market will allow the venture to make large-scale production investments. This would induce a quantum step down the PV learning curve towards substantial cost reduction. The venture will eventually enable Solarex to be the largest world PV module manufacturer, with 10 MWp annual output capacity expected in 1997, to be doubled again by the year 2000.
A final reason for optimism is that several types of laboratory test cells show much higher efficiencies than those of present commercial modules (see table 7.1 above), which still remain far below the theoretical limits. These very encouraging results will probably be transferred to the market within the next few years. For example, the 24 per cent efficient buried contact technology silicon cell is expected to go into mass production before 20006. The first commercial modules, using low-cost, low-purity polycrystalline feedstock material, will have 15 per cent efficiency, with the potential for later improvements up to 22 per cent. Thus, observing that a wide range of technological patterns are in contention for the PV market, it can be concluded that some or several of them will achieve commercialization at competitive prices in a reasonable time-span.
More efficient modules need less area to produce the same amount of energy. As a consequence, efficiency gains will lead to a reduction in both specific (US$/Wp) module costs and "Balance-of-System" (BOS) costs, which also account for a high percentage of current costs.
There are reasons for substantial optimism that PV could soon enter a "virtuous circle" leading to competitiveness as far as the "Balance-of-System" too is concerned. For instance, "learning by doing" in large demonstration projects led to cost reductions of 50 per cent between 1984 and 1994 (see table 7.3). The world's largest operating PV power plant in Serre, Italy, showed BOS costs in the order of US$3.2/Wp.7 An assessment of further possible improvements in industrial replications suggests further reduction potential in the range of 40 per cent, leading to an eventual cost of US$2/Wp (lliceto et al. 1994).
Table 7.3 BOS cost reduction in large demonstration projects
PV power plant
Year of construction
Total power installed (kWp)
"Balance-of-system" costs (US$/Wp)
|Carrisa Plant (USA)||1984||760||6.56|
|Kerman Project (USA)||1992||500||3.02|
|Possible replicas||1,000 5,000||2.01)|
Sources: adapted from Cunow (1994), Iliceto et al. (1994).
a. Original data in ECU; a 1994 exchange rate of I ECU = US$1.2 was assumed.
Table 7.4 Total system costs for different PV technologies and applications
Module price (US$/Wp)
BOS cost (US$/Wp)
Total system cost (US$/Wp)
|1994||Crystallized silicon||Large power plant||4.25a||3.20||7.45|
Sources: estimated on the basis of Kelly (1993),Iliceto et al.
(1994), TERES (1994).
a. Large modules in quantity are usually sold at a price 20 25 per cent lower than the average selling price of all modules.
Potentially far more effective BOS cost reductions are likely to be achieved by the high diversification process of PV market products related to PV integration in buildings. At present, BOS costs of building-integrated PV demonstration systems are generally higher than those of power plants. However, as soon as more standardized market products and installation procedures are introduced, BOS costs will drop. Substantial improvements in both performance and cost reduction are also expected through the adoption of solid-state electrical inverters. The best building-integrated system costs could be as low as half the costs of some conventional centralized PV power plant systems as early as the year 2000 (TERES 1994). This would lead to significant total system cost reductions, as summarized in table 7.4.
Moreover, the installation costs of PV in buildings have to be considered within the framework of total building construction or retrofit costs. As regards conventional PV power plants, there are savings both because already existing (or at least planned) supporting structures are used and because PV panels are used instead of planned cladding components. In fact, it was recently estimated that PV panels on highway sound barriers in Switzerland could be cost effective in the immediate future (Strategies Unlimited 1994).8
Finally, PV integration in buildings also allows useful thermal energy recovery. In addition, the PV daily power supply curve coincides with the typical electricity demand curve, where that is driven by buildings-related energy final uses, such as air-conditioning.9 Both of these "services" (co-generation and an electricity supply where and when it is needed) actually increase the value of the energy provided by PV systems in buildings.
In fact, the integration of PV in buildings could be enormously effective in providing the momentum needed for the PV industry to move along the learning curve.
A PV market diffusion strategy
In many energy "future" studies it has been assumed that PV will be competitive with other energy sources only if or when the long-term module price goal of US$1/Wp is reached. At that price, 15 per cent efficient PV modules in a sunny area could provide electricity for around US$0.10/kWh (Kelly 1993).10 However, if US$1/Wp is the "break-through" point for buse load electricity production, there are other market segments in which PV could become competitive at an earlier developmental stage, because the actual cost of conventional electricity for these applications is substantially higher than average electricity prices.
Following this approach, the International Energy Agency (IEA) has developed a PV market diffusion strategy. This approach envisages PV entering and diffusing through a series of six expanding markets, namely remote customer applications, remote communities and islands, grid connected building-integrated systems, local utility grid support, peak power supply, and (lastly) bulk power supply.
Two competitiveness parameters are defined, namely the "entry price" and the "deployment price." The "entry price" is the lowest price at which PV is likely to find "niche" opportunities to enter the market. As soon as PV reaches the "deployment price," it is fully competitive in the market and large-scale diffusion will begin to occur. For completeness of information, both system and module prices are indicated in figure 7.5.
Fig. 7.5 Entry and deployment pricing of PV power applications (Source: Vigotti 1994a)
The resulting market diffusion strategy for six different PV system applications is summarized in figure 7.6.
Fig. 7.6 Diffusion strategy of PV system applications (Source: Vigotti 1994a)
PV systems have already proved to be cost effective for a wide range of small applications to power remote communications, safety, and control devices (in the range of 10W to 10 kW).
Remote communities and islands
PV systems are also cost effective, or at least very close to competitiveness, when supplying power for local grids in remote villages and small islands (10 kW to 1 MW power range). Both applications have been successfully demonstrated in industrialized countries and have an enormous application potential in developing countries because of missing or incomplete electricity supply and distribution infrastructures in those regions. PV is one of the most appropriate technologies to meet the increasing demand for rural electrification. In fact, there has been a very high recent growth in remote solar home systems in India. Indonesia is already a proven market for these stand-alone systems. The potential for this kind of application is also very high in China, Central Asia, southern Africa, and the Maghreb region. Hybrid diesel-PV systems for supplying villages have recently been successfully applied in Brazil and have proved to be cost effective. The overall potential of a decentralized PV systems market in developing countries has been estimated to be up to 140 GW (Vigotti 1994a). This huge market potential could certainly have a positive impact on the development of PV technology, and also on grid-connected applications in industrialized countries.
Grid-connected, building-integrated systems
In the short term, PV could be competitive in building-integrated grid connected systems. In certain cases (for instance, in sunny places and when expensive building cladding materials are replaced by PV modules), such systems have entered the market already. This is the fastest-growing market sector for PV in Europe. The IEA estimates that further substantial system cost reductions down to US$4/Wp will be necessary for PV building integrated systems to be fully competitive and become widespread. However, the IEA model does concentrate only on electricity production. It does not take into account the possibility of recovering useful heat from PV panels, or the possibility of PV-energy-saving coupling measures (e.g. using PV panels as sun-shading systems, PV integration into bioclimatic architecture design, or using PV as a means of promoting demand-side management). As a consequence, PV competitiveness as a wider "energy-service supply technology" in the building sector could actually be greater than it appears at first glance.
This has significant general implications. In the first place, the share in energy final use of the building sector is quite high. In Europe, it accounts for almost 40 per cent of total energy consumption. Second, direct solar energy conversion technologies (both active and passive) are the only renewable energy technologies that can be used in urban areas. This very simple consideration is relevant for both industrialized and developing countries. For the latter, electrification as a means of increasing the quality of life in rural areas remains a top-priority issue, but urban pollution in developing countries' megalopolises raises strong environmental concerns as well. Moreover, integration of PV into buildings greatly increases the potential applicability of PV in areas of high population density and substantially reduces the main environmental obstacle to the diffusion of PV, namely land requirements. The theoretical potential of PV on rooftops is impressive indeed. Although there has been no systematic detailed assessment of the potential as yet, a rough estimate of 2,800 km2 and 8,400 km2 of "available" rooftops has been reported for European and OECD countries, respectively (van Brummelen and Alsema 1994). Assuming a (future) module efficiency of 20 per cent, this would correspond to an installed capacity of 560 GWp and 1,680 GWp, respectively. Consequent annual electricity production would be, at an overall 15 per cent system efficiency, around 475 TWh and 1,950 TWh, corresponding to around 27 per cent and 29 per cent of Europe's and OECD countries' 1990 electricity production, respectively (CEC 1993).
A very detailed study on rooftop PV potential in Puglia, a southern Italian region, has shown that PV on roofs could cover 27 per cent of the current total electricity demand of the region, corresponding to 94 per cent of demand in the residential sector (Vigotti 1994b).
Local utility grid support and peak power supply
At a system cost of US$3/Wp, PV would be a cost-effective option for electricity utilities for both local utility distribution grid support and peak power supply. At this cost, the more than 80 electricity utilities that have formed the Utility Photovoltaic Group (UPVG) estimate a short-term 7,500 MWp PV potential in this market sector in the United States (Moore 1994). The IEA estimates that PV will be competitive in this sector by 2005.
Bulk power supply
Once the most promising market sectors have been at least partially exploited, scale economies would further reduce module and system costs down to the target US$1/Wp threshold, making PV fully competitive in the bulk power sector as well. This is not likely to occur before 2010, but is very likely to come about soon after that.
The IEA diffusion strategy is only one of the possible paths - albeit a plausible one - along which PV could become a significant option in the energy market. The previously mentioned strategy of Enron Corporation for cheap baseload electricity production is an example of another possible parallel path. However, the IEA model clearly emphasizes the diversification of the actual energy system. Carefully taking into account the local realities of actual energy system infrastructures, it certainly opens more opportunities for PV diffusion.
Contents - Previous - Next