Performance and Cost Discussion

As indicated in Tables 1 and 2, the physical size of an individual residential PV system is assumed to remain fixed a t 20 m2, fitting within the unobstructed space available on the south-facing slope of a typical residential rooftop. D C unit ratings increase from 2.8 kW in 1997 to 3.2 kW in 2000 to 4.0 kW in 2030. The rated dc module efficiency an d rated dc power are for standard reporting conditions (1 kW/m2, 25^C/77°F). The rated ac power is the product of tie dc module rating and the inverter efficiency. The system operating efficiency is the product of the module efficiency , the inverter efficiency, and an additional factor of 0.9 to account for operation away from standard rating condition s [16].

The PV output at any given time is directly proportional to the available solar energy (insolation). The cost o f producing PV solar energy is therefore inversely proportional to the solar insolation. The solar insolation depends upon latitude, local climate, and PV module mounting. PV module mounting refers to positioning of the PV module wit h respect to the position of the sun - a tracking PV array collects the maximum available sunlight by pointing the arra y at the sun as the sun changes position in the sky, while, with a fixed array, the solar intensity changes continuousl y during the day. Residential systems generally use fixed arrays. Insolation varies between 1.6 and 2.4 MWh/m 2 -yr for a south-facing, fixed array. This report considers both average-insolation (1.8 MWh/m 2-yr) and high-insolation (2.3 MWh/m2-yr) locations. The high insolation location is of particular interest for early cost-effective applications . The annual energy production is the product of the system efficiency and the solar insolation. The ac capacity facto r is defined as the annual energy production divided by the product of the rated ac power and the number of hours in a year (8,760).

For Table 1, the PV module, power-related BOS, and area-related BOS costs for the base year were based on the first few large utility-sponsored residential PV system projects (SMUD's PV Pioneers), where houses were widely dispersed. These costs were compared to costs independently estimated using standard construction-industry project estimatio n procedures [17]. The independent estimate considered both low-voltage and high-voltage dc systems, and considere d ac PV modules (PV modules with integrated inverters). At present, low-voltage inverters cost less per rated capacity than high voltage inverters since similar inverters are already manufactured commercially at low volumes for othe r applications (uninterruptible power supplies). However, low-voltage systems have higher area-related BOS costs due to increased wiring requirements. The ac PV modules have the lowest area-related BOS cost since there is no longe r a separate dc system, but the inverters for ac PV modules presently have a higher cost. A large manufacturing volume and some technology improvements (e.g., integrated circuits for power supplies) will be required to reduce the cost o f inverters for ac PV modules. Despite these differences, the net result is that the three types of systems had similar total BOS costs. The independent estimate yielded costs similar to the large utility-sponsored project. Most of the systems installed to date use a low-voltage system, which was considered in this report. It should also be noted that the power-

related BOS costs include the utility costs for the interconnection, such as replacing a home's meter and adding the disconnect switches to allow for net metering.

For Table 2, a compact neighborhood of houses with rooftop PV systems is assumed. Beginning in 1985, NEE S installed 60 kW of PV on existing residential rooftops in Gardner, MA, plus 40 kW in commercial applications in three nearby states [10]. NEES did not sell the PV systems when it divested its generating assets [18]. A larger series o f projects was undertaken by SMUD with their "Residential PV Pioneer" projects, which ranged from 87 kW on 2 5 homes to 400 kW on 119 homes [11]. In Table 2, for 1997, plant size is assumed to be 0.299 MW based on placin g 2.3 kWac systems on 130 homes. For 2000 and later, plant size is estimated at 1.0 MW, assuming systems installe d on 385 houses in 2000 to 294 houses in 2030. Experience will lead to an optimal number of homes in the grouping . The compact neighborhood and bulk purchases translate into lower PV module, BOS, and O&M costs relative t o similar values in Table 1.

Estimation of costs for highly evolving products like photovoltaic modules and systems over several decades is a very difficult task. One method is to extrapolate from historical data. A useful tool for performing extrapolations of th e costs of manufactured products from historical data is the learning curve [19-21]. This method is derived fro m examination of cost data for many different industries, which has found that the cost of the product in constant dollars is a geometric function of the product's cumulative volume. The price reduction expected for a doubling of volume is known as the learning curve factor. The learning curve may be combined with an annual projected growth rate t o estimate the annual reduction in product cost.

Data for the price of PV modules, as a function of cumulative volume, has been analyzed by several groups, and the y reported learning curve factors between 0.68 and 0.82 [19-21]. The more conservative learning curve factor of 0.8 2 was used in this study because analyses of many other industries have found similar values [21]. This value means that a doubling of the cumulative volume of PV modules sales will reduce the cost of PV modules to 82% of its previou s value. The annual growth rate in PV module sales has been between 15-20% in recent years [22,23]. Given the strong demand for PV modules and the broad interest in accelerating adoption of PV energy (e.g., Million Solar Roofs Initiative), an annual growth rate of 20% can be conservatively assumed. A learning curve factor of 82% and assumed growth rate of 20% yield an estimated price reduction of 5% per year. An annual growth rate of 20% and annual cost reduction of 5% is used to generate the projections for the years 2000-2030 (Table 3). The price of $3.15 in 1997 is based on the estimated module price of one of the lowest recent bid system prices ($5.76/Wp for SMUD PV Pioneer residential PV systems). The average wholesale price of crystalline-silicon PV modules has stayed around $4.00/W p in recent years because of increased demand and constrained capacity. Table 3 illustrates the potential of th e technology, given a more mature market.

Table 3. Projections of crystalline-silicon photovoltaic module sales and prices.

Module Effic.

Annual Sales

Price

Sales

Module

Year

(%)

(MW)

($/Wp)

($M)

($/m2)

1997

14

84

3.15

265

441

2000

16

174

2.55

444

408

2005

17

433

1.97

853

335

2010

18

1,078

1.51

1,628

272

2020

19

6,678

0.90

6,010

171

2030

20

41,347

0.53

21,914

105

The prices in Tables 1, 2, and 3 are all in constant 1997 dollars, excluding inflation. Therefore, if the average inflation rate also happened to equal our average annual cost reduction of 5%, the price of PV modules in 2030 would be $3.15

in current-year dollars. Also note that Price does not refer to the manufacturing cost and as such reflects overhead factors as marketing, distribution, and research and development.

The validity of using the learning curve to extrapolate PV module costs to the low values after year 2010 should b e assessed because the nature of the industry might change at the larger sales volumes or other more fundamental (i.e. , physical) limits might arise. A second type of cost extrapolation was used to check the validity of the preceding table. This cost estimate used a "bottom up" analysis of the industry; i.e., the manufacturing cost is estimated at different production volumes for a specific proposed factory and manufacturing process. A detailed study was recentl y completed by a European research group [24]. The study estimated the manufacturing cost of crystalline-silicon an d of thin-film PV m odules at a production level of 500 MW per year. The European study estimated a manufacturing cost of $1.30/W for both the crystalline-silicon and thin-film PV at a production level of 500 MW per year. Th e manufacturing cost of $1.30/W compares well with our learning curve-based, extrapolated price of $1.92/W at a production level of 433 MW per year. This comparison gives confidence in using the learning curve to extrapolate PV module costs.

There is less data available for BOS components to estimate learning curve factors. Substantial cost reductions are still possible in the small inverters used for residential systems through design changes (reduce high-cost ferromagneti c materials with silicon d evices), technology improvements (e.g., integrated circuits for power supplies), and high-volume manufacturing [25]. Improvement in system design and standardization of components will reduce area-related BO S (i.e., installation and wiring) costs, and a substantial impact would be expected with the successful development of an ac PV module. Some observers suggest that there is little learning improvement available in BOS due to the maturity of the industry; for example, the costs of installation and wiring are well known from the much larger constructio n industry [26]. Nevertheless, a recent project achieved a 50% reduction in BOS costs for ground-mounted PV systems through improvements in integration of the system components [27]. As was the case for modules, a learning curv e factor of 0.82 and a growth rate of 20% were used, and these correspond to an estimated cost reduction per annum of 5%, for both power- and area-related BOS. The uncertainties in BOS costs in later years are larger because of th e difficulty in projecting the performance of a maturing industry with multiple technology options.

As pointed out earlier, PV systems have very low operation and maintenance costs. A recent study examined th e performance of a residential PV energy system after ten (10) years of operation [28]. This study found that the system, with the exception of some of the power conditioner components, was highly reliable and had minimal O&M costs . The report found an average annual O&M cost of only $52. The O&M cost represents a maintenance contract in Table 1 when the system is owned by the homeowner. In Table 2, it represents the cost of system monitoring an d maintenance if the system is owned by the utility or a third party. The components and system are anticipated to have 20-year warranties, so no cost for component replacement was included.

Solar Stirling Engine Basics Explained

Solar Stirling Engine Basics Explained

The solar Stirling engine is progressively becoming a viable alternative to solar panels for its higher efficiency. Stirling engines might be the best way to harvest the power provided by the sun. This is an easy-to-understand explanation of how Stirling engines work, the different types, and why they are more efficient than steam engines.

Get My Free Ebook


Post a comment