Photovoltaic (PV) systems are generally divided into two major categories: grid‐connected (also known as grid‐tied) systems that are interfaced to an electricity grid and stand‐alone systems that are self‐contained. Over the years it has been customary for books on PV to describe stand‐alone systems first, probably because they are seen as “pure PV.” Also we should remember that stand‐alone systems, including those launched into space and the solar home systems (SHSs) that supply electricity to individual families in developing countries, accounted for much of the PV industry in its early days. But since the 1990s the market has shifted decisively toward PV power plants and installations on buildings connected to an electricity grid. In 2000 grid‐connected PV had overtaken stand‐alone systems in global market share, and in 2016 more than 98% of solar cell production was being deployed in grid‐connected systems. In many ways such systems are simpler to design and describe than their stand‐alone cousins. For both these reasons our own story begins with grid‐connected PV.
Since most people have seen PV arrays mounted on the roofs of homes, this seems a good place to start. Figure 4.1 shows the elements of a domestic PV installation, typically with an array power between 3 and 8 kWp, interfaced to the local electricity grid. The major advantage of this arrangement is that the output from the PV array is fed into the grid when not required in the home; conversely, when the home needs power that cannot be provided by the PV (especially at night!), it is imported from the grid. In other words the PV system and grid act in harmony and there is an automatic seamless back‐and‐forth flow of electricity according to sunlight conditions and the electricity demand.
Figure 4.1 Home PV system connected to the grid.
In more detail the various items numbered 1–8 in the figure have the following functions:
The adoption of domestic rooftop installations is mushrooming in developed countries in response to the falling prices of PV modules, the support of governments, and the enthusiasm of citizens to do something positive about global warming. Larger grid‐connected systems, for example, those installed in schools, offices, public buildings, and factories, extend the power scale from hundreds of kilowatts up to megawatts (MW). All have the advantage of generating solar electricity where it is needed, reducing the losses associated with lengthy transmission lines and cables. And at the top of the grid‐connected power scale come multi‐megawatt power plants, generally remote from individual consumers, which send all their power to the grid.
The inverter is the key item of equipment for converting DC electricity produced by a PV array into AC suitable for feeding into a power grid. Inverters use advanced electronics to produce AC power at the right frequency and voltage to match the grid supply. While a single inverter may well be sufficient for a domestic installation such as that illustrated in Figure 4.1, multiple units become the norm as we advance up the power scale and their efficiency, reliability, and safety are major concerns of the system designer.
Inverters must obviously be able to handle the power output of a PV array over a wide range of sunlight conditions. Normally they do this using maximum power point tracking (MPPT) to optimize the energy yield. DC‐to‐AC conversion efficiencies up to 98% can be achieved over much of the range, although efficiency tends to fall off if an inverter is operated below about 25% of its maximum power rating. In larger systems with multiple inverters, it can make sense to switch all the power into one unit at sunrise and then, as the sun rises in the sky and the array power increases, bring other inverters successively into play, keeping all working optimally. The switching sequence is reversed toward sunset. Overall, inverter system design is quite a challenge, especially with high‐capacity units; few electronic systems are expected to maintain high efficiency over such a wide power range.
Figure 4.2 Raising the power level: a 17.6 kWp grid‐connected roof installation on the Oslo Innovation Centre, Norway.
(Source: Reproduced with permission of IEA‐PVPS)
From the technical point of view, there are two main classes of inverter: self‐commutated, where the inverter’s intrinsic electronics lock its output to the grid, and line‐commutated, where the grid signal is sensed and used to achieve synchronization. Inverters are also classified according to their mode of use, with four main types:
Figure 4.3 This Korean power plant uses four 250 kWp inverters to connect 1 MWp of PV arrays to the grid. The modules are mounted on horizontal single‐axis trackers.
(Source: Reproduced with permission of IEA‐PVPS)
Several factors influencing the choice of inverters for small‐ and medium‐size systems can be explained by referring to Figure 4.4. For simplicity we have shown arrays with just a few modules although most systems contain more—and some a great many more. The array in part (a) consists of two strings of three modules each. In this case all the modules are assumed to be of the same type and rating, with the same orientation and without shading, so the strings are paralleled in the combiner/protection units (1/2) and fed to a single central inverter (3). The inverter is presented with an input voltage equal to three times the individual module voltage and an input current equal to twice the individual module currents. Since the modules are well matched, the MPP selected by the inverter for the whole array ensures that all modules work at, or close to, their maximum output.
Figure 4.4 PV arrays served by: (a) a single central inverter; (b) two individual string inverters.
In part (b) of the figure, the two strings are dissimilar. They may have different numbers of modules (as shown), or different module types or orientations, or one string may suffer partial shading. For whatever reason they do not produce similar outputs and cannot be efficiently characterized by a single MPP, so each string has its own inverter and is operated at its own MPP. An alternative is to use a single multi‐string inverter. And as we have already pointed out in the previous chapter, manufacturers are now offering power optimizers, one to be connected to each module in a string, allowing every module to work at its own MPP. There are various options for extracting the maximum amount of power from strings and arrays.
To put our discussion in the context of a practical system, suppose we need to specify an inverter for a PV array of about 5 kWp on the roof of a suburban house. In a sunny climate an array of this size may well generate, over a complete year, electricity equal to the annual requirements of the household. In the summer months the PV will be a net exporter to the grid; in the winter months the solar deficit will be made up from the grid. We will assume that monocrystalline silicon modules rated at 180 Wp have been selected, so 28 will be needed, yielding 5.04 kWp (the module specification given in Section 3.2.2 will be used in the calculations later). Fortunately, they can all be mounted on the roof at the same tilt angle, and there is no shading, so we may specify a central inverter. Since an array rarely generates its nominal peak power, an inverter rated at slightly less than 5.04 kWp should be adequate as long as its maximum input voltage and current are never exceeded. We will therefore investigate the suitability of a 5 kWp central inverter with the following manufacturer’s ratings:
Nominal DC input power | 5.0 kW |
Peak instantaneous input power | 6.0 kW |
Maximum DC input voltage | 750 V |
Voltage range for MPPT | 250–650 V |
Maximum DC current | 20 A |
We first need to estimate how many modules can be connected in a series string. The maximum number is given by the MPPT voltage of 650 V divided by the MPP voltage of an individual module. The latter is 35.8 V at 25°C, but increases by 0.33% for every degree drop in temperature. Therefore if we allow for sunny winter days with temperatures down to −5°C, the MPP voltage could reach 10% above 35.8 V, that is, 39.4 V. The maximum number of modules in a string for effective tracking is therefore 650/39.4 = 16.5, say, 16.
We should also check that the maximum DC input voltage of 750 V is never exceeded. Once again, the danger condition is a cold winter day with bright sunshine. The module open‐circuit voltage of 43.8 V at 25°C rises by 10% to 48.2 V at −5°C. So to keep within the 750 V limit, the maximum number of modules is 750/48.2 = 15.6, say, 15.
The minimum number of modules in a string is dictated by the need to keep the MPPT voltage above 250 V. The module’s MPP voltage falls with rising module temperature, which could reach 70°C and cause a 15% drop in MPP voltage to 30.4 V. The minimum number of modules is therefore 250/30.4 = 8.22, say, 9.
To keep within the inverter’s voltage limits, we conclude that strings may have any number of modules between 9 and 15. Since the array contains 28 modules, two strings of 14 are acceptable, but not four strings of 7. Finally the array current supplied to the inverter should be checked to make sure it does not exceed the permitted maximum. In this case the peak short‐circuit module current is 5.5 A and is little affected by temperature. So two parallel strings of 14 modules will give a peak DC current of 11 A, well below the permitted maximum of 20 A. The inverter is therefore suitable for the job.
Back in Section 3.3.1 we discussed the problem of shading and suggested reducing the effects of recurrent shadows by confining them to as few strings as possible. Where shading is unavoidable, it may be appropriate to use a number of string inverters rather than a single central inverter, giving flexibility to connect the modules in a favorable configuration, perhaps with strings of different lengths.
We have already mentioned the need for inverters to operate efficiently over a wide power range. Some inverters include transformers, and these reduce efficiency slightly. High efficiency is not purely a question of economics; it also relates to keeping inverters cool. For example, if a 5 kWp inverter is working at full stretch and converting 96% of its input power to AC, the other 4% (200 W) must be dissipated as heat. It is hardly surprising if the manufacturer recommends mounting the unit on an outside, north‐facing wall with plenty of air circulation! The cooling issue becomes more and more significant as inverter power‐handling capacity increases.
If the electricity grid is turned off for maintenance purposes, or due to a fault, it is very important for an inverter to disconnect itself automatically to avoid putting a voltage on the grid. Otherwise it can endanger personnel working on the grid and may deceive other local inverters into believing that the grid is still operating normally. Sophisticated electronics are included to prevent this potentially hazardous situation, which is referred to as islanding.1,2
As we gaze at a domestic rooftop system rated at a few kWp, it is hard to appreciate the engineering challenges posed by scaling up inverters for multi‐megawatt power plants. There are major issues of technical performance to be considered including lightning and surge protection, safety, reliability, inverter sequencing, and the mode of connection of tens or hundreds of thousands of PV modules into strings and arrays.1 The waveform purity and power factor of the inverter output must be satisfactory to the grid operator. Grid‐connected inverters are sensitive to fluctuations in grid voltage, frequency, and impedance and will shut down automatically if these parameters stray outside the agreed specification. Islanding, which could be disastrous in a large installation, must be avoided. High‐power inverters provide a major challenge to today’s electrical and electronic engineers, but these challenges have been largely resolved as discussed in Section 4.8.2.
Figure 4.5 Scaling up: this 1.6 MWp inverter weighs over 20 tons.
(Source: Reproduced with permission of Padcon GmbH)
Figure 4.6 The Moura power plant in Portugal, rated at 45.6 MWp.
(Source: Reproduced with permission of IEA‐PVPS)
Various items are required to complete a grid‐connected PV system. They may be less eye catching than solar cells and PV modules, but they are essential to a properly engineered installation. Costs, long‐term reliability, ease of maintenance, and sometimes appearance are important considerations. They are generally referred to as balance‐of‐system (BOS) components.
As the prices of solar cells and modules continue to fall and PV manufacturers achieve the cherished long‐term objective of “less than one US dollar per watt,” the cost of BOS components can, unless carefully controlled, seriously inflate total system costs. In the past a figure of about 50% has often been quoted for BOS, including inverters. One of the main problems has been a proliferation of components supplied by many manufacturers in small quantities, lacking the economic benefits of scale. As the PV industry continues to grow, there is perhaps a better chance that volume production will drive costs down.
We mentioned and illustrated various BOS components for a domestic PV installation in Section 4.1. It is now time to give a more complete list and add further comments:
Figure 4.7 Array mountings at the Kings Canyon power plant in Australia.
(Source: Reproduced with permission of IEA‐PVPS)
Figure 4.8 Key functions of the combiner/protection units in a domestic PV system.
As we move up the power scale toward larger grid‐connected systems, the importance of accurate performance monitoring grows. Large PV power plants have a full range of instrumentation typical of modern high‐tech industrial facilities. And their full complement of BOS subsystems and components account for a substantial part of overall costs.
We have seen how a grid‐connected system is built up using PV modules, inverters, and BOS components. In previous chapters we included several photographs showing PV roofs and vertical facades. So what exactly is implied by the term building‐integrated photovoltaics (BIPV), and what more is there to be said about giving buildings a “face to the sun?”
PV technology is unique among the renewable energies in its interaction with the built environment. Future generations will find it entirely natural to see PV arrays on roofs and facades, in gardens and parks, on bus shelters and car ports, and as electricity‐generating windows and screens inside homes, schools, offices, and public buildings. Most will be grid‐connected. And hopefully they will bear testament to the trouble our generation has taken to blend them visually and aesthetically into their surroundings. PV will become increasingly a part of the urban experience.
By contrast, most wind power is generated in wild open country or offshore, and whatever one thinks of the visual impact of large turbines, they rarely impinge on the urban and suburban scene. Wave and tidal power do not affect the daily visual experience of office workers or families—unless they go on trips to marvel at large‐scale renewable energy in action. Large PV power plants may be impressive and even beautiful in their own way, but they are not generally noticed by city dwellers.
BIPV is different. It proclaims a message about our care for the environment, it can be anywhere and everywhere, and it matters what it looks like and how people feel about it. Public enthusiasm and support are vital, not least to the PV industry. Architects will, or should, be involved with engineers in the design of solar buildings so that PV is integrated into the fabric in ways that marry technical function with aesthetics. A modern factory producing solar cells, or an exhibition center for renewable technologies, offers an ideal opportunity to create a striking building that makes a highly visible statement about our technological future; a family living in a low‐energy timber‐framed eco‐home may see their PV modules as a symbol of sustainability, an alternative lifestyle. In their very different ways, all wish to proclaim a message about renewable energy that can only be successfully communicated by high‐quality BIPV.
Of course there are difficulties. Countries including England, France, Italy, and Spain have a huge stock of old and historic buildings. It would be difficult or impossible to modify most of them to accept PV modules in aesthetically pleasing ways. PV arrays tacked on to existing roofs hardly ever increase their visual attraction. We may like to see them because of what they represent—the owner’s commitment to renewable electricity—but our enthusiasm stems from what the PV does, not how it looks.
The main opportunity for successful BIPV, as opposed to PV that is simply superimposed on existing roofs and structures, lies in the creation of new buildings that, from the very start, treat PV as an integral part of the design, full of exciting possibilities. It is very encouraging to see architects in some countries—Germany and the Netherlands are good examples—realizing that PV offers far more than a way of producing electricity. While appreciating its technical possibilities and limitations, their primary goal as architects is to ensure that PV enhances the human environment. For them PV is neither add‐on nor afterthought, but an important part of the building and a pointer to its function and personality. It must inspire as well as serve its utilitarian purpose. As old housing and building stock is gradually replaced, we may expect PV to exert a growing influence on architectural design, opening up hitherto undreamed‐of possibilities.
Figure 4.9 Proclaiming a message: the Solar Showcase in Birmingham, England.
(Source: Reproduced with permission of EPIA/BP Solar)
Apart from aesthetics BIPV has several important economic advantages:
Figure 4.10 Proclaiming a message: an eco‐home in Denmark.
(Source: Reproduced with permission of IEA‐PVPS)
So how well do photographs included in earlier chapters (Figures 2.2, 2.25, and 3.18) square up to the expectations of successful BIPV? You may like to refer back and make your own judgments. From the purely technical perspective, it is clear that all these PV installations are integrated on to and into the buildings. But from the architectural point of view, overall appearance is key and a PV array should be a harmonious part of the overall design. If these examples underline the difficulty of defining and agreeing architectural aesthetics, this certainly does not absolve us from trying!
The aesthetics of successful BIPV may be hard to define and judgments are inevitably subjective—yet most of us know instinctively when a solar building feels right for its setting and context. In this section we consider a number of examples to illustrate the wide range of recent international BIPV. Since a picture “is worth a thousand words” in the field of visual impressions, our focus is on photographs accompanied by short explanatory captions.
All PV installations are “outside” in the sense that they must receive sunlight. Building façades and sloping roofs are often highly visible to the public; flat roofs are more likely to be hidden. Any PV array on public display should appeal to passersby and bystanders as well as users and owners of a building. Its environmental statement is offered to the world at large.
Although many PV installations are visible only from the outside, some are also “inside” in the sense that people within the buildings are highly aware of them, and if well designed they can both inspire and delight. Modules may be interspersed with glass windows or arranged as louvers to provide internal shade and ventilation. Some crystalline silicon modules have glass at front and back, allowing light to enter through the gaps between wafers. Thin‐film modules can be semitransparent, producing partial shade and generating electricity at the same time. Modules on rooftops that are invisible from the outside may be highly visible on the inside—indeed, this is usually the architect’s intention. The advent of tinted and flexible thin‐film products means that architects can be increasingly bold and imaginative about incorporating PV into their designs.
It is clear that aesthetic judgments should depend to a considerable extent on whether PV is on the “outside” or “inside.” Outside, it interacts with the neighboring buildings and the local landscape and affects a great many people, some of whom are probably skeptics. Inside, it is more self‐contained and speaks only to the users of the building who, in most cases, are enthusiastic supporters of renewable energy. It may be helpful to bear these points in mind when assessing the following photographs. They are arranged in two groups labeled PV outside and PV inside. The selection is designed to show a good international range of solar buildings with different personalities, acknowledging the efforts that many architects are making to enhance the built environment by incorporating PV imaginatively into their designs.
The PV on these buildings and installations is highly visible from the outside.
Figure 4.11 This building in Tübingen, Germany, proudly proclaims its solar identity.
(Source: Reproduced with permission of EPIA/BP Solar)
Figure 4.12 Traditional stone and PV in harmony: a building at the Technical University of Catalonia, Spain.
(Source: Reproduced with permission of EPIA/BP Solar)
Figure 4.13 Architects in countries with a tradition of social housing can spread their influence widely. This example is in Amersfoort, the Netherlands.
(Source: Reproduced with permission of IEA‐PVPS)
Figure 4.14 A huge solar pergola at the World Forum of Culture in Barcelona, Spain, supports a 4000 m2 PV array.
(Source: Reproduced with permission of EPIA/Isofoton)
Figure 4.15 The Sydney Olympic Games brought PV to the attention of millions with solar‐powered lighting and more than 600 1 kWp arrays on athletes’ houses.
(Source: Reproduced with permission of EPIA/BP Solar)
Figure 4.16 This eco‐home in Oxford, England, uses PV modules, water‐heating panels, and passive solar design to reduce its external energy requirements almost to zero.
(Source: Reproduced with permission of EPIA/BP Solar)
Figure 4.17 PV louvers replace standard glass shading to provide a dual function.
(Source: Reproduced with permission of EPIA/BP Solar)
Figure 4.18 A PV‐covered walkway at an exhibition center in Japan.
(Source: Reproduced with permission of IEA‐PVPS)
Figure 4.19 A 1.6 km PV array gives added purpose to a highway sound barrier in Germany.
(Source: Reproduced with permission of EPIA/Isofoton)
The PV systems on these buildings add a great aesthetic value in addition to energy savings.
Figure 4.20 Sunlight and shadow: a striking interior at the Energy Research Centre of the Netherlands.
(Source: Reproduced with permission of EPIA/ECN)
Figure 4.21 In harmony with nature: 30 kWp of glass/glass modules at the National Marine Aquarium, Plymouth, England.
(Source: Reproduced with permission of IEA‐PVPS)
Figure 4.22 Thin‐film semitransparent modules allow dappled light into this building in Germany.
(Source: Reproduced with permission of EPIA/Schott Solar)
Figure 4.23 Customer satisfaction: a shop in Tours, France.
(Source: Reproduced with permission of EPIA/Total Energie)
These photos show that PV has aesthetic value in addition to its apparent environmental and economic value and the great service that provides to the people in developing countries that lack access to reliable electricity. However, cost reductions in grid‐integrated systems, mostly in developed countries, have been the major driver for a phenomenal growth for PV.3 At this point it makes sense to examine the growth of the markets before we discuss large‐scale grid‐integrated PV.
Over the last 10 years, the market for photovoltaics as measured by their cumulative installed capacity has been growing by about 45% per year. Between 2005 and 2015, global solar PV capacity increased from approximately 5 to 220 GW, and it was estimated that it would exceed 320 GW at the end of 2016 (Figure 4.24). This strong growth can be attributed to cost reductions, induced by innovations in a market environment that was created by national and state subsidy programs. Early on most of the market growth happened in Germany catalyzed by a renewable portfolio standard (RPS) that provided feed‐in tariffs (FITs) to solar and wind installations in the country.
Figure 4.24 Global growth of PV deployment.
(Source: A. Jäger‐Waldau; PV Status Report 2016; EUR 28159 EN)
The scale and profitability of the German market enabled efficient scaling up of PV manufacturing worldwide. Italy started a strong incentives program in 2010 and China and the United States introduced FITs a year later. As of 2016, China has both the largest annual deployment and the largest PV manufacturing capacity in the world (Figures 4.25 and 4.28 correspondingly).
Figure 4.25 Annual growth of PV deployment.
(Source: A. Jäger‐Waldau; PV Status Report 2016; EUR 28159 EN)
The decrease in rooftop installed costs can almost entirely be attributed to the drop in module prices, which fell from an average of $5/W in 1998 to approximately $0.6/W in 2014, whereas utility‐scale system cost reductions were also enabled by increased efficiencies in installation. Further cost reductions are targeted on marketing, administrative, and permitting costs (called “soft costs”), which make up approximately 50% of residential and commercial installations in the United States but only 10% in Germany. Investment in solar PV installations was encouraged recently by substantial fall in the costs of solar PV that resulted largely from its widespread deployment and lately by substantial overcapacity/oversupply (Figure 4.26).
Figure 4.26 Top 14 countries in terms of cumulative capacity as of 2014; shown in order of increasing annual capacity from top to bottom.
(Source: Data from Report IEA PVPS T1‐26:2015)
As shown in Figure 4.30 the initially booming markets in Greece, Belgium, and Spain almost disappeared, and the market in Germany, which has been the greatest consumer of PV during 2003–2012, has slowed down a lot; currently the strongest markets are in China, Japan, and the United States. On the positive side, Italy, Greece, and Germany have now enough installed capacity to produce 7–8% of their annual electricity demand with PV.
So far, the increase in solar PV installations has been supported by FIT and purchase power agreements (PPA) that reduce a project’s risk as long‐term returns are guaranteed, typically for 10–20 years. Such incentives and financing mechanisms are discussed in Chapter 7. Beyond Europe, the largest PV markets are in China, the United States, Japan, and India. The market in China rose to the top in 2013 and remained there, largely in response to the introduction of a national FIT. In 2015, PV installed capacity in China exceeded 38 GW, among which 32 GW is PV power plants and 6.2 GW is distributed PV systems.
In the United States, falling prices combined with state incentives and the extension of federal investment tax credits (ITC) doubled the market, bringing the total operating capacity to more than 40 GW at the end of 2016. The highest growth in the US market has been on utility‐scale systems as those are the least costly and can compete with conventional power costing (Figure 4.27). California remains the nation’s largest market, followed by New Jersey and Arizona. One would expect this high growth for the sunny south California and Arizona; in New Jersey the growth was enabled with efficient state policies. The US PV installed capacity increased from 2 to 40 MW in just 6 years (2010–2016). This impressive growth was catalyzed by the extension of the federal ITC and a rapid decrease of the utility‐scale installed cost, as shown in Figure 4.31. An increase by another order of magnitude by 2030 is feasible provided that large‐scale storage at costs of approximately 2–4 cents/kWh will be available.3,4
Figure 4.27 Trends in deployment increase and price decrease in the United States.
(Source: Adapted from Solar Energy Industries Association)
Japan continues to rank third globally for total operating capacity. On the other hand, most of the growth in Japan has been in residential systems reflecting the difference in land and solar resource availability between the two countries. India has been another recent solar power success story; the installed capacity grew to 5 GW within 5 years and is projected to grow to 50 GW or more by 2024. The Indian case is especially interesting for the large populations in developing countries worldwide who do not have access to reliable electricity. The PV market in India started with small solar systems replacing diesel generators in remote areas, and when the reliability of PV was established, it grew to large‐scale PV deployment across the country. Chile is another fast‐growing market, having installed about 0.4 GW in 2014; the north of Chile has the richest solar resources of the world, and there is a vision prevailed among academics that solar electricity from the Atacama Desert there can serve, in addition to Chile, large loads in Rio de Janeiro, San Paolo, and Buenos Aires.
How can this growth be maintained and even accelerated? Let us look at the United States, which is a difficult market because of low electricity prices. If the United States averages 15 GW of new solar per year from 2016 to 2030, the 2030 target of the Grand Solar Plan will be materialized. Catalysts for further growth include:
However, increased penetration of solar makes it more difficult to compete with conventional generation, as it would have to become dispatchable to displace base power. The solutions for making PV dispatchable (thus available on demand) include (i) investing in high‐voltage direct current (HVDC) transmission to transfer solar from the SW to the rest of the country4; (ii) increasing the size of the grid balancing areas to harvest the benefit of regional geographical diversity in supply and demand; (iii) combining solar and wind resources; (iv) increasing the flexibility of the grid, with, for example, natural gas turbines displacing coal and nuclear power plants; (v) improving forecasting and thus certainty of PV output; (vi) adding energy storage; and (vii) implementing demand management.
While energy storage serves its purpose on the supply side of the equation, demand management can provide additional support on the demand side. As more devices in buildings become controllable, smart, and networked, customers and service providers will gain the ability to shift load away from traditional peak periods to periods of higher solar and/or wind production.
PV manufacturing capacity has grown rapidly in response to booming global demand, initially in Europe, Japan, and the United States; thereafter, leadership in production shifted to China, which expanded its manufacturing capacity massively to meet growing international solar PV demand (Figure 4.28). In recent years manufacturing capacity expanded much more quickly than demand for PV panels; fortunately for the Chinese manufacturers, a big domestic market was quickly developed there and it absorbed the overcapacity. Annual worldwide solar cell production reached 40 GW in 2014 with about 80% of it being multicrystalline and monocrystalline Si‐based PV. China and Taiwan accounted for approximately 75% of global cell production based entirely on Si. In the United States approximately 32% of the module production was thin film, mostly CdTe PV.
Figure 4.28 Annual PV global production.
(Source: A. Jäger‐Waldau; PV Status Report 2016; EUR 28159 EN)
2011 was a transition year for the PV industry; due mainly to an oversupply of modules from China, module prices fell more than 40% during the year. Installers of solar PV systems and electricity consumers greatly benefited from falling solar PV prices, but solar PV manufacturers around the world, and particularly those in the United States and Europe, experienced financial losses or shrinking profits with long‐lasting impacts. Cell, module, and polysilicon manufacturers struggled to make profits or even survive amid excess inventory and falling prices, declining government support, and slower market growth for much of these years. During this period, there was significant industry consolidation worldwide to lower costs and become more competitive; several large companies became bankrupt. Among the US and European manufacturing firms that survived, several shifted their production to Asia where incentives are given and labor is cheaper. Trade tensions have arisen between the United States, Europe, and China, resulting in the imposition of import tariffs by the United States and the EU on solar panels from China. Many solar PV manufacturing firms continued their vertical integration by expanding into project development to remain competitive. Large companies developed new business models and partnered with electric utilities, real estate developers, sports teams, and retailers.
We next discuss the large PV systems that have brought prices down to “grid parity” in several regions of the world. Until a few years ago, the idea of a PV plant generating megawatts seemed unlikely to most people, but in 2008 there were around 1000 plants worldwide rated at 1 MWp and above and in 2015 many plants in 100–500 MW range were added. The great driver of this revolution has been the generous financing of PV electricity in certain countries, most notably Germany, Spain, and the United States. Germany and the United States had seen steady increases in capacity for many years; then, in 2007–2008, a remarkable surge took place in Spain due to its government’s introduction of a highly attractive tariff of 0.44 euro cents/kWh. In 2008 alone Spain installed 2.7 GWp of PV, including some 700 MWp of power plants rated above 10 MWp. The fast evolution to greater plant sizes is shown in Figures 4.29 and 4.30. Spain’s achievement in a single year was remarkable. It must be added, however, that the Spanish government reduced the power plant tariff substantially toward the end of 2008, slanting the future more toward roofs and facades, and placed a cap of 500 MWp on annual PV installation for the following few years. Even though the immediate boom was over, Spain’s experience, although short‐lived, changed international perceptions of what is possible and provided a massive boost to the PV industry.
Figure 4.29 A 7.2 kW system in Marchal, Spain.
(Source: Reproduced with permission of First Solar)
Figure 4.30 The 23 MWp La solar farm in Magascona, Spain.
(Source: Reproduced with permission of IEA‐PVPS)
Other countries active in PV power plant installation are pushing global cumulative capacity into the multi‐gigawatt era. Germany and the United States are especially prominent, but Japan, Italy, Portugal, France, Greece, Korea, and as of 2015 Chile and India all deserve mention.5 About three‐quarters of plants have static arrays; the rest use single‐ or double‐axis tracking, the great majority without sunlight concentration.
In the rest of this chapter we will discuss the economic drivers and the system integration challenges of two growing segments of large PV deployment, namely, commercial and utility installations.
Installations on public and commercial roofs proliferate in Australia (Figures 4.31 and 4.32) and recently started taking off in the Unites States and other countries, catalyzed by Investment Tax Credits (ITC) and demand charge reductions in other countries. Utility tariffs for commercial and industrial customers are commonly divided into demand and consumption. Many utilities in the United States charge demand based on the peak 15–30 minutes of load each month, and these charges apply to the whole month. Thus substantial demand charges can be incurred due to load variability. Demand charge reduction can be achieved by storage alone, PV alone, and even more effectively with a combination of the two. Peak demand reductions also benefit the utility by reducing overloads in congested distribution grids. An example of such a system is the “JFK Solar Park” (Figure 4.33) located on roofs of commercial buildings near JFK International Airport, which is operated under New York State’s “remote net metering” program; the electricity produced at JFK Solar Park is consumed by Bloomberg LP at its headquarters in Manhattan.
Figure 4.31 The 1.2 MW PV rooftop system at the University of Queensland, Australia.
(Source: Reproduced with permission of Trina)
Figure 4.32 The 1.17 MW PV rooftop system at the Adelaide Airport, Australia.
(Source: Reproduced with permission of Trina)
Figure 4.33 Section of the 1.6 MW “JFK Solar Park” located on the roofs of three airfreight logistics buildings near JFK International Airport in New York.
(Source: Reproduced with permission of JFK Solar Enterprises LLC)
Over the last 5 years the deployment of large utility‐scale scale PV power plants has been impressive. They first made their appearance about 10 years ago in Germany and Spain, and since 2010 they have become the fastest‐growing segment of the PV market in China and the United States.
Figure 4.34 The 1.3 MW PV power plant at Dimbach, Germany.
(Source: Reproduced with permission of First Solar)
Large ground‐mount PV plants have grown in the United States more than residential applications due to drastic cost reductions and increased familiarity of utilities with the reliability of PV power plants, which, in turn, led to power purchase agreements (PPA) between the plant owner and the utility. China and the United States also have the largest PV plants (a 500 and 520 MW in China, two 550 MW in the United States). As of 2015, other emerging markets for large‐scale multi‐MW PV plants include Chile and South Africa.
Figure 4.35 Section of the 10 MWac PV power plant at Tibet Sangri, China.
(Source: Reproduced with permission of Trina)
Figure 4.36 The 300 MWac PV power plant at Yunnan Jianshui, China.
(Source: Reproduced with permission of Trina)
Figure 4.37 Section of the 290 MWac PV power plant, Agua Caliente, AZ, USA.
(Source: Reproduced with permission of First Solar)
The biggest advantage of large ground‐mount PV plants is their easy installation (shown in Figures 4.38, 4.39, and 4.40) and economies of scale that make the total costs much lower than those of rooftop installations. Also these plants can be controlled according to the utility requirements and help the stability of the grid. On the other hand, smaller distributed PV systems have the advantage of being at the point of demand so that transmission and distribution power losses are eliminated.
Figure 4.38 Start of PV plant construction; support and mounting structures.
(Source: Reproduced with permission of First Solar)
Figure 4.39 During PV plant construction, mounting the modules on fixed‐tilt system.
(Source: Reproduced with permission of First Solar)
Figure 4.40 Section of the 52.5 MWac one‐axis tracking PV plant during construction, Shams Ma’an, Jordan.
(Source: Reproduced with permission of First Solar)
At the time of writing the biggest PVPS in the world are the 550 MWac Desert Sunlight, 550 MWac Topaz, and 579 MWac Solar Star plants in south California. The first two were constructed by First Solar using their CdTe PV modules, and the third was constructed by SunPower using their monocrystalline silicon modules. Each of these projects was completed ahead of a 3‐year schedule and is operating under PPA with regional utilities (Pacific Gas and Electric (PG&E) and Southern California Edison). It is noted that two of these plants are now owned by a Warren Buffett company; the big capital is behind PV now!
Figure 4.41 52.5 MWac one‐axis trackers, almost completed, Shams Ma’an, Jordan.
(Source: Reproduced with permission of First Solar)
Figure 4.42 Section of 550 MWac Desert Sunlight, California, PV plant during construction stages.
(Source: Reproduced with permission of First Solar)
Figure 4.43 Section of the 550 MWac Desert Sunlight PV power plant.
(Source: Reproduced with permission of First Solar/NEXTera Energy)
Figure 4.44 Close‐up on another session of the 550 MWac Desert Sunlight PV plant.
(Source: Reproduced with permission of First Solar/NEXTera Energy)
These and other projects have taken the price of PV electricity under US Southwest irradiation conditions down to 6–7 cents/kWh, ahead of the projections made in 2008 by one of the authors4 and also ahead of the US Department of Energy (US‐DOE) projections made in 2010 (Figure 4.45), which materialized into the SunShot Vision Study.6 Figure 4.45 shows estimated and projected price reductions for utility‐scale PV expressed in levelized cost of electricity (LCOE). The LCOE averages all the costs during the operating life of a project; it is used for comparing the cost of energy technologies with different operating conditions, as technologies like solar have higher capital costs (and consequently higher financing costs), whereas fossil fuel‐based power technologies have higher operating (and fuel) costs. The LCOE is further discussed in Section 7.1.
Figure 4.45 Levelized cost of electricity (LCOE) for utility conventional grid and PV power.
(Source: Reproduced with permission of US‐DOE Solar Technologies Program)
The electricity grid is a complex network of power lines, designed to transport energy from suppliers to loads. In its normal configuration, power plants feed electricity into the high‐voltage (>100 kV) transmission grid, which transports the energy to demand centers where the voltage is stepped down to deliver power to individual customers on the distribution grid. Both the European and the US electricity grids are more than a century old. The US electricity grid is a conglomerate of many smaller grid systems that were built in the late 19th century, each regulated by separate utility companies. The grid eventually grew into three major “interconnects” on which all power plants operate synchronously. They are the Eastern Interconnection, Western Interconnection, and Electric Reliability Council of Texas (ERCOT); there are very few links between the three interconnects. Figure 4.46 shows these interconnects, the North American Electric Reliability Corporation (NERC) interconnect subregions, and about one hundred generator, load, and transmission balancing areas. Balancing authorities integrate power resources to meet demand within balancing areas; they are responsible for maintaining interconnection frequency and controlling the flow of power so that overloading of transmission lines is avoided. The US power transmission grid consists of 300 000 km of lines operated by 500 companies.
Figure 4.46 The US electric grid: three major interconnects, eight NERC subregions, and about 100 balancing authorities
(Source: http://www.eesi.org/briefings/view/021617wires. CC BY 4.0).
The largest synchronous (by connected power) electrical grid is that of continental Europe connecting over 400 million customers in 24 countries. Although synchronous, some countries operate in a near island mode, with low connectivity to other countries, and there are plans to increase connectivity and host more renewable energy while enhancing the reliability of the grid. To this effect, the European Electricity Grid Initiative (EEGI) has the following objectives: (i) to transmit and distribute up to 35% of electricity from dispersed and concentrated renewable sources by 2020 and a completely decarbonized electricity production by 2050 and (ii) to integrate national networks into a market‐based, truly pan‐European network to guarantee a high‐quality of electricity supply to all customers and to engage them as active participants in energy efficiency.
Now let us look at the basics of electricity grid operation to better understand the challenges and benefits of integrating renewable and distributed energy in the grid. Supply and demand of active power on the grid must always be in balance, or the grid frequency will deviate too far from its set point (60 Hz in the United States, 50 Hz in Europe), causing connected appliances to shut down or get damaged. The frequency of the system would vary as load and generation change. If there is more generation than demand, frequency goes up; if there is less generation than demand, frequency goes down. During a severe overload caused by tripping or failure of generators or transmission lines, the power system frequency will decline, due to an imbalance of load versus generation. Loss of an interconnection while exporting power will cause system frequency to rise. Also, temporary frequency changes are an unavoidable consequence of changing demand. Automatic generation control (AGC) is used to maintain scheduled frequency and interchange power flows, and the presence of many generators and a large distributed load allows for easy frequency management. Control systems in power plants detect changes in the network‐wide frequency and adjust mechanical power input to generators back to their target frequency. Modern PV power plants can also respond to frequency regulation by curtailing or increasing power instantaneously (see Section 4.8.2).
As PV generation grows to the point of making a significant contribution to the grid, the PV industry is developing large PV power plants that support grid stability and reliability. A modern utility‐scale PV power plant is a complex system of large PV arrays and multiple power electronic inverters, and it can contribute to mitigate impacts on grid stability and reliability through sophisticated “grid‐friendly” controls (Figure 4.47).7 Components of a typical multi‐MW, utility‐scale PV power plant are shown in Figure 4.48 including power conversion and electrical equipment, such as PV panels, inverters, switchgear, grid interconnection, power plant controller (PPC), supervisory control and data acquisition (SCADA), and communication systems.
Figure 4.47 Utility‐scale PV operations center in Mesa, Arizona, USA.
(Source: Reproduced with permission of First Solar)
Figure 4.48 PV plant grid integration and control system.
(Source: Reproduced with permission of First Solar)
“Grid‐friendly” PV plants help to stabilize the grid by incorporating voltage regulation, active power controls, ramp‐rate controls, fault‐ride through, and frequency response.7 These services are outlined in the Box below:
A plant‐level control system, which controls a large number of individual inverters to affect plant output at the grid connection point, is a key enabler of such features (Figure 4.48). The PPC monitors system‐level measurements and determines the desired operating conditions of various plant devices to meet the specified targets. It manages the inverters, ensuring that they are producing the real and reactive power necessary to meet the desired settings at the point of interconnection (POI). For example, when the plant operator sends an active power curtailment command, the controller calculates and distributes active power curtailment to individual inverters. In general, the inverters can be throttled back only to a certain specified level of active power without causing the DC voltage to rise beyond its operating range. Therefore, the plant controller dynamically stops and starts inverters as needed to manage the specified active power output limit.
The actively controlled plant also has the ability to minimize the impact of cloud cover. Accommodation of the reduction of power output due to partial plant shading is done by increasing the output of the inverters from the unaffected sections of the plant (Figure 4.49).
Figure 4.49 Impact of cloud passage in utility PVPS operation; the plant comprises of eight power blocs.
(Source: Reproduced with permission of First Solar)
These “grid‐friendly” capabilities, essential for increased penetration of large‐scale PV plants into the electric grid, are operational and available today for utility‐scale PV plants ranging from several megawatts to several hundred megawatts. These advanced plant features enable solar PV plants to behave more like conventional generators and actively contribute to grid reliability and stability, providing significant value to utilities and grid operators. They also use the active power management function to ensure that the plant output does not exceed the allowed ramp rates, to the extent possible. It cannot, however, always accommodate rapid reduction in irradiance due to cloud cover and storage may be needed in such cases. Figure 4.50 shows field data from a PV plant operating at around 90 MW power.
Figure 4.50 Power curtailment at different levels: this figure shows field data from a PV plant operating at around 90 MW power. The brown lines show the power set points and the blue shows the supplied power.
(Source: Reproduced with permission of First Solar)
The brown lines show the power set points and the blue shows the supplied power. The plant controller turns down the inverters (and turns off some of them if required) to achieve the new set point. Note that the turndown of power is gradual to meet the specified ramp‐rate limit.
A discussion of electricity markets is necessary to better understand the effects of variable generation on the grid. The National Renewable Energy Laboratory (NREL), AES, the Puerto Rico Electric Power Authority, First Solar, and the ERCOT have conducted a demonstration project on two utility‐scale PV plants to test the viability of providing important grid ancillary services from these facilities. This demonstration showed that active power controls can leverage PV’s value from being simply a variable energy resource to providing additional ancillary services that range from spinning reserves, load following, ramping, frequency response, variability smoothing, and frequency regulation to power quality. Specifically, the tests conducted included variability smoothing through AGC, frequency regulation for fast response and droop response, and power quality.8
Prices in energy markets are set by the dynamics of supply and demand under the geographical constraints imposed by the reach and limitations of the transmission grid. For this reason, large grid systems are split into smaller markets at major nodes or zones, each of them establishing a price for each hour of the day according to supply and demand. This is called nodal pricing, or locational‐based marginal pricing (LBMP). Imbalances of supply and demand on single nodes can be overcome by importing electricity from other zones, as long as there are not transmission constraints.
In the USA, balancing of supply and demand on the grid is achieved by independent service operators (ISO) using seasonal, week‐ahead, day‐ahead, and hour‐ahead load forecasts. The ISO ranks all generator bids based on their bidding price and fills up the load forecast of each hour of the day starting from the cheapest bidder, taking into account transmission line limits. Any differences between the forecast and actual load are settled, in real time, in the ancillary services market, which consists of regulation and reserve resources. Regulation resources can quickly adjust their output to accommodate changes to the balance of supply and demand, upon receiving a signal from the ISO. To protect against the risk of a plant outage, the ISO also has in‐service spinning reserves that can be on full capacity within 10 minutes and non‐spinning reserves that can respond within 30 minutes.
The grid operations outlined earlier are important when we consider the impacts of high penetration of variable generation (i.e., solar and wind) into the grid. Let us now discuss the different types of power plants according to the services they provide. Nuclear and large coal‐fired and gas‐fired power plants that operate 24 hours a day serve a baseload, typically the demand that is required throughout the year (8760 hours). Peaker plant are smaller units, typically natural gas‐ or diesel‐fired power plant that operate only during the high demand days of the year, for example, hot summer afternoons in tropical and subtropical climates (depicted as up to 1000 hours in Figure 4.51). Medium‐size natural gas and coal power plants satisfy the balance of daily loads (depicted as intermediate load in the same figure).
Figure 4.51 Example of a load duration curve.
The market clearing price of electricity is set by the marginal price for the last MW needed to meet the load. During peak load hours, a large percentage of the generator fleet is dispatched, including “peaker plants,” which are more expensive, less efficient, and more polluting than conventional generators.
These plants set a high electricity clearing price for all energy delivered in that time period, and it is therefore economically desirable to prevent their dispatch. Solar generators have the potential to minimize the need for peakers, as peak demand is typically AC driven, and therefore effectively reduce the market clearing price while mitigating emissions.
The power output of hydroelectric power plants, gas turbines, and combined‐cycle gas turbines (CCGT) can be effectively adjusted, and they are used to follow the variation of demand load throughout the day. Load‐following power plants run during the day and early evening when the demand is higher and are either shut down or greatly curtailed output during the night, when the demand for electricity is the lowest. Coal power plants with sliding pressure operation of the steam generator can generate electricity at part‐load operation up to 75% of the nameplate capacity and can also be used for load following although their ramping rates are slower than those of gas turbines.
Hydroelectric power plants can be efficient for both intermediate load‐following and baseload applications. Thermoelectric baseline generators (nuclear, large coal) use the Rankine cycle (discussed in previous text box) and are built to operate at their maximum or near‐maximum output 24 hours a day and are expensive to ramp or cycle as this incurs physical wear due to their high thermal inertia. CCGT power plants are also used for serving baseload, and as the price of natural gas is being reduced, they are displacing coal in baseline power plants.
Intermediate and peak load power plants can also provide important voltage and frequency stabilization services in addition to supplying the required loads.
These reserves are normally supplied by generators that have the ability to be dispatched up or down remotely (commonly referred to as automatic generation control).
These can be supplied by generators that are online (spinning) and by “quick‐start” generators that can be started and turned up within 15 minutes. The amount of power that a generator can contribute is limited by its ramp rate and the difference between its current dispatch level and its maximum capability. Contingency can also be supplied by demand response controlled by the system operator.
Figure 4.52 Power generators for load balancing and regulation.
The electric grid system and its market operations were designed to deal with variability of demand and supply on different timescales, mainly by dispatching controllable generators and, to a lesser degree, by using electricity storage systems. A high penetration of variable solar and wind electricity into the grid creates challenges for the grid operators who need to reliably satisfy load demands every hour of the day. Thus it is important to closely investigate the variability of the solar resource on seasonal, diurnal, and cloud‐induced time domains (Figure 4.53).
Figure 4.53 Solar resource fluctuations and options to mitigate them.
The seasonal and diurnal variability are precisely described with a set of geometric equations, describing the Earth’s rotation and its elliptical movement around the sun. Clear sky irradiation can be predicted with very high accuracy, but stochastic variability due to cloud coverage is cause of concern for grid reliability in high solar penetration scenarios. The following solutions are available for reducing or mitigating such variability: (i) geographical diversity/transmission interconnections, (ii) solar forecasting, and (iii) energy storage. A combination of these solutions would in most cases provide the minimum cost solution as variability is reduced with geographical diversity and controlling it is easier and less expensive when we have accurate forecasting. Also demand management has a significant role in handling the diurnal variability.
In the following we discuss solar forecasting and geographical diversity and high‐voltage transmission lines; energy storage is the subject of Section 4.11.
Irradiation data are available in real time from satellite measurements, but their time and scale resolution is rather coarse. For individual PV plants, this can be improved with ground‐based sky‐imaging hardware and software that can provide local irradiation information 1–15 minutes ahead by estimating the movement of the clouds and associated shading. An irradiance sensor network, together with multipoint optical imaging systems, can provide three‐dimensional spatial information about clouds and their motions. As shown in Figure 4.54, this information leads to greatly improved accuracy and system reliability in forecasts, especially for cloudy and partly cloudy conditions wherein the fluctuation of solar energy is the largest, and the value‐added gain in forecasting is the highest. With this information PV operators could foresee a forthcoming rapid ramp‐rate change and handle it with providing extra power, gradually curtailing output, or dispatching power from storage. For large ramp rates, use of storage would be required, but this is minimized by considering the changes in total output aggregate of the PVPS or of the aggregate of PV plants in the same balancing area.
Figure 4.54 Forecast mean average errors (MAE) are reduced by 39–24% when sky imaging is integrated with satellite data9.
(Source: Courtesy of Dantong Yu, NJIT)
Irradiance measured at a single point can change drastically; however, the aggregate output of geographically dispersed PV systems is much smoother than that of a single point. Such smoothening is also observed within large multi‐MW PV, which typically occupy larger areas that are not totally shaded in a partially cloudy day. This effect is shown in the histogram shown in Figure 4.55 of daytime minute‐to‐minute irradiation changes (called ramp rates) observed at a 5 and 80 MW plants near each other in Ontario, Canada, over the course of a month. The 80 MW plant exhibits relatively less variation compared with the 5 MW plant. Figure 4.55 shows how the aggregate power production from the whole 80 MW plant is smoother than the irradiation profile on one point of the plant.10 The smoothening of the fluctuations is even greater in the multi‐hundred MW PV plants constructed in the United States, China, and other countries.
Figure 4.55 One‐minute ramp‐rate histograms for a 5 and 80 MW plants near each other in Ontario, Canada. The larger the plant, the smaller the ramp rates.
Figure 4.56 Impact of cloud passage on an 80 MW plant power output in Ontario, Canada. The green line shows irradiation measured on one point (W/m2), and the orange line shows the smoothening effect from the aggregate of the inverters (MW).
(Source: Reproduced with permission of First Solar)
Understanding the spatial and temporal characteristics of solar resource variability is important because it helps inform the discussion surrounding the merits of geographic dispersion and subsequent electrical interconnection of photovoltaics as part of a portfolio of future solutions for coping with this variability. Unpredictable resource variability arising from the stochastic nature of meteorological phenomena (from the passage of clouds to the movement of weather systems) is of most concern for achieving high PV penetration because unlike the passage of seasons or the shift from day to night, the uncertainty makes planning a challenge. A detailed discussion of these variability aspects is given by Perez and Fthenakis.11 The drastic effect of geographical diversity on reducing the stochastic variability of solar irradiation is shown in Figure 4.57.
Figure 4.57 Daily variability over the course of 2 years in (a) Los Angeles and (b) an area of 190 × 190 km2 around the city.
Geographical diversity can greatly reduce the fluctuations of solar energy. As discussed before, increasing the geographical area of PV, by either connecting an aggregate of dispersed small systems or by constructing large power plants, decreases the cloud‐induced fluctuations of their total output. In addition, long interconnects can reduce diurnal intermittency, by taking advantage of different time zones; in theory, if we could connect anti‐diametric locations of the globe, we could resolve the diurnal intermittency (e.g., by connecting China with the United States) as well as the seasonal variability (e.g., northern Chile with the Southwest United States). The long distances involved will necessitate HVDC lines, a technology that exists and can be further improved if there is a market. The transmission of electricity over 2000 miles of HVDC lines typically entails a 10% electromagnetic loss versus a 22% or higher loss using high‐voltage AC power lines of the same distance. Also construction of HVDC power lines typically requires 37% less land area than constructing high‐voltage AC ones. The technology is well established, but there are cost and siting challenges that need to be addressed. Currently, HVDC transmission lines with a capacity of 5 GW are operating in China utilizing 800 kV technology, and in the future a doubling of this capacity is expected.12 The array converters will be boosting array voltages of around 1 kV DC to a “gathering” voltage of around 50 kV DC. A second DC‐to‐DC converter will be used to boost the “gathering” voltage to the transmission voltage of 800 kV DC. These converters are already in common use throughout the electric utility industry in power‐conditioning devices and HVDC power lines in Europe and the United States. Transmission experts McCoy and Vaninetti reported in 2008 a cost of $0.02/kWh for constructing a network of HVDC lines with 800 kV.12 The same experts foresee a dramatic improvement in technology and in unit cost with a plan for large deployment and domestic production of the large quantities of heavy cables required. Future developments in superconducting cables would make the vision even more feasible. The biggest challenge regarding long‐distance transmission is that would need approvals from a plethora of regional and national jurisdictions. For the United States, Fthenakis and collaborators Zweibel and Mason have defined the needs of HVDC from the SW for solar and the High Plains for wind to the rest of the country.3,4 Crossing national borders may be more involving, but we have plenty of boundary crossing transmission lines in Europe, United States–Canada, and elsewhere. Developing large global electricity grids is not a far‐fetched idea; it could follow the paradigm of global telecommunications networks, which were enabled by fiber‐optic technology connecting continents with underwater cables; it is expensive but it is technologically feasible.
The flexibility of a power system, that is, its ability to vary its output to meet the demand, depends on the mix of its generators. Its flexibility is constrained mainly by the baseload generators, usually nuclear and coal power plants. The operation of nuclear power plants is quite inflexible, and that of coal power plants is flexible only within a narrow range.
We discussed earlier the need for adjusting power output for load following and for satisfying peak demand. In addition to these duties, in order to enhance the reliability of the grid, a number of generators are operated at partial load with their surplus megawatts available to ensure reliability under rapid changes on demand (frequency regulation) or an unplanned outage of a unit or a transmission line (contingency reserves). Frequency and contingency reserves operating at partial load are called spinning reserves. There are also units that provide grid stability (voltage, frequency, reactive power). Overall, a number of generators operating at partial load account for the functions of load following, frequency regulation, and spinning reserves. Solar and wind penetration into a power system displaces generation from conventional units, and often their combination can increase such displacement. Figures 4.58 and 4.59 shows a result from simulations based on hourly wind, solar, and load data in NY State.12 The figures show the load and PV and wind outputs for a summer day where a peak of the load for that year occurred; it is shown that PV and wind together serve the load better than each of them separately. From a utility’s perspective, the main advantages of this penetration are fuel economy, emissions reduction, and reduced need for increasing overall system capacity.
Figure 4.58 Synergy of PV and wind in New York State.
(Source: Nikolakakis and Fthenakis13. Reproduced with permission of Elsevier)
Figure 4.59 Effect of grid flexibility on PV energy delivery.
(Source: Nikolakakis and Fthenakis13. Reproduced with permission of Elsevier)
The integration of large amounts of variable renewable energy supply into the grid may necessitate additional means for frequency regulation and increase the requirements for ramping rate of load‐following units. Thus, such integration on the grid may translate into additional costs to cover the greater amount of flexibility, ramping capability, and operating reserves needed in the system. However, such costs are minimized when PVs are dispersed geographically over large regions.
The limit of the penetration of renewable energy into an electricity grid depends on the mix of generators of the system. As described previously, there are two main types of generators: inflexible baseload units and the more flexible cycling units. The former units are designed to operate at full output, and they often provide most of the winter demand or 35–40% of the annual peak capacity. The penetration of wind and solar power cannot drive the net load below the limit imposed by the number and the type of baseload generators and the amount and type of reserves. This limit depends on the ability of conventional baseload generators to reduce significantly their output, and on economic and mechanical constraints. For example, coal plants can vary their output from full to half capacity, but if this is done frequently, it would demand costly maintenance. (The flexibility limit that separates the flexible from the inflexible capacity and, hence, the flexibility of a system is defined as the percentage of the annual peak capacity that is flexible.)
High levels of wind and solar energy penetration may stress the system because of its flexibility limit; there will be hours throughout the year where the net load is brought below it. Then, the amount of energy below the flexibility limit cannot be absorbed and must be curtailed. This is more of a problem for incorporating wind power than solar power, since winds are stronger during the night when the load levels are low. The more flexible a power system is, the higher is the penetration achievable, and the less the restraint on renewable electricity. The amount of energy to be cutback can be determined with a cost analysis; often it makes economic sense to curtail small amounts of energy as a trade‐off in penetration.
The irregularity of renewable resources and the limit on the grid’s flexibility both pose restrictions on the maximum penetration achievable in a system. There are some interesting studies on the maximum renewable penetration that can be realized without storage. It was shown that the maximum annual energy penetration attainable from solar energy alone in the Texas grid is around 10% if no energy is curtailed, but it increases to 22% if we allow for 10% of the PV energy to be curtailed.13 Studies focusing on the New York yielded almost identical results and also showed a great synergy between wind and solar resources in meeting the NYISO loads.14
Energy storage can increase the flexibility and reliability of the system; it can offer the following services:
Electric storage technologies are differentiated by various attributes, such as rated power and discharge time. In general, there are three major categories of large‐scale energy storage technologies: power quality, bridging power, and energy management. The main difference between them is the timescales over which they operate and their power and energy capacities (Table 4.1).15
Table 4.1 Categories of electricity storage technologies.
Categories | Applications | Operation timescale | Technologies |
Power quality | Frequency regulation, voltage stability | Seconds to minutes | Flywheels, capacitors, superconducting magnetic storage, batteries |
Bridging power | Contingency reserves, ramping | Minutes to ~1 hour | High‐energy‐density batteries |
Energy management | Load following, transmission/distribution deferral | Hours to days | CAES, pumped hydro, high‐energy batteries |
Power quality refers to the set of parameters (e.g., voltage and frequency regulation, reactive power, fault‐ride through) that must be continuously satisfied for electrical systems to operate as expected. The storage technologies that are best suited for ensuring this continuity must provide a large power output on very short timescales, typically seconds. Since power delivery occurs in such a brief period, large storage capacities are not necessary. These technologies include superconducting magnetic energy storage (SMES), electric double‐layer capacitors (EDLCs), flywheels, and batteries (Figure 4.60).
Figure 4.60 Comparison of storage systems in terms of discharge times and rated power.15
On the other hand, technologies that provide power over longer timescales for applications such as load leveling and peak shaving are used for energy management. Whereas power‐quality applications deal with short‐term and unpredictable fluctuations in power output, energy management technologies address variability that largely is predicted by peak and off‐peak demand. Emphasis is placed on storage capacity and less so on instantaneous power, and the timescales involved are much longer than those needed for power‐quality technologies. The upper region of Figure 4.60 provides a few examples of storage mechanisms used for energy management: pumped hydro, compressed air energy storage (CAES), and flow batteries. Between these two boundaries lie storage technologies used for bridging power to ensure continuity when switching from one source of energy to another.
Let us now discuss individual technologies.
Among the most efficient storage technologies are SMES systems. They store energy in the magnetic field created by passing direct current through a superconducting coil; because the coil is cooled below its superconducting critical temperature, the system experiences virtually no resistive loss. Four components comprise a typical SMES system: the superconducting coil magnet (SCM), the power‐conditioning system (PCS), the cryogenic system (CS), and the control unit (CU).
The major disadvantage of SMES is the high cost of refrigeration and the material of the superconducting coil. Research is being made into so called “high‐temperature” superconductor (TSC) technology that utilizes liquid nitrogen operating at 65–77°K (about −200°C) rather than the costly liquid hydrogen required for a very‐low‐temperature superconductor; nevertheless, the costs of HTSC material remain high. SMES systems with large capacities of 5–10 GWh involve large coils, several hundred meters in diameter that must be kept underground, adding to the expense of the system.
However, SMES have very high round‐trip efficiencies (e.g., ~95%). In addition, SMES systems can discharge almost all the energy stored in the system with a high power output in a very short time, making them ideal for power‐quality applications. SMES systems improve power quality and system stability in several ways. Following an interruption, such as a downed power line or generator, a SMES unit can dampen low‐frequency oscillations and mitigate voltage instability by providing both real and reactive power to the power system. On the demand side, an SMES can balance fluctuating loads by releasing or absorbing electricity according to demand. They also can be used as a backup power supply for critical loads that may be sensitive to disturbances in power quality; their fast response time allows them to inject power in less than one power cycle.14
EDLCs, also known as supercapacitors or ultracapacitors, offer another solution to ensure quality and short‐term reliability in power systems. Like a conventional capacitor, electricity in an EDLC is stored in the electrical field between separated plates; the capacitance is a function of the plates’ area, the distance between them, and the dielectric constant of the separating medium. However, whereas standard capacitors employ two plates of opposite charge separated by a dielectric, EDLCs consist of two porous electrodes immersed in an electrolyte solution, a structure giving a highly effective surface area and minimal distance between electrodes.
Compared with regular batteries, EDLCs have lower energy densities and higher power densities, making them suitable for power‐quality applications. For short‐term high‐power applications, electricity discharge in an EDLC is not limited by the rates of chemical reaction rates as is the case with batteries. In addition, EDLC–battery hybrids that incorporate the benefits of both technologies often are used for distributed energy storage.
This hybrid storage offers a dynamic solution to problems related to off‐grid PV: the EDLC provides the power necessary for large fluctuations in power demand, while the battery remains the source of continuous electricity over long periods. In this arrangement, the battery’s size is geared for a constant load rather than for peak current demand, which can be up to 10 times the normal operating current, and may only need to be satisfied for a few seconds at a time. Because the EDLC handles high currents, the battery does not experience deep discharges, and thus its life is extended.
Batteries in off‐grid PV hybrid storage systems, when paired with EDLCs, experience less discharge depth, which translates to longer lifetimes and smaller batteries. In addition, the hybrid arrangement increases the reliability of the PV system on both large and small timescales. Increased power quality ensures that fluctuations in load demand will not adversely affect the stand‐alone system, making off‐grid PV systems a viable option where they might not be without energy storage technology.14
Energy in a flywheel is stored in the form of rotating kinetic energy, in contrast with batteries and SMES where energy is stored in chemical and electrical form, respectively. Peak power for flywheels depends on the application, ranging from kilowatt to gigawatt scales. One major advantage of flywheels is long life, longer than 20 years and independent of depth of discharge; that is, unlike electrochemical batteries, flywheels operate equally well whether discharges are few and deep or frequent and shallow.
The most mature commercial application for flywheels is providing an uninterruptable power supply, taking advantage of the flywheel’s high power density and fast recharge time. Short bursts of power are administered when power line disturbances occur, 80% of which last for less than a second. For some applications, a flywheel can coast for over an hour to zero charge.
Batteries can be used for both power‐quality and bridging power applications. The most common ones are lead‐acid batteries.
Sealed, valve‐regulated lead‐acid (VRLA) batteries have been the most common type of batteries in PV residential systems (Figure 4.61). There are two types of VRLA batteries: absorbed glass mat (AGM) and gelled electrolyte. The former store the electrolyte on a glass mat separator composed of woven glass fibers soaked in acid. The latter immobilize the electrolyte in a gel. Hybrid VRLA batteries encompass the power density of AGM design and the improved thermal properties of the gel design. Of all types of lead‐acid batteries, the hybrid VRLA proved to be the technology best suited for PV stand‐alone lighting systems.
Figure 4.61 Pb‐acid battery operational principles; shown anode, cathode, electrolyte, and associated reactions16.
(Source: Mertens16. Reproduced with permission of Wiley)
PV panels are not ideal sources for charging lead‐acid battery because they generate power intermittently. One proposed method to extend the lifetime of the VRLA batteries is combining them with supercapacitors into a hybrid storage system utilizing the high power density, longer life cycle, and fast charge–discharge times of the supercapacitor to supply short bursts of power during times of peak demand and motor starting. The more energy‐dense VRLA battery supplies energy continuously over longer periods. Incorporating a supercapacitor allows the battery to be sized according to the demands of normal operating current rather than to that of peak current while avoiding deep discharge, maintaining a high state of charge (SOC), preventing sulfation and stratification, and extending the battery’s life. Though supercapacitors are expensive, cost reduction through technology development and market growth may enable such supercapacitor–VRLA battery hybrids affordably to provide more reliable storage for PV systems.
At this point of time (mid‐2016) among all battery technologies, lithium‐ion (Li‐ion) batteries have the largest potential for future development and implementation. Both their energy density and power density are high compared with other battery technologies (Figure 4.62), making them ideal for portable applications and a good candidate for PV‐supporting applications as well.
Figure 4.62 Battery types: energy density comparisons.15
Li‐ion batteries have long lifetimes and low self‐discharge rates. Electricity can be charged and discharged very quickly with high power output, with no memory effect. Round‐trip efficiency is 90% or higher. A Li‐ion battery stores electricity when voltage is applied to it causing Li‐ions to travel from the metal oxide cathode through the electrolyte separator to the graphitic carbon anode. Electricity is discharged when the ions travel in the opposite direction; Figure 4.63 depicts this mechanism. The disadvantages of this technology include its high cost and sensitivity to extreme conditions.17
Figure 4.63 The basic working mechanism of a Li‐ion battery.
(Energy and Power Group, University of Oxford)
Despite the large power density of a Li‐ion battery, like any battery it deteriorates when exposed to deep discharging and overcharging. In fact, much of the cost of Li‐ion batteries lies in the overcharge protection units to prevent such events from occurring. High temperatures can further decrease the battery’s life.
However, despite its drawbacks, Li‐ion battery technology offers an interesting solution to grid‐scale energy storage. Due to their high storage capacities, fast charging rates, and relatively small sizes, Li‐ion batteries are popular for use in electric vehicles. If implemented on a large scale, fleets of plug‐in electric vehicles can offer not only as cleaner modes of transportation but also distributed sources of stored energy for the grid. In time, when their battery capacity is increased, large numbers of electric vehicles powered by batteries would add flexibility and stability to a smart grid and enable further penetration of renewable energy.
Tesla, which has been using Li‐ion batteries in their cars, started in 2015 marketing a battery for stationary applications at a cost of $350/kWh for residential (10 kWh systems) down to $250/kWh for commercial (100 kWh) systems. The latter can also be used for community systems that connect several residential ones and take advantage of both the smoothening of the solar and the demand variability and the economies of scale from the larger battery systems.
Another rechargeable Li‐based technology with potential automotive applications is the lithium‐ion polymer (Li‐poly) battery that is like a Li‐ion battery, wherein the liquid electrolyte is replaced by a solid polymer electrolyte. The cost of Li‐poly batteries currently is prohibitive, but their increased production may lower the cost in future. Hyundai announced plans to use Li‐poly batteries in its HEVs, and an Audi A2 powered by Li‐poly batteries recently set the record for distance traveled on a single battery charge.
In flow batteries, unlike conventional batteries, the charged electrolyte is stored in a separate tank and circulated when needed. This configuration essentially decouples the power and energy aspects of the batteries and allows them to be sized independently: the power is determined by the size of the electrochemical cells, and the energy storage is limited only by the size of the external storage tank and volume of electrolyte. It also avoids the problem of self‐discharge present in most battery technologies. However, flow batteries typically have lower energy densities than most portable batteries when the storage and reactor tanks are accounted for; they also require using additional components, such as pumps and sensors.
There has been limited deployment of at least two types of flow batteries, vanadium redox and zinc bromine; other types are under development, such as polysulfide bromide ones. Redox flow batteries consist of two half cells, each containing dissolved species in different oxidation states, separated by an ion‐exchange membrane. Hybrid flow batteries such as zinc bromine batteries (Figure 4.64) contain metallic species deposited in one of the half cells.
Figure 4.64 Schematic of a flow‐assisted battery.
VRB Power (currently Prudent Energy) invented the vanadium redox battery energy storage system (VRB‐ESS™) built on a 175 kW modular basis. Large‐scale installations include a 1 MWh per 500 kW storage system for the China Electric Power Research Institute (CEPRI) as part of a project that includes 78 MW of wind capacity, 640 kW of PV capacity, and 2.5 MW of energy storage. In addition, the company is building in California vanadium redox battery systems with SunPower’s PV systems, in cooperation with PG&E, KEMA, and Sandia National Laboratories.
Flow batteries are being developed on the residential PV system scale. HomeFlow is a 30 kWh/10 kW zinc bromine flow battery intended for such usage. The product is being developed by Premium Power, which strives to become the “Dell computers of flow batteries” by bringing the technology directly to homes through modular design and inexpensive manufacturing. Its product line also includes larger zinc bromine batteries with capacities/rated powers of 45 kWh/15 kW, 100 kWh/30 kW, and 2.8 MWh/500 kW that can be installed at the community level to curb peak power demand.
The key characteristic of flow batteries is their independent scalability: reactors can be scaled up in response to increasing power demand, while storage tanks and electrolyte solution can be added for more energy storage capacity to accommodate additional renewable generators. At the utility scale, where capacity and cost are more influential than volume requirements, flow batteries are a promising technology for renewable energy systems.
Pumped hydro energy storage (PHES) is the most widely used large‐scale energy storage technology. It utilizes elevation difference between natural (or man‐made) reservoirs to increase the potential energy of water by pumping it into the higher reservoir and later produce electricity by reversing the operation of the pump running it as a turbine. The schematics of a pumped storage plant are shown in Figure 4.65. In general, PHES plants have a round‐trip efficiency of around 75% and can have discharge capacities of more than 20 hours. Projects may be practically sized up to 4000 MW and operate at about 76–85% efficiency, depending on design. Pumped hydro plants have long lives, on the order of 50–60 years. As a general rule, a reservoir having 1 km in diameter, 25 m deep, and an average head of 200 m would hold enough water to generate 10 000 MWh.
Figure 4.65 Schematic of the Raccoon Mountain Pumped Hydro Plant (Wikipedia).
The earliest plant in the United States was built in the late 1920s, and the last pumped storage plant commissioned was in the 1980s, when environmental concerns over water and land use severely limited the ability to build additional pumped hydro capacity. Most of the PHES storage facilities were constructed during the 1970s and the 1980s, and their capacity just in the United States grew to 20 GW. The oil crisis of 1973–1976 and associated fuel price increases catalyzed this growth as PHES were planned to use nighttime power from nuclear plants to satisfy day peak loads. Even though the capital cost of PHES was always higher than conventional generation, the difference was small till late 1980s. PHS started being less competitive since the 1980s as nuclear power deployment became standstill and CCGT capital costs and the price of natural gas were reduced. At this time, the capital cost of PHES is almost twice that of CCGT (the overnight capital cost of CCGT in 2006 was around $800–1100 compared with the estimated cost of $2100 of a 10 hours conventional PHES plant in 2009). An order of magnitude estimation of energy storage costs can be found in a handbook produced by the Electric Power Research Institute18; these are summarized in Table 4.2.
Table 4.2 Energy storage technology cost estimates (Akhil et al.20).
Source: Ibrahim et al.17. Reproduced with permission of Elsevier.
Storage type | $/kW | $/kWh | Hours | Total capital ($/kWh) |
CAES (100–300 MW underground storage) | 590–730 | 1–2 | 10 | 600–750 |
Pumped hydro (conventional 1000 MW) | 1300 | 80 | 10 | 2100 |
Battery (10 MW) | ||||
Lead‐acid, commercial | 420–660 | 330–480 | 4 | 1740–2580 |
Sodium–sulfur (projected) | 450–550 | 350–400 | 4 | 1850–2150 |
Flow battery (projected) | 425–1300 | 280–450 | 4 | 1545–3100 |
Lithium‐ion (small cell) | 700–1250 | 450–650 | 4 | 2300–3650 |
Lithium‐ion (large cell, projected) | 350–500 | 400–600 | 4 | 1950–2900 |
Flywheel (10 MW) | 3360–3920 | 1340–1570 | 0.25 | 3695–4313 |
Superconducting magnetic storage | 200–250 | 650 000–860 000 | 1 second | 380–489 |
Supercapacitors (projected) | 250–350 | 20 000–30 000 | 10 seconds | 300–450 |
The reader is advised to look at the source for details of those estimates and keep in mind that cost may vary with the price of commodity materials and the location of a project.
Compressed Air Energy Storage (CAES) converts grid electricity to mechanical energy in the form of compressed air stored in underground (or surface) reservoirs. The source of input energy can be excess off‐peak electricity or renewable electricity coming from wind or solar farms. To convert stored energy back to electricity, the compressed air is released through a piping system into a turbine generator system after having been heated. When compression and expansion are rapid, the processes are near adiabatic; heat is generated during compression, and cooling occurs during expansion. The first is associated with large energy losses as compression to 70 atm can produce temperatures of about 1000°C, so necessitating cooling.
For large CAES plants, a large storage volume is required and underground reservoirs are the most economically viable solution. Such reservoirs can be a salt formation, an aquifer, or depleted natural gas field. When the volume confining the air is constant, pressure fluctuates throughout the compression cycle. Constant pressure operation in hard rock mined caverns is achievable by using a head of water applied by an aboveground reservoir. For smaller CAES plants (e.g., <5 MW), air can be stored in aboveground metallic tanks or large on‐site pipes, such as those designed for carrying natural gas under high pressure.
A typical CAES power plant comprises a compression and a generation train connected through a motor/generator device. During the compression mode, electricity runs dynamic compressors that compress air at pressures of 70 bars or more. Because of the high pressure ratio required, compression takes place in a series of stages separated by cooling periods. Cooling the air is necessary to reduce power consumption and meet the cavern’s volume requirements. The higher the number of stages, the greater is the efficiency attained; however, this increases the cost of the system. During the expansion mode, motor operation stops and clutches engage the generator drive. Air is released to run the expanders after having first being heated in properly designed combustors. Heating the air assures high efficiency and avoids damaging of the turbomachinery due to low temperatures resulting from the rapid expansion of air and the Joule–Thomson effect. A recuperator sited after the exit from the expanders recovers some of the energy of the heated air before it is released to the atmosphere. Even though fuel is needed to run a CAES power plant, the input for a certain power capacity is around 65% less than the amount required to run a Gas Turbine (GT) because around two‐thirds of the energy produced by a GT is used to run its compressor. Thus, when the compressors are fed by renewable electricity, the emissions of a CAES power plant are 35% of those produced by a GT of the same capacity.
Currently, two CAES power plants are operating. The world’s first facility is the Huntorf CAES plant that has operated since 1978 in Bremen, Germany. It is a 290 MW facility, designed to provide black‐start services to nuclear power plants located nearby, along with spinning reserves and VAR support as well as cheap off‐peak electricity. It stores air up to 1000 psi (68 atm) in two depleted salt caverns located 2100 and 2600 ft under the ground; it offers up to 4 hours of power generation. The second CAES plant is a 110 MW power plant operating in McIntosh, Alabama, since 1991 (Figure 4.66). It pressurizes air to 1100 psi (75 atm) and has electricity generation cycle of up to 26 hours between full charges. The McIntosh plant also has a heat recuperator in the expansion train that reduces fuel consumption by 25% compared with the Huntorf plant that does not include recuperation.
Figure 4.66 Schematic of a CAES plant source.
(Source: Compressed Air Energy Storage (CAES) Conference19)
Deregulation and the current structure of electricity markets now allow storage technologies to participate in the electricity market and profit from their operation. As an example, the NYISO includes markets for installed capacity, energy, and ancillary services and for preventing transmission congestion. CAES power plants have the following attributes that make them suitable for large‐scale, diurnal, multiday, and seasonal energy storage:
3.141.201.26