the-eye.eu · 1 State of the Art 1.1 INTRODUCTION This chapter is structured intwo parts. Firstly, the available knowledge with regard to low-powerenergysourcesisexamined(section1.2).Thisreview

1State of the Art

1.1 INTRODUCTION

This chapter is structured in two parts. Firstly, the available knowledge with regard tolow-power energy sources is examined (section 1.2). This review indicates that solar (pho-tovoltaic) cells have a number of advantages within this group. For the example of indoorphotovoltaic (IPV) products, therefore, the intellectual property (section 1.3) and IPV tax-onomies (section 1.4) are investigated. These are presented with respect to their genericparameters. The links between the different taxonomies are summarised in Table 1.3 andshow the wealth of information that may be available at the beginning of an IPV designproject.In the second part of this chapter, the areas where information is less complete are

considered as well as how this book proposes to fill these gaps (see section 1.5).

1.2 LOW-POWER ENERGY SOURCES

The trend of the number of electronic devices per person, in the first world at least, hasbeen steadily increasing for decades. From ostensibly electronic products, such as the mobilephone, to the more imperceptible, such as embedded systems, the issue of providing sufficientenergy is ubiquitous. In some cases it is relatively simply resolved by the use of exisitinginfrastructure: for example, automotive products have their own power supply and thisis usually already fed throughout the vehicle. However, for each incremental device, theextra wires must be balanced against their extra cost and weight. Providing each electronicdevice with an independent power supply may therefore be an attractive option in order toavoid cost, weight or design complication. Unfortunately, existing technologies generallyrequire recharging or refuelling (see the lower part of Table 1.1). An example of this is thepower source for a mobile telelphone, which users are accustomed to recharging with theinconvenience of reduced flexibility of use during this period.

Designing Indoor Solar Products – Photovoltaic Technologies for AES J. F. RandallPublished in 2005 by John Wiley & Sons, Ltd

2 State of the Art

Table 1.1 Energy comparison of low-power �<1W� energy sources

Typical power Typical annualLow-power density energy densitytechnology ��W/cm3� �J/cm3� References

Ambient energy sourcesSolar (indoor – near window) 1000 10510 Table 3.3Solar (indoor – electrical source) 100 1050 Table 3.3Shoe insert 300 200 [232]Vibrations (mechanical) 45 120 [215]Push button 300 80 [119]Acoustic noise (100 dB) 1 32 [215]Peltier thermoelectric (10 �C gradient) 15 20 [218]

Low-power energy sources requiringrefuelling

Wind up clockwork generator 210000 27600 [119]Hand squeeze generator 120000 15700 [119]Microheat engine (unlimited fuel) 400 12700 [215]Fuel cell (PEMFC) (unlimited fuel) 400 12700 [215]Fuel cell (PEFC) (unlimited fuel) 70 2210 [174]Electrochemical (single-use lithium) 45 1420 [215]Microbial fuel cell (unlimited input) 15 470 [192]Electrochemical (rechargeable lithium) 7 220 [215]

Low-power energy sources withoutrefuelling

Nuclear 15 470 [133]Radio frequency 10 130 [209]

Another trend that has persisted much longer than the use of electronic devices is the ten-dency of humans to spend a greater proportion of time in protected environments, especiallybuildings. The confluence of these two trends is the automation of the built space.An approach that avoids the need for user intervention and extra wiring and, in principle,

assures that the product is available whenever required is the collection and use of ambientenergy in the environment. The word ambient means ‘encompassing on all sides’ and energyis the ‘capacity for performing work’ [264]; therefore, ambient energy systems seek tocollect, in their immediate environment, those energetic flows that have the potential to beput to use. Both this energy (and this potential!) may or may not be perceptible to humans.This definition can be embellished by other aspects of ambient energy. For some it maymean the convenience of reduced maintainance while for others it may be the elegance ofbeing self-sufficient from resources that will not be depleted.A number of ambient energy technologies exist, compared in the top part of Table 1.1.

One of the main reasons that they are not more widely used is that their energy density isgenerally relatively low (less than 200 J/cm3 per annum). Also, no single technology can beapplied in all environments [217].One exception nevertheless stands out – the solar (or in the jargon ‘photovoltaic’ cell).

As can be seen, the annual energy densities of solar technology under different indoor

Intellectual Property Rights 3

conditions have the same orders of magnitude as those of other technologies that requireregular recharging. In the next sections, solar technology is therefore explored.

1.3 INTELLECTUAL PROPERTY RIGHTS

Patents may serve the product designer at two stages of a project. At the outset, the generaltechnological area can be checked. Once the definition of the product is clarified, a morefocused search ensures that no prior art is contravened. Both of these stages often provide awealth of indirect knowledge and ideas which serve to inspire the designer. They also allowthe designer to form an opinion on the extent of protection that their work will require.

1.3.1 Methodology

In this section, the first stage (technological area) is assessed. This area was considered withregard to four criteria that form the basis of an indoor photovoltaic (IPV) sensor product.Charge storage was included as this indicates a higher-power product than with solar alone.However, a ceiling of 100mW was set on the consumption of the application. Each criterionwas associated with keywords that were used as the basis of the search (Table 1.2) [59].

1.3.2 Results

The prior art analysis returned a total of 137 abstracts. Of these, 74 were obtained in hardcopy. These, prioritised by relevance to the IPV concept, showed that 59 (80%) madereference to a photovoltaic system and, within these 59, the largest four groups of patentssatisfied the following criteria:

1. Photovoltaic (PV), low light, low power, storage.2. Photovoltaic (PV), low power, storage.3. Photovoltaic (PV), low power.4. Photovoltaic (PV), storage, mid power �<100mW�.

The same results are presented graphically in Figure 1.1. The patents that could present themain challenge to the further implementation of the IPV concept are in group 1 (PV, low

Table 1.2 Basis for author’s intellectual property rights research

Criteria Condition Keywords

Photovoltaic Is technology mentioned? Photovoltaic, photoelectric,solar, photocatalytic,photoinduce

Indoor light Is 200 lux specified orimplied?

Domestic, indoor, low light, lowintensity, lux

Charge storage Is any mentioned? Portable, batteryApplication Is power less than

100mW?Low power

4 State of the Art

24%

35%16%

25%

PV, Indoor light, Low-power app., storage (13)

PV, Low power, Storage (19)

PV, Low power (9)

PV, Storage, power > 100 mW (14)

Figure 1.1 Proportion of the 55 most appropriate patents read and the criteria that they satisfied

light, low power, with storage) on the basis that all other patents reviewed were lacking oneor more criteria. Of the 13 patents in group 1, the patents could be further sorted by relevanceto storage technology, for example (i.e. the nine accumulator-based and four capacitor-basedsolutions protected).Taking the nine accumulator-based patents, the following comments can be made:

FR 2 618 242: Dispositif émetteur à cellule photovoltaïque – may be important for sensorand communication application.FR 2 606 912: Dispositif formant émetteur à cellule photovoltaïque – may be importantfor sensor and communication application.EP 0 681 549: System for identifying, searching for and locating objects – may beimportant for specific application.JP 03015168: Photo energy storage device – makes no mention of application in theabstract.GB 2320 356: Combined liquid crystal display and photovoltaic converter – may beimportant for specific application.WO 99/34337: Chip card for paying freeway tolls – may be important for specificapplication.EP 0 856 738: Extern gespeiste Anordnung mit einer elektronischen Anzeige und/odereiner Kommunikations-Schnittstelle – may be important for specific sensor and commu-nication application.EP 0 918 212: Verfahren zur Erfassung und Auswertung von temperaturabhängigenVerbrauchswerten oder Messwerten anderer physikalischer Grössen – may be importantfor specific sensor application.EP 0 935 099: Elektrisches Haushaltgerät mit Solarzelle – may be important for upgradinghousehold equipment applications.

The four capacitor-based design patents are as follows:

EP 0 871 018: Electrically isolated interface for fluid meter – may be important for suchfluid meters, but does not stock the energy.JP 08036070: Solar cell timekeeper – may be important for horological applications.

IPV Taxonomies 5

FR 2 694 109: Ticket réutilisable d’affichage de données temporaires, notamment ticketde parking – may be important for display applications.US 5 208 578: Light powered chime – may be important for such applications.

1.3.3 Conclusion

Based on the above research, it can be concluded that the IPV concept has been protectedto some extent, but in an application-specific way. This implies that, for extending the useof the IPV concept in other applications, no intellectual property protection is in place.Furthermore, even if applications of the same kind as those found in the patents were tobe targeted, this does not imply that the above patents would be automatically contravened.Once the targeted IPV concept product(s) has been specified, it would then be possible toperform a more focused prior art investigation.

1.4 IPV TAXONOMIES

As the product design of just one type of application is considered, expected requirementsand typical characteristics may be predicted. The latter form the basis of the followingclassifications which may serve to guide the designer as they come at the price of somesubjectivity and simplification. The purpose of such taxonomies for the designer is toappreciate the implications of their initial assumptions and decisions which, as will be seenin Section 2.3, may have a significant impact on the final design.

1.4.1 Product Use Taxonomy

IPV product or application descriptions may be sorted into three groups: toys, tools andsensors. Each of these has typical design priorities and useful life:

1. Toys are for amusement, especially of the young. These are typically low-cost and arenot expected to last more than one or two years.

2. Solar tools include the pocket light, the calculator, the watch and the automatic stapler.They are expected to be available for use at all times, even if their use phase is relativelyshort compared with product life. As the customer benefits from not needing to changeor actively recharge the battery, they may be more expensive than their battery-poweredequivalents. Typical product life will be between two and five years.

3. Sensors measure one or more variables and communicate these, increasingly by wireless.For sensors in the security or comfort market sectors, reliability can be expected to havepriority over cost. Typical expected life is between three and ten years, despite the factthat the useful life of the building in which they are installed is at least 20 years.

1.4.2 Function (or Circuit) Taxonomy

While the product use taxonomy may be useful at a relatively high level, the applicationfunctionality required forms a basis for product specification, such as the typical electroniccircuit required. Functional categories ranked in roughly increasing energy consumption

6 State of the Art

include respond to light, display, light, move, compute, amplify and communicate (in thiscase understood as wireless communication).The respond to light function is typical of the simplest solar-powered products: toys,

mobiles, moving shop window presenters and simple light meters. It is associated with anelectronic circuit with no charge storage between the solar cell and the application, RA

(see Figure 1.2).Sense and display products often use a (non-backlit) liquid crystal display (LCD); such

applications, RA, include calculators, temperature gauges or electronic weighing scales.Given that LCDs require a certain amount of light to be read implies that only a smallamount of charge storage is required. This is often achieved with a capacitance (seeFigure 1.3).Light, move, compute or amplify functions are required of pocket torches, some toys,

microcomputers and hand-held televisions respectively. The charge storage associated withthese for solar-powered systems is generaly a rechargeable battery, such as Figure 1.4. Inthis diagram, the Zener diode is used as an overcharge protection for the battery. The Zenerdiode may be used in the same way and for the same purpose in other circuits with chargestorage, such as Figure 1.5.Wireless communication even for short-range devices (SRD) with a range of a few hundred

metres may not be possible using the power of battery alone. This is because storage systemhigh-current responsiveness and low-storage self-discharge are generally mutually exclusive.

RA

Figure 1.2 Typical circuit diagram for respond to light functionality

RA

Figure 1.3 Typical circuit diagram for sense and display functionality

IPV Taxonomies 7

RA

Figure 1.4 Typical circuit diagram for light or movement functionality

RA

Figure 1.5 Typical circuit diagram for wireless functionality (RA represents the application)

To compensate for this, a hybrid solution (see subsection 6.6.2) may be used, e.g. a capacitorand a rechargeable battey, such as the circuit in Figure 1.5.

1.4.3 Radiant Energy Application Taxonomy

By considering where and how the product will be used within the built space, it may bepossible to determine the amount of radiant energy it will receive.The typical location of the product within the built environment will determine the

prevalent spectral type of radiant energy as well as typical intensity. In many cases there willeither be a majority of daylight or electric light, see the ‘crossover point’ in subsection 3.4.3.Another classification can be made as to the solar module orientation with respect to

the prevalent radiant energy. In the case of a mobile device (e.g. a wristwatch), it may beimpossible to determine this fully [36]. However, for those products that have a relativelyfixed orientation such as by being wall mounted or set in a holster against the window,a typical solar module orientation with respect to the prevalent energy source may bespecified.Given typical radiant energy (see Chapter 3), the mean electrical energy that will be

available to the application may be estimated. As an example, assume that a proposed producthas a solar module active surface of 10 cm2 which on average converts 5% of the incomingradiant energy into electrical charge; let the mean ratio of charge used by the applicationto charge available be 50%. From the figures in Table 3.4, the electrical energy that maybe used by the application in an average month from a typical fluorescent source might be7mWh/month; daylight might deliver 125mWh/month.

Table 1.3 Typical energy consumption of indoor consumer electronic devices as measured by the author

Application description[Reference]

Product usetaxonomy

Main function (orcircuit) taxonomy Voltage (V)

Runningpower (mW)

Mean useper day(min)

Meanenergy perday (mWh)

Mean energyper month(mWh)

Suited to fluorescent sourcein example

Basic arithmeticcalculator

Tool Display 14 001 20 0005 014

Wristwatch Tool Display 15 00005 720 001 021Weighing scales (Maultronic) [210]

Tool Display 30 005 1 01 2

Wireless security key Tool Wireless 120 56 02 02 6Suited to daylight source inexample

Wireless door movementdetector

Sensor Wireless 103 52 1 1 33

Alarm clock Tool Display 15 8 5 2 62Fire sensor Sensor Amplify 90 58 0 2 65Desktop automatic stapler Tool Move 34 578 03 2 72TV/Video remote control Tool Wireless 30 27 5 3 80

PDA Casio PV S250 [229] Tool Compute 32 15 20 5 150Wireless presence detector Sensor Wireless 96 36 1 8 230Door lock remote controller Tool Wireless 30 60 10 10 300Hand-held radio Tool Amplify 12 18 30 12 260PDA Palm Pilot Series 1 Tool Compute 30 75 10 13 400Laser pointer Tool Light 30 66 15 17 500

Table 1.3 (Continued)

Application description[Reference]

Product usetaxonomy

Main function (orcircuit) taxonomy Voltage (V)

Runningpower (mW)

Mean useper day(min)

Meanenergy perday (mWh)

Mean energyper month(mWh)

Pacemaker Tool Amplify 60 07 1440 17 520PC mouse (wired) Tool Amplify 50 75 15 19 560Pager Tool Wireless 15 09 15 22 650White LED hand torch [159] Tool Light 36 72 20 25 750Bicycle lamp (5 LEDs) Tool Light 30 60 30 30 900Alice microrobot Tool Move 30 75 120 35 1040Hearing aid Tool Amplify 13 18 1440 43 1296Wireless mouse (Logitech) [32] Sensor Wireless 32 19 5 45 1360PDA Palm III xe [229] Tool Compute 32 42 20 53 1600Solar torch Tool Light 10 130 30 65 1950Light sensor (bluetooth) Sensor Wireless 24 96 48 77 2320PDA Psion Revo [229] Tool Compute 32 109 20 89 2680Hand lamp Tool Light 20 400 30 200 6000Portable radio Tool Amplify 30 150 240 600 18000Furby (toy) Toy Move 60 600 60 601 18030Translator Tool Compute 30 1800 30 900 27000Mobile phone Tool Wireless 36 720 45 1210 36300Khepera mobile robot Tool Move 50 1250 60 1262 37900Portable personal computer(laptop)

Tool Compute 40 960 120 1929 57900

Hand-held television Tool Amplify 45 2385 60 2395 71900

10 State of the Art

1.4.4 Mean Energy Taxonomy

While there are those for whom power and energy can be considered synonymous, e.g. [34],for IPV design this might be a serious oversight. This is because, to achieve the lowenergy consumption described in subsection 1.4.3, the application should be on standby (notrunning) for the majority of the time. With this in mind, applications must be consideredwith regard to their mean energy consumption and not their power rating when running. Thetypical actual consumption of potential applications is shown in Table 1.3. These are sortedby increasing mean energy consumption.It is noteworthy that many applications that users may intuitively think would have

potential to be powered from ambient energy consume, in their present designs, orders ofmagnitude more energy than is available. A good example of this is the mobile telephone.The latter suffers from the further disadvantage that its outer surface may be covered forthe majority of its useful life. The energy consumption values in Table 1.3 are indicativeand may be reduced by improved design. Other techniques for designing applications withimproved energy feasibility are considered in the ensuing chapters, especially Chapter 7.Table 1.3 forms the basis for comparisons between the various taxonomies. For the

examples shown, no overriding trend can be seen between energy consumption and productuse (subsection 1.4.1) or main function (subsection 1.4.2) taxonomies. However, for theexample mentioned in subsection 1.4.3, it can be seen that the ‘display’ function may bepowered by fluorescent sources; also, wireless applications may be powered by a daylightsource. Those applications for the example mentioned above that might be powered bydaylight sourced energy but not by electrically sourced energy are indicated.In reviewing the mean use per day, it can be seen that the majority of applications function

for less than one hour per day. This implies that they are on standby for the rest of the timeand underlines the importance of standby power or current.

1.5 IPV GAPS IN KNOWLEDGE

While the importance of information gleaned at the outset of a design project cannot be toohighly emphasised (see also subsection 2.3), some will inevitably be lacking. These gaps indesigner knowledge form the directions of this work.

1.5.1 Radiant Energy Available

Many professionals are involved with the radiant energy found in the built space, but few areinterested in the wavelength range collected by solar cells. The most common assessmentsare made in the units of a standard human eye (photometric or lux). Other analyses coverthe thermal radiation range, for human comfort, for example. Radiant energy in the builtspace to which solar cells are sensitive is therefore characterised in Chapter 3.

1.5.2 PV Solar Cells

The literature supporting the technical understanding of how PV solar cells work is usuallycouched in the terms of physics. This may discourage some designers from exploring

Conclusion 11

the parameters used to characterise PV and ultimately appreciating what opportunities ofperformance improvement are possible, especially for indoor products. Both of these issuesare addressed in Chapter 4.When collecting information on solar modules from suppliers, there is a noticeable dearth

of comparable data of performance at indoor light intensities and spectra. This is thereforeinvestigated in Chapter 5. A corollary of the lack of industrial information is that thereis little research interest in characterising photovoltaics within the ranges of intensity andspectra found indoors. This is reflected in the lack of modelling in this area. Models aretherefore developed in Chapter 5 that match the measured performance of 21 solar cellsrepresenting eight different photovoltaic technologies.

1.5.3 Charge Storage

Most renewable energy systems are not practical without some storage of the energy. In IPVthis is in the form of charge storage. Given that this element may be the most expensive,voluminous and heavy within the IPV energy system, in Chapter 6 a number of ways ofensuring that it is correctly specified are presented, including a model of the relationshipbetween storage cost and capacity required.

1.5.4 Energy Source Guidelines

Finally, in Chapter 7, the lack of a practical guide to ambient energy source design isaddressed.

1.5.5 Applications

Although nanowatt standby applications are technically feasible, such as regulators andmicrocontrollers (e.g. [162]), their emergence on the market is limited.

1.6 CONCLUSION

This chapter opened by considering the low-power energy systems that are available, withspecial focus on ambient energy systems (see section 1.2). This section shows the advantagesof indoor photovoltaic solar cells for powering such systems. The next section (section 1.3)therefore assesses what intellectual property already exists in the area of indoor photovoltaicsand concludes that in general there is little protection. However, for the specific casesmentioned, more patent research would be necessary once the details of the proposed productwere known.With this is mind, in section 1.4, the available photovoltaic products are investigated and

catagorised into a number of taxonomies. These focus chiefly on the product with respectto its function, use pattern and consumption. The product environment is also considered interms of the ambient energy that will be available to power the device.In the final section (section 1.5) the boundaries of available information to the engineering

designer are set. The way in which the following chapters contribute to filling these gaps issummarised.

2Engineering Design

2.1 INTRODUCTION

The majority of this book is given over to ensuring the IPV product will conform withtechnical requirements. However, these tasks are not performed in abstracto; the successof a product development project is unlikely to rest solely on technical issues. Many otheraspects can cause failure and some of these are reviewed in this chapter.The overview of the essentials of design (section 2.2) shows the breadth of definitions

associated with the word, as well as how it is used in this case. The generic components ofthe so defined design process are then presented, followed by a more detailed and thereforemore satisfying representation of the engineering design process.Within this context, the importance of the early stages of the design process is identified

with respect to typical trends in design (see section 2.3). This indicates the contribution thatmay be brought to an indoor photovoltaic (IPV) product designer by providing guidance,ideally from the outset.One design trend is that of increasing consumer concern with the environment. It may be

that this will motivate the further use of ambient energy sources. A useful way to deal withthis issue from the designer perspective is to use some form of life cycle analysis (LCA). Insection 2.4, LCA is defined and those aspects that may support the designer are explained.

2.2 DEFINING DESIGN

The word design is virtually indefinable without reference to the context in which it is usedor the ‘product’ that is envisaged. Take an architect, an engineer, a legislator and a policymaker and probably the only consensus that will be found among them is that all considerthemselves to be designers (of a kind!). One of the reasons for this is that human beingshave been resolving problems that form part of the design process for millennia.


14 Engineering Design

The French translation for engineering design is ‘conception’. This word comes from theLatin ‘capere’ (to catch) and ‘con’ (meaning together) – one might say to gather. Design istherefore a fusion of a number of elements. Where this definition falls short, however, isthat the elements may be abstract (mental) and the resulting concretisation is new.One solution is to be more specific by the juxtaposition of an adjective such as ‘engi-

neering’ in front of the word design. In this book, the word design is understood to meanengineering design.Another way to think of design is as a number of interconnected problem resolution

cycles, such as the simplified process in Figure 2.1.A generally recognised categorisation [188] [87] of such cycles for engineering design

finds four phases which may be referred to as analysis of the problem, conceptualdesign, embodiment and detailing. A brief description of a design project might read asfollows.

1. Analysis of the problem is the step that takes a recognised need and converts it intoa formal definition of the problem including boundaries and goals of the possiblesolution(s).

2. This is followed by conceptual design in which broad solutions or schemes areprepared.

3. At the embodiment stage, the schemes are further developed and may result in prototypingor some other result that will allow selection of the ideal candidate.

4. The latter then undergoes detailing which includes the many but essential points thatmust be decided upon for production, for example.

What this description fails to highlight is that design never follows such a simple sequencewithout some (and usually much!) iteration. This is because not all problems encounteredcan be resolved without reviewing and possibly changing the work of a previous stage. Forthis reason and for its greater completeness, the description of engineering design proposedby VDI directive 2221 [254] (see Figure 2.2) is more comprehensive. This diagram includesthe above-mentioned phases on the left-hand column.Figure 2.2 brings out both the iterative nature of design (the ‘forwards and backwards’

through the stages mentioned in the leftmost box) and the dynamic nature of design (the‘fulfil and adapt requirements’ mentioned in the rightmost box). It also includes certainstages (boxes numbered 1 to 7) that in practice are often neglected. One of these is box 2(‘determine functions and their structure’).

Cycle 1

Problem

Idea

Problem

Resolution

Concretisation

Resolution

Cycle 2

Figure 2.1 Simplified schema of a monolinear design process showing two problem resolutionmodules or cycles

Defining Design 15

Stages

Task

Clarify and definethe task

1

2

3

4

5

6

7

Determine functionsand their structure

Search for solution principlesand their combinations

Divide into realisablemodules

Develop layout ofkey modules

Iter

ate

forw

ards

and

bac

kwar

ds b

etw

een

prev

ious

and

fol

low

ing

stag

es

Complete overalllayout

Prepare production andoperating instructions

Further realisation

Product documents

Definitive layouts

Preliminary layouts

Module structures

Solution principles

Function structures

Specification

Phases I

PhaseResults

Phases II

Fulf

il an

d ad

apt r

equi

rem

ents

Phases III

Phases IV

Figure 2.2 Generic engineering design process [18]. English translation available in [188] and[255], reproduced with permission of Verein Deutscher Ingenieure e. V.

A function in this context can be defined by answering the question ‘What does it do?’with one or more couples of an action verb plus a substantive. Examples of such couplesinclude ‘collect energy’ and ‘store charge’.Such functions for products will exhibit both usable and aesthetic functions. Usable

functions satisfy requirements such as ease of use, effectiveness and product longevity.Aesthetic functions relate to appearance and the impression a product gives. Dependingon the kind of product being designed, the share of these two kinds of function variesfrom capital investment (which is usable function intensive) to jewellery (which is aestheticfunction intensive), as shown in Figure 2.3.For a designer, functional analysis is a necessary precursor for a cost–function analysis

both of the design alternatives and competitor products where these exist. This is based oncomparing the cost of each design component with the functions that it can perform. Thegoal of such an analysis is to satisfy the required functionality with least cost.


0

20

40

60

80

100

Prop

ortio

n of

func

tions

(%

)

Capitalinvestment

Consumerproduct

Fashion orjewelery

Aesthetic functions

Usable functions

Figure 2.3 Examples of product share of usable and aesthetic functions, adapted from [224]

2.3 TRENDS IN ENGINEERING DESIGN

Few authors when discussing the product development process fail to refer to the dispro-portionate amount of project costs, say 60–85%, which are fixed in the first stage(s) of theengineering design process. They will also recognise the miserly proportion of costs incurredin these first stages, maybe 5% (see Figure 2.6). They therefore argue that, the earlier theproblems can be prevented, the less the projects will cost. While this explanation is correct,it is incomplete. A further trend is the cost of failure which grows exponentially as theproduct development unfolds (see Figure 2.4).

Fault prevention Fault detection

Procurementand production

Developmentand planning

Plan

ning

Dev

elop

men

t

Cos

t per

fau

lt

Prod

uctio

n sc

hedu

ling

Man

ufac

ture

Fina

l tes

ting

Cus

tom

er

100

10

10.1

Figure 2.4 Failure cost with product development progress [62], reproduced with permission ofHMSO (Her Majesty’s Stationery Office)

Trends in Engineering Design 17

Such costs clearly damage profits and may become business fatal owing to ever increasingproduct liability. The latter, although prevalent principally in the United States, is alsoincreasing in Europe (see the case in the United Kingdom, Figure 2.5).However, while the importance of design is not in doubt, the problem is that knowledge is

usually just as scarce as is it necessary at the start of the new product cycle, thus leaving thedesigner to deal with what Fabrycky [76] refers to as the ‘knowledge gap’ (see Figure 2.6).

1970 1980 1990 2000

Year

0

1

2

3

4

5

Prod

uct l

iabi

lity

cost

s (b

illio

n or

109 G

BP)

Figure 2.5 Product liability costs and prediction for the UK (based on [243])

Ease of change

Knowledge

Kno

wle

dge

gap

Kno

wle

dge

gap

Cost incurred

Cost/technicalcommitment

Conceptual/preliminary

design

Detaildesign and

development

Constructionand/or

production

System use,phaseout and disposal

Contribution

100%

85%

60%

5%

Need

Figure 2.6 Knowledge gap and other trends of the product development process [22], reproducedwith permission of Butterworth-Heinemann


This is made up of lack of vision as to what the real issues are, as well as salient data.This book contributes to bridging this gap for a specific product type: autonomous (self-recharging) energy supplies based on ambient energy. The example taken focuses on theuse of photovoltaic solar cells to convert indoor radiant energy into charge which is thentypically stored for later use. The proposed system would typically replace one or moresingle-use batteries.

2.4 LIFE CYCLE METHODS

2.4.1 Introduction

This section briefly explains what life cycle assessment (LCA) is and how it may beused to support the practising IPV product designer; associated suggestions for productimprovements are also provided.Evidence from a number of LCA studies is presented that coincides in indicating that the

main environmental issue of small electronic devices is the battery, in particular single-use(or primary) batteries. This lends support to the case for ambient energy, the principlesof which seek to aid designers to avoid primary batteries by the use of limited secondary(rechargeable) storage, for example. Other design enhancements that can accrue from LCAare also mentioned.

2.4.2 Definition

LCA is also referred to as eco-balancing or cradle to grave assessment. It is the assessmentof environmental performance of a product or activity. This performance is based on aninventory of the product or activity inputs and outputs within a defined system boundary,such as the example in Figure 2.7. For IPV products, this will lie between their conceptionand eradication, although more limited time frame studies may also be considered.The impact assessment is based on the inventory. For example, the energy required

for product manufacture may be calculated by integrating for each product component theweight (kg) and the related effect energy (e.g. MJ/kg). The result may be taken as beingproportional to part of the environmental impact. Values for these effects and others may be

Energy Energy Energy Energy Energy

Rawmaterials

acquisitionMaterials

manufactureProduct

manufactureProduct use

or consumption

Fianl disposition-landfill, combustion

recycle, or reuse

Wastes Wastes Wastes Wastes

Reuse

product recycling

Figure 2.7 Simplified product Life flow chart [86], reproduced with permission of FranklinAssociates Ltd

Life Cycle Methods 19

Framework

DirectApplicationsGoal & Scope

definition

Inventory Analysis Interpretation

Impact Assessement

• Product development, improvement

• Strategic planning

• Marketing

• Other

Figure 2.8 Phases of an LCA, based on ISO 14040 [40]; diagram [49] reproduced with permissionof CIT Ekologik

found in databases such as those mentioned in the UNEP world review [179]. The carbondioxide emissions, ozone layer damage and so on can thus be estimated.The fourth stage of an LCA is interpretation (previously improvement analysis). The

four phases are summarised by the flow chart in Figure 2.8. Note the resemblance with thedesign process in Figure 2.2, especially with regard to the iterations.

2.4.3 LCA for IPV Designers

The effort required for complex LCA implies that it is ‘not a solution for the busy productdesigner, who has many other design considerations to deal with. There are, however, impor-tant lessons to be learned’ [171]. The process of understanding an LCA builds awareness ofthe complexity of the environmental issues at stake. While LCA helps the product designerto appreciate that there are no simple answers, LCA can contribute to the decision-makingprocess by indicating where the major issues lie.Further limitations of LCA are enumerated in ISO 14040 [40], including difficulties with:

• determining the boundaries and assumptions;• transferability of LCAs between domains and geographies;• statistical issues related to the variety of input data.

Given the concerns with full LCA for the product designer, more limited approaches havebeen proposed such as ALCA (Abridged LCA) used by Motorola. Research in the areaof simplified LCA is on-going [208]. Such methods allow more usable information to beavailable earlier in the design process which it is recognised increases the information value(see section 2.3).Benefits of the LCA perspective can ‘boost creativity’ [25] and alert the designer to

other concerns, such as legal issues. An example of the latter is that primary and secondarybatteries are regulated differently depending on the country. For example, in the UnitedStates ‘Alkaline (primary) batteries are not listed as hazardous waste under the Resource


Conservation and Recovery Act (RCRA)’ [72] while NiCd (secondary) batteries are [137]. Inanother example, Switzerland collects over 60% [33] of its primary and secondary batterieson a voluntary basis, and recycles both kinds of battery [12].A further benefit is that LCA can provide formal methodologies in which to compare

different design solutions, with the proviso that the absolute figures should be treated withcaution.LCA may therefore support the European Commission goal of ‘more environmentally

friendly product design’, as mentioned in the framework description of the integrated productpolicy [39].

2.4.4 LCA in IPV Design

In practice, despite the variety of products upon which an IPV designer may work, con-sideration may be given to a weight assessment, limited effects assessment for energy andemissions, the use cycle and a number of ways to make improvements.

2.4.4.1 Product weight assessment

For IPV products, as for most mobile electronic devices, the charge storage device typicallyrepresents a significant fraction of the global weight (e.g. 1/3) and is therefore important ifwaste minimisation is a goal. Batteries contribute 1% of the weight of United States refuseand significant toxic waste (see subsection 6.2.2). Therefore, the designer should includethe battery in environmental impact calculations of such products, despite the fact that somereports do not do so, e.g. [182].

2.4.4.2 Energy and emissions

Given detailed weight data, the energy and emissions of different scenarios may be compared.The boundaries of the assessment should be defined, such as considering for IPV only theproduction energy of the batteries. It has been shown that the production energy of a single-use (Zn-alkaline) battery is around 5MJ, and for a rechargeable R6 or AA (NiCd) cell ofsimilar size it is 6MJ [167].Typically, LCA of IPV products will show that battery storage is a major contributor with

regard to most energy and emissions metrics [91].Battery impact can be confirmed indirectly with respect to another main component of

the IPV power solution: the solar module. Making the reasonable hypothesis that a thin filmtechnology (see section 4.6) will be used, this will represent only a small weight increase ofthe total product. Also, the PV module will not have a significant impact on energies as thebulk of its weight (around 90%) is made up of glass which has a low (primary) energy of14MJ/kg compared with the photovoltaic material, the (primary) energy of which is around100MJ/kg [268]. Therefore, a typical IPV solar module of 20 g has a primary energy ofaround 0.4MJ, i.e. an order of magnitude less than the batteries. Polymer or organic substratesolar modules (such as Flexcell produced by VHF Technologies, Switzerland) are lighter,and therefore represent even less waste.

Life Cycle Methods 21

2.4.4.3 Use phase

It was seen above that the primary or production energy of the rechargeable cell power solu-tion is higher than a single-use cell. This is to be expected because such an LCA excludesthe product use cycle and therefore the advantages of rechargeable storage. As Lankey putsit, ‘We find that materials use, energy use, and emissions can be quantified over the entireproduct life cycle to quantitatively show that resource use and emissions are substantiallylower if a rechargeable battery can be substituted for a primary battery. However, consumeruse patterns will affect the relative environmental benefits of rechargeable batteries’ [135].An example of ‘bad’ consumer use patterns is leaving the battery recharger plugged intothe mains when not in use: even though no charging is occurring, the charge transformerdraws current which is wasted. If such ‘bad habits’ could be avoided, daily energy con-sumption savings of 25% for mobile telephones and 80% for wireless telephones wouldaccrue [82].A further life cycle comparison of batteries (both primary or secondary and fuel cells)

indicates that the best alternative would be rechargeable polymer exchange membrane fuelcells [83]. However, as mentioned in subsection 6.3.7, such cells, especially at the scale ofIPV applications, are not market available.The finding that over 40% of (financial) life cycle costs of another kind of stand-alone

outdoor PV system are due to the batteries [23] is understandable. Battery manufacturerscan contribute by producing products whose life cycle costs are lower.

2.4.4.4 Improving IPV product LCA

The Eco-design Guide [66] recognises eight ways of improving a products eco-performancewhich are:

• selection of low-impact materials;• reduction in materials usage;• optimisation of production techniques;• optimisation of the logistics system;• reduction of impact during use;• optimisation of the initial life stage – design for upgrade and reuse;• optimisation of the end-of-life system;• new concept development.

Some of the ways that the IPV designer can make improvements have been mentionedin the paragraphs above. Further approaches include labelling the application to encouragerecycling and proper disposal, using labels of material content to aid battery sorting, designingproducts so that batteries can easily be removed and taking into account the level of recyclingavailable for certain materials such as lithium [150].Consumer choice may also be influenced prepurchase by eco-labels. Examples can be

found in electricity supply [216], to ‘promote products which have a reduced environmentalimpact compared with other products in the same product group’ in Europe with the use ofthe ‘daisy’ label [67] (see Figure 2.9), the United States energy star label used at present forelectrical and electronic goods [263] and lastly in construction [143].


Figure 2.9 Examples of ‘eco-labels’. European ‘daisy’ label (left), United States energy star (right)

2.4.5 Designer Responsibility

The importance of product design is reflected in the fact that the equivalent of ‘about 1.5%of Germany’s total carbon dioxide emissions in 1995’ were necessary to power consumerelectronic devices that the user had ostensibly turned off [93]. For example, a PC andperipherals consume 20W when turned off but are connected to the mains [141]. Giventhat the same equipment consumes around 300W in function [61], then it must run forover 1.5 h/day in order for the energy consumed in function to exceed that consumeddaily in standby. This problem may be associated with ‘user’ bad habits, as mentioned insubsection 2.4.4, but the designer should also accept partial responsibility for the typical useof their products. This is for three reasons: firstly, only a minority of users are fully aware oftheir environmental impact; secondly, a number of technical solutions exist such as puttingthe transformer on reduced current mode when it is not in use; thirdly, legislation can beparlously sluggish. The transformer standby current issue was recognised by the EuropeanCommission in 1991 [181], yet the code of conduct [60] that sets the ‘no-load powerconsumption’ target of 1W by 1 January 2001 took almost a decade. So far, 15 companieshave signed this agreement [74]. Motorola has a (mobile phone) charger that now runs at0.12W in standby [195].

2.4.6 Summary

LCA is a tool that can serve the IPV designer by indicating the environmental burden ofdifferent choices. A typical example is to check whether the energy saved in the use phaseand at end of life exceeds the extra environmental cost during the production phase. Ingeneral this can be expected to be the case for IPV products when compared with a similarproduct that uses primary batteries.

2.5 CONCLUSION

In this chapter, the product engineering design process is explained. It is shown that what isknown at the early stages has a significant influence on the final design.Another influence on the designer, not least for ambient energy sources, is environmental

concerns. Ways in which these can be handled are presented in subsection 2.4.

3Radiant Energy Indoors

3.1 INTRODUCTION

One of the first logical steps in ambient energy system design is to establish the availableenergy resources. This chapter takes the example of photovoltaic solar cells indoors, andaims to characterise the energy whose wavelength lies between ultraviolet and infrared onthe electromagnetic spectrum that solar cells can collect. This energy is governed by thebuilding physics parameters presented in section 3.2.Such a characterisation is necessary owing to the inability of the human eye fully to

perceive the radiant energy that a solar cell collects. Much of the previous indoor environmentwork cannot be applied directly as it is in the photometric domain (sensitivity of the humaneye), not the radiometric domain. Photometric findings nevertheless provide qualitativeinformation which is investigated in section 3.3.To allow a quantitative assessment, a number of radiometric experiments were carried

out. These consider the obstacles and window transmissivity as well as the PV productorientation and location (see section 3.4).Radiant energy is affected by many more variables including time and geographical

location, for example. A simulation can be used to confirm calculations, but few suchprograms are applicable to the radiometric characterisation of the built space (as most arein the photometric domain). A ray-tracing approach that adheres to the laws of physics istherefore presented in section 3.5.A summary of the various parameters and their importance is covered in the discussion

(section 3.6). The ways in which IPV design can be optimised are also covered in thissection.As product applications may be permanently fixed to the walls or ceiling, particular

attention will be given to these areas of the indoor space. It will nevertheless be possible toinfer the overall energy distribution from what follows.


Malestrom

enjoy!!!

24 Radiant Energy Indoors

3.2 PHYSICS OF BUILDINGS

While a great number of variables can be used to characterise radiant energy, those associatedwith irradiance (see subsection 3.2.1) are usually the most significant for indoor photovoltaicpurposes, as this parameter can be expected to vary over the greatest range. The radiantenergy spectra (see subsection 3.2.2), in comparison, usually have a smaller impact. Theseparameters, while absolute in theory, interact in practice with the built space, in particularwith respect to optical parameters such as those described in subsection 3.2.3.

3.2.1 Radiant Energy

The fundamental unit of radiant energy in the spectral range of interest for IPV is the photon.Energy and frequency are related by the Planck equation:

EQ =hcf (3.1)

where

EQ = quantum energy (J)hc = Planck’s constant = 663× 10−34 �J s�f = frequency (Hz)

and

f = c

w�Hz� (3.2)

where

c= speed of a photon in a vacuum = 299× 108 �m/s�w= wavelength (m)

Taking the wavelength range 0.4–0.8�m on the electromagnetic spectrum (see Figure 3.1)gives an energy range for the quantum between 5× 10−19 J (blue) and 2× 10−19 J (red). This

108

700 600 500 400

109 1010 1011 1012 1013 1014 1016 1017 1018 1019 1020 1021

10 1 10–1 10–2 10–3 10–4 10–5 10–6 10–7 10–8 10–9 10–10 10–11 10–12 10–13

Microwave

Infrared X-rays

UV Gamma rays

Frequencies (Hz)

Red Orange Yellow Green Blue Violet

Wavelength (nm)

TV

Wavelength (m)

1015

Figure 3.1 Electromagnetic spectra

Physics of Buildings 25

infinitesimal amount of energy is required in significant quantities for IPV applications. It isalso evident that an ultraviolet photon may carry 3 times the energy of an infrared photon.Once large quantities of photons are available, they can be described in relation to a

number of ‘processes’. Following a typical photon path for ease of understanding, four suchprocesses can be identified:

1. Light Creation. The rate at which radiant energy is output from a point source is calledthe radiant intensity (W/sr). Many radiant energy source volumes can be modelled aspoint sources at sufficient distance; this is the case for a spherical source such as the sun,as the Earth’s orbit �150× 1011 m� is three orders of magnitude greater than the solarradius �696× 108 m�. A common IPV case is the window of a room. From the indoors,this may be considered as a planar radiant intensity source made up of many point sources.

2. Translation. The resulting field formed by the radiant intensity is called the radiantflux or (radiant) power and is measured in W (or J/s). Such flux can be associated withemission or transmission.

3. Incidence. If the radiant flux reaches anobject, then the total received radiant flux (or power)per unit area is the irradiance �W/m2�. For a flat surface, this is analogous to the radiant fluxin a particular direction. In the jargon, irradiance may be referred to as radiant flux density.

4. Reflection. If this object reflects, the resulting radiant flux per steradian is referred to asradiance �W/�m2 sr��. For the special case of a uniform radiant flux surface, radianceis constant independently of the distance from it. This is because the angle of a spheresubtended by a steradian is constant.

These four processes are summarised in simplified form in Figure 3.2.Further physical concepts such as reflection, absorption and transmission are presented in

subsection 3.2.3.

3.2.1.1 IPV irradiance

For IPV, of interest is the irradiance on solar cells from radiant energy sources whose volumecan be approximated to a point (e.g. the sun), a line (e.g. a fluorescent tube) or a surface(e.g. the window on a cloudy day).

RadiantIntensity

RadiantFlux

Irradiance Radiance

W/sr W W/m2 W/(m2 sr)

‘‘LightCreation’’

‘‘Translation’’ ‘‘Incidence’’ ‘‘Reflection’’

r r

Figure 3.2 Schema of radiant energy ‘processes’


Generally, as the distance between the emitter and receiver changes, so does the irradiance(received). Such irradiance, Irps, from a point source emitting radiant energy with intensityIps at a distance r is related by the inverse square law

Irps =�constant�Ips

r2(3.3)

However, if the radiant energy source volume can be approximated to a line source ofinfinite length and of intensity Ils, then the irradiance Irls will vary with the reciprocal of thedistance, not the inverse square:

Irls =�constant�Ils

r(3.4)

By corollary, if a further spatial dimension is added to the radiant energy source such thatit has a surface of infinite size and constant intensity, the irradiance is then independent ofthe distance from the surface. A window is a good example so long as the solar module issuitably close and oriented towards the window.Another mechanism with which irradiance varies is the cosine law

Irpa = Ipa cos� (3.5)

This is presented in Figure 3.3, which shows the following: when radiant energy is incidenton a surface with a non-zero angle, �, to the normal, the irradiance Irpa is spread over a largersurface, and therefore the intensity at a point on the surface is lessened by the cosine of �.The incident radiant flux in this case is assumed to be a parallel beam, identified as Ipa.

Such an assumption becomes valid once the distance between the source and surface ismuch greater than the source radius (for the case of a spherically shaped source). For IPV,a minimum distance/radius ratio of 10 is usually appropriate.

3.2.1.2 Daylight

The most significant source of radiant energy for planet Earth is the Sun. Outside theatmosphere, the average intensity received in recent decades has been 1366± 15W/m2

(see Figure 3.4). The slight variation in annual average has a period of approximately 11years [271]. A standard for such solar radiation is AM0 described in [7].

Ipa

Irpaφ

Figure 3.3 Cosine law


Year

1363

1364

1365

1366

1367

1368

1369

78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04

Year

1363

1364

1365

1366

1367

1368

1369

Sola

r Ir

radi

ance

(W

/m2 )

0 2000 4000 6000 8000

HF

AC

RIM

I

HF

AC

RIM

I

HF

AC

RIM

II

VIR

GO

0.1%

Difference between minima: –0.020 ±0.007 W/m2

Figure 3.4 Total solar irradiance daily values from different originating experiments; on-line [89]and published [88], reproduced with kind permission of Springer Science and Business Media

Owing to the relative rotation of the Earth and the Sun, the irradiance on Earth is variablein both diurnal–nocturnal and annual cycles. This results in a preferential orientation of theirradiance both with respect to the azimuth angle (or compass bearing) and the elevation(angle above the horizon). Poor orientation about the vertical can reduce annual irradianceby approximately 75% of the ideal [253].The atmosphere absorbs a part of the global (synonymous with total) radiant energy so

that the vector received on Earth is made up of two components: the direct and indirectradiations. For a cloudless sky, the direct instantaneous radiation incident perpendicular toa horizontal surface will reach a maximum of around 1000W/m2, while, as the weatherworsens, the radiation will be more indirect. In European locations such as Switzerland,daytime terrestrial instantaneous irradiance perpendicular to a horizontal surface in the range200–600W/m2 can be expected.Considering this radiation in cumulative terms, an ideally positioned outdoor application

in a European location facing due south (0� azimuth for architects) and having a 45� elevationwill receive a cumulative daily irradiance in the range 1000–5000W h/m2 day [247].In practice, a more complete set of values for a particular location and orientation can be

obtained from a local meteorological station (see Figure 3.5) or an interpolation model ofsolar radiation such as Meternorm [156]. Also, a number of standard skies [48] have beenspecified that allow laboratory comparisons to be made [160]. Typical standard skies includeovercast conditions (deep stratus clouds and sun disc invisible from the ground) and clear(cloudless) conditions.


0

50

100

150

200

250

J M A M J J A S O N D

Month

Cum

ulat

ive

irra

dian

ce(k

Wh

/m2

mon

th)

0

300

600

900

Cum

ulat

ive

irra

dian

ce(M

J/m

2m

onth

)

F

Figure 3.5 Cumulative irradiance incident perpendicular to a horizontal surface at Payerne, CH,1998 (courtesy of World Radiation Center, Davos)

3.2.2 Radiant Energy Spectra

Light can be obtained from both natural sources (e.g. sunlight, lightning, bioluminescenceand pyroluminescence) and artificial sources (e.g. incandescence and luminescence). Lightenergy in the IPV environment can be categorised in the same way. Artificial light canbe considered in the majority of IPV cases as synonymous with electrically poweredlight.

3.2.2.1 Thermal radiation

Each source can be identified by its spectral signature. Figure 3.6 shows how (natural)sunlight reaching the Earth’s atmosphere (AM0) is then filtered by Rayleigh scatteringat some wavelengths more than others before reaching sea level (AM1). The resultingabsorption troughs are associated with particular molecules that attenuate mostly above700 nm (red). This explains why sunlight varies in colour with latitude, as the light mustpass through a greater distance of atmosphere to reach the ground at the poles than at theequator. For example, areas north of the fortieth parallel at midday therefore receive a bluersunlight than at the equator.From Figure 3.6 it can be seen that a surface just outside the Earth’s atmosphere (AM0)

receives an irradiance (spectrum) similar to that produced by a black body at 5760K. Anideal black body absorbs all radiation in the electromagnetic spectrum incident on it withoutreflecting or transmitting this energy. This typically raises its temperature with respect to thenon-illuminated surroundings. To return to equilibrium, it radiates energy at a power Pr (W)with a spectral distribution that Planck found to be fixed (see Figure 3.7). Stefan discoveredthat this distribution is related to temperature as follows:

Pr = sAeT 4 (3.6)


Figure 3.6 Spectra of natural light with wavelength in micrometres, based on [189]

where

s= Stefan–Boltzmann constant �567× 10−8W/m2 K4�A= surface area of the object �m2�e= emissivity constantT = temperature (K)

The radiant energy sources of interest to IPV that can be approximated by a black body theoret-ically require a temperature of at least 3000K, such as daylight and incandescent light sources(see Figure 3.7), in order to emit with a peak in the spectral response of PV (see Figure 4.21).Given that indoor lit objects will absorb radiant energy, their resulting temperature can be

used to approximate the wavelengths at which they re-emit. As can be seen in Figure 3.7 forobjects at room temperature (300K), peak emission is around 10�m which is much longerthan the typical PV spectral response. The process of absorption and re-emission equatesto a wavelength (lengthening) shift of the radiant energy. Fortunately, for IPV purposes, aswill be seen in subsection 3.2.3, objects located indoors that receive radiant energy do notonly re-emit but also reflect and transmit.

3.2.2.2 Other radiation spectra types

Not all indoor light sources have black body type characteristics. Another common spectrumtype is fluorescent (see Figure 3.8). Similarly to shop-sign neon lamps, such sources function


108

106

104

102

10–4

10–2

1.0

0.1 101.0 100

Wavelength (µm)

Spec

tral

rad

iant

exi

tanc

e W

/m2

µm

Visible radiant energy band

Approximate suntemperatureat the surface

0 °C

Incandescent lamp3000 K approx.

6000

K

2000 K

1000 K

800 K

400 K

273 K

600 K

200

K

4000 K

Figure 3.7 Spectral distribution of energy radiated from black bodies for a range of temperatures,based on [262] and reproduced with permission of Prof. J. Watson

0

2

4

6

8

10

12

14

16

18

380 420 460 500 540 580 620 660 700 740 780

Wavelength (nm)

Arb

itrar

y un

its

Figure 3.8 Typical spectrum of an indoor fluorescent tube (author’s measurements)

by the ionisation of mercury in an inert gas (such as argon) at low pressure. The mercuryre-emits this energy in the form of ultraviolet light which in turn excites phosphor on theinner surface of the lamp tube. As a result, the phosphor fluoresces in the visible range witha spectrum that is generally bluer than incandescent bulbs.


It is of note for IPV that fluorescent sources lose 9–30% of their luminous flux in the first10 000 h of use [187].Many other artificial light source types exist that are not dealt with here as they are less

common in the indoor environment. These include discharge lamps (high-intensity sources,carbon arc, flash), semiconductor light-emitting diodes (LED) and coherent sources (laser). Itis likely that LEDs will be increasingly used in buildings of the future as they have relativelylow volume and voltage, higher energy efficiency and a wide choice of intrinsic colours.

3.2.3 Basic Optical Parameters

Quantum models as mentioned in subsection 3.2.1 are one way to describe radiant energy.Another relatively older approach based on wave theory is also useful for understandingmaterial properties indoors. A simple example that summarises a number of effects is shownin Figure 3.9 in which radiated energy passes from a transparent medium (e.g. air) througha second transparent medium (e.g. glass) and back out to the first transparent medium. Anexample of this is glass in air, such as for window applications. The arrows in the diagramare drawn with different thicknesses to indicate typical intensity.As can be seen, the incident radiant energy (IRE) undergoes reflection, absorption and

transmission. For a point on the top surface of the second material in Figure 3.9, the IRE iseither reflected or penetrates the layer. The IRE that penetrates is refracted which indicatesa change in refractive index described by Snell as

RI� sin ��=RI� sin �� (3.7)

where

RI = index of refraction of the materials�= angle of the radiant energy to the normal of the surface� and �= first and second materials respectively (e.g. air and glass)

A glass of increased refractive index may be synonymous with increased reflection and lossof energy.The third property is absorption, usually in the form of heat to the material bulk. Such

absorbed radiant energy can be re-emitted. However, for the temperatures (300K) and

Transmitted radiantenergy

Emitted radiantenergy

Emitted radiantenergy

Incident radiant energy Reflected radiantenergy

e.g. Air

e.g. Air

Absorption e.g. Glass

Figure 3.9 Light passing through transparent materials


Table 3.1 Typical fenestration transmissivity values

Glass window type Single plate glass Double glazing

Transmissivity ⊥ to surface 0.8–0.9 0.5–0.7

Spread Mixed Scattered

DiffuseLambertianSpecular

φ φ

Figure 3.10 Reflection types

Light incident angle from normal to glass plane0° 10° 20° 30° 40° 50° 60° 70° 80° 90°

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

Rat

io o

f lig

ht r

efle

cted

to li

ght i

ncid

ent

56°40′

(a)(b)

(c)

θ1 =

Figure 3.11 Ratio of light reflected to light incident as a function of the angle of incidence on glasswhere �a�=R⊥� �c�=R and (b) is an equal mix of perpendicular and parallel light [24], reproduced

with permission of CUP

Photometric Characterisation 33

Table 3.2 Typical reflectivity values found indoors [142][47]

Room surface Walls Floor Ceiling Working planes

Reflectivity, R 0.3–0.6 0.2–0.5 0.5–0.8 0.2–0.6

materials (glass, air) of the applications considered in this work, the emitted energy is bothslight and has a spectral distribution well away from the visible (see Figure 3.7).The radiant energy that exits the bottom surface of the second material in Figure 3.9 does

so with a reduced energy to the IRE. The ratio of transmitted radiant energy (TRE) to theIRE is called the transmissivity ratio, RT:

RT =TREIRE

(3.8)

Similar ratios can be found for absorption and reflection.Typical transmissivity values for window glass are shown in Table 3.1. In practice, the

transmissivity of windows may often be significantly lower �<02� owing to films or tinting.Assuming the surface is specular, the IRE at an angle to the normal is reflected at the same

angle to the normal (see top left of Figure 3.10). However, the proportion of light reflected isproportional to the angle of incidence (see Figure 3.11). In practice, the majority of surfacesare not specular and exhibit further reflection types such as those shown in Figure 3.10. Themost common type in practice is the mixed reflection.The range of reflectivity values that can be expected for IPV is shown in Table 3.2.

3.3 PHOTOMETRIC CHARACTERISATION

Most indoor light characterisations are photometric (synonymous for IPV purposes withphotopic), i.e. they measure only that portion of the electromagnetic spectrum to which thehuman eye is sensitive (visible spectrum 380–750 nm or 400–789GHz) and have the unitlux. As this falls within the spectrum of sensitivity of photovoltaic cells (see Figure 3.12),for qualitative purposes the photometric measure may be used as a representative sample

100%

80%

60%

40%

20%

0%300 400 500 600 700 800

Wavelength (nanometres)

1.0

0.8

0.6

0.4

0.2

0.0400 600 800 1000 1200

160

120

80

40

0

Wavelength (nm)Daylight spectrum

Rel

ativ

e sp

ectr

al r

espo

nse

(A/W

)

CdTea-Si monoX-Si CIS

Irra

dian

ce (

W/m

2 )

Rel

ativ

e re

spon

sivi

ty

Figure 3.12 CIE standard photometric response (left) [238], reproduced with permission of FrankBason, SolData Instruments, Denmark, and some typical solar cell spectral responses (right) based

on [176]. For abbreviations of the latter, see Figure 4.16


of the spectrum of interest for radiant energy characterisation (see section 3.4) in which theunit of radiant energy is W/m2.The Commission Internationale d’Eclairage (CIE) have standardised an average human

eye response to daylight with respect to wavelength, as shown in Figure 3.12, with a peak at555 nm (green). Real human eyes cannot be applied as reliable optical sensors since they donot respect this curve exactly. As the nineteenth-century physicist and physiologist Hermannvon Helmholtz categorically put it: ‘If anyone offered me an optical device with such seriousflaws, I would reject it in no uncertain terms.’ [104].

3.3.1 Characterisation Methodology Issues

A number of approaches for assessing indoor radiant energy can be applied, some developedby architects and lighting designers. These, which include simplified methods, in situ mea-surements, scale modelling and computer simulation, are described with their advantages anddisadvantages elsewhere [160]. Key parameters for IPV needs are the daylight factor (subsec-tion 3.3.2), the aspect ratio (subsection 3.3.3) and the glazing/floor ratio (subsection 3.3.4).Each of the methods has specific issues with respect to the characterisation of indoor

radiant energy. For example, simplified methods are applicable to the mass of photometriccalculations required in lighting installations; no such method was found for radiometricmeasurements because of the lesser general need in this area.The in situ approach has the disadvantage of being specific to location, room design

and orientation, weather, time of year and the many other variables that must be taken intoaccount in such global assessments. This approach should therefore be applied for specificrelative comparisons only.The convenience of a scale model is attractive and is used by architects [228]. For radiant

energy simulations for IPV it has a key disadvantage with respect to finding sensors ofthe right scale. Taking a conservative 1:10 ratio for the reduction in room size impliesthat the measurement devices require a surface area of a few square millimetres. This is atleast an order of magnitude below that of commonly available photovoltaic minimodules.Such modules could be scanned over each surface using an x–y table for data collection.The scanning method would alter the wall reflectivity (and therefore the results) less thanpositioning multiple sensors on each wall surface.Lastly, computer simulation (see section 3.5) cannot exactly reproduce reality but has

the advantage of being able to simultaneously deal with sufficient variables. It also offers thebenefit of averaging significant quantities of data (i.e. years of meteorological data), makingthe final result more robust and understandable.

3.3.2 Daylight Factor

Many great architects (including Frank Lloyd Wright and Louis I. Kahn) have describedthe importance of natural light in the built environment. Architects and building physicistsdescribe the effectiveness with which a particular built space transmits natural light indoorsby a parameter called the daylight factor, D (see Figure 3.13). This is the ratio betweenthe light level inside a building, Lint, to the light level outside the building, Lext, incidentperpendicular to a horizontal surface in both cases. The ‘indirect component’ is light thathas undergone reflection.

Photometric Characterisation 35

Interior indirect component Typical DF (%)1086420

Exteriorindirectcomponent

Directcomponent

Lint

Lext

Figure 3.13 Components of the daylight factor, based on [193][213]

The daylight factor is therefore

D= Lint

Lext

(3.9)

where Lint is measured at workplane height, taken (in Europe) as 0.85m above the floor.As Figure 3.13 shows, the reduction in light intensity follows a curve quite similar to the

inverse square law [equation (3.3)] as the distance from the window increases. In this case,the window can be considered as a multitude of point sources, but the points do not all haveequal intensity.Despite its limitations,D provides a useful indication of how little outdoor light is typically

found indoors. Figure 3.14, which compares a number of different window techniques,shows that peak D is around 15%, with an average around 5% or below. As will be seenin section 3.4, similar values can be found for radiant energy. This is significant for IPVbecause the majority of photovoltaic products are designed for outdoor-use where light levelsat least an order of magnitude higher than indoors can be expected.From Figure 3.14 it is clear that D peaks in the vicinity of the window. It is also of

note that light penetration is greater for the same surface of glass with a taller window;

D(%)10

05

D(%)10

05

Ouvertures en Toiture

Ouvertures en FaçadeD

(%)10

05

D(%)10

0

5

D(%)10

05

D(%)10

05

Figure 3.14 Typical daylight factor values for a number of window configurations in the facade orroof (toiture) [193], reproduced with permission of OFQL


this is indicated by the increased D of the top right diagram compared with the case at thetop left.

3.3.3 Nearby Obstacle Aspect Ratio

Architects are also concerned with the impact of neighbouring structures, to which they referin terms of the aspect ratio (AR) of a particular ‘valley’ created by buildings on two sidesof a street, for example (see Figure 3.15). The aspect ratio in this case is the height of thebuildings, H , divided by the distance between two facades, B.

AR= H

B(3.10)

The impact of this is examined experimentally in subsection 3.4.1. Further data on therelationship between daylight factor (DF), aspect ratio and office layout can be foundelsewhere [166].

3.3.4 Glazing/Floor Ratio

The amount of daylight entering a room is related to the surface area of its windows. Bydividing the window surface by the floor surface, the glazing/floor ratio is found, which can

Sun

Building 1 Building 2

Flat 2

Flat 1

Flat 3 Breadth(B)

Height(H)

Figure 3.15 Aspect ratio parameters for the case of two facing buildings

Radiometric Characterisations 37

be useful for IPV design both for comparing absolute radiant energy available in a roomand for assessing the opportunity of positioning the application close to a window.

3.3.5 Lighting Recommendations

A related approach that provides insight into the relative radiant energy may be based ona reference such as the proposed international standard ‘Lighting of indoor work places’[47]. This standard provides recommendations for minimum lighting levels (in lux) for over30 kinds of industry by type of interior, task or activity. The applicability of these valueswas confirmed experimentally (but not published) in July 2002 by the author in two typesof building (educational and medical in Switzerland) by comparing measured data withthe above standard. In the majority of cases (over 2/3), values measured exceeded thoserecommended by the standard, suggesting that values from the standard may be taken as aminimum for indoor lighting intensity estimation.Other surveys in the area of lighting exist, but they usually consider fewer kinds of interior

than are covered by the above standard, e.g. [69].

3.4 RADIOMETRIC CHARACTERISATIONS

While previous work in IPV design [214] [199] has mentioned many of the issues above,the analysis of some parameters can be improved upon, either because the whole IPVradiant energy spectral range is not included or because the specific case of wall-mounteddevices is not investigated. The complementary parameters covered in this section relate toobstacles (subsection 3.4.1), windows (subsection 3.4.2), IPV location (subsection 3.4.3), IPVorientation (subsection 3.4.4) and various profiles (subsection 3.4.5). These are consideredwithin the 400–1000 nm range with relatively constant radiant energy sensitivity. As shownin Figure 4.16, this is the spectral response range of many photovoltaic technologies.

3.4.1 Obstacles

For IPV, an obstacle is any object that hinders radiant energy reaching the IPV application.The case of a direct light obstacle is familiar because it creates distinct shadows in thephotometric domain.

3.4.1.1 Outdoor obstacles

Some basic theory of nearby obstacle aspect ratio has been presented in subsection 3.3.3 butis too limited to predict the radiant energy reaching an obstructed window, especially in thehorizontal plane which may be perpendicular to the PV module of a wall-mounted IPV appli-cation. The issue of predicting the radiant energy available is further complicated by the pro-portion of direct to indirect light available and the reflectivity of the obstruction, for example.In order to investigate the impact of nearby buildings, the total irradiance normal to the

window of a number of rooms of the same building was measured as indicated in Figure 3.16.The rooms all faced east but had different distances to the nearest buildings; these buildings


Radiometer

Breadth, B

Height,H

Obstacle

Figure 3.16 Schema of irradiance measurement

had the same exterior material and were designed similarly. The measurements in nine roomswere taken within 33min under constantly overcast conditions (diffuse light).As Figure 3.17 shows, the radiant energy intensity suffers rapid decline as the nearby

obstacle aspect ratio increases. From the unobstructed case (aspect ratio of 0) to a pointwhere the inclination from the measurement point to the top edge of the obstacle is 45�

(aspect ratio of 1), a relative intensity drop of over 75% is measured.Architects typically position buildings with an aspect ratio of no more than 1, i.e. building

height is a guide to the minimum distance to the next building. An example for IPV of such

0 0.5 1.0 1.50

5

10

15

20

Aspect ratio (H/B)

Tot

al in

stan

tane

ous

irra

dian

ce(W

/m2 )

Figure 3.17 Total instantaneous irradiance normal to the window with respect to the aspect ratiobetween the measurement point and the top edge of the building directly opposite (author’s

measurements)


an ‘obstacle’ might be caused by either trees and buildings outside or by descending a fewfloors of the building. Typical aspect ratio ranges for homes and offices may be 0.5–2, whilefor industrial sites the range is greater (e.g. 0.5–6), given the imperatives of contiguousprocesses, for example.One of the ways to compensate for the aspect ratio effect with regard to daylighting, used

in particular before the advent of electric lights, is to include an ever increasing surface ofwindow per floor from the top to the bottom of the building. An example of a building thatmay demonstrate this technique in Fribourg is shown in Figure 3.18.Evidently it will be beneficial to improve the accuracy of the design if the expected

building (aspect ratio) type is known. As not all IPV applications will be positioned on theground floor (worst case), it can therefore be expected that the typical obstacle aspect ratiowill be below 1.In the fortunate case of having more detailed building layout data, a more complete radiant

energy scenario can be achieved by preparing a Waldram diagram such as Figure 3.19.The diagram above is a graphical representation of the view from a particular point; the

bold line may represent a new building or a window opening. As the variable scale of thediagram shows, it allocates different weights to the sky depending on elevation, i.e. moreweight to the mid-range elevations than the highest and lowest elevations as the extreme

Figure 3.18 House on the Pont de Berne, Fribourg, with the greatest window surface on the groundfloor (author’s photograph)


90° 80 70 60 50 40 30 20 10 0° 10 20 30 40 50 60 70 80 90°

90°807570

65

60

55

50

45

40

35

30

25

20

Figure 3.19 Waldram diagram from the perspective of a room showing the surrounding obstacles[112], reproduced with permission of HMSO

elevations contribute the least light to the viewer. For example, such a chart can be used tocalculate total light entering a building through an aperture of known surface.

3.4.1.2 Indoor obstacles

Obstacles indoors are objects that create shading for the IPV application within the builtspace. These include furniture and plants in the office or home, but may also be formed bymore permanent elements such as screens and pillars. The built space is dynamic. Therefore,‘rogue’ obstacles that shade the IPV can also be expected. An example of malicious behaviourmight be the positioning of duct tape on the solar cell, while an equally pernicious effectcan be produced quite unwittingly by the thoughtless positioning of a new bunch of flowers.For the case of wall-mounted IPV application, the impact of indoor obstacles can in largepart be mitigated by careful installation, for instance at a suitable height above the floor.

3.4.2 Window Transmission

The transmissivity of the interface between the outdoors and indoors is determinant infinding the daylight radiant energy available for IPV. As will be seen below, the effects ofthe elements around the window can have an impact as great as the glass properties.

3.4.2.1 Furnishings

One class of IPV radiant energy obstacles that may be positioned either just in front of orjust outside the window is window furnishings. It includes (venetian) blinds, curtains and


Figure 3.20 Net curtain (author’s photograph)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

380 420 460 500 540 580 620 660 700 740 780

Wavelength (nm)

Rel

ativ

e T

rans

mis

sivi

ty

1 layer

2 layers4 layers

Figure 3.21 Transmissivity of the net curtain shown in Figure 3.20 (author’s measurements)

shutters. Each furnishing, although designed with visual comfort in mind, inevitably has animpact on the radiant energy available.Take, for example, the common (home) net curtain such as shown in Figure 3.20. To the

naked eye it appears to filter little, but, as can be seen in Figure 3.21, even a single layerof net curtains transmits only 56% �±2%� over the 450–700 nm band. For the case of twolayers, transmittance over the same range decreases to 32%, and for four layers it reaches16% (±2% in both cases).In practice, the true impact of a net curtain can be expected to be the equivalent of more

than one layer as the folds of the material imply that some photons will travel through twoor more layers. Equally, the values presented in Figure 3.21 were measured perpendicularto the netting, while in practice photons are usually incident with a non-zero angle to thenormal of the netting. They must therefore cross a greater amount of material, with thecorollary of reduced transmission.In comparison, consider a venetian blind. For the outdoor model measured in Figure 3.22

it can be seen that, when the slats are angled so as to touch each other (slats closed),the transmissivity may be reduced to as low as 0.06, while with the slats open it may be


0.0

0.2

0.4

0.6

0.8

1.0

0.0 0.2 0.4 0.6 0.8 1.0Proportion of window covered by blind

Rel

ativ

e T

rans

mis

sivi

ty

slatsopen

slatsclosed

Figure 3.22 Relative transmissivity with proportion of blind surface deployed. Measurements takenat 2m from the window (facing it) at a height of 1.5m (window sill height 1.10m) with the sensor at

30� from the vertical plane (author’s measurements)

reduced only down to 0.38. It can be deduced that closed shutters will have a similarly lowtransmissivity.

3.4.2.2 Glass properties

Glass has been used for making windows for centuries [201], as the stained glass in churchesattests [161]. The subject has fascinated many authors [170]. The importance of glass inmodern architecture goes well beyond traditional window requirements. Glass today mayhave a safety remit (antifire and antiballistic) and a structural contribution to the buildingas well as a noise abatement role. In terms of radiant energy transmissivity properties, acompromise between the two often contradictory needs of visual and thermal comfort issought. Visual comfort requires sufficient photopic transmissivity, while thermal comfort isrequired in two cases: firstly, when the outdoors is colder than the indoors, as an insulatinglayer; secondly, when the aim is to keep the indoors cool by preventing the ingress ofinvisible radiation (UV and IR range photons) and hot air. The different designs use one ormore sheets of glass. The relatively linear reduction in radiant energy transmissivity for theexample of a few sheets of plate glass perpendicular to the surface can be seen in Figure 3.23to be around 0.83 per sheet.

0

1

2

3

4

5

6

7

0 1 2 3Number of sheets of plate glass

Inte

nsity

(W

/m2 )

0.0

0.2

0.4

0.6

0.8

1.0

Tra

nsm

issi

vitiy

IntensityTransmissivity

Figure 3.23 Measured intensity and transmissivity of a number of sheets of plane glass (author’smeasurements)


Such plate glass radiant energy transmissivity loss is also a component of the loss mecha-nism of thin film solar modules (see subsection 4.5.1). It varies with the difference betweenthe spectral response of the particular photovoltaic technology and the spectrum transmittedby the glass (see Figure 3.24).The complex and developing area of window design has already produced a number of

solutions such as the sun protection designs shown in Table 3.3. As can be seen, the ratio ofthermal energy transmissivity to photopic transmissivity varies over a considerable range.Other factors (such as conductivity) being equal, glass with a ratio above 1 can be consideredto provide less thermal insulation than the visual transparency may suggest. Modelling suchproperties in more detail can be involved [220].To appreciate the difference between thermal and photopic transmissivity, a short

Gedanken experiment is useful. Imagine you are standing before a roaring fire in a coldroom. Think what you see and feel. If you then hold a large sheet of colourless glass upbetween you and the fire what do you now see and feel? You should now see the fire asit is, but it will feel much cooler behind the glass. This is because the transmissivity ofglass, although high for visible wavelengths, drops markedly for wavelengths above the near

Antisun

Calorex

Au coated

Ni coated

Rel

ativ

e T

rans

mitt

ance

1.0

0.8

0.6

0.4

0.2

0100 1000 10 000

Visible Wavelength (nm)

Ordinary

Figure 3.24 Transmission spectra for five kinds of glass (Au=gold, Ni=nickel), based on [146]and reproduced with permission of Square One Research

Table 3.3 Comparing the energetic and photometric transmissivity of a number of examples ofsolar protective glass (based on [193])

Glass typeThermaltransmissivity (%)

Photopictransmissivity (%)

Thermal trans./Photopic trans. ratio

Normal transparent 77 81 0.95Infrastop silver 48 48 1.00Infrastop neutral 39 51 0.76Parélio clair 50 43 1.16Calorex A1 42 38 1.11Stopray blue 36 50 0.72Antisun green 48 66 0.73


infrared band (in the 2.5–3�m range) which we can sense with our skin. These propertiesof glass (and polycarbonate material [103]) are fundamental to the greenhouse effect.The greenhouse effect is based on four steps. Taking the example of the built environment,

they are as follows.

1. (Solar) irradiation in the visible to near infrared range incident on a window is largelytransmitted indoors.

2. The indoor objects are warmed by absorbing this energy.3. In the third step, they re-emit with a much longer wavelength than the incident radiation

(see Figure 3.7). Such long-wave infrared cannot be collected by the PV solar cell (seeFigure 4.21). This wavelength shift is described in more detail in subsection 3.2.2.

4. In the fourth and final step, the glass both physically blocks the warm air from escapingand reflects the long-wave infrared back into the room. The resulting indoor temperaturerises so long as the radiative heat gain is not balanced by conduction and convection(i.e. heat loss). The ratio of the solar heat gain entering the room via the windows tothe incident solar radiation is called the solar heat gain coefficient (SHGC). Heat gainin this case includes transmitted solar heat as well as radiation absorbed which is thenreradiated, conducted or convected into the room.

3.4.3 IPV Location

The ideal position for an IPV application will depend on a number of factors including theavailability of artificial radiant energy sources such as Figure 3.8. The case of a classroomwas characterised for irradiance on the walls. The windows were fitted with blinds making itpossible to seal out the majority of the daylight. As can be seen in Figure 3.25, measurements

fenêtres

tableaunoir

plafond

portes

zones‘‘significatives’’

tubesfluorescents

A′B′

D′′

E′

D

AE D′′′

D′

C

C′

B

Figure 3.25 Diagram of the room characterised showing the areas where the measurements weretaken (zones ‘significatives’), the doors (portes), the light sources (tubes fluorescents), the windowswith blinds fully deployed (fenêtres) the ceiling (plafond) and the blackboard (tableau noir) [8]


were taken on two sections of the wall which it was hypothesised were representative of theother parts of the walls. This is due to the fact that the position pattern of the light sourceswas repeated in both the length and breadth of the room.The light sources can be seen to form discontinuous lines down the length of the room.

For this reason it was not expected to find inverse square law behaviour [equation (3.3)] butrather intensity variation with the reciprocal of distance [equation (3.4)].As the graphs in Figure 3.26 show, the irradiance on the wall parallel to the tubes (C–D)

was found to be generally less than that found along the wall to which the tubes areperpendicular (D–E). This is to be expected as the tubes are installed further from C–D thanfrom D–E.

0

1

2

3

Irra

dian

ce (

W/m

2 )

0 2 4 6 8 10Distance along wall parallel to tube axes (m)

0.8 m above floor1.5 m above floor2.2 m above floorPosition of tubes

0

1

2

3

0.8 m from floor1.5 m from floor2.2 m from floorPosition of tubes

2 3 4 5 6 7Distance along wall perpendicular to tube axes (m)

Irra

dian

ce (

W/m

2 )

Figure 3.26 Irradiance at three heights above the floor: left graph shows values along wall C–Dparallel with tubes and right graph shows values along wall D–E perpendicular to tubes


3.4.3.1 Superposition

The relationship between tube position and intensity can also be seen. The position of themaxima indicates that there is superposition between adjacent light sources at 1.5m fromthe floor on both graphs while no such superposition is found at 2.2m from the floor. Thegeneral case is schematised in Figure 3.27 which indicates the importance of vertical positionon the wall for an IPV application. If the vertical position is specified and the light sourcesare suitably close together, then a related ideal horizontal position with respect the lightsources is evident; i.e. in the case of a vertical position in the ‘superposition’ area this wouldbe equidistant between two sources, and for a vertical position in the ‘no superposition’ zoneit would be directly below one of the light sources.

3.4.3.2 Crossover point

The various sources of irradiance have so far been reviewed individually. In practice, ofcourse, they overlap. Given the artificial light source irradiance distribution mentioned andcombining it with the daylight data (subsection 3.3.2) allows a comparison of the two (seethe general case in Figure 3.28).

No super-position

Super-position

SuperpositionLight incident on wall

Light sources

Figure 3.27 Superposition of two adjacent light sources

1

10

100

0 5 10Distance from window (m)

Irra

dian

ce (

W/m

2 )

Daylight sourced irradiance

Fluorescent tube sourced irradiance

Total irradiance

Crossover point

Figure 3.28 Crossover point of daylight and artificially sourced irradiance (author’s measurements)


As can be seen, the greatest irradiance is usually found at the window (see Figure 3.32)and is independent of weather conditions during daylight hours. This irradiance rapidlyattenuates with distance from the window so that its intensity equals that of the electricallysourced light at a ‘crossover point’ (see Figure 3.28). The reason for such attenuation canbe explained by separating the direct and indirect daylight vectors. The strongest direct lightwill come from a direction overhead and will have a relatively shallow angle to the planeof the window. Some of the indirect light will have a more ideal angle for penetration intothe built space but at an intensity of around an order of magnitude less than the direct light.The remaining indirect light will also have a vector with a relatively shallow angle to theplane of the window and can therefore not penetrate deeply into the room either, thus alsocontributing to the intensity near the window. Only when the sun is low in the sky candirect light penetrate into the room, but with an intensity below the maximum for directlight.Therefore, so long as an IPV design can cope with the reduced intensity available from

artificially sourced irradiance, it can be located in a much larger range of positions withinthe room. On the other hand, if it cannot be adapted, then the position must be within ashort distance (less than 2m) from the window.

W/m2 Symbol

>2.5

2.00–2.24

1.75–1.99

1.50–1.74

1.25–1.49

1.00–1.24

0.75–0.99

0.25–0.49

0.50–0.74

<0.25

2.25–2.49

0.58m 0.63m 0.62m 0.59m 0.59m 0.6m 0.63m 0.63m 0.82m0.6m

0.6m 0.73m 0.56m 0.56m 0.69m

2m

1m80

1m60

1m30

1m

0m75

2m

2m20

1m80

1m60

0m75

1m30

1m

2.7 m

2.7 m

D′′

D′′′

D

D

Figure 3.29 Irradiance maps for both areas measured with the sensor parallel to the wall (author’smeasurements in collaboration with [8])


3.4.4 IPV Cell Orientation

The importance of cell orientation with respect to the irradiance has been presented [equa-tion (3.5)]. However, only one wall-mounted IPV application has been found (MigrosIndoor/Outdoor temperature gauge) that has an orientation of the solar cells other thanparallel with the wall. Consider the irradiance maps in Figure 3.29; these were measuredin the room shown in Figure 3.25 with the sensor parallel to the wall. The irradiance isgenerally below 1W/m2.

If the sensor is then incrementally oriented over the range shown in Figure 3.30, thena maximum value can be found (see Figure 3.31). These values are generally over doublethose found in Figure 3.29, such as those measured on the wall perpendicular to the tube axisof 1.5–3W/m2. This is evidently an important parameter in IPV design as it can translateinto reduced solar cell surface or battery size.The angle chosen will depend on typical installation height and the extent to which the

product (in the case of a retrofit) can be modified to allow the solar cells to be more ideallyoriented. In the study on which the results are based [8], the angle at which maximumirradiance was measured was found to be in the range 35–70� from the vertical.

E′E

0.75 m

1.0 m

1.3 m1.6 m

1.8 m

2.2 m

2.7m

2.7m

2.0 m

C′C

0.75 m

1.0 m

1.8 m

2.2 m2.0 m

x

y

1.3 m

1.6 m

Figure 3.30 Orientation range over which measurements were made [8]


W/m2 Symbol

>2.5

2.00–2.24

1.75–1.99

1.50–1.74

1.25–1.49

1.00–1.24

0.75–0.99

0.25–0.49

0.50–0.74

<0.25

2.25–2.49

2.0 m

0.58m 0.63m 0.62m 0.59m 0.59m 0.6m 0.63m 0.63m 0.82m

2.7m

2.7m

0.6m

2.2 m

1.8 m1.6 m

0.75 m

1.3 m

1.0 m

D′′ D

D D′′′

2.0 m1.8 m

1.6 m

1.3 m1.0 m

0.75 m

0.6m 0.73m 0.56m 0.56m 0.69m

Figure 3.31 Maximum irradiance at the wall found within the orientation range shown inFigure 3.30

3.4.5 Further Profiles: Buildings, Users and Applications

In the specification of an IPV product, a number of parameters cannot be specified exactlyand are therefore treated with profiles. Profiling can allow a complex problem to be convertedinto a small number of simplified representative cases which can provide input to the overallanalysis.The importance of building profile, for example, is evident when comparing the typical

distribution of artificial light sources. In United States surveys [138], 98% of households usedincandescent lighting while commercial floor space was lit in 77% of cases by fluorescentsources [120]. This affects IPV as solar cell technologies have different response to spectra(see Figure 5.4). A trend today is the growing acceptance of long-life compact fluorescentlamps (CFL). As these are lower rated power than the incandescent bulbs they replace, thisimplies that less emitted power or radiant energy is available to IPV.Equally, the building users do not necessarily have predictable lighting use behaviour.

This might be due to a number of factors including climate, culture, typical hours workedand whether extra light is required for performing the task.Further profiles (e.g. product use pattern) are covered in subsection 7.2.2.


3.5 COMPUTER SIMULATION

The attraction of simulation software in the area of lighting design can be gauged by thenumber of products available (23 are mentioned for Switzerland in [194]). However, thesehave issues, chief of which for IPV design is their low physical accuracy. Typically, inorder to maximise computational speed, the models pay little heed to the exact laws ofphysics over the range in which they are applied. This naturally compromises the scientificvalidity of the results produced. Further pros and cons of simulation in the area of lightingare covered in [55].The software that to the author’s knowledge best respects the physical laws is the UNIX-

based ‘Radiance’ by Ward [260] [261]. It has the disadvantage of being one of the leastaccessible to an unskilled user, requiring of the order of 3 months for proficiency. However,as can be seen in Figure 3.32, a scene can be fully modelled and characterised.Such a representation of an indoor environment is typically made for a single time. It is

also interesting to IPV designers to have cumulative data over a year for example. ‘Radiance’studies of cumulative solar energy for photovoltaics have already considered outdoor loca-tions with simulated [51] [53] and real-data [52] skies. The application of either kind of skyto indoor studies has, to the author’s knowledge, not been previously described.Ideally, such daylight data will be combined with appropriate electrically powered light

irradiance data so that the total energy reaching an IPV product can be computed. Suchelectrical light sources can be simulated in Radiance.

3.6 DISCUSSION

Reports dealing with the irradiance available for solar cells exist but are incomplete forIPV purposes as, for instance, they deal only with outdoor radiant energy [253] [130].Another reference is the ISO standard ‘Lighting of indoor work places’ [47] which suggests

Figure 3.32 Rendering of a simple office augmented with iso-intensity lines (red=high,blue= low) [54], reproduced with permission of Raphael Compagnon

Discussion 51

appropriate lighting levels for a large variety of indoor locations. Experiments to validatethis standard in a number of location types (e.g. schools, hospitals and homes) indicate thatthe lighting levels mentioned in the standard are generally exceeded. This means that thedesigner may use them as an expected minimum light level.

3.6.1 Summary of Parameters

In order to provide an overview of this chapter, the radiant energy that can be expectedoutdoors and indoors is set out in the top six lines of Table 3.4. The remainder of the tableshows the radiant energy parameter ranges that can be expected inside built spaces. Fora particular product case, this would be augmented by the (statistical) distribution of eachparameter. The tighter these parameter tolerances can be specified by the designer, the morelikely it is that the resulting product will perform as required.

3.6.2 IPV Designer Recommendations

A parameter that in IPV applications is often poorly implemented is the orientation of the solarmodule (see subsection 3.4.4). This parameter is often overlooked and could save both PV

Table 3.4 Built space radiant energy parameter range (based on author’s findings)

Parameter Typical IPV rangeSubsectionreference

Radiant energy: daylight outdoors 0–1000W/m2 5.133–200kW h/m2 month 3.2.1

Radiant energy indoors:electrical (fluorescent) 0–3W/m2 3.4.4

0–700W h/m2 monthdaylight (worst month) 0–10kW h/m2 month

Distance from effective radiant energysource (including windows) to IPVapplication 0–100m 3.2.1

Horizontal orientation −180–180� 3.2.10.25–1.0

Orientation above horizontal 0–90� 3.4.40.5–1.0

Spectra Daylight, fluorescent,incandescent

3.2.2

Built space daylight factor 0–15% 3.3.2Obstacle aspect ratio 0–6 3.3.3Glazing/floor ratio 0–0.5 3.3.4Window transmissivity (single anddouble glazing) 0.5–0.9 3.2.3

Surface reflectivity 0.2–0.8 3.2.3Window furnishings transmissivity 0.06–1 3.4.3


surface and battery size. In practice, for a wall-mounted device this saving can be achieved bydesigning the product casing to accommodate an elevation of the module (see Figure 7.9).With respect to the IPV location, it seems worthwhile recommending better installation

of wall-mounted applications with respect to IPV location (see subsection 3.4.3). For thosethat the user installs, brief instructions could be included with the product. For the installer,only limited training could ensure more reliable performance.

3.7 CONCLUSION

An easy mistake for designers to make in assessing the applicability of a location for IPVis to rely too heavily on what they see. The human visual systems suffer two disadvantageswhen determining the radiant energy capital of a location: they are relative light meters andthey do not have the same spectral sensitivity as solar cells. The results presented here maytherefore seem counterintuitive at first. This chapter should enable designers to overcometheir human (photometric) perception and appreciate the radiant energy resources availablein a built space.The new perception is based on the many parameters that affect indoor radiant energy.

These may be qualitative (photometrics in section 3.3) or quantitative (radiometric experi-ments in section 3.4). Considered together, such as in section 3.6, they may serve the IPVproduct designer at a number of levels such as providing the range of values of the radiantenergy found indoors. As can be seen in Table 3.4, a well oriented and located product mayreceive of the order of 1kW h/m2 month of radiant energy. At least 10 times this value mayalso be achieved, on a window ledge, for example.This chapter also provides an appreciation of the most significant parameters found along

the ‘photon chain’ between the light source and the PV device. A number of recommendationsare provided in order to allow the designer to achieve the best performance from a proposedIPV application. Further recommendation for the practitioner can be found in Chapter 7.

3.8 FUTURE WORK

Ambient energy in the built environment is available at other frequencies than the radiantenergy chiefly considered here. These include mechanical vibrations in the 50–200Hz range[215]. It would be interesting to perform a similar study of the built space for such fre-quencies. Figures for 12 vibration source examples are presented in [219]. By combiningthe results of the present chapter and this future study, a more general mapping of the ideallocations for ambient energy products could be developed.The well known mathematical non-diffusive billiard ball model could be applied to IPV

photon modelling if the energy diffusion created by the imperfect reflectivity of the wallswere included. At present, while the billiard ball model has attracted hundreds of journalcontributions, none deals with diffusion.

3.9 FURTHER READING

For more detail on many aspects of building physics, especially light distribution in the builtspace, see [112].

4Fundamentals of Solar Cells

4.1 INTRODUCTION

Understanding the way photovoltaic (PV) solar cells work is non-trivial for most people.Therefore, designers without prior training are likely to find understanding the concepts onwhich solar cells rely uninviting. One reason is that the quality texts on semiconductorsgenerally require physics understanding.It is the aim of this chapter to address the lack of widely accessible information about

photovoltaic solar cells. In tandem with the radiant energy information (chapter 3), thefoundations will thus be laid for the following chapter (chapter 5) which considers thecharacterisation of solar cells indoors.An appreciation of how solar cells work is essential to the designer as it forms the basis

for the improvement of solar cell performance (treated in section 4.7 and section 5.5).This chapter may serve a broad spectrum of readers, from the scientifically inclined

novice up to the photovoltaics specialist. For those who are relatively new to the area ofsolar cells, the first two sections set PV in its historical (section 4.2) and technologicalcontext (section 4.3). The more experienced may prefer to skip to section 4.4, where thefundamental solar cell parameters are briefly explained. In section 4.5, the reasons why solarcell efficiency is relatively low are reviewed. The most common PV materials available arereviewed in section 4.6, with special consideration to IPV applications.In summary, this chapter aims to answer the following questions.

1. How was PV developed? (section 4.2).2. How does PV work? (section 4.3 and section 4.4).3. What are the weaknesses of PV? (section 4.5).4. What PV solar cells are available? (section 4.6).5. How may indoor PV modules be improved? (section 4.7).

To the author’s knowledge, no other single published document combines all these elementsas succinctly and with reference to indoor ambient energy systems.


54 Fundamentals of Solar Cells

4.2 BRIEF HISTORY OF SOLAR COLLECTORS AND PV

All inhabitants of planet Earth have been collecting solar energy since the beginning of theirexistence. Evolution (and more recently civilisation) can be considered as a progression ofever more efficient means of collecting and converting this energy which is both bountiful –delivering 1.5 ZWh per annum [zeta (Z) is the SI unit for 1021] – and consistent, havingbeen burning for an estimated 4.6 billion years. Relative to such timescales, the discoverythat solids can convert solar energy directly to produce that most flexible of energy media,electricity, is a very recent one. In its infancy, however, photovoltaic technology was basedon a solid–liquid system.

4.2.1 The ‘Selenium Years’

An event of note in the history of solar cells was the discovery of selenium by Berzeliusin 1817 [267]. Incidentally, he was also the first to prepare elemental silicon a few yearslater. However, it was over two decades before Alexandre-Edmond Becquerel (1820–91), aFrench physicist, observed the photovoltaic effect in 1839. The nineteen year old made thisdiscovery by immersing two metal plates in a conductive fluid and observing a small voltageacross them on exposing the apparatus to the sun. It is a twist of fate that his son, Antoine-Henri (1852–1908), shared the 1903 Nobel Prize for Physics for the discovery of spontaneousradioactivity, a competing technology to solar energy in modern power production!Thirty-four years after the discovery of the photovoltaic effect, Willoughby Smith (born

in Great Yarmouth, UK, 1828, died Eastbourne, UK, 1891) made another significant dis-covery. Already a pioneer in the installation of submarine telegraphic cables, he reportedthe photoconductivity of selenium in 1873 [237]. This was to inspire W. G. Adams andR. E. Day to make the first selenium cell in 1877.One of the first to envisage that such cells could be applied to power production was

an American, Charles Fritts. Having produced his own selenium solar cells, which wereverified by Werner Siemens [234], he predicted their use in what are today called buildingintegrated PV (BIPV). He wrote in 1886 that the sun ‘is both without limit and without costand will continue to stream down on earth after we exhaust our supplies of fossil fuels’[197]. How familiar such arguments for renewable energy sound even today! Yet, as today,the reason for the lack of expansion of photovoltaic cells was affordable efficiency. Typicalcell efficiency was less than 1%, and therefore unrealistic for large-scale power production.This remained the case for over half a century, but did not stop further discoveries in

photovoltaics. New materials were found to be photosensitive, such as the combination ofcopper with cuprous oxide, discovered by Hallwachs in 1904. The science of photovoltaicsalso evolved with Milikan’s experimental proof of the photovoltaic effect in 1916 and AlbertEinstein’s Nobel Prize of 1921 for his contribution to explaining this effect [6].

4.2.2 The ‘Silicon Years’

It was to be another material on which the progress of photovoltaics was to rely: silicon. Thefirst solar cell of silicon was reported in 1941 by R. Ohl. Thirteen years later, the efficiency ofsuch cells had reached 6%, as reported by D. M. Chapin, G. L. Pearson and C. S. Fuller [42]

Brief History of Solar Collectors and PV 55

from Bell Telephone Laboratories. Such cells were applied to space applications within 4years aboard America’s second satellite, Vanguard 1 [266], which continued to transmit until1964! Indeed, the then space industry culture of ‘performance at any cost’ contributed to thedevelopment of solar cell technology which was to favour terrestrial applications. The Clubof Rome report ‘The Limits to Growth’ [154] and the 1973 oil crisis also raised awarenessof the importance of renewable energies in general and photovoltaics in particular, but, ascan be seen from Figure 4.1, the most significant market growth for PV has only come since1995.This is reflected in the present PV annual production rate of the order of 400MW [81]

and a consistent growth rate of the order of 30% per annum in recent years, which is farsuperior to a typical business cycle.These impressive figures should be taken in context. The majority of outdoor-use PV

applications benefit from subsidies; also, at the global scale, PV is still a minority powersource, i.e. 400MW represents 0.004% of the global power installed capacity of around10 TW [1TW (or 1× 1012W) is equivalent to 1 billion kW; therefore, assuming 7 billion�7× 109� inhabitants of the Earth, average installed capacity is around 143W/inhabitant].This is due to the much larger contributions of other technologies indicated by global energysupplied in 2000 (see Figure 4.2).PV will remain a minor power source even if such an ambitious target as that set by

the European Commission [56] of 3GW of PV production by 2010 are met. The lattertarget represents 100 times the 1995 value – the most aggressive relative growth of all therenewable energies considered in this White Paper [56, Table 1A, 4.8]Exceptional progress has also been achieved in the cost of PV. In 1970, modules cost $100

per rated watt, while now one would expect to pay less than $4 for the same power rating [94].However, the true cost of producing electricity using PV includes the installation and mainte-nance cost (collectively, balance of system or BOS costs). When comparing power-producingtechnologies from the perspective of installed cost (see Figure 4.3), it can be seen that PV isuncompetitive with respect to existing centralised power technologies and price structures.An exception to this is for locations that are outside the electrical power distribution net-

work (in the jargon ‘off-grid’), where the cost of installing power cabling may be prohibitive.

100

0

200

300

1980 1985 1990 1995 2000

PV M

odul

es s

hipp

ed (

MW

P)

Figure 4.1 Global production of PV 1980–2000, based on [250]


Gas 21%

Oil 35%

Coal 35%

Nuclear 7%

Hydro 2%

CombustibleRenewablesand Waste11%

Wind 0.03 %Solar 0.04 %

Geothermal 0.4%

Renewables 14%

Figure 4.2 Global total primary energy supply (9958 Mtoe) in 2000 by fuel (NB solar includessolar thermal). Data from [115]

Sola

r PV

0

10

20

30

40

50

60

70

80

Typ

ical

cos

t (c/

kWh)

Nuc

lear

Gas

- c

ombi

ned

cycl

e

Win

d

Hyd

ro

Geo

ther

mal

Coa

l - S

team

Bio

mas

s

Sola

r T

herm

al

Oce

an

Figure 4.3 Typical minimum and maximum electricity generation cost for a number of powergeneration technologies [1]

For the moment, such decentralised PV applications represent only a minor fraction of thePV installed base.Decentralised energy production is conceptually equivalent to the target applications of

this project. While the majority of research in PV is aimed at making power productionapplications economically viable, the following chapters focus on indoor applications, anarea that is already economic and, as submitted here, can be further developed.

4.2.3 History of IPV

IPV (or indoor PV) has been successful despite much less governmental support. One ofthe first applications of solar cells to indoor devices came in the mid-1960s in the form ofa table-clock with a solar panel on the dome (see Figure 4.4).

Brief History of Solar Collectors and PV 57

Figure 4.4 Cheveaux table-clock courtesy of Patek Philippe Museum, reproduced with permissionof V. de Mallmann

Watches followed in the first half of the 1970s, such as the Nepro Solar LED watch(Figure 4.5) and calculators, for example those produced by Royal (Solar 1), Teal (LC812)and Sharp (EL-8026).Calculators have been the most successful IPV application to date, seeing the strongest

growth in sales during the 1980s and reaching annual sales of 10 million pieces by 1990 [108],thus avoiding millions of single-use batteries. This volume of solar cell sales would produce4MW under AM1.5 (1 sun), which at the time (see Figure 4.1) represented approximately10% of the global PV shipments. Since then, the latter market has grown significantly whilethe solar calculator market has saturated. Therefore, in order for the IPV sector to grow,other applications such as those considered herein (e.g. Figure 4.6) are necessary.

4.2.4 IPV Today

The present-day situation is one of relatively ubiquitous application of amorphous siliconfor many calculators and some watches. An ever growing range of indoor-use solar productsis now available, such as those shown in Figure 4.6 including, from left to right, calcula-tors, watches, alarm clocks, scales (e.g. Maul®), meters, moving shop window presenters(e.g. Dual Promoter®), decorations (e.g. Solar Seerose with the lid removed to show theLED), office staplers (e.g. Skrebba®), torches and a kitchen whisk (e.g. SoLait®), not tomention radios (not shown).Given the lower radiant energy intensities indoors, and therefore reduced voltage (see

section 5.2.2), it is all the more important to create series-connected modules. For suchapplications, thin-film technologies are appropriate as they can be monolithically series


Figure 4.5 Nepro Solar LED watch, the first Swiss watch of its kind. Notice the solar module madeof ‘tiled’ cells. This example was loaned by the designer, Dr Paolo Spadini (author’s photograph)

Figure 4.6 present-day IPV products (author’s photograph)

Photonic Semiconductors 59

Others8%

MonocrystallineSilicon

36%MulticrystallineSilicon

44%

AmorphousSilicon

12%

Figure 4.7 Photovoltaic cell production by technology in 1999 [75]

interconnected (see subsection 4.6.6). Indoor applications therefore contributed to the pro-duction growth of such technologies as amorphous silicon, which today represent 12% ofsolar cells manufactured (see Figure 4.7).

4.3 PHOTONIC SEMICONDUCTORS

All solar cells are based on semiconductors, more specifically the class of photonic semicon-ductors. A general description of semiconductors is therefore provided in subsection 4.3.1;the fundamentals of photonic semiconductors may be found in subsection 4.3.2. These maybe categorised in at least four different ways (see the ‘solar cell categories’ described insubsection 4.3.3).

4.3.1 What is a Semiconductor?

A simplified approach to material conductivity categorisation is to graph the available energylevels that electrons can occupy for each material. These are in bands, unlike the case foratoms which have discrete energy levels. Each material can then be sorted into one of thethree categories shown in Figure 4.8.These bands can be further summarised on the energy E scale into the valence band (VB)

and the conduction band (CB). Materials with a greater than 5 eV energy gap between the

Energy

Insulator Semiconductor Conductor

Cond-uctionBands

ValenceBand

Eg > 5 eV0 < Eg < 3.5 eV

Figure 4.8 Band gap diagrams for a typical insulator, semiconductor and conductor, showingapproximate excitation energy Eg


VB and the CB are called insulators; those with a gap in the range 0–3.5 eV are referred toas semiconductors. If the CB overlaps the VB, the material is said to be a conductor.Conduction requires electrons to be available in the CB. Semiconductor materials can fulfil

this criterion with controlled variation of heat, light, magnetic field or impurities (doping).Their resistance may therefore be adjusted across many orders of magnitude, for exampleroughly within the range 1m"–1M".

4.3.2 Photonic Semiconductor Properties

Semiconductors can be used to convert light into charge, and vice versa; such semiconductorsare called photonic devices. A few examples are shown in Table 4.1.

4.3.2.1 The p–n junction

Solar cells therefore convert light into charge. They achieve this across a junction (seeFigure 4.9), which is an interface between an appropriately p-doped (positive) and an n-doped(negative) area. This doping is characterised as being fixed in the material lattice and havinga different amount of electrons in the outer orbit than the bulk material. It can therefore beionised to provide either an electron or a hole (an effective positive charge due to the absenceof electrons). In physics, such dopants are called electron donors and electron acceptorsrespectively.On formation of the junction, the mobile charge carriers (electrons and holes) next to the

junction jump across it, being attracted by the opposite charge on the other side, and thenrecombine. In so doing, the electrons from the n-doped area leave positively ionised donorsbehind. Equally, holes expose ionised acceptors. This leaves a net negative charge on the

Table 4.1 Examples of photonic devices

Conversion Process Light → Charge Charge → LightExamples Photodiode, solar cells LED, OLED, LASER

p-doped semiconductor

hole Depletion zone Junction electron

n-doped semiconductor

A–

A–

A–

A–

A–

A–h+ h+

D+

D+

D+

D+

D+

D+

D+

D+

D+

D+

D+

D+

D+

D+

D+

D+

D+

D+

D+

D+

h+ h+

h+ h+

h+ h+

A–

A–

A–

A–

A–

A–

A–

A–

A–

A–

A–e– e–

e– e–

e– e–

e– e–

A–

A–

A–

Figure 4.9 Simplified schema of p–n junction inspired by [272]. A− is an ionised acceptor, D+ isan ionised donor


p-doped side and a net positive charge on the n-doped side, thus creating an electric fieldwhich separates the remaining mobile charge carriers. The area where this occurs is calledthe depletion zone (see the middle vertical stripe in Figure 4.9).Such a p–n junction can be used as a diode, and therefore Figure 4.9 can be summarised

with the standard symbol for a diode as shown in Figure 4.10 which maintains the same p–norientation. When a voltage is applied across the junction in forward bias (left-hand diode),the depletion zone is reduced in width allowing current to pass. In reverse bias (right-handdiode), the electrons are drawn apart, the depletion zone is widened and it is thus muchmore difficult for electrons to pass the junction.When the junction is made of photovoltaic material and illuminated, the arriving photons

can excite electrons from the valence band into the conduction band, leaving a ‘hole’. Eachelectron–hole ‘pair’ may be separated under the influence of the electric field at the junctionand contribute to the photocurrent.

4.3.2.2 Light absorption modes

The absorption process is material dependent and may be direct or indirect. As a result,semiconductor materials are said to have a direct band gap or an indirect band gap respec-tively. The simpler (first-order) process of direct band gap absorption is so called becausethe maximum valence band energy (band edge) has the same crystal momentum as the min-imum of the conduction band. This means that absorption of the photon can occur withouta change in momentum (see Figure 4.11). A material that exhibits such properties is GaAs.

+ – – +

Figure 4.10 Forward bias (left-hand diode) and reverse bias (right-hand diode)

Eg

E4

Ev

Crystal momentum, p

Ene

rgy,

E

Figure 4.11 Energy–crystal momentum for a direct band gap semiconductor [107], reproducedwith permission of Dan Hessman


Crystal momentum, p

Energy, E

Phonon absorption

Phonon emission

Ec

Ev

Eg – Ep

Eg + Ep

Figure 4.12 Energy–crystal momentum for an indirect band gap semiconductor, showing atwo-step absorption process [99], reproduced with permission of M. A. Green

Indirect band gap semiconductors have a conduction band edge that has a different crystalmomentum (or wave vector) to the valence band edge (see Figure 4.12). The process ofabsorption is therefore second order, requiring a phonon to be absorbed and emitted to allowthe absorption to take place.Simply put, a phonon is the fundamental (quantum) particle corresponding to the crystal

structure vibration. Just as light can be considered as a wave or a quantum (fundamentalparticle) which is the photon, sound can also be treated as a wave or a phonon. The twoquanta differ in that the photon has high energy and low momentum, while the phonon haslow energy but high momentum. This may be understood more intuitively by consideringthe large difference in the velocity of light and sound through a material.

4.3.2.3 Electrical contacts

Post absorption, the current is collected via front and back contacts. The rear contact typicallycovers the entire surface, while the front contact must have low surface area for non-transparent materials (achieved by the use of a fine grid of metal) to allow the photons accessto the PV material. Semitransparent contacts are commonly produced using a conductiveoxide.

4.3.3 Solar Cell Categories

Solar cells can be categorised in a number of ways:

(a) by thickness of semiconductor material;(b) by group number(s) in the periodic table;


(c) by specific compound;(d) by being based on plastics (organic polymer compounds).

4.3.3.1 Semiconductor thickness

The solar cells most used at present are technologically based on the silicon wafers ofthe microchip industry. They have a thickness of approximately 300–400�m and includecrystalline (single-crystal), polycrystalline and multicrystalline silicon cells. They are manu-factured in blocks and sawn into wafers except in the case of edge-defined film-fed growth(EFG), in which wafers are drawn directly out of the melt by capillary action.The other principal category with respect to thickness is thin-film solar cells, so called

because the typical photosensitive semiconductor material thickness is of the order of 3–5�m.These are produced by deposition of the photovoltaic material onto a substrate such as glass,metal and, relatively recently, polymers.

4.3.3.2 Periodic table group number(s)

Semiconductors can be made from single-element materials or compounds. Group IV ele-ments such as silicon (Si), germanium (Ge) and tin (Sn) are used to produce elementalsemiconductors. These elements can also be combined (e.g. SiGe) to form compound semi-conductors. More common compound semiconductors are made from combinations of groupIII and V elements (e.g. GaAs) or group II and VI elements (e.g. CdTe).A simplification of the group number categorisation is to consider the chemical compo-

sition of the bulk material on either side of the junction. If it is the same, it is called ahomojunction; if it is different, it is called a heterojunction.It is of interest that diamond (C), although a group IV element, has non-ideal semicon-

ductor properties for at least two reasons: it has a high band gap (around 5 eV [212]) and,while it can be doped positively (B), it is more difficult to dope negatively (e.g. with N)which jeopardises the formation of p–n junctions. With present technology, such as chem-ical vapour deposition (CVD), this material cannot presently be used for solar cells.Research is nevertheless going ahead to make solar cells with diamond thin films for spaceapplications [252].

4.3.3.3 Specific compounds

The term chalcogenides (group VI and, in the future, group 16 when including thetransition metals) is most often used to refer to various compounds containing at leastone member of the group in the period range III–V (i.e. sulphur, selenium or tel-lurium). These therefore include the compound CuInSe2 (typically abbreviated CIS), theband gap Eg of which is around 1.0 eV, and its alloys with CuInS2 (Eg 1.5 eV) orCuGaSe2 (Eg 1.7 eV), for example. CdTe (Eg 1.4 eV) is one of the other common chalco-genides. Ternary chalcopyrites also include compounds of (Cu,Ag)(Al,Ga,In)(S,Se,Te)2form [19].


4.3.3.4 Organic solar cells

The fact that certain organic materials have semiconductor properties has been known sincethe 1970s [148, 233] and led to the Nobel Prize for Chemistry of 2000 [178]. Early examplesinclude melanins and polyacetylene.While such electrically conductive plastics are already being applied to flat screens [84],

they could in principle be used in the ‘reverse direction’, i.e. as solar cells [78].Photonic semiconductor research based on organics continues [230]; one of the Nobel

laureates (Heeger) is involved in the development of photodetectors [78] based on organics.The production of organic semiconductor solar cells, however, is still at the research phase

[65]. These may be interesting for IPV applications as the peak fill factor occurs at relativelylow light intensities, in one case at 30W/m2 [211].

4.4 PHOTOVOLTAIC TECHNOLOGY

Solar cells can be characterised by a significant number of parameters, both common to thebranch and specific to a particular material. It is not the aim of this thesis to investigatesuch details, as a large number of specialists are already occupied in this pursuit. However,in order to appreciate the following chapters, a basic understanding of the four commonparameters is essential, especially inasmuch as they relate to IPV. These are presented oneby one in the next four subsections. Notably, the effect of temperature is neglected as indoortemperatures are generally more stable than outdoors.

4.4.1 Electrical Efficiency Calculation

The parameter that is of chief concern to researchers and practitioners alike is electricalefficiency, from here on referred to as efficiency. Although in principle this can be calculatedby the ratio of output power PO to input power PI, in practice it is only calculated under theideal condition that PO is a maximum. To satisfy this condition, the resistance across thesolar cell or module under test is varied across a wide enough range for the relationship ofvoltage against current to take the form of the curve in Figure 4.13.At zero resistance the current is at its highest possible value, called the short-circuit

current, ISC, while at maximum resistance the voltage is at a maximum, referred to as theopen-circuit voltage, VOC. Returning to ISC and increasing the resistance above the impedanceof the solar cell, the voltage increases quickly and the current decreases slowly until the‘knee’ of the curve. At this point, known as the maximum power point (MPP), the power(product of voltage and current) is at a maximum and the related values of voltage andcurrent are referred to as VM and IM respectively (see Figure 4.14).

4.4.2 Fill Factor

The latter values can also be used to calculate the electrical efficiency by the equation

%=VMIM/PI (4.1)

Photovoltaic Technology 65

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

0 0.5 1 1.5 2 2.5 3

Voltage, V (V)

Cur

rent

,I (

mA

)

ISC

VOC

Figure 4.13 Typical solar module I-V characteristics (author’s measurements)

Voltage, V (V)

Cur

rent

,I (

mA

)

VM

IM

Figure 4.14 I–V curve showing maximum power point current and voltage (author’smeasurements)

but in practice a further parameter is used to summarise both VM and IM in one. This is thefill factor (FF), so called as it represents the ratio of how fully the area delimited by VMIMcovers the ideal power VOCISC. It can be found graphically (Figure 4.15) but is more usuallycalculated from the equation

FF= VMIMVOCISC

(4.2)

Substituting equation (4.2) into equation (4.1), the efficiency can be found by

%= VOCISCFFPI

(4.3)


Voltage, V (V)

Cur

rent

,I (

mA

)

VM VOC

IM

ISC

VMIM

Figure 4.15 Graphical calculation of the fill factor (author’s measurements)

Efficiency is therefore related to the three parameters VOC, ISC and FF, which are related tomany more variables (see the next two sections). In practice, actual efficiency will always beless than or equal to the efficiency in equation (4.3), as the solar cell is not likely to be usedat the maximum power point at all times. Actual voltage and current therefore determine theactual efficiency.

4.4.3 Short-circuit Current

For a range of incident radiant energy intensity well beyond that of IPV, the short-circuitcurrent is assumed to be directly proportional to the radiant energy intensity. For IPVpurposes, this is considered a given. Ideally, it is also equal to the light-generated current IL.

Current IL is calculated by integrating the product of the incident radiant energy spectrumand the solar cell spectral response; examples of both spectra are shown in Figure 4.16. The

1.0

0.8

0.6

0.4

0.2

0.0400 600 800 1000 1200

160

120

80

40

0

Wavelength (nm)Daylight spectrum

Rel

ativ

e sp

ectr

al r

espo

nse

(A/W

)

CdTea-Si monoX-Si CIS

Irra

dian

ce (

W/m

2 )

Figure 4.16 Spectra of daylight and spectral response of solar cells representing fourPV technologies: a-Si = amorphous silicon; CdTe = cadmium telluride; monoX-Si =

crystalline silicon; CIS = copper indium diselenide (based on [176])

Photovoltaic Technology 67

latter is the ratio of charge carriers collected per incident photon. Therefore, IL can be seento be for a unit bandwidth:

IL�&�= �1−R�&��SD�&�SR�&� (4.4)

where R is the portion of incident radiant energy reflected from the surface of the solarcell, SD is the spectral distribution of the incident radiant energy flux and SR is the spectralresponse. Integrating across the SD wavelength range gives the total light-generated currentfor a particular device.

4.4.4 Open-circuit Voltage

The ideal diode law (in the dark for the junction region) has been shown to be [99]

I = IS�e�qV�/�kT�− 1� (4.5)

where I is current, IS is dark current, q is electronic charge, V is voltage and k is Boltzmann’sconstant. Current IS is found by the same procedure as that used to collect I–V curves suchas Figure 4.13, with the exception that illumination is kept as close to zero as possible. Thisproduces a graph of the form shown in Figure 4.17.Under illumination, equation (4.5) becomes [99]:

I = IS[e�qV�/�kT�− 1

]− IL (4.6)

where IL is the light-generated current. Setting I to zero and resolving for V gives VOC

(for a ‘perfect’ junction) as [244]

VOC =kT

qln(ILIS

+ 1)

(4.7)

–5.0

–4.0

–3.0

–2.0

–1.0

0.0

1.0

–1 10 2 3 4

Voltage, V (V)

Cur

rent

,I (

mA

)

IS

Figure 4.17 I–V characteristics of a solar cell in the dark (author’s measurements). As IS is notstrongly voltage dependent, it need not be measured at a specific V


0.0

5.0

10.0

15.0

20.0

0 0.5 1 1.5 2 2.5 3

Voltage, V (V)

Pow

er,P

(mW

) 80% of maximum power

Figure 4.18 Power–voltage graph for the module in Figure 4.13 (author’s measurements)

Other than temperature, one of the principal influences on the dark current IS is the materialband gap. This represents the minimum energy required by an incoming photon to producean electron–hole pair that is collected. Therefore, if the band gap is reduced, IS will increaseand, from equation (4.7), VOC will decrease.

4.4.5 Power Curve

While standard test condition efficiency (see section 5.1) is important for comparing solarcells, real systems are more than likely not to be running at this ideal power, especially IPVones, without the benefit of a maximum power point tracker. This is not only because theintensity may be less than AM1.5 (1 sun) but also because power is the product of voltageand current; PV cells have a single maximum power point – see the power–voltage graphin Figure 4.18 for the same module and values as shown in Figure 4.13. For an IPV systemcharging a battery, for example, the battery voltage will tend to control the PV voltage. Inthis case, there is no reason why this voltage will be either constant or ideal; as a result,some power will be lost.

4.5 SUBOPTIMAL SOLAR CELL EFFICIENCY

The relatively low efficiency of solar cells is a cause of surprise to those who know littleabout the technology and of concern to PV specialists. Assuming constant temperature,which is reasonable for indoor applications, there are three important reasons for this:

(a) the incompatibility of the solar module with the incident light which is associated withreduction in ISC (see subsection 4.5.1);

Suboptimal Solar Cell Efficiency 69

(b) recombination of the charge carriers which is associated with reduction in VOC

(see subsection 4.5.2);(c) effects that are associated with deterioration in FF (as well as VOC and ISC)

(see subsection 4.5.3).

4.5.1 Optical Issues

The principal cause of low efficiency in solar cells is the spectral mismatch, or the proportionof overlap between the incident light spectrum and the spectral response. An example oftypical spectra can be seen in Figure 4.16. The mismatch is due to two energetic modes.Firstly, incident photons that have an energy, h), below the solar cell band gap, Eg, cannotcontribute directly to the photocurrent. Secondly, while all photons having more energythan the solar cell band gap contribute to the photocurrent, the excess energy (greater thanthe band gap) is typically not collected. The dissipation mechanism in this case is to heatthe solar cell (thermalisation in the jargon). The combined effect of these two spectralmismatches accounts for energy conversion losses of the order of 50% in the 400–1000 nmrange.Another source of loss is the electrical contact on the front (light-facing) surface of the

cell or module. This is especially evident for conventional (silicon) solar cells, as contactingis achieved using a metal grid. While this grid is necessary for appropriately low-resistancecurrent collection of these relatively high-current cells, by its presence it creates a shadowon the solar cell, blocking light from the cell. Losses due to such coverage are around5% using conventional techniques. For devices that use alternative front contacts, such astransparent conductive oxide (TCO), which have a higher resistance than a grid, the currentmust therefore be lower. The lower current is achieved using a monolithic series connection.This also implies coverage loss as the series interconnections take the place of active solarmaterial. A further implied loss of TCO solutions is that the incident photons pass througha sheet of glass which will have a non-unity transmissivity.As already indicated in subsection 4.4.3, the calculation of Isc takes into account the

reflectivity of the front surface facing incident light. From optics the reflection of normalincident light is related to the refractive index by the complex refractive index nc = n− i*as follows:

R= �n− 1�2 +*2�n+ 1�2 +*2 (4.8)

Using this equation with the appropriate values for silicon, for example, gives a reflectivityof around 30% [99], p. 41. In practice this can be reduced by antireflection coating toabout 10%, or light trapping to about 3% (by preferential chemical etched texturing), or acombination of both to around 2%.The radiant energy that is transmitted into the semiconductor material then suffers expo-

nential attenuation of the form

S�x�)�= S�)� exp�−,cx� (4.9)

with the absorption coefficient

,c�)�=4-*)c

(4.10)


where

S�)�= number of incident photons (per unit time, area and energy)x= distance from the semiconductor surface (m))= frequency of the incoming radiant energy,c = absorption coefficient

In theory, therefore, there is an optimal minimum thickness of material required to collectall the transmitted light. For direct band gap material, it is of the order of 1–3�m, whilefor indirect band gap material it is 20–50�m. An example of the latter is conventionalcells whose 300–400�m material thickness is not a limiting factor. However, for thin-filmsolar cells, the typical thicknesses of a few micrometres are much more critical, especiallyfor indirect band gap material. In the case of amorphous silicon, this is countered by thelight-trapping techniques mentioned above in this subsection.

4.5.2 Material Quality Issues

The recombination of charge carriers (electrons and holes) is a normal and unavoidableaspect of solar cells, as it is the process by which an illuminated solar cell returns to thermalequilibrium when the light is turned off for example. This is another reason for the interestin making an I–V characterisation in the dark, e.g. Figure 4.17. The dark current indicatesjunction characteristics which in turn determine how efficiently the solar cell will deliverthe photocurrent to the terminals.The process of recombination may be simply the reverse of the charge separation (radiative

recombination) process shown in Figure 4.11 and Figure 4.12 or by other processes suchas Auger recombination or recombination in traps. Traps can be thought of as being adeterioration in the ideal solar cell which only has energy bands (permitted states) in thevalence and conduction bands (see Figure 4.8) and none in between (in the forbidden zone).Practical solar cells, therefore, for example owing to imperfections or impurities in thematerial lattice, have further allowed energy states (traps). When such states occur in theforbidden zone, they create further opportunities for electrons and holes to recombine. Theseeffects can be located both at the junction (depletion zone) or surface and in the bulk of thesemiconductor.Such imperfections are therefore associated with reduced charge carrier lifetimes, or

the average time from charge carrier creation to recombination. Material imperfections areespecially important for conventional solar cells where material thickness implies that thecharge carriers may have to pass through relatively more material to reach the junction.

4.5.3 Parasitic Resistance

In the same way as electrical sources can be symbolised with internal resistance losses, solarcells suffer from two parasitic current losses (ISR and IPR) caused by series and parallelresistance respectively (see Figure 4.19). The series resistance, RS, is due principally to bulkresistances of the semiconductor and its metallic contacts as well as the contact resistancebetween these two (materials). The parallel resistance, RP, can be caused by leakage at

Suboptimal Solar Cell Efficiency 71

RP

RS

RA

ISR

U

IPRIDiode

IL

+

–

Figure 4.19 Solar cell equivalent circuit and load RL, based on [58, p. 307]

the cell edge or extended defects in the depletion zone. Extended defects associated withlow-cost solar cells include grain boundaries, dislocations and large precipitates. These sitesbecome areas likely to suffer (doping) diffusion or fine metallic bridges from the contactmetallisation.When a resistance is connected to a solar cell, such as an application of resistance RA

(represented by the dashed lines in Figure 4.19), two parasitic effects are found: currentdeteriorates the higher the RS, while voltage deteriorates the lower the RP. Both thesedetrimental cases reduce the FF as IM and VM (see Figure 4.15) respectively are shiftedtowards the origin on an I–V graph.

4.5.4 Efficiency Losses Summary

It is easier to appreciate the previous section by summarising the different efficiency lossesin a single chart (Figure 4.20).As can be seen in Figure 4.20, the principal effects are the incident photon spectral mis-

match having either more or less energy than the band gap (h)>Eg or h)<Eg respectively).A third significant power loss is due to the collection efficiency of photogenerated chargecarriers being less than unity. This implies that the VOC is less than ideal �qVOC<Eg�.

Less significant power losses include front surface losses (such as reflection or area loss)as well as non-unity quantum efficiency �nQ< 1� and FF �FF< 1�.

Front surface

FF < 1 nQ < 1

hυ > Eg

hυ < Eg

qVOC < Eg

Figure 4.20 Efficiency loss summary for a conventional silicon cell at AM1.5 [based on76, pp. 242–3]


4.6 IPV MATERIAL TECHNOLOGIES

Solar cells can be produced in a large number of ways, as presented in subsection 4.3.3 andelsewhere [99, Chapter 9]. The aim of this section is to provide a more detailed review ofthose technologies that have been or could be applied to IPV.IPV solar cell technologies in common use today are therefore reviewed individually in

subsections 4.6.2 to 4.6.4. Other technologies that can be considered but are not presentlyin use for IPV are covered in subsection 4.6.5.Few IPV applications have a voltage requirement of less than a few volts, while cells

deliver less than 1V each; the series connection of cells to form modules is therefore brieflydescribed in subsection 4.6.6. This is followed by a section on efficiency improvements(section 4.7) that apply to the majority of solar cells.

4.6.1 Spectral Response

Given that spectral mismatch is the most significant loss of efficiency in subsection 4.5.4, it isworthwhile comparing the spectral response of a number of technologies (Figure 4.21) withthe spectra encountered indoors (Figure 4.22). Naturally, it should be remembered that thosephotons with the highest energy (UV) are on the left of the graphs, and the lowest-energyphotons, therefore of less relative interest, are on the right.

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

400 600 800 1000 1200

Wavelength (nm)

Spec

tral

Res

pons

e (A

/W)

Zno/CIGS

monoX-Si

a-Si

CdTe

PolyX-Si

Figure 4.21 Typical spectral response for solar cell technologies: a-Si = amorphous silicon;CdTe = cadmium telluride; Poly-Si = polycrystalline silicon; monoX-Si = monocrystalline

silicon; ZnO = zinc oxide; CIGS = copper indium gallium diselenide (based on [80])

IPV Material Technologies 73

400 600 800 1000

1.0

0.8

0.6

0.4

0.2

0

a-Sisinglecell

Fluorescent lamp

Sunlight behindwindow

Incandescentlamp

Wavelength (nm)

Spec

tral

Res

pons

e

Rel

ativ

e In

tens

ity

Figure 4.22 Qualitative comparison of light source spectra and an amorphous silicon (a-Si) singlecell, based on [214, p. 167] and reproduced with permission of Werner Roth

The better match between the fluorescent source and amorphous silicon (a-Si) compared withthe other samples can be seen in Figure 4.22, which helps to explain why this technologyis well adapted to IPV. However, the technology of choice for indoor applications dependson many other factors such as cell cost, application location in the room, colour of cellrestrictions and application use pattern.

4.6.2 Amorphous Silicon Thin Film

This thin-film technology has been applied to IPV for decades and is still the most significanttoday. It has a material structure without long-range order and a band gap, Eg, of 1.7 eV,which is much higher than that of crystalline silicon. Deposition of the successive layersof silicon requires silane gas �SiH4� mixed with around 10% hydrogen in the presence ofa plasma in vacuum. The hydrogen passifies the amorphous material structure, reduces itsresistivity and allows it to be doped.The two doped areas need only be very thin (of the order of 10–20 nm) for charge

separation, but, as silicon is an indirect band gap material, greater material thickness isrequired. The latter is achieved by depositing an undoped (intrinsic) layer between the p-layerand n-layer. Typical i-layer thickness is in the 1/m range. The result is a p–i–n structure.This i-layer deteriorates with illumination of the solar cell, reducing initial module effi-

ciencies of around 6% by as much as one-third after extended outdoor illumination (calledthe Staebler–Wronski effect [240]). Given the much lower radiant energy intensity foundindoors, it is known that such a loss of efficiency will usually be negligible.For outdoor applications, a partial solution to the degradation problem has been found by

depositing successive p–i–n junctions one on top of the other; these are known as tandemmodules. Apart from the higher resulting voltage, this solution has the advantage of enablingthe Eg of each layer to be preferentially adjusted to a different spectral response. Thisincreases efficiency (8–10%). Moreover, such cells degrade less (around one-sixth of startingefficiency).


4.6.3 Polycrystalline Thin Film

There are two main compounds of interest in this group: the II–IV CdTe [184] and theI–III–VI CuInSe2 (or CIS) [246]. These cells are both produced as heterojunctions withcadmium sulphide (CdS). As CdS absorbs light, it is deposited thinly (around 50 nm).Where the two compounds differ is in band gap. CIS has an Eg of 1 eV which is relatively

low for solar cells. It is usual to alloy the indium with gallium to increase the gap. Suchcells are called CIGS (CuIn GaSe2). CdTe on the other hand has a near-ideal band gap foroutdoor application of 1.4 eV.One of the ‘bugbears’ of these technologies is the cadmium content which it is frequently

claimed to be a risk to health or the environment. Such statements are rarely supported byevidence and seem to be motivated by fear and uncertainty.Risk is defined as the product of hazard and the probability of occurrence. With regard

the hazard, while it has been shown that the elements cadmium and tellurium can be toxic,cadmium telluride is relatively inoffensive [163, p. 223]. The highest hazard found is occu-pational (to production staff) [90], whose working environment should be controlled by goodhousekeeping, for example. The potential hazard to the public is related to the productionmethod used (some involve less toxic waste than others) and accidental combustion of themodules in a fire.Both these hazards should be considered in the context of their likelihood of occurrence;

there is natural pressure on manufacturers to use material-efficient processes, as much forregulatory as for economic reasons. The options for achieving this exist [169]; therefore,manufacturers are motivated to keep waste and pollution to a minimum.Fire-related hazards would require temperatures above the CdTe melting point of 1041 �C.

Such a situation is unlikely for a large installation (>100kWP such as a PV farm). For asmaller domestic system (e.g. 5 kWP), those in the immediate vicinity of the fire would beat risk for a few minutes and there would be no hazardous ground level contamination [163,p. 226].As another report [2] states, ‘the risks from cadmium or selenium use in CdTe and CIS

modules respectively seem acceptable in comparison with some existing products or serviceslike NiCd batteries or coal-fired electricity production’.

4.6.4 Conventional Silicon Cell

Although for the last half-century this has been the technology most applied for outdoorapplications, such cells are rarely used for IPV applications. An exception is the Neprowatch in Figure 4.5. This is due not only to the relatively low band gap of 1.1 eV, whichoffsets the spectral response towards the infrared range, but also to the inconvenience ofcreating series connections between relatively small �<1cm2� cells. Furthermore, such cellsare typically too expensive for the low currents found indoors.Conventional cells are mainly based on an ingot of silicon which can be produced in a

number of ways. This ingot is then cut into wafers. Usually it is pre-doped p-type (e.g. withboron). The outside of the wafer is then thinly n-doped (e.g. phosphorus POCl3); this layeris then removed from one side and the edges of the wafer to leave a p–n junction.

IPV Material Technologies 75

4.6.5 Other PV Technologies

A number of other PV technologies have been considered in this report but, because theyare relatively new on the market, cannot be recommended for IPV applications at present.

1. Dye Solar cells. This liquid-state technology (as with Becquerel’s first solar cell, seesection 4.2) is based on a semiconductor (titanium dioxide, TiO2) layer impregnated witha photosensitive dye. This layer along with a liquid electrolyte is ‘sandwiched’ betweentwo conducting glass electrodes. As the semiconductor band gap of TiO2 of 3.2 eV is toohigh for light collection, the dye acts as to promote electrons into the TiO2 conductionband. These may then make their way to the electrodes under the influence of an electricfield inside the bulk material. The electrolyte (redox solution, typically of I−/I−3 ) maybe based on a molten salt, butyrolactone or acetonitrile. Owing to its liquid and reactivenature, one of the challenges of bringing dye products to market is sufficiently long-lasting sealing. It is believed that LCD technology might be an interesting avenue ofresearch at this level [241]. It would be of interest to designers of IPV applications if dyecells were more widespread, as they seem well adapted to indoor light conditions [31];further references can be found in [200].

2. Thin-film crystalline Si. Other thin films of silicon, not to be confused with amorphoussilicon, are microcrystalline [127] and polycrystalline thin films. These can be depositedby chemical vapour deposition (CVD) or from solution onto low-cost substrates. Thesecells are especially thin (1�m for polycrystalline, 3�m for microcrystalline) and there-fore, like amorphous cells, require optical trapping of the light. Theoretically, these cellscould achieve better efficiencies than conventional cells, owing in part to the reduced bulkcarrier recombination. The bulk refers in this case to the semiconductor material apartfrom the junction through which the charge carriers must pass to reach the electrodes.Because of its non-ideal conductivity, this area is a cause of electrical losses.

3. Space cells. A number of material alloys are generally considered only for space applica-tions on account of their cost. These include amorphous silicon with germanium [100] andgallium-based [123] alloys. A special case for ‘deep space’ (low irradiation) explorationis ‘low-intensity/low-temperature’ (LILT) solar cells. Given that typical temperatures of−135 �C (much lower than IPV) and irradiances of 50W/m2 (relatively high for IPV)are considered [242], little can be concluded here from such work. Should their pricereduce, these cells would also be of greater interest to the IPV designer.

4.6.6 Thin-film Modules

The commercial thin-film technologies described in section 4.6 are only available on themarket as ‘monolithically’ (literally ‘one stone’, in practice ‘one piece’) series-interconnectedmodules. Such an approach is scalable from high-volume manufacturing of high surface area(of the order of 1m2) products down to minimodules (around 100 cm2) [126] and miniatureproducts (less than 1 cm2) [131] [223].The processing begins with cleaning of the substrate sheet (typically glass), followed by

deposition of the transparent conductive oxide (TCO) if this is not already present. The TCOis then scribed into long straight bands of around 1 cm in width (typically by laser) before thedeposition of the solar cell material over the entire surface. A second scribing then followswhich matches the first except for a small �<1mm� offset in a direction perpendicular to the


Glass

Moly

CIS

ZnOCdS

Figure 4.23 Example of a series-interconnected thin-film technology solar module. In this caseeach CdS/CIS PV cell has a molybdenum (Moly) back contact and a zinc oxide (ZnO)

transparent front contact ([245], reproduced with permission)

strips. Lastly, the back contact is deposited, often by sputtering if it is metallic. It is scribedin the same way as the second scribing step. This results in a solar module where the topof each cell (one of the strips) is electrically connected with the bottom of the adjacent cell,forming a series-connected module such as Figure 4.23. This has the advantage that eachmodule has much higher voltage than would have been the case without series connection;however, if the cells do not deliver similar current, the lowest-performing cell will lower thecurrent of the entire module.During the remainder of the manufacturing process, ribbon or wire contacts are added, the

entire module is sealed by lamination (typically with epoxy and often with a second sheetof glass) and the necessary quality tests are performed.Thin-film modules have several advantages for IPV use over conventional cell modules as

they can be deposited with the required voltage and surface area design. IPV products alsobenefit from the progress made for outdoor cells with respect to the series interconnection.As mentioned above, outdoor modules require uniform deposition over areas of the order ofm2. As IPV products have surface areas measured in cm2, it can be expected that currentmatching issues due to deposition variation can be ignored.A further benefit of thin films is that they can be deposited on flexible substrates. These,

although not usually transparent (like glass), have the promise of both continuous (roll-to-roll) manufacturing and new applications without the constraint of using a flat glass surfacemodule. This may well be important to IPV as it may be advantageous to have the solar cellbetter respect the shape of the product; also, the flexible substrate is usually thinner than aglass solution, which may be beneficial where space is limited.

4.7 EFFICIENCY IMPROVEMENTS

Efficiency can be altered both at the design stage and during manufacturing of PV. At thedesign stage it is still reasonable to proceed with respect to the scientific principles and idealsmentioned above to which PV can aspire. These are hard to maintain on the production line

Efficiency Improvements 77

where further imperatives such as units produced per unit time and low-cost compete forlimited resources.When considering the improvements possible from the scientific perspective, each of the

heading level parameters in section 4.4, namely VOC, ISC and FF, can be reviewed. Efficiencyis related to the product of all three [equation (4.3)], so an improvement in any one withoutdetriment to another is beneficial to overall efficiency.

4.7.1 Current and ISCJust as the principle loss area for outdoor PV is optical (subsection 4.5.1), this can beexpected to be the more so for IPV given that indoor use is characterised by an increasedvariety of light sources (i.e. fluorescent, incandescent, window filtered daylight). An avenuefor improving efficiency via the current would be to review the PV spectral response atthe radiant energy intensities and spectra found indoors. Reduced spectral mismatch wouldimprove ISC and might be possible by an adjustment of band gap Eg by changing the materialchemical composition, for example. This application-specific approach would be hinderedby the mix of spectral signatures that IPV applications may receive.With regard the parasitic resistance, at IPV radiant energy intensities it can be assumed

that series resistance is relatively unimportant as the current flow is relatively low.Other technical solutions for the improvement of current performance are also of interest,

e.g. tandem (stacking) of cells, antireflection coating (subsection 4.5.1), improved lighttrapping and improved TCO, but these need to be balanced against how cost effective theyare for the specific application.

4.7.2 Voltage and VOC

While current is proportional to intensity, and therefore directly related to IPV radiant energyintensities, voltage is less affected, i.e. it varies with the log of intensity [see equation (4.7)].At indoor intensities, improvements of the voltage can be achieved by maximising theparallel resistance, as the related leakage current may otherwise be comparable with thephotocurrent.Improvements are also possible by improving the photocell ideality. The case of an ideal

junction is described mathematically in equation (4.7). In practice, equation (4.11) can beused [113]. Here, A represents how ‘ideal’ the junction is and takes a value between 0 and 1:

VOC =A[kT

qln(ILI0

+ 1)]

(4.11)

Therefore, maximising ideality A, or ‘junction perfection’, will also benefit VOC.

4.7.3 Fill Factor

The fill factor is also dependent on parasitic resistance (subsection 4.5.3). Like the voltage,at low radiant energy intensities it is more affected by low parallel resistance.


4.7.4 Concentration

A technique for efficiency improvement that has been tried outdoors is concentration. Thisis the PV jargon for the use of mirrors or lenses, such as to focus the radiant energy onto asmaller surface [270][180]. The main disadvantage is that the resulting device is ‘directional’,i.e. it mainly collects direct light [139]. This is mitigated outdoors with a ‘sun-tracking’device which ensures that the solar cell is facing as close to the sun as possible. Clearly,such tracking devices are hard to conceive indoors. However, some focusing of the radiantflux can be imagined in the case of a majority of direct radiation.

4.8 CONCLUSION

‘Do not re-invent the wheel’ summarises the message of this chapter to IPV designers.The idea that they should be familiar with a technology such as PV is much more than anacademic nicety. The aspects considered in this chapter support this as follows.Firstly, the history of PV technology (see section 4.2) provides the designer with an

impression of what has been achieved. Equally, what remains to be done can be inferred.For example, no wireless (communicating) PV-powered indoor products are as yet availableon the market. The rest of this thesis shows what would be required for such products to seethe light of day.Secondly, a grounding in the science of a technology is essential. It may help the designer

to understand how solar cells work (sections 4.3 and 4.4) and the advantages and limitationsof PV (section 4.6), and to see the improvements that can be achieved (section 4.7).The main limitation to PV performance is related to a mismatch between the spectrum of

the incident radiant energy and the spectral response of the solar cell. Other issues includethe recombination of charge carriers and effects which are associated with the deteriorationin the fill factor (FF). Ways of improving solar cells for indoor use therefore aim principallyto reduce the spectral mismatch and increase the parallel resistance.To what extent the improvements mentioned in section 4.7 will be beneficial in practice

will depend on designers’ success in inducing the PV manufacturer to make the necessarymodifications. Manufacturing structures appropriate for the production of outdoor-use mod-ules may hinder changes that would serve indoor products. Reading sections 4.3 to 4.7 willhelp equip the designer to face this challenge.

4.9 FURTHER READING

For those wishing to further their understanding of PV, the two references from this chapterthat will prove most useful are by Green [99] and Hovel [113]. For more history, see Perlin[198] [197].

5Solar Cells for Indoor Use

5.1 INTRODUCTION

This chapter explores the correct design of indoor photovoltaic (IPV) cells and modules.Firstly, existing solar cells under indoor conditions are investigated (see section 5.2), mod-elled (see section 5.3) and discussed (see section 5.4). Then, in section 5.5, the ways inwhich future IPV modules may be improved are considered.As has been seen in Chapter 3 (see Table 3.4), the parameter that may vary the most

in the built space is radiant energy intensity. Therefore, having characterised a particularenvironment, the IPV product designer will then wish to establish how the various solarmodules will respond under these conditions.Some reports on the use of PV indoors have been published, including [214] [199] [173]

[92]. In comparison with these, this chapter provides the IPV designer with a more completeappreciation of PV module performance indoors, especially with respect to technologychoice. This work is both novel and important for the same reason: solar cell comparisonis generally based on an arbitrary maximum terrestrial intensity and spectrum (of 1 sun or1000W/m2) at 25�C perpendicular to the cell plane. In practice, no solar cell experiencessuch standard test conditions (STC) [116] that ‘combine the irradiation level of a clearsummer day, the module temperature of a clear winter day and the spectrum of a clear springday’ �26�, yet few alternative bases for comparison exist �27�.One way is to compare real modules outdoors [37] [68] [4] [45], but these conditions, when

compared with those indoors, are characterised by much higher radiant energy intensities(a factor of 10 or more), fewer spectrum types (such as the various electrical light sources)and more temperature variation. Further issues for solar cells used outdoors are coveredin [129].IPV products generally receive both daylight and electrically sourced spectra. Given the

lack of comparable data in existing reports of PV indoors, it was elected to test a wide rangeof possible technologies under both these spectrum types and a range of radiant intensitiesfrom AM1.5 �1000W/m2� down to those found indoors �1W/m2�.


80 Solar Cells for Indoor Use

The models presented in section 5.2.2 simulate PV response across a range of intensitiesand under both spectra. This is complemented by a heuristic model which shows greateraccuracy still. These may permit the designer to infer properties based on more freelyavailable AM1.5 data.As a complement to the above information and section 4.7, certain potential performance

improvements of photovoltaic modules are described in section 5.5.The reader is assumed to be familiar with photovoltaic technology in this chapter. For a

more basic introduction, see Chapter 4.

5.2 TECHNOLOGY PERFORMANCE AT INDOOR LIGHTLEVELS

5.2.1 Experimental Procedure

‘Standard test conditions’ (STC) [116] similar to 1 sun directly overhead on a cloudless dayat 25 �C were reproduced with a Wacom solar simulator connected to a Keithley 238 high-current source, a Keithley 705 scanner and a PC running I/V measurement software. Beforetesting began, the lamps of the solar simulator (a Xenon XBO 1000W and a Halogen OsramCP/72 2000W) were switched on and allowed to stabilise for 20 min. The room temperaturewas controlled to 22 �C�±3� with air conditioning. The light intensity was controlled withone or more wire mesh filters between the light source and the sample (see Figure 5.1).Three samples of each cell code mentioned in Table 5.1 were measured individually withfour point contacts (two positive and two negative contacts) except for the samples fromEdmund Scientific owing to lack of contact space. Current/voltage �I/V � measurementswere made for the following percentages of 1 sun: 100%, 58.2%, 39.7%, 19.1%, 11.0%,4.1%, 1.1%, 0.439%, 0.211%, 0.08%. As precautions, the filters were positioned on thesupport in the same orientation and the light source shutter (which otherwise prevented

Temperaturecontrolledenvironment

IV measurementPC, current sourceand scanner

LightWACOM solarsimulator

Wire meshfilter(s)

Lens

Sample

Figure 5.1 Experimental apparatus

Table 5.1 Technologies and sources of the 21 types of solar cell/module that the author collected and tested, showing whether the manufacturer wasa laboratory or industry, the active area and the number of cells in the module

Technological classification Supplier or laboratory name Cell codeIndustrial = ILaboratory = L

Active area�cm2�

Number ofcells in module

Silicon (crystalline) BP Solar (via IWS) xSi-BP I 936 1Silicon (crystalline LGBC) BP Solar, UK xSi-LGBC I 090 1Silicon (crystalline) Spacecells, Edmund

Scientific, USAxSi- EdSi I 038 1

Silicon (crystalline) Unknown (via distributor) xSi-Dist I 1095 1Silicon (polycrystalline) MAIN, Tessag, Germany pSi-MAIN I 1247 1Silicon (polycrystalline) EFG, Tessag, Germany pSi-EFG I 1025 1Silicon (polycrystalline) Unknown (via distributor) pSi-Dist I 288 1III–V cells (GaAs) NREL, Golden, CO, USA 3-5-NREL L 025 1Polycrystalline thin film (CdTe) Matsushita/Panasonic, Japan CdTe-Mats I 580 5Polycrystalline thin film (CdTe) Parma University, India CdTe-Parm L 079 1Polycrystalline thin film (CIGS) ZSW, Stuttgart University,

GermanyCIGS-ZSW L 046 1

Other (GalnP) NREL, Golden, CO, USA GIP-NREL L 025 1Amorphous silicon Tessag, Putzbrunn, Germany aSi-Tess I 495 5Amorphous silicon Sanyo Electric, Hyogo,

JapanaSi-Sany I 371 4

Amorphous silicon Solems, Paris, France aSi-Sole I 176 3Amorphous silicon VHF Technologies, Le Locle,

SwitzerlandaSi-VHF L 336 4

Amorphous silicon Sinonar Corporation, Taipei,Taiwan

aSi- Sino I 126 4

Amorphous silicon Millenium, BP Solar aSi-BP I 020 1Photochemical (nanocrystalline dye) Greatcell SA, Yverdon,

SwitzerlandPC-GCSA L 100 1

Photochemical (nanocrystalline dye) EPFL ICP2, Lausanne,Switzerland

PC-ICP2 L 090 1

Multijunction cell (GaAs–GalnP tandem) NREL, Golden, CO, USA MJ-NREL L 025 1


light reaching the sample) was only allowed to be open during the time that a measurementwas being made. The latter contributed to minimising the heat transfer from the lightsource to the sample. Furthermore, the sample area temperature was homogenised witha small fan.Some samples were also tested under a separate artificial Erad source (the fluorescent

Philips Ecotone PL-L, 830/4P HF, 40W) over a smaller range of intensities �1–5W/m2�.The intensity from this artificial light source on the solar cell was varied by controlling thedistance �>1m� between the source and the solar cell under test.Solar cell temperature outdoors is a factor of both the surrounding climate and the fact that

solar cells thermalise a significant proportion of the incoming radiant energy (see section 4.5).The indoor environment is characterised by a smaller temperature range and much lowerradiant energy intensity which might warm the module. Therefore, these experiments wereperformed at a temperature of �22± 3 �C�. This contributed to reducing the uncertaintyrelated to changing more than one variable at the same time, often the case in outdoorcomparative testing.Samples were selected by the author to be representative chiefly of those cells that the

IPV designer would be likely to use at present (e.g. amorphous and crystalline silicon).Some samples of more ‘exotic’ technologies (e.g. based on gallium) were also included forcomparison purposes. Samples whose active area exceeded 5cm× 5cm�25cm2� were notconsidered, firstly to afford relative comparability between samples and secondly as manyIPV product PV modules are no larger than this size.

5.2.2 Results

xSi-BP xSi-LGBC xSi-EdSi xSi-Dist

pSi-MAIN pSi-EFG pSi-Dist CIGS-ZSW

0

2

4

6

8

10

12

14

16

0 200 400 600 800 1000

Rel

ativ

e E

ffic

ienc

y (%

)

Light intensity (W/m2)

Rel

ativ

e E

ffic

ienc

y (%

)

0

2

4

6

8

10

12

14

16

1 10 100 1000Light intensity (W/m2)

0.1

Figure 5.2 Author’s measurements of efficiency of all samples under a wire mesh filtered AM1.5source at 1000W/m2, showing the same results against intensity on the base 10 scale (left) and

natural logarithm (right). The log scale slope has been used to sort the results between the top andbottom graphs (see phenomenological model)

Technology Performance at Indoor Light Levels 83

MJ-NREL 3-5-NREL GIP-NREL CdTe-Mats CdTe-Parm

aSi-Tess aSi-Sany aSi-Sole aSi-VHF aSi-Sino

aSi-BP PC-GCSA PC-ICP2

Rel

ativ

e E

ffic

ienc

y (%

)


0

2

4

6

8

10

12

14

16

0 200 400 600 800 1000

Rel

ativ

e E

ffic

ienc

y (%

)


0

2

4

6

8

10

12

14

16

1 10 100 10000.1

Figure 5.2 (Continued)

5.2.3 Efficiency with Intensity

The two graphs on the left of Figure 5.2 show that solar cell efficiency in the highest-intensitydecade, 100–1000W/m2, varies less than in the lower decades. The ranking by 1 sunefficiency is almost completely maintained down to 200W/m2, as has been found elsewhere[28]. For intensities below 100W/m2 (see Figure 5.2, right-hand graphs), which are typicalof indoor conditions, a much more marked change is found and the ranking by technologyis altered when the lowest intensities are reached, so that some of the highest-performingcells at 1 sun were the weakest at 1W/m2.For cells of some technologies (including amorphous silicon) the loss of efficiency from

1000 to 1W/m2 was less than 50%, while for other technologies (such as mono- andpolycrystalline silicon samples) the drop was greater than 65%.

5.2.3.1 Open-circuit voltage with intensity

The effect of intensity change on VOC has already been presented [206]. The open-circuitvoltage VOC over the range tested is relatively linear on the logarithmic scale with a negativegradient. In Figure 5.3 the open-circuit voltage has been normalised to one cell for ease ofcomparison.It is noteworthy that the samples appear to form two distinct groups. The ‘upper group’,

with open-circuit voltage in the 0.7–0.9V range at 1000W/m2, is formed exclusively ofamorphous silicon, cadmium telluride and dye cell samples. The ‘lower group’ members areall of crystalline, polycrystalline or CIGS technologies and have an open-circuit voltage at1000W/m2 in the 0.5–0.6V range.


0.0

0.2

0.4

0.6

0.8

1.0

1 10 100 1000

Daylight intensity (W/m2)

Ope

n-ci

rcui

t vol

tage

(V

)Upper group

Lower group

Figure 5.3 Open-circuit voltage per cell with respect to daylight spectrum intensity for 16 of thesamples from Table 5.1 (author’s measurements)

5.2.4 Spectral Response

The graphs in Figure 5.4 compare the efficiencies under filtered AM1.5 with those foundunder the fluorescent source for selected samples representing three technologies. Theseindicate, for the samples tested, an increase in efficiency for the amorphous silicon, a slightchange for the dye cells and a decrease in efficiency for the crystalline silicon samples. Theresult for the crystalline silicon samples is suggested elsewhere [173].

0

2

4

6

8

0 2 4

Intensity (W/m2)

Rel

ativ

e ef

fici

ency

(%

)

aSi-Tess Filtered AM1.5aSi-Sino Filtered AM1.5aSi-Tess FluorescentaSi-Sino Fluorescent

increase

4.0

4.2

4.4

4.6

4.8

5.0

0 2 4

Intensity (W/m2)

Rel

ativ

e ef

fici

ency

(%

)

PC-ICP2 Filtered AM1.5PC-GCSA Filtered AM1.5PC-ICP2 FluorescentPC-GCSA Fluorescent

slightchange

0 0

2

4

6

0 2 4

Intensity (W/m2)

Rel

ativ

e ef

fici

ency

(%

)

xSi-EdSi Filtered AM1.5xSi-BP Filtered AM1.5xSi-EdSi FluorescentxSi-BP Fluorescent

decrease

Figure 5.4 Author’s measurements of efficiency difference going from filtered AM1.5 to thefluorescent spectrum for two samples of amorphous silicon cells (left) and dye cells (middle) and

crystalline silicon (right)

Indoor Light Level Model Presentation 85

The fluorescent intensity was measured using a lux meter and then converted to W/m2

using the (simplified) relationship

Erad�lux�120000

= Erad�W/m2�

1000(5.1)

or

Erad�W/m2�= Erad�lux�

120(5.2)

5.3 INDOOR LIGHT LEVEL MODEL PRESENTATION

5.3.1 Phenomenological Model

The modelling of the maximum power point on the I–V curve under the influence ofeither RS or RP has been modelled [95]. However, neither of these factors nor which ofthe respective parasitic currents (see section 4.5.3) will be prevalent is known in advance.A model is therefore proposed below that requires none of this information, and focuses onthe efficiency rather than maximum power.For the samples in Figure 5.5 it can be seen that FF is approximately constant in the range

1–100W/m2. This was found to be relatively valid for some samples; for example, the FFmaximum–FF minimum for the range 8–100W/m2 was 2% for CdTe and 5% for a-Si anddye cells. Polycrystalline silicon and monocrystalline silicon varied over the same intensityrange by 22–23%.It is also well accepted that ISC is directly proportional to G over a range wider than is

considered here; in this case ,fG is used instead of ISC, where ,f is a constant:

ISC =,fG (5.3)

The solution of the ideal diode [equation (4.5)] for VOC has been shown to be [244]

VOC �kT

qln(ISCIS

)(5.4)

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.00.1 1 10 100 1000

FF

Tessag Solems BP Mill

Intensity (W/m2)

Figure 5.5 Fill factor (FF) with respect to intensity on base 10 logarithmic scale for selectedamorphous silicon samples (author’s measurements)


where IS is the saturation current and kT/q is the thermal voltage. If IL/IS>>1, then a termfor RP is not required, such as proposed elsewhere [157]. Substituting equations (5.4) and(5.5) into equation (4.3) yields

%�FF,f

kT

q�ln�G�+ ln�,f�− ln�IS�� (5.5)

IS can be found [164], p. 29 [99] using the approximate formula

IS �1f exp(−Eg

kT

)(5.6)

where Eg is the band gap, 1f is relatively constant and temperature, T (K), is held constantfor the experiments. Constant 1f is of the form

1f = eACSNCNV

{DnFp

LnNA

+ DpFn

LpND

}(5.7)

where

ACS = cross-sectional areaNC�NV = effective densities of states in conduction and valence bandsDn�Dp = diffusion coefficientsLn�Lp = diffusion lengthsNA�ND = densities of acceptors and donors

For non-research purposes, it will suffice to know that 1f has been estimated [99], p. 88as having a value of 105 A/cm2 for crystalline silicon. Substituting equation (5.6) intoequation (5.5) gives

%�FF,f

kT

q

[ln�G�+ ln�,f�− ln�1f�+

Eg

kT

](5.8)

which in the form

%= a ln�G�+ b (5.9)

has

a=FF,f

kT

q(5.10)

and

b=FF,f

kT

q

[ln,f + ln1f +

Eg

kT

](5.11)

From the right-hand graphs of Figure 5.2 it can be seen that the overall trend is a straight lineon a logarithmic scale and therefore a good fit with an equation of the form of equation (5.9).This is particularly the case in the range 1–100W/m2 and for the lower right-hand graph.


y = 1.6ln(G) + 5.2 CC = 0.99 y = 0.4ln(G) + 5.5

CC = 0.97

y = 0.49ln(G) + 3.7 CC = 0.99

y = 0.2ln(G) + 3.2 CC = 0.99

0

2

4

6

8

10

12

14

1 10 100


Rel

ativ

e E

ffic

ienc

y (%

)

aSi-Sole aSi-BP CdTe-Parm xSi-LGBC

Figure 5.6 Fit of author’s measurements for selected samples from Figure 5.2

In Figure 5.6 the efficiency range 1–100W/m2 is magnified and compared with a linewhose equation is of the same form as equation (5.9). This latter relationship was appliedto the data for all samples in the range 08–100W/m2, and the results shown in Table 5.2suggest a good fit [linear correlation coefficient (CC) over 0.98].An ideal cell for IPV use therefore has a low a (gradient on the logarithm graph) and high

b (intersect of the efficiency axis). Such ideal characteristics are displayed by those samplesthat perform best in these experiments under indoor light conditions (e.g. Ga compounds).For convenience, the parameters a and b can be combined to form a single parameter that

is equal to a/b by adjusting equation (5.9) as follows:

%= b[(ab

)ln�G�+ 1

](5.12)

where

a

b= 1

ln,f − ln1f + EgkT

(5.13)

As can be seen from equation (5.13), with ,f�1f and T constant, a/b is inversely related toband gap. In Table 5.2, the values of a/b are shown for all samples tested. They are rankedby a/b, and, although this relationship to band gap is not found, it can be seen for thetop eight lines that for a/b greater than 0.3 the approximate material band gap is less thanor equal to 1.1. The remaining cases, where a/b is always less than 0.2, the approximateband gap (where known) is greater than 1.4. It may be inferred that the latter values of a/b


Table 5.2 Parameters of the author’s phenomenological model [equation (5.9)] inthe 1–100W/m2 range for all samples tested by the author under AM1.5

Cell code aAM bAM CC aAM/bAM ∼Eg

pSi-EFG 233 048 099 487 110pSi-MAIN 205 �−�084 092 243 110xSi-EdSi 150 216 100 069 110xSi-BP 171 258 100 066 110CIGS-ZSW 173 283 098 061 090pSi-Dist 167 280 100 060 110xSi-Dist 140 288 099 048 110xSi-LGBC 162 518 100 031 110aSi-Sino 040 206 095 019 170aSi-VHF 033 230 096 014 170aSi-BP 049 366 100 013 170CdTe-Mats 038 330 100 011 140PC-IPC2 042 425 095 010 —PC-GCSA 040 437 096 009 —MJ-NREL 074 902 099 008 —aSi-Tess 043 541 099 008 170CdTe-Parm 042 551 097 008 140aSi-Sany 017 239 098 007 140aSi-Sole 022 321 099 007 1403–5-NREL 057 872 098 007 150GIP-NREL 036 859 100 004 140

Note: aAM and bAM are values of a and b under AM1.5 source; cell codes are defined in Table 5.1; averageCC for all 21 samples = 098

and band gap may be appropriate for IPV; in general, the lowest a/b is recommendable forIPV design.Two distinct technological groups are also found when ranking the results with respect

to a. Those cells with a value of a greater than 1.4 (including monocrystalline silicon,polycrystalline silicon and CIGS) and a in the range 0.17–0.74 (amorphous silicon, CdTe,gallium compounds and dye cells). These two modes have already been identified [206].However, this is the first time that numerical variables have been associated with them.Neither a nor a/b therefore have been shown to be related to band gap. This can be

attributed not only due to the comparison of both cells and modules in the same experimentbut also to the variety of processes and methods used to produce the samples.Table 5.3 shows the parameters of equation (5.9) for the results in Figure 5.4. These

are then compared in Table 5.4 with the values under AM1.5 (Table 5.2). As can be seen,for the amorphous silicon the parameters increase by 13–34%. For the dye cell samples,parameter b changes little and there is an 18–36% increase in parameter a. For the crys-talline silicon samples the parameters decrease by 51–62%. This indicates that the lattersamples are affected more by spectral mismatch (product of the incident spectrum and thespectral response of the cell integrated over the response range) than the amorphous anddye cells.


Table 5.3 Phenomenological model [equation (5.9)] parametersover the 1–100W/m2 range for samples under fluorescent source

Cell code aF bF CC aF/bF

xSi-BP 082 126 100 065xSi-EdSi 057 095 100 060aSi-Sino 046 276 100 017aSi-Tess 057 611 100 009PC-IPC2 057 426 100 013PC-GCSA 048 429 099 011

Note: aF and bF are values of a and b under fluorescent source; average CC for allsix samples = 100

Table 5.4 Ratio of fluorescent parameters(Table 5.3) to AM1.5 parameters (Table 5.2)

Cell code aF/aAM bF/bAM

xSi-BP 048 049xSi-EdSi 041 033aSi-Sino 115 134aSi-Tess 132 113PC-IPC2 136 100PC-GCSA 118 098

5.3.2 Heuristic Model

Another equation from [64] that for the moment remains without complete explanation isnevertheless noteworthy given its good fit to the results presented here. The full equation isas follows:

%=p[qsGn

Gn0

+(Gn

Gn0

)ms][

1+ rs7

70

+ ssAMAM0

](5.14)

where the following constants have the values Gn0=1000W/m2�70=25 �C and AM0=15.

The variables light intensity Gn, temperature 7 and spectrum (AM) are accounted for;p�qs�ms� rs and ss are parameters. Given that, for the present experiment, temperature andspectrum were kept constant, the above equation may be simplified to

%= as�bsGn +Ghn� (5.15)

However, a slightly improved fit is obtained with

%=,s�1sG8sn +G9sn � (5.16)

As Table 5.5 shows, the fit of equation (5.16) is satisfying, and this over a range of intensityone order of magnitude greater than that applied to equation (5.9), i.e. 1–1000W/m2. Thisis therefore of interest to IPV practitioners.


Table 5.5 Fit of equation (5.16) for all samples tested by the author over the range1–1000W/m2 under the AM1.5 spectrum

Cell Code ,s 1s 8s 9s CC

xSi-EdSi 19.605 −0861 0.422 0.405 0996xSi-Dist 5.130 −0351 0.466 0.339 0993xSi-LGBC 12.267 −0554 0.342 0.285 0999xSi-BP 9.012 −0641 0.439 0.389 0993pSi-Dist 6.867 −0535 0.466 0.393 0997pSi-MAIN −14324 0959 0.094 −0033 0986pSi-EFG 44.670 −0956 0.563 0.558 09903-5-NREL 11.855 −0266 0.157 0.090 0992CdTe-Mats 3.419 −0044 0.484 0.136 1000CdTe-Parm 7.257 −0246 0.250 0.127 0996CIGS-ZSW 12.325 −0798 −0218 0.029 1000GIP-NREL 10.791 −0210 0.152 0.069 0999aSi-VHF 3.420 −0322 0.342 0.206 0998aSi-Tess 5.558 −0035 0.436 0.099 1000aSi-Sole 6.068 −0474 0.194 0.134 1000aSi-Sino 3.217 −0352 0.365 0.243 0983aSi-Sany 2.648 −0113 0.402 0.133 0996aSi-BP 5.019 −0267 0.257 0.156 0999PC-GCSA 4.499 −0052 0.514 0.140 1000PC-ICP2 11.473 −0630 0.265 0.209 0987MJ-NREL 8.644 0.061 0.064 0.064 0993

5.3.3 Technology-specific Models

While it is ideal for the IPV practitioner to have such pan-technological models of per-formance prediction, it is clear that each individual technology (e.g. amorphous silicon,cadmium telluride or dye cells) can and usually is modelled using mutually exclusive models.Equally, the modelling is at the level of a single and more fundamental parameter (e.g. VOC)rather than that of a combined parameter such as efficiency.Similar technologies can sometimes be modelled using similar approaches such as the

variable illumination method (VIM) for amorphous silicon [151] [152] and microcrystallinesilicon [153].

5.4 DISCUSSION

The effect of series resistance, RS, and parallel (shunt) resistance, RP, on solar cell I–Vcurves is known [99, pp. 96–7], and is explained in subsection 4.5.3 and can be modelledwith one or more diodes [185][186][20]. At high Erad intensity, high RS creates high seriescurrent IS which reduces FF, while at low Erad, low RP implies a high parallel current IPwhich reduces FF. The effect of IS and IP on efficiency can be seen in the gradient ofthe curves in Figure 5.2. This is consistent with the hypothesis [158] that, where RS issufficiently high for the shunt current IP to be negligible, the efficiency below 1 sun willfirst increase as radiant energy intensity is reduced. As can be seen, some samples in the

Designing PV Modules for Indoor Use 91

intensity decade 100–1000W/m2 indeed have such a negative gradient. Where the gradientapproaches zero, the efficiency is maximised. For IPV products in particular, knowledge ofthese efficiency variations can improve product and cell design as well as help to explainvariations in performance of ostensibly similar modules.As IP can be related to current paths through the solar cell or at its edges, the perimeter

length was compared with the estimated RP, but no clear relationship was found.For IPV solar cell design, the low Erad intensity implies that high RS should not have a

significant effect. This is fortunate as one of the contributions to RS is the contact resis-tance, which can be thought, of as a ‘leaky Schottky diode with non-linear characteristics’[77, p. 229]. The relatively small effect of high RS (and overly low RP) for the presentexperiments is indicated by the ‘flatness’ of the fill factor curves of certain samples such asthose shown in Figure 5.5; this suggests that these cells are in the range of radiant energyintensity in which they will best perform. Such an ideal range has also been described foroutdoor conditions with respect to efficiency [28].The effect of the incident radiant energy spectrum is of interest for the IPV practitioner.

It has been demonstrated in Figure 5.4 that amorphous silicon sample efficiency was higherwith the fluorescent spectrum (up 14–34% at 2W/m2) while crystalline silicon sampleefficiency deteriorated (down 51–57% at 2W/m2) compared with filtered AM1.5. It wouldbe interesting to test other samples as well as corroborate the results with comparisonof the spectral response. The latter was not pursued, as the selection of samples did notinclude a single-cell sample (a prerequisite for the spectral response equipment) for eachtechnology.

5.5 DESIGNING PV MODULES FOR INDOOR USE

Given that certain thin-film technology solar cells perform better at low radiant energyintensity, it is worthwhile reviewing other factors that can further improve the performanceof such modules. These include the transparent conductive oxide (TCO) in section 5.5.1and the dimensioning of the cells within the module (see section 5.5.2). Solar modulesdesigned for a specific indoor use may be able to benefit from cost savings associated withthe less stringent indoor environment compared with the outdoors. The benefits might beachieved by reducing the glass thickness; another way would be to use less or cheaperencapsulation.

5.5.1 Transparent Conductive Oxide

Thin-film solar cells use a TCO as the front contact through which the light passes (seeFigure 5.7). The design of this layer is a trade-off between its thickness, which will filterincoming radiant energy, and its resistance (e.g. Asahi U-type SnO2:F coated 1mm glasshas a sheet resistance of 125"/cm2 and a transmissivity of 82% for a TCO thickness of1�m [273]). At high radiant intensities, e.g. AM1.5 (1 sun), it may be reasonable to takea (thicker) TCO with a lower resistance while accepting that it filters more radiant energy.However, for indoor purposes, with the current to be conveyed orders of magnitude lower, athinner layer may well be sufficient and hold two further advantages. Firstly it will transmit


Glass

TCO

PV material

Rear contact

Incident radiant energy

Figure 5.7 Position of TCO ‘front’ electrical contact in a typical thin-film PV cell

more light to the photovoltaic material, and secondly, as TCO cost is usually proportionalto thickness, it will allow some savings.

5.5.2 Dimensioning of the Module

The width of the cells in a module (distance from one series connection to the next) has beenmodelled by Burgelman [29] and combined with an outdoor climate model [30]. This worksuggests that efficiency at low-intensity radiant energy is related to the cell width. An initialexperiment made using indoor-use cells from TESSAG lends support to the hypothesis thatwider cells are more appropriate for indoor use (see Figure 5.8). Further tests are required.

R2 = 0.89

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

0 10 20 30 40

Cell Width (mm)

VO

C/c

ell (

V)

Figure 5.8 Relationship between VOC/cell and cell width for RWE indoor solar cells

Further Work 93

5.6 CONCLUSION

The lack of comparable low-intensity solar cell data has been identified in subsection 1.5.2and section 5.1. Results for 21 samples representing eight material technologies tested by theauthor under two spectrum types and radiant energy in the range 1–1000W/m2 have beenpresented. Thereby, it has been demonstrated that an AM1.5 (1 sun) efficiency referencecannot be projected reliably down to indoor radiant energy intensities for all solar cells eitherbecause peak efficiency does not necessarily coincide with STC or the gradient of efficiencywith intensity is non-zero. Also, efficiency varies markedly with intensity for many samplesin the decades 1–100W/m2 (see Figure 5.2), the limiting factor being the parallel current, IP.Furthermore, the results confirm that efficiency is affected by spectra. This indicates thelimitations of STC solar cell comparisons for indoor PV (IPV).The lack of appropriate models has also been described and addressed; two models, one

developed by the author, have been shown to correlate well with experimental results in theintensity and spectrum ranges of interest for IPV design. These can be of service to IPVresearchers and designers alike.The above results form the basis for two relative classification criteria of the samples

tested. The first classification is determined from the gradient for each sample on thelogarithmic scale graphs (see the right-hand side of Figure 5.2). Two groups appear, basedon photovoltaic technology:

(a) group 1: crystalline silicon, polycrystalline silicon and CIGS samples tested;(b) group 2: amorphous silicon, cadmium telluride, gallium-based and dye cell samples

tested.

For the 10–1000W/m2 range, group 1 generally performs better than group 2. The oppositeis the case for an intensity of 1–10W/m2. This shows the importance of knowing whatradiant energy capital is available (see Chapter 3). Based on radiant energy available,recommendations can be made as to which sample will have the best performance.A second relative classification can be made with respect to spectra. The group 1 samples

tested performed better (or efficiency was little affected) under a fluorescent spectrum,while the group 2 samples performed noticeably less well in the same experiment. Giventhat fluorescent sources generally deliver radiant energy in the 1–10W/m2 range, group 2samples should be selected for such cases. The fact that they have lower STC efficiencymay be a bonus to the designer as this will reduce the need for overcharge protection shouldthe product be exposed to outdoor radiant energy intensity levels. Members of group 1 willbe better suited to daylight spectra in the 10–1000W/m2 range. This classification can beof service to the designer who will nevertheless take other factors into account such as cost,facility of series connection, surface area limitations and component availability. Cost, forexample, will at present rule out the gallium-based samples for indoor use.

5.7 FURTHER WORK

An ideal outcome from the IPV practitioner’s perspective would be a model based on eas-ily accessible data (such as AM1.5 efficiency) which would provide a prediction of cellperformance over the full range of intensities tested here. The phenomenological model


Crystalline Silicon η = 0.21ln(RP) + 2.3R2 = 0.86

Amorphous Si. η = 0.48ln(RP) – 4.7

R2 = 0.93Multicrystalline Si. η = 0.61ln(RP) – 0.33

R2 = 0.98

0

2

4

6

1.E + 00 1.E + 02 1.E + 04 1.E + 06 1.E + 08 1.E + 10

Estimated RP (ohms)

Eff

icie

ncy

at 4

.4 W

/m2

(or

0.44

% s

un)

Figure 5.9 Efficiency under low illumination (filtered AM1.5 with intensity of 44W/m2) vs. RSC(approx. RP)

[equation (5.8)] is valid for only some of the samples across the full range tested,08–1000W/m2, namely Solems, BP Millenium and Tessag. In order to extend the valid-ity, more terms are required to model that part of the efficiency–intensity curve where thegradient becomes negative. It is also necessary to investigate the physical meaning of thecontrol parameters a [equation (5.10)] and b [equation (5.11)].Another area requiring further understanding is the physical mechanism that induces RP.

It is thought to be related to intrinsic leakage (junction resistance) and extrinsic leakage(perimeter). Assuming it is independent of light intensity, the RP at low light intensitiescan be related to efficiency. The RP was estimated by the common technique of taking thegradient of the I–V curve at ISC (see Figure 4.13) at a low Erad�44W/m

2�. The samplesin Figure 5.9 suggest that there is a natural log relationship between RP and efficiency pertechnology. Other techniques for measuring RP exist, such as [149].

6Indoor Ambient Energy Charge Storage

6.1 INTRODUCTION

Many IPV systems require charge storage to fulfil their specified function. Storage is impor-tant for a number of reasons. Firstly, it is often a determinant of application ‘lifetime’. Sec-ondly, it may be the costliest component of an autonomous power system (see Figure 7.18).Thirdly, the trend of power and energy consumption of consumer electronic devices isfar outstripping the advancement in battery storage technology [236] despite significantscientific effort [128].The above three reasons focus attention on charge storage. However, finding the ideal stor-

age solution for a particular application is made non-trivial by the many storage parametersthat must be appreciated and balanced by the designer; these are mentioned in sections 6.3and 6.4. In section 6.3 the general requirements of IPV charge storage are explained, as wellas the technologies at hand.One of the determinant factors of charge storage cost and feasibility is the storage capac-

ity (J). In order to ensure this is correctly specified, a model of the relationship betweencharge storage and the probability that the application will have sufficient charge to functionis developed. The relationship is composed of both deterministic (see subsection 6.5.1) andnon-deterministic elements (see subsection 6.5.2). These are combined in subsection 6.5.3to form a global model.The indoors is characterised by a less harsh environment (i.e. lower temperature and

humidity ranges) than outside. The ways in which such charge storage requirementsmay be met in practice are described in general in section 6.3. Of these, electrochem-ical charge storage (section 6.6) is the most likely to be applied to ambient energysystems.This chapter, however, opens by setting the storage problem in its wider context

(see section 6.2), in particular with respect to the trend of recent decades towards greaterdecentralisation of a number of associated fields.


96 Indoor Ambient Energy Charge Storage

6.2 TRENDS IN CHARGE STORAGE

The last few decades have been witness to a general trend of decentralisation leading todistributed networks in various areas including data processing, electrical power generationand electronic devices (e.g. mobile telephony and computing). It is also recognised thatthis trend will continue for the coming decades, as can be inferred for electrical powergeneration from a forecast made in Osaka by the International Energy Agency. This statedthat non-hydroelectric renewable energy ‘will grow faster than any other primary energysource, at an average rate of 3.3% per year for the next 30 years’ [117].One factor on which all these trends rely is storage, be it programs on the hard disk drive

of a computer, water stored in a hydroelectric dam of an electricity network [259] or theelectrochemical potential of the battery in a personal digital assistant (PDA).A related parallel can also be seen between the technical needs of distributed electrical

networks and those of electronic products. Both must cost effectively balance supply anddemand of electrical energy – a non-trivial problem. A number of policies may be considered,including reducing or smoothing global energy requirements (consumption) as well as storingenergy for later use, such as with batteries for ‘peak shaving’ (e.g. the 40MWh NiCdin California [50]). Typical IPV storage technologies are considered in section 6.6, whileFigure 6.1 shows the breadth of electrical power storage technologies currently available.

Metal–Airbatteries

Flow batteries Pumpedhydro

Compressedair

NAS battery

Ni–Cad

Li-ion

Other Adv. batteries

Lead-acid batteries

Super capacitors

High-energy

flywheels

EnergymanagementPower quality

& UPS

1 kW 10 kW 100 kW 1 MW 10 MW 100 MW

Seco

nds

Min

utes

Dis

char

ge ti

me

Hou

rs

System power ratings

SMES

Bridging power

Figure 6.1 Simplified summary of electrical power storage technologies [144], reproduced withpermission (SMES = superconducting magnetic energy storage; NAS = sodium sulphur; UPS =

uninterruptible power system) from Elsevier

Trends in Charge Storage 97

6.2.1 Electrical Storage Parameters

The conceptual graph in Figure 6.1 places discharge time on the y axis. This isof interest because it indicates not only how long the storage medium will function(e.g. provide back-up) but also the responsiveness of the medium. Given the variety of prop-erties, no one technology can cover all requirements. As a result, another trend affectingboth electrical and electronic storage technologies can be brought out: hybridisation of tech-nologies. An example of this in electrical power production is cogeneration (simultaneousproduction of electricity and heat). A similar case for indoor electronic applications is theuse of a battery and a capacitor.Many other parameters must be considered by the system designer requiring such a storage

device; these include energy density (the volume per unit charge stored), safety, maintenance,recycling, charge/discharge efficiency, cycle life and, in some cases, portability. However,under present economic conditions, capital cost is most often a priority.Some reversible charge storage is essential for IPV applications. A reversible battery,

for example, is one that can both deliver charge in one polarity and be recharged with theinverse polarity. This class of rechargeable (referred to from here on as secondary) storageshould not be confused with refuellable secondary technologies such as the way in whichzinc–air and fuel cells are often used in practice. By inference, primary storage is synony-mous with devices that can be used only once, and are therefore neither rechargeable norrefuellable.Taking the example of electrical systems again, it can be seen from the conceptual graph

in Figure 6.2 that metal–air batteries hold an apparently ideal position (and are in fact often

Bet

ter

for

ener

gy m

anag

emen

t app

licat

ions

Cap

ital c

ost p

er u

nit e

nerg

y –

$/kW

h

10 000

1000

100

10100 300 1000 3000 10 000

Li-ionSupercapacitors

NAS battery

flow batteries

Pumped hydro

Comp.air

Better for UPS & Power quality applications

Capital cost per unit power – $/kW

Metal–airbatteries

Lead-acidbatteries

Ni–Cad

Otheradvancedbatteries

Figure 6.2 Simplified summary of capital cost of electrical storage media (NAS = sodium sulphur)[144], reproduced with permission from Elsevier


used for hearing aid applications, e.g. zinc–air). An important reason why such storagehas not replaced other technologies is that it is not used reversibly and can therefore beconsidered in practice only as a refuellable. Equally, the ‘fuel’ is not costeffective at the kWscale.

6.2.2 The Market

The scale of the worldwide battery market is remarkable. Take the automotive sector(see Figure 6.3) alone, which represents a demand of the order of G units (109).

The fact that most individuals buy a car less often than a consumer product, and thatthe latter often have a shorter usable life, explains why the consumer electronic market(of which IPV products are part) represents much larger volumes (e.g. 1.6 G-units for theportable sector for Germany alone in 1996) [101]. It is predicted that the sales of secondarybatteries for the consumer sector will exceed those in the automotive sector in at least oneEuropean country – Germany – in 2003 [101].

6.2.3 The Waste

So what happens to all these batteries once used? In the United States, batteries in generalrepresent ‘less than one per cent of total municipal solid waste generated’ but ‘accountedfor nearly two-thirds of the lead, ninety per cent of the mercury, and over half of thecadmium found in the waste’ [150]. This supports the case for battery recycling structures aswell as switching from primary to reversible secondary batteries where possible. Secondarybatteries, when used correctly, contribute to reducing waste [136] and therefore pollution.

West Asia

North America

Latin America andthe Caribbean

Europe andCentral Asia

Asia andthe Pacific

Africa

Million units

700

600

500

400

300

200

100

1980 1990 19960

2.3

5.5

6.4223.2

44.2

256.5

127.3

18.6

184.7

208.6

17.4

32.3

191.0

93.2

11.1

129.1

52.35.3

Figure 6.3 World car market growth for the years 1980, 1990 and 1996 [251], reproduced withpermission of UNEP

Charge Storage Technology 99

Related pollution at the worldwide level comes from small consumer electronic productsthat are disposed of still containing their batteries. This indicates the importance of ensuringminimal environmental impact of storage devices in general, be they primary or secondary.An ideal goal is to avoid electrochemical storage altogether, which is already technicallyfeasible in low-power electronics [196], an area that may be ideal for IPV applications.

6.3 CHARGE STORAGE TECHNOLOGY

6.3.1 Introduction

The necessity for storage implies both significant investment and lifetime costs for out-door stand-alone PV systems [23]. Significant investment costs for charge storage mayequally be the case for IPV autonomous power systems (see Figure 7.18). It is thereforeworth considering the various technologies available for fulfilling this role, both mechanical(subsection 6.3.3) and electrochemical (subsection 6.3.4).Given that electrochemical technologies are used for the majority of applications, these

are reviewed in turn (see subsections 6.3.5 to 6.3.7).All such storage technologies may be compared with respect to common units and prop-

erties which are presented first in subsection 6.3.2.

6.3.2 Technology Characterisation

Storage technology comparisons may be based on many factors. One of these, the dischargetime, has been mentioned above (see subsection 6.2.1) and may be measured as a rate(e.g. per cent per unit time) or the time to reach a set level of discharge (see Figure 6.1).This should not be confused with the inherent parasitic discharge of the majority of

storage devices, usually referred to as the self-discharge. This is usually quoted as a rate.For sensor applications, with wireless communication requiring product life of some years,self-discharge can be anticipated to be one of the most important issues associated withensuring satisfactory product life.Four further factors that are often quoted with regard to storage technology relate energy

and power to weight and volume (see Table 6.1).As can be expected, ideal storage technologies will have asmuch power and energy as possi-

ble with the least possible volume and weight. All of these values are for constant temperature,humidity and pressure. Although extremes of such parameters can be found in other environ-ments, indoor use can be expected to be relatively less harsh for charge storage devices.A weakness suffered by some battery technologies, such as nickel cadmium, for IPV

applications is the charging ‘memory effect’. Such batteries are unable to accommodate

Table 6.1 Four interrelated storage technology comparison factors

Units Volume (l) Weight (kg)

Power (W) Power density (W/l) Specific power (W/kg)Energy (Wh) Energy density (Wh/l) Specific energy (Wh/kg)


pauses in the charging process which are typical of the intermittent nature of renewableenergies. The consequence is that actual capacity is reduced, in the case of NiCd owing tothe formation of Ni5Cd21 which renders the active material inaccessible [136]. The resultantloss of storage capacity should be allowed for if using such a storage device for IPV.The storage technologies considered here may be categorised into two classes: mechanical

and electrochemical. In the following two sections, the principles on which these classes relyare explained. At present, electrochemical technology is usually used for IPV applicationsrequiring charge storage. However, this assumes that they have sufficient usable calendarlife. For building applications, use life might be of the order of decades, which is greaterthan the approximate ten years maximum life that electrochemical battery technologies canoffer at present.

6.3.3 Mechanical Storage Technology

One solution for extending usable life of the storage device beyond 10 years is mechanicalstorage. It is generally characterised by a store of potential energy linked to an electricalmotor. The motor may be used both to convert the potential energy into charge and, whenrunning in the opposite direction, to recharge the potential energy. In principle, the factthat the motor is required is a disadvantage owing to its volume, the added complexity andincumbent inefficiency.

6.3.3.1 Springs

Of the available mechanical storage technologies, springs have a long history of use forenergy storage. Their use in horology (e.g. timekeeping) since the fifteenth century [134]paved the way for both miniaturisation and portability. Their main advantage is havinga virtually nil self-discharge rate, making them appropriate for applications requiring anautonomy of greater than 10 years (roughly the limit of electrochemical technology life).This may be appropriate for building sensor applications where ideally the device wouldhave a usable life in proportion to the expected life of the building (i.e. decades). Thekey disadvantage of using springs for storage is that their energy density is two orders ofmagnitude lower than that of batteries [269].

6.3.3.2 Flywheels

Flywheels are in themselves not a new concept. They have long been used to reduce varia-tions in engine operating speed. Their use for energy storage is being researched for spatial[114], automotive [109], micropower distributed energy source [35] and power quality [122]applications. The orders of magnitude of capacity are for the moment much larger thanIPV scale, even at the laboratory level, such as a 50Wh design [125]. Also, the typicalpractical time for which a charge is held is of the order of hours [106], indicating signif-icant self-discharge. For IPV, if the necessary miniaturisation can be achieved, flywheelshave the potential of higher power with rapid response. Another advantage is their rela-tively lower sensitivity to temperature and charge–discharge cycling, both often the bane ofelectrochemical technologies.


6.3.3.3 Shape memory alloys

An area with potential for storing small amounts of energy owing to limited expectedefficiency [13] is shape memory alloys (SMAs). They are sensitive to heat rather thanelectrical energy, so are typically associated with thermal cycle systems. Therefore, use inan IPV system implies a hybrid solution such as a catch or a piston and spring.

6.3.4 Electrochemical Storage Technology

Electrochemical technology is characterised by at least two electrodes (i.e. anode and cath-ode) between which is a separator and an electrolyte; the resulting cell is housed in acontainer. When battery electrodes are connected to a suitable resistance, the cathode ischemically reduced, releasing electrons, and the anode is oxidised by receiving electrons.Such charge carriers pass through the electrolyte from one electrode to the other and throughthe exterior circuit. From the outside of the circuit, the positive electrode is therefore theanode, the negative electrode the cathode.In principle, electrochemical devices have a number of properties in common, including

thermodynamics, electrode kinetics and transport phenomena [249]. Temperature sensitivityof electrochemical devices can be an issue; however, this is usually less the case indoorsgiven the relatively lower expected temperature variance.Electrochemical device capacity is directly related to the surface area of anode and cathode

that overlaps, and therefore their construction seeks to maximise this area. This is achievedvia a spiral section or a stack of parallel layers leading to the cylindrical or prismatic shapesshown in Figure 6.4.In the future, cells will also be available in sachets [236]. Spiral (wound) cells are the

most established technology; prismatic shapes are newer and, while they offer the promiseof tighter packing density, easier series connection and bipolar electrodes (which can act asboth anode and cathode), they have the disadvantage of requiring the perimeter of each layerto be sealed. Compared with a spiral design, the latter is more demanding to manufactureand gas venting is also more complex.Members of the electrochemical storage technology class include primary and sec-

ondary batteries (see subsection 6.3.5), capacitors (see subsection 6.3.6) and fuel cells(subsection 6.3.7). These are compared on the graph of specific power vs. specific energyin Figure 6.5.This shows how batteries are generally the most well matched technology to IPV appli-

cations. However, some specifications require a supercapacitor (or double-layer capacitor,

Spiral wound Stacked Folded

Figure 6.4 Electrochemical storage device configuration, based on [183] and reproduced withpermission of Henry Oman


Capa-citors

SupercapacitorsBatteries Fuel

cellsTypical IPV application

needsSpec

ific

pow

er (

W/k

g)

Specific Energy (W h/kg)

107

106

105

104

103

102

10–2

101

0.1

0.01 0.1 1 10 100 1000

Figure 6.5 Simplified electrochemical storage technology comparison with IPV storagerequirements (author’s graph based on [71])

DLC) or fuel cells. Combinations of technologies are also feasible, leading to hybrid storagesolutions such as battery and DLC for increasing instantaneous power [222].

6.3.5 Batteries

While electrochemical storage devices have certain aspects in common, a battery may usuallybe distinguished from a DLC by the fact that the battery anode and cathode are of differentmaterials; these determine the battery electrochemical couple. Ideally, the difference instandard electrode potentials of the two materials should be as great as possible as they havea direct influence on output voltage (V). Another determinant of voltage is the number ofsuch cells in series.A further difference of batteries, when compared with fuel cells or DLCs, is that with

batteries the anode and cathode participate in the electrochemical reaction; for DLCs, thecharge separation is distributed through the volume of the electrodes. This means thatthe limiting factor determining battery capacity (mAh) is related to the amount of activeelectrode material available to react. This quantity of electrons available for discharge isdelivered at a rate referred to as current (A).Some battery chemistries are readily reversible which allows the battery to be electrically

recharged. The range of current output from a battery gives an indication of what currentrange may be used for recharging. This range must also be compatible with the applicationcurrent requirements.The unit used to describe charging and discharging of a storage device is C (note that C in

this context is not a coulomb). The measure is determined as the ratio of current to capacity,more specifically a current of 1 C is required to fully ‘fill’ (from 0% charge) or ‘empty’(from 100% full) the battery in 1 h. Typical charging currents in IPV will be 0.1 C or less.A related battery technology parameter is the depth of discharge (DoD) sensitivity, usually

measured as the number of cycles that can be achieved for a certain specification. In the longterm, repeated deep (>80% of rated capacity) discharge is known to have a deleterious effecton the capacity lifetime of some battery technologies such as lead–acid. Other technologiessuch as nickel–cadmium are better suited to a single continuous recharge from empty.


Figure 6.6 Typical storage charge cycles for a daylit (left) and electrically lit (right) IPVapplication with lead–acid technology storage

The designer may seek to ascertain how many such cycles are likely to occur and whetherpredicted capacity loss will reduce application functionality. For IPV applications positionedclose to the window, a life cycle similar to outdoor applications can be expected, as shownin the left-hand graph of Figure 6.6.For those applications that collect electric light, the cycles may not show such a deep

discharge in winter (see right-hand graph of Figure 6.6) but are likely to suffer more frequentshallow cycles (e.g. an office or school lit in the week and unlit at the weekend).For portable applications in particular, it is attractive if batteries have the lowest possible

weight. This is covered further in subsection 6.6.2.

6.3.6 Double-layer Capacitors (DLCs)

These devices, of interest for IPV autonomous power sources, may be described as a super-capacitor, ultracapacitor, electrochemical capacitor, gold capacitor or double-layer capacitor(DLC) [203]. Such capacitive storage relies on opposite charges being separated by aninsulating medium. A basic example of this is two parallel conducting plates separated by adielectric. Such a device can be readily charged and discharged, and it is for this property thatcapacitors are applied in electronics (for power smoothing circuits [221] or blocking directcurrent). For IPV storage applications, they are used to support a battery by coping with peakenergy demand or as storage devices in their own right, for example for solar calculators(see subsection 4.2.4). A key parameter is the amount of electrostatic charge that is stored (F):

CF =KAO

d(6.1)

where

CF = capacitance (F)AO = surface area of the overlap between the plates �m2�d= distance between the plates (m)K= permittivity of the separation material (at a fixed temperature)


It is of note for storage solutions using only capacitors that the actual voltage tends tovary over a wider range than for batteries in normal charging or discharging; for this reason,such devices should be used for charge storage rather than as voltage source [274].For peak energy demand, it is rather the maximum power, PF, that can be delivered from

a capacitor that is of interest; this is directly related to the square of the maximum voltage[222] as follows [191]:

PF =�Vmax�

2

4Res

(6.2)

where

Vmax = maximum voltage deliverable from the capacitor (V)Res = effective series resistance �"�

DLCs have greater energy density than traditional dielectric capacitors, although this densityremains approximately one order of magnitude below that of batteries. Recent research hasconsidered these devices combined with batteries [9] [11] and as an electrical energy storagesolution in their own right, e.g. [10][274]. In the latter article it is observed that DLCs havesimilar characteristics to a number of increasingly small capacitors connected in parallel.A simplified model of this is shown in Figure 6.7.This is confirmed in a DLC catalogue [175] which shows that more total charge can

be extracted at low discharge currents (e.g. 0.1–0.001mA) than at high discharge currents(e.g. 1mA–1A). The rechargeable nature of all capacitors suggests that they would likewisestore relatively more charge when trickle charged, a case typical of the energy deliveredfrom PV under indoor electric light sources.With regard to the useful life of a DLC, it was found [175] that load life at 85 �C is at

least 1000 h. Such conditions are expected to be extreme for IPV products, and thereforelonger useful life can be expected.Devices with a diameter of around 2 cm (i.e. the size range of some IPV applications)

with capacities of tens of farads are available on the market, indicating that they are likelyto play an increasing role in consumer electronic charge storage; with this in mind, of thecapacitive devices, only DLCs are considered as storage devices below.

6.3.6.1 Pseudocapacitance

One way of increasing the charge storage compared with the simple parallel conducting plateexample above is to encourage charge to intercalate in the plates. This charge comes from

R1

C1 C2 C3

R4

R2 R3

Figure 6.7 Simplified DLC model where R1>R2>R3 and C1>C2>C3

Charge Storage Parameters 105

the electrolyte, i.e. there is still no reaction with the electrodes. While this idea is unlikely toimprove the power characteristics, pseudocapacitance can be expected to give higher energydensity [172].

6.3.7 Fuel Cells

Fuel cells might in the future be used as a solar rechargeable storage technology; however,for the moment, they are being mostly considered as ‘engines’, converting a fuel into charge.Given that, once the fuel reservoir is spent, it needs to be recharged, such an approachtherefore corresponds to a primary cell. In view of the importance for IPV of having someelectrically rechargeable storage, present-day fuel cells cannot by envisaged for IPV chargestorage. Their use for IPV in the future will depend not only on their reversible use but alsoon whether the fuel reservoir is sufficient for the life of the product.Of the six classes of fuel cell, only two have appropriate operating temperature for IPV

(approximately 80 �C [83]). These are direct methanol fuel cells (DMFCs), which, as thename suggests, consume methanol, and polymer exchange membrane fuel cells (PEMFCs),which run on hydrogen. DMFCs benefit from the fact that, as methanol is a liquid, it is easierto contain than hydrogen, especially in the small quantities (ml) required. PEMFCs, althoughconsidered safer and ‘closest to commercial production’ [140], suffer a price disadvantage.DMFCs are thus under development [38] for mobile phone applications [41], although thelatter will not reach the market before 2005 [165]. PEMFCs have also been announcedon the market, but at over $US 7000/unit (with three recharge packs) [57] they are out ofproportion for expected IPV applications.

6.4 CHARGE STORAGE PARAMETERS

Given the energy stored and the voltage of a storage device, the capacity may be determined,the units of which for IPV are generally mAh (3.6 coulombs). Manufacturers quote therated capacity which corresponds to the amount of energy that can be discharged at therated current (at a set temperature which is usually close to indoor conditions). However,practical applications do not constantly consume rated current, so for each application anactual capacity is also found. This may be greater or smaller than rated capacity.Actual capacity will be a compromise between the many (and sometimes conflicting)

properties that suitable devices and technologies should exhibit. In general, some desirableproperties may be as follows.

1. be cheaper than the other components – rather than the most expensive element of theIPV system, as may often be the case.

2. be designed for lower life cycle costs than presently used solutions – this may includebeing (a) readily producible with low energy, (b) recyclable, (c) composed of non-toxicmaterials and (d) in a structure of material reuse.

3. be lower volume with respect to power and capacity than existing solutions – given thatthe storage device may be the most voluminous IPV power source component.


In charging, some desirable properties may be as follows.

1. be chargeable over the several orders of magnitude of current delivered by the IPVdevice – from slow/trickle charge to fast charge with constant charging efficiency and/orwithout damage to the internal structure. They will not suffer loss of capacity owing toinconsistent charging.

2. be sufficiently insensitive to overcharging - continuing to charge when a storage device isfully charged is associated, for some technologies, with the creation of heat and gas, bothof which contribute to increased pressure in the storage device. This can cause damageand reduce useful life if the device is sealed. If it is not, the device may need the lostcontents to be replaced, whether this is practical or not. The solution found for systemsmuch larger than IPV where overcharge for extended periods is likely [256] is a chargecontroller which among other functions might reduce current delivered to the storagedevice once it is fully charged.

In discharge, some desirable properties may be as follows.

1. have an appropriately low self-discharge rate – this may be important for an IPV productif it should function immediately the customer receives it, and therefore the storage should‘hold’ charge from manufacturing to first use. The same parameter may be importantonce the product is in use for periods of low radiant energy intensity. The process of self-discharge can be related to an electrochemical reaction, and the rates of all such reactionsrise with temperature [17]; therefore, if the IPV product is subject to temperatures outsidetypical average room temperature (e.g. 20± 5 �C) this may be an issue.

2. be able to deliver sufficiently high and constant current without reduction in voltage orcapacity (in discharge). With respect to the latter, batteries are known to suffer fromthe rate capacity effect which is observed if the current magnitude drawn is greater thanthe rated current. The result is that the ratio of delivered energy to stored charge isreduced. This may be compensated for in idle periods by an increase in actual capacity,and therefore in useful life; this is referred to as the ‘recovery effect’, which may be ofinterest for intermittent use IPV applications.

3. be insensitive to deep discharge (>80% capacity) and not suffer from being 100% dis-charged with respect to technical characteristics such as polarity reversal. IPV applicationscan expect such conditions at least annually (holiday weeks when windows are shutteredand no lights are used). Deep discharge may be important when more than one kind ofstorage device is used (i.e. in a hybrid system), as the most used device tends to sufferdeepest cycling. Another similarly undesirable example may occur for cells connected inseries when the weakest cell is discharged below 0V; this can have the consequence ofside chemical reactions in the case of batteries.

4. not require maintenance – they may have reliable and sufficient service life with respectto the required charge–discharge cycling, for example.

For portable devices, some desirable properties may be as follows.

1. have sufficiently low weight with respect to power and capacity – given that the storagedevice is often the heaviest IPV power source component.

2. be appropriately insensitive to temperature, which can induce premature ageing of thestorage device.

Determining the Storage Capacity 107

3. be usable in any three-dimensional orientation.4. be sufficiently resistant to shocks and vibration and not leak if punctured or misused.

These requirements may be contradictory: for example, a design that can deliver high currentimplies relatively low internal resistance with the incumbent likelihood of increased self-discharge rate. The difficulty of satisfying all the functional needs of storage lends supportto the case that storage should be minimised or, better still, eradicated [196].Usually, this is not possible and the designer will prioritise the charge storage parameters.

Two factors that are likely to be important to most ambient energy systems are low self-discharge and cost, treated in subsections 6.5.1 and 7.4.3 respectively. This is because theaverage self-discharge of most batteries rivals the ambient power collected. Generally, thosestorage devices that are used for standby (float-service duty) will be the most applicablefor IPV.

6.5 DETERMINING THE STORAGE CAPACITY

An essential consideration for the technical feasibility of an IPV system is that there is suf-ficient energy for the required functionality (see Figure 6.8). Evidently, the energy availablemay be sourced from either daylight and/or electrical sources (see Chapter 3).Assuming that the inequality in Figure 6.8 is always satisfied, no storage may be necessary,

e.g. a solar toy that only functions with sufficient radiant energy. If the equality is onlypartially satisfied, two cases can be imagined:

(a) the total available energy is insufficient for the application to function satisfactorily, inwhich case the present product configuration cannot be pursued;

(b) total available energy does exceed total application energy, but some charge storage willbe required to compensate for the phase difference between supply and demand.

In the latter ‘energy buffer’ case, a charge storage specification can be developed includingcapacity (A h), voltage (V) and maximum discharge current (A). It is assumed that the totalenergy available and the total energy consumed are similar. The necessary capacity maybe specified with respect to a number of parameters, some of which are relatively stable

Sum of energy available > Sum of energy consumed

Collected daylightCollected electrical light

T (S)

E (W)

T (S)

E (W)

Application profile

Figure 6.8 Schematic of energy constraint


and may be considered deterministic (see subsection 6.5.1) while others are stochastic andrequire more involved modelling (see subsection 6.5.2).

6.5.1 Relatively Stable Factors

In this subsection, the specification of storage size is considered with regard to those factorsthat are expected to be sufficiently stable to be modelled deterministically.

6.5.1.1 Application standby current

The application will typically have a constant standby current demand which requires asuitable storage capacity for standby consumption, hSBY. Note that the global consumptiondue to standby mode may be greater than the total energy consumed when ‘in-function’.

6.5.1.2 Storage actual capacity

The storage device will also have issues. For example, it is rarely possible to extractall the charge at a fixed voltage (see Figure 6.9) and the application minimum voltagerequirement will usually be reached before the storage device is fully discharged. A symptomof this problem is that around one-third of batteries collected for recycling in Switzerlandstill contain 90% of their charge! [124]. Typical culprits for such waste are high-currentapplications (e.g. cameras) which fortunately for the present case consume much more thantypical IPV applications.The two discharge curve examples in Figure 6.9 show that the minimum application

voltage fixes the proportion of the storage capacity that may be discharged in practice. Fortechnology 1, only 50% of the rated capacity can be used, while for technology 2, 80% of therated capacity may be extracted. An approximation of the extra storage capacity, hMAV, thatwill have to be specified to use a particular storage device is the reciprocal of the proportiondischarged on reaching the minimum application voltage.

Minimumapplicationvoltage

Technology 2

Technology 1

Voltage (V)

0 50 80 100

Proportion discharged (%)

Figure 6.9 Examples of discharge profiles for two battery technologies


6.5.1.3 Storage self discharge

Equally, a certain self-discharge of the storage system is typical and should be catered forwith storage capacity, hSD. This rate can be modelled by a constant loss of charge with time.It is especially important for applications with occasional but large current demands or forapplications that will receive reduced radiant energy for extended periods. Also, given thatthe self-discharge rate for many technologies is directly proportional to the storage capacity,it will become more of an issue as capacity is increased. The designer should therefore guardagainst excessive storage capacity as it is technically feasible that the self-discharge currentcan be greater than the average energy delivered from the photovoltaic module.Two of the above three parameters are schematised graphically in Figure 6.10. To sum-

marise, all three (hSBY� hMAV and hSD� losses may be considered deterministic and be mod-elled appropriately, e.g. constant consumption with time. Assuming that these parametershave normal distributions, it can be expected that such values will suffice 50% of the time. Inorder to increase the confidence level, it is advisable to scale them up by a suitable heuristicconstant. From manufacturing experience, a confidence level of 98% might be achieved witha safety factor of �1+ 2=�, where = is the standard deviation [96].The deterministic parameters can therefore be set aside until the total storage required is

calculated in subsection 6.5.3.

6.5.2 Non-deterministic Parameters

This leaves the fact that the energy collected by the PV and the application in-functionconsumption of this energy, while usually exhibiting some periodicity, are not deterministic.Energy collected may be available by day and not by night, for example, but to model usinga diurnal cycle only is to ignore annual variation, for example. Equally, it may be assumedthat the application will be in function without a constant frequency and independently ofthe energy available at the time.In order to specify the capacity required, more involved modelling is necessary. Four

classes of modelling have been identified as analytical, electrical circuit, stochastic and elec-trochemical in [132]. Lahiri et al. conclude that, while electrochemical models are desirableowing to their increased accuracy, they tend to rely on proprietary data (which are rarelyavailable to the designer) and be computationally intensive. More general models have the

Application standby

Storage self-discharge

Application in function

Cur

rent

mA

µA

Time (ms)

Figure 6.10 Global consumption of IPV system


Table 6.2 Summary of an IPV dipole compared withtwo storage models

Generic terms

IPV energy key

IPV energy blockdiagram

Conceptual bath

Manufacturingdipole key

Manufacturingdipole blockdiagram

Input

Radiantenergy

Chargestorage

Applicationconsumption

Storage Ouput

‘Tap’ Bath ‘Plughole’

Machine 1 Machine 2Buffer

M1 M2B12

RE Sto Ap. Co.

advantage of being applicable to a wider range of storage devices and being based onparameters set by the IPV system designer; for these reasons a general model is chosen here.An IPV energy system where the PV module and the IPV application are connected by a

charge storage device is shown in Table 6.2. Conceptually, this can be thought of as a bathwith a tap and a plug. When the tap (the current from RE) is opened, the bath may fill; whenthe plug is lifted (or the application is used), the bath contents can flow out (the storagedevice empties). This analogy can be used to describe the functioning of water reservoirs,e.g. [121], production processes, e.g. [265], communication systems and electronic systemssuch as IPV.Such analogies to storage systems all imply a flow, be it information, water, electricity or

products. Such flows are made up of fundamental elements which, when in sufficient number,may be approximated by discretisation into packets for ease of modelling. An example ofthis might be the night and day cycle of energy available (see Figure 6.11). Equally, theapplication in-function mode may be likewise simplified for modelling purposes.The discretisation allows the parameter to be considered as having binary mutually exclu-

sive states and so a storage model from queueing theory may be applied, implying Markovianchain mathematics. An example of a well known queueing theory model is the simplesingle-server queue referred to in the shorthand as the ‘M/G/1 queue’, [177, Chapter 7].Another example from queuing theory based on [63] is the manufacturing dipole consisting

of two manufacturing processes (M1 and M2� separated by a buffer stock �B12�. It can be

daynight

Cur

rent

(µA

)

day

Time (h)

Figure 6.11 Example of discretisation of energy available


seen in Table 6.2 that the radiant energy available (converted by the PV) may be representedas the throughput of a machine (M1), the IPV storage device may be thought of as bufferstock �B12� and the application energy consumption may be considered to be a machine�M2� that consumes the ‘contents’ of the buffer stock.The functioning of machine M1 therefore represents the PV charging the storage

(e.g. battery) and the functioning of M2 represents the in-function application consumingthe charge stored. Given this close comparability between these stochastic aspects of an IPVstorage and this simple manufacturing system, and that the mathematical equations for themanufacturing dipole model exist [111], the seven hypotheses (H1 to H7 respectively) onwhich the latter equations rest may be investigated in detail. Here it will be shown that allthe latter may be respected by an IPV dipole and therefore the model may be applied.

H1 That the flow through the circuits in Table 6.2 may be approximated by a hydrodynamicflow. This is better respected by an electronic circuit than a manufacturing process, asin electronics the fundamental unit is an electron, not for example a semi-assembledproduct. The number of fundamental units typically flowing through an electronic circuitis therefore significantly higher than a typical assembly process. This implies thateither there is flow or there is not, which forms the basis of Markovian approximation.Furthermore, by assuming a constant voltage throughout the dipole, current may beconsidered proportional to energy flow.

H2 That when B12 is empty and M1 is not functioning, M2 is ‘starved’. Likewise, whenB12 is full and M2 is not functioning, M1 is ‘blocked’. These two states are equivalentto zero flow at the corresponding machine–buffer interface. For the IPV system, thestarved case represents the application failing on account of lack of energy; the blockedcase is analogous to energy being lost to overcharging the battery.

H3 That machine malfunctions are considered, i.e. a blocked or starved machine cannotsimultaneously malfunction. While a manufacturing process may suffer a certain down-time due to malfunction, in IPV all components are expected to function throughout thelife of the product. Therefore, while it is possible for the application or PV module tofail, irrespective of whether it is starved or blocked, designs considered will typicallyavoid blocked or starved states, and the mean time between failures (MTBF) of thecomponents is expected to be of the same order as the product life. Therefore, anyresulting reduction in the specified capacity of the storage will be insignificant.

H4 That the time for flow in the system is negligible; this is the case for IPV.H5 That the in-function duration of machines M1 and M2 is random and exponentially

distributed with parameters &1 and &2=,&1. The probability density function is thereforefk�t� = &k exp �−&kt�, where k = 1�2. Strictly speaking, a PV module ‘machine’ isconstantly in function throughout its useful life. It is the radiant energy incident onthe module that is the main factor in determining its output. For convenience, theapproximation that the radiant energy intensity is constant for relatively long periods(e.g. 1/&1 hours of the day) and otherwise zero (e.g. 1//1 hours at night, explainedfurther in H6) is permitted. The PV in-function state can therefore be modelled by aMarkovian renewal process taking values ?0� IA@. The mean frequency of the in-functionstate is &1 (or the mean duration of one cycle is 1/&1�. The application in-function modecan be modelled in the same way as for the PV in-function mode. The mean frequencyof the in-function state is &2 (or the mean duration of one cycle is 1/&2).


IA

ID

IIF (mA)C

urre

nt (

µA)

1/λ1

1/µ1

1/λ2

1/µ2

(µA)

Cur

rent

Time (h) Time (ms)

Figure 6.12 Energy delivered by photovoltaic and consumed by application, where IA is theaverage current available (equivalent to the M1 nominal capacity), IIF is in-function current and ID is

deterministic current consumption

H6 That the non-functioning duration of machines M1 and M2 is random and exponentiallydistributed with parameters /1 and /2=1/1, i.e. fk�t�=/k exp�−/kt�, where k= 1�2.As in H5, these conditions are respected for IPV, with PV and application equivalent toM1 and M2 respectively (see Figure 6.12).

H7 That the average rates of production (U in parts/s) of machines M1 and M2 (referred toas U1 and U2 respectively) are equal. This hypothesis cannot normally be assured forIPV, as the instantaneous current collected will generally be lower (e.g. �A) than theapplication in-function current (e.g. mA). To correct for this, a normalisation may beapplied as follows. Given that the non-availability, N , of a machine is the ratio betweenmean time between two consecutive in-function phases and the mean time in function,this may be written as

N = 1//1/&

= &

/(6.3)

thenU1

1+N1

= U1

1+&1//1

= U2

1+ &1//1

= U2

1+ N1

(6.4)

Therefore, a machine M1 defined by parameters U1�&1 and /1 is transformed into an effectivemachine eM1 with parameters U2� &1 and /1. The two machines, M1 and eM1, have identicaloutput, but the instantaneous production of eM1 is U2.

If the above hypotheses are fulfilled, and as N2 = �,/1�N1, the dipole non-availability Ndip

with relation to the battery capacity, h, can be found using the equation

Ndip�h�=N1

[�,/1�2 eBh− 1�,/1� eBh− 1

](6.5)

where the non-availability of M1 and M2 (referred to as N1 and N2 respectively) is not equal,, is not equal to 1 and where

B = ,−1U

(/1

1+, + &11+1

)(6.6)


While the non-availability is a useful concept, it cannot be applied directly to aid the designerin determining the storage capacity required. The latter is usually based on the proportion ofapplication requirements that should be satisfied, or, in networking vocabulary, the qualityof service. The latter is the ratio of the proportion of time the dipole runs to the proportionof time the application is in function:

PDS =P2

Pdip

(6.7)

where

PDS = probability that application demand will be satisfiedPdip = proportion of time the dipole runsP2 = proportion of time the application is in function

From Figure 6.12, the proportion of time that a machine is in function, P (%), for the casewhere C and � are constant (reasonable for certain IPV dipoles) may be found as follows:

P= 1/&1//+ 1/&

= 1&//+ 1

(6.8)

Substituting equation (6.3) into equation equation (6.8) gives

P= 1N + 1

(6.9)

and therefore equation (6.7) becomes

PDS =Ndip + 1

N2 + 1(6.10)

or, with respect to the storage capacity, h, substituting equation (6.5) into equation (6.10)yields

PDS =N1

[�,/1�2 eBh−1�,/1� eBh−1

]+ 1

,1N1

[�,/1�2 eBh−1�,/1� eBh−1

]+ 1

(6.11)

As storage capacity, h, tends to infinity (and is full), the probability that the demand willbe satisfied, PDS, tends to unity, indicating that in IPV terms the application will neverbe starved. This can be visualised in Figure 6.13 which shows equation (6.11) with someexample values. In practice, designers will avoid excessive storage capacity, as mentioned insubsection 6.5.1, and they cannot specify an infinite storage capacity but will chose a suitablevalue, hIF, based on the trade-off between the probability that demand will be satisfiedand storage capacity. In the example in Figure 6.13 it can be seen that a 97% probabilitycorresponds to a storage capacity for the in-function demand of 200mAh.A minor issue is that the model is not applicable for very small values of storage, as can

be seen by the intercept on the PDS axis. This is acceptable as such low capacity is unlikelyin practice.This model shows that the designer may make use of meaningful frequency distributions,

as also noted by [227].


1.00

0.95

0.90

0.85

0 100 200 300

Prob

abili

ty th

at a

pplic

atio

nde

man

d w

ill b

e sa

tisfi

ed, P

DS

Battery storage capacity (mA h)

Figure 6.13 Example for the dipole model in equation (6.11)

6.5.3 Heuristic of Total Storage Capacity

The total capacity given by the model can now be found by summing the energy collected/in-function capacity (covered in subsection 6.5.2) with the function independent factors men-tioned in subsection 6.5.1. For a simple example based on the above-mentioned factors, thetotal theoretical capacity is therefore:

hTTC =hIF +hSBY +hSD +hMAV (6.12)

where

hTTC = total theoretical storage capacityhIF = in-function storage capacityhSBY = standby consumption storage capacityhSD = storage self-discharge capacity requiredhMAV = extra storage due to the minimum application voltage

In practice, designers make such a calculation before choosing an available product, as theymust also take other factors into account (e.g. cost).Based on the actual storage capacity, the system autonomy can be calculated (days); it is

the ratio of storage capacity to average daily load. This is useful to the IPV designer withrespect to periods of reduced charging because of ‘bad weather or failure in the system’ [239].

6.5.3.1 Differences with reality

It is of note that this model is a Markovian approximation and that other distributions maymatch real-data more closely but are likely to yield more complex mathematics.Another issue not treated by this model is the characteristic of electrochemical storage

to require relatively tight voltage tolerances for charging, for example, accumulators [226]and DLCs. It may be that some of the energy delivered by the photovoltaic device will belost owing to out-of-limit voltage. This will depend on the system design and the range ofincident radiant energy intensity.

Electrochemical Storage Technologies 115

6.6 ELECTROCHEMICAL STORAGE TECHNOLOGIES

While the storage function has been investigated above, it remains to consider the varioustechnologies that may satisfy these needs. In this section, two classes of electrochemicalstorage technology are considered: DLCs and secondary batteries. The way these work canbe found in subsection 6.3.4. Here, the individual chemistries and associated properties aredescribed.

6.6.1 Double-layer Capacitors (DLCs)

Capacitor technologies have seen three technological phases. In the first phase, symmetric(identical) carbon electrodes separated by an aqueous electrolyte were used (late 1970s). Inthe second phase, non-aqueous electrolyte was used instead which increased the cell voltage(1980s). In the third phase, asymmetric electrodes (e.g. one of carbon and one of nickeloxyhydride) with an aqueous electrolyte (e.g. potassium hydroxide) have been applied [73].DLCs combined with certain batteries of low current range, such as those with low self-

discharge current, offer the benefit of reducing the response time of the resulting hybridsystem.

6.6.2 Secondary Batteries (Accumulators)

Although many chemical pairs are suitable for electrochemical secondary storage, in IPVlow-power electronics three groups of technologies are most common: these are based onnickel (with cadmium or a metal hydride), lithium (ion or polymer) and reusable alkaline(manganese). Some of their half-reactions are shown in Table 6.3. The theoretical voltageof a cell is the difference between the potential for the anode and the cathode.Lithium sets itself apart by having a low relative atomic mass. It in fact has the lowest

equivalent weight of any practical negative electrode [172]. Similarly, the high atomic massof lead, and to a lesser extent cadmium, may be a disadvantage. However, the usability of

Table 6.3 Some IPV storage device anode and cathode half-reactions

Name Cell type Electrode Anode �−� / cathode �+� half-reactionsLithium/

manganesePrimary Anode Li�s�→Li+�aq�+ e−

Lithium/manganese

Primary Cathode MnO2�s�+ 2H2O�l�+ e− →Mn�OH�3�s�+OH−�aq�

Zn/Mn(alkaline)

Primary Anode Zn�s�+ 2OH−�aq�→ZnO�s�+H2O�l�+ 2e−

Zn/Mn(alkaline)

Primary Cathode MnO2�s�+H2O�l�+ e− →MnO�OH��s�+OH−�aq�

Lead/acid Secondary Anode Pb�s�+HSO−4 �aq�→PbSO4�s�+H+�aq�+ 2e−

Lead/acid Secondary Cathode PbO2�s�+ 3H+�aq�+HSO−4 �aq�+ 2e− →PbSO4�s�+ 2H2O�l�

Nickel/cadmium

Secondary Anode Cd�s�+ 2OH−�aq�→Cd�OH�2�s�+ 2e−

Nickel/cadmium

Secondary Cathode NiO�OH��s�+H2O�l�+ e− →Ni�OH�2�s�+OH−�aq�


Table 6.4 Secondary electrochemical battery materials

Name Anode Cathode Electrolyte Container

Lithiumpolymer

Lithium (Li) Composite Solid polymer Steel

Lithium ion Lithiatedcarbon

LiX intercalationcompound

Non-aqueous salt inorganic solvent

Steel

Nickel–metalhydride

Metal hydridealloy

Nickel (Ni) Potassium hydroxide(KOH)

Stainless steel

Nickel–cadmium

Cadmium (Cd) Nickel oxide Potassium hydroxide(KOH)

Ni-plated steel

Lead–acid Lead (Pb) Lead dioxide Sulphuric acid Plastic or metal

a storage device is not based on mass alone. With respect to the standard electric potential,lithium is again notable for its high reduction potential [205] (synonymous with being astrong reducing agent). The materials for typical IPV electrochemical batteries are shown inTable 6.4.In general, the above details are not considered in the choice of technology, but rather the

factors mentioned in Table 6.1. Their values for the technologies of interest are shown inTable 6.5 [132] [136] [110] [16] [102] [175], although this serves only as a guide becausethe data have been compiled from a number of authors. To deal with differences betweensources owing to the diverging interests of these authors, a range rather than a specific valuehas been indicated. Wherever possible, data applicable to consumer electronics are used.Technology choice cannot be made without reference to the application. In abstracto, how-

ever, the importance of the characteristics described in Table 6.5 can be discussed. For all IPVapplications, the storage self-discharge is often of principal concern (see subsection 6.5.1).The nickel-based batteries show the worst properties in this respect, and therefore cannotbe recommended except for applications that have regular radiant energy or an alternativeenergy source (e.g. mains transformer).

6.6.2.1 Useful life

The number of cycles may also be of importance. The figures shown in Table 6.5 are typicallyfor an 80% depth of discharge (DoD), which for IPV applications is deeper than expected(see Figure 6.6). Therefore, further tests are required to indicate what the DoD would befor the shallower but more frequent discharges typical of IPV. Nevertheless, the relativecharacteristics of the technologies can be expected to be maintained even in lower depth ofdischarge, and therefore care should be taken when using reusable alkaline manganese, as,of the batteries considered here, these have the lowest cycle life.

6.6.2.2 Weight

For IPV building sensors in particular, which are usually wall mounted, it can be assumedthat it is possible to compromise on weight (i.e. specific power and energy). This is fortunateas some believe that with lithium ion, the technical limits of specific energy are close tobeing reached [21].

Table 6.5 Comparison of secondary electrochemical storage technologies

Cell type Name

Ratedvoltage(V)

Energydensity(Wh/l)

Power density(W/l)

Specificenergy(Wh/kg)

Specificpower(W/kg)

Self-discharge(%/month)

Cycle life(number ofcycles)

ElectrochemicalsecondaryBattery

Lithium polymer 3.0 150– 385 >350 100–200 >200 1–2 200–1000


Lithium ion 3.6 165–300 400–500 60–127 200–250 5–10 500–1200


Reusable alkalinemanganese

1.5 77–290 2 25


Nickel–metalhydride

1.2 175–260 475 30–90 130 20–30 300–600


Nickel–cadmium 1.2 60–200 220–360 23–60 140–220 10–25 300–2000


Lead–acid 2.0 70–85 ∼400 25–35 ∼200 4–10 200–300

For comparison

Electrochemicalsecondary

Supercapacitor 2.5 5 300 5 1800 20–200 >50000

Mechanicalsecondary

Flywheel 370 185 >10000

Electrochemicalprimary

Alkaline 1.5 375 35 150 14 0.3 1


Zinc–Air 1.2 204 190 146–175 150 ∼5 ∼200


Fuel cell (DMFC) 300–767 0


6.6.2.3 Volume

Volume (l) is often more of an issue for IPV, although it is to be expected that a secondarystorage volume will not be greater than the space required by primary batteries. As can beseen in Table 6.5, lead–acid, which is associated with automotive applications but is alsoavailable in suitable sizes for IPV applications, suffers relatively low specific energy. Onthe other hand, overall the lithium technologies have good characteristics.

6.6.2.4 Hybrid solutions

Batteries are associated with relatively slow discharge (hours or more) with relatively con-stant voltage and current. Capacitors typically are used for a rapid discharge (e.g. minutesor less) but with higher power. Some storage system requirements may only be satisfied bya combination of both of these characteristics, e.g. at least one battery and one capacitor.This is called hybrid storage.Another conceivable hybrid solution is that of primary and secondary storage technologies

being combined. Given that such an approach is exceptional, this section has dealt mostlywith secondary storage technologies available to the IPV system designer.

6.6.2.5 Cost

Last but by no means least, consideration must be given to the cost of the storage. As can beseen in Figure 6.14, price per kJ capacity (and, by inference, cost) can be seen to vary over

0

2

4

6

8

10

12

(Sorted by price per kJ)

Pric

e ($

US)

Capacity (kJ) 3.89 3.50 2.59 9.50 5.18 4.75 2.38 0.73 4.54 1.08 2.59 1.94

Price ($US) 1.78 1.63 1.37 3.89 4.68 4.05 1.83 11.74 3.89 10.87 8.57

Price per kJ ($US/kJ) 0.457 0.465 0.530

6.98

0.735 0.750 0.985 1.703 2.485 2.589 3.600 4.194 4.408

NiCdAA

NiCd3.6V

NiCdAA

NiMHA

NiMHAA

NiMHAAA

NiCdAAA

NiMH4.8V

NiMH3.6V

NiCdAA

NiCd9V

NiCdbouto

n

Figure 6.14 Price capacity comparison for nickel-based batteries [159], reproduced with permission

Future Work 119

an order of magnitude. This can be explained by the various sizes of the batteries compared,as well as by the other functionalities, such as the parameters mentioned above, that theymay offer.In a similar way to the data in Table 6.5, the values in Figure 6.14 are less valuable if

the context is not mentioned. In this case, the data were collected for a report on the designof a solar (PV) powered LED hand torch and are therefore relatively applicable to the IPVdesigner.

6.7 CONCLUSION

The importance of the storage function in ambient energy systems has been seen in thischapter at the economic level (e.g. cost of component in Figure 6.14) and the technical level(e.g. balancing stochastic inputs and outputs in section 6.5) as well as at the environmentallevel (see subsection 6.2.3). Advantages of ambient energy systems have been shown atthe economic and environmental levels. Further advantages of ambient energy devicescompared with primary storage devices may relate to greater safety and robustness. Safetyis improved with less charge storage capacity as the risk of fire from accidental shortcircuit is reduced. Greater robustness can be expected in that, if the product is left on byaccident, so long as the product is not needed immediately, it will ‘recover’ without furtherintervention.Assuming that sufficient radiant energy is available for an IPV application, in the majority

of cases correct functionality will depend on charge storage. The latter balances the availabil-ity of energy from the ambient radiant energy and the demand of energy from the application.A model has been developed in section 6.5 that determines the relationship between thestorage capacity and the confidence level that the application will have sufficient energy tofunction. This treats the deterministic factors (subsection 6.5.1) and the non-deterministicfactors (subsection 6.5.2) separately; these factors are combined in subsection 6.5.3. Thenon-deterministic factors are modelled using queuing theory (Markovian chain).This model will typically be complemented in practice by many more parameters such as

those in sections 6.2 and 6.4. The number of these parameters alone indicates the complexityof the charge storage problem and its importance to the IPV designer. The most importantof these parameters is likely to be component cost.Having calculated the storage required, the IPV designer must also find suitable compo-

nents. These are treated globally in section 6.3 and with regard to typical components usedtoday in section 6.6.

6.8 FUTURE WORK

The model presented in section 6.5 could be improved in a number of ways, some of whichhave already been mentioned. One approach, by which the model would be brought closerto the designer selection, would be to combine it with cost functions for the various IPVsystem components in order to optimise the surface of PV material and storage capacitywith respect to cost and, by inference, the quality of service.


Ideally, batteries would be avoided altogether, and some believe this may be possible inthe future, especially for low power range electronic applications such as IPV systems [196].However, while reduced storage capacity for autonomous devices is an elegant concept,applications are rarely suitable from the outset today.

6.9 FURTHER READING

More information about electrochemical technologies can be found in [136] (nickel–cadmiumbatteries), [258] (lithium batteries) and [102] (rechargeable batteries).

7Ambient Energy Power Source Design

7.1 INTRODUCTION

7.1.1 Chapter Goals

While the design process by which any two products are developed will rarely be exactlythe same, the designs of a product type generally do have commonality. For this reason, andin view of the prodigious general product design literature, this chapter is not an exhaustivemethodology for all ambient energy power source design but rather a product-specificcomplement to it. This takes the form of guidelines and, where appropriate, heuristics; theseare based on practical experience gleaned from experiments, research and working with anumber of companies on a variety of products.As with the rest of the book, the main focus is indoor photovoltaic (IPV) products.

However, the approach may be generalised to other ambient energy sources.For brevity, cross-references are made to the rest of the book as some of the principles are

presented elsewhere. It is the author’s intention that this chapter may serve the practitioneras a key to the rest of the work.

7.1.2 Chapter Structure

Design guidelines can be categorised in a number of ways including by design processphase, design phase outcome, actor concerned, selection level and chronology. The firstfour of these are compared in Table 7.1 which indicates that these categorisations may berelatively equivalent. Given the general recognition [188] [87] [98] of the design processphase categorisation, it is used in this case. The up-and-down arrows in the left-hand columnof Table 7.1 indicate that the phases are interdependent, e.g. a result in one phase may alterthe conclusion drawn in a previous phase.


122 Ambient Energy Power Source Design

Table 7.1 Design process after[188]

Design process phaseTypical end-of-phase ‘go/no go’conditions

Phaseoutcome

Actor(s)concerned

Selectionlevel

Designspecification

Marketing, architect, designer

‘Ball park’Is energy equality [equation (7.3)]satisfied?

Have the options for maximisingavailable energy and minimisingapplication consumption beenexhausted?

Concept Designer Application

Does the customer find the productvisually attractive?

Can the producer make the product?

Preliminary layout

Designer producer

Component

Definitive layout

Installer customer

Final product

Are the user instructions welladapted?

Planning and clarification

Conceptual design

Embodiment

Detailed design

The design process phases cannot be defined exhaustively, as previously mentioned insection 2.2. They indicate different types of activity that are ideally consecutive and punc-tuated with evaluation against criteria (e.g. technical and economic criteria).During the clarification of design (see section 7.2), product requirements are specified

largely on the basis of existing information. Companies may consider both internal aspects aswell as the external (contextual) aspects such as market and legal factors. Conceptual design(see section 7.3) extracts from the resulting design specification those essential functions andproblems on which the product specification (or concept in Table 7.1) may be based. Theembodiment phase (see section 7.4) is associated with first concretisation of this conceptand may result in one or more functional prototypes. In detailed design (see section 7.5)the final product properties are specified for production. This typically includes materialand dimensional specifications, documentation and cost estimates. Each of these phases istreated in this chapter and reviewed for two representative cases in section 7.6.

7.1.3 Guideline Caveats

Design decisions are often associated with the optimisation of numerous variables. This isno less the case with these guidelines which may be partially contradictory. It is thereforein the hands of the product designer to apply them appropriately.The design phases in Table 7.1 read from top to bottom are increasingly product specific.

The guidelines below therefore aim to respect this categorisation. However, the author’schoice of design phase in which each guideline is found should not be considered prescriptive;the practitioner should use each guideline when it is deemed necessary.

7.2 CLARIFICATION

Clarification is the first phase of the design process (see Table 7.1) and considers the ‘ballpark’ figures. The designer will want to ensure technical feasibility. To allow an applicationto function at all times, the electrical energy available to it and the optional charge storage,EA, must always exceed the energy it consumes, EC. This may be summarised as

EA>EC (7.1)

Clarification 123

Both sides of this fundamental equation are covered in subsections 7.2.1 and 7.2.2 respec-tively. Those legal considerations that are necessary at the start of an IPV design project arethen covered in subsection 7.2.3.

7.2.1 Realistic Functions

For electronic functionality, a finite average power level will be consumed over a periodof time. The product of these two variables determines the energy consumed EC, which isgenerally calculated with respect to both in-function power and standby power:

EC =∫ tIF

0�PIF dtIF�+

∫ tSBY

0�PSBY dtSBY� (7.2)

where

PIF = power in function (W)tIF = time in function (t)PSBY = power on standby (W)tSBY = time on standby (t)

Therefore, the fundamental energy inequality (7.1) may be rewritten as

EA>∫ tIF

0�PIF dtIF�+

∫ tSBY


Two categories of IPV products may thus be identified:

1. Low storage capacity: those products that consume the majority of the energy availablein function. Therefore, to allow the fundamental energy inequality to be true, the in-function power must be less than or equal to the power converted from the radiant fluxand available to the application �PA�. In IPV, this is commonly of the order of /W.Examples of products that function at such power levels are some electronic wrist watchesand solar calculators. For an explanation of the radiant flux, see subsection 3.2.1:

PA>PIF (7.4)

2. High storage capacity: those products that consume the majority of the energy availablein standby. Typically, such products consume intermittent ‘peaks’ of energy when infunction at power ratings in excess of what the instantaneous radiant energy can deliver.In order for equation (7.3) to be true, therefore, the ratio of in-function time to standbytime will be suitably low. A heuristic for wireless communication products is a ratio oftime in function to time on standby of 1:300 or less. Furthermore, the average standbypower must be lower than the average radiant energy available, i.e.

EA>∫ tSBY


Given that most applications that the IPV designer will consider are in the second category,a general guideline is ‘at first, be less concerned with the product function than seeking


0.001

0.01

0.1

1

10

100

0 1000 2000 3000 4000

Time (ms)

Cur

rent

(m

A)

Figure 7.1 Typical charge consumption of datalogging or sensor device

products that need only function intermittently’, such as Figure 7.1. For such products, whichspend the majority of time on standby, the designer should then ensure that the standbyconsumption is suitably low, or can be reduced to comply with inequality (7.5). To enablethe designer to recognise suitable applications for IPV, typical values for the parameters ofinequality (7.3) are as shown in Table 7.2.Appropriate functions for IPV have been identified in subsection 1.4.2. With regard to the

wireless communication function, it is of note that mobile phones, whose in-function powerrange (0.5–2W) is acceptable for IPV, typically suffer too high a standby power (5–20mW).For short-range devices (SRD), the range of which is hundreds of metres or less, the valuesof in-function and standby power can be expected to be less and are therefore better adaptedto IPV.Examples of present-day IPV products are shown in Figure 4.6. Within these examples

are products that have relatively high in-function power (0.1–0.5W) such as the solar stapler(mechanical) and the lily in the water (light) shown in Figure 7.2. These support the casethat such high-power functions may be achieved; naturally, standby energy consumptionis kept to a minimum (a few �A). In the case of the stapler this is achieved by using amechanical switch, triggered only when pages are to be stapled (see Figure 7.3). Despitethis, the batteries will dissipate energy in self-discharge.

Table 7.2 Parameter values for equation (7.3)

Applicationdescription

Typicalfunction PSBY ��W�

PIF (mW) tSBY(min/day)

tIF(min/day) tSBY /tIF

Mean energy(mWh/day)

Mean energy(mWh/month)

Fluorescentsource

Display <3 <60 >1439 <1 >1440 <0.3 <10

Daylightsource

Wireless <90 <600 >1435 <5 >300 <3 <100

Clarification 125

Figure 7.2 Relatively high in-function power IPV products (author’s photograph)

Figure 7.3 Mechanical switch on a solar stapler (author’s photograph)

7.2.2 Realistic Use Pattern

The fundamental energy inequality (7.1) cannot be satisfied unless sufficient radiant flux iscollected. While radiant energy in the built environment is characterised in Chapter 3, thedesigner should also consider how the product will be used. A typical question is ‘will theproduct be fixed or mobile?’ The answer to this question will affect how deterministically itis possible to specify the radiant energy resource. This is important to the designer in that,the more tightly the energy system tolerances may be set, the more precisely (and typicallycost effectively) the product may be designed.The energy available to fixed products is generally easier to specify than that available to

mobile ones. Two examples of ‘fixed’ product are the higher-power products (stapler andlily in Figure 7.2) which are not intended to be moved over more than 1m or so duringuse. Furthermore, such products may be assumed to be in areas of human activity whereminimum incident radiant energy can be expected for hours at a time. Cases where this ismore difficult are personal mobile products – e.g. watches [36], phones, digital assistants andcameras, electronic keys – as in each case their outer surface is likely to be shielded from


the available radiant energy, e.g. up a sleeve or in a draw, pocket or holster. Furthermore,their economic value makes it unlikely that users will be prepared to leave them uncovered(let alone unattended) for hours at a time to collect radiant energy.Further heuristics based on optimising radiant energy are as follows:

1. Seek products that will be positioned within 1m of an unobstructed window (ideallywhere the module can be oriented towards the window) (see subsection 3.4.3). In such acase, the PV will output around 50mWh/cm2 month.

2. Seek product locations in areas of frequent (daily) human activity (see subsection 3.4.5).These are associated with a PV output of 15mWh/cm2 month.

3. Seek product locations with constant (24 h/day) lighting (e.g. public multistorey carparks). While the radiant energy is of lower intensity than in the home or the office, asit is available all day, the PV will output around 2mWh/cm2 month.

4. Seek to wall mount products. In any location, wall-mounted products have the advantagethat, once installed, they require no user intervention. A well positioned PV cell candeliver 1mWh/cm2 month.

It is also possible to satisfy equation 7.1 by using charge storage. The use pattern willinfluence the charge storage capacity required.

7.2.3 Prior Art

Given that a number of PV technologies exist, the protection of one in particular has thedisadvantage that it encourages the product designer to transfer to a competing technology,be it PV or otherwise. Furthermore, patents relating to the majority of established PVtechnologies can be expected to have been filed over twenty years ago, and will have thereforeexpired [248]. This may not be the case for the intellectual property rights protecting specificIPV applications; these should always be investigated as demonstrated in section 1.5.

7.3 CONCEPTUAL DESIGN

Conceptual design is the second phase of the design process in Table 7.1, during which aproduct concept will be developed on the basis of the required product functions. Consider-ations of importance during this phase are how to ensure the concepts are as energy efficientat all levels of application design. This may be at the level of the electronic device (sub-section 7.3.1), how the available energy may be maximised (subsection 7.3.2) or optimised(subsection 7.3.3) and how to optimise likely components (subsections 7.3.4 and 7.3.5).

7.3.1 Energy-efficient Design and Responsiveness

Low-power device designers must often resolve one or more ‘standby dilemmas’. While itis simpler from the functionality point of view to have a product in function constantly, thiswill tend to drain the energy resources rapidly, in particular with batteries. A compromiseis to put the product on standby (or quiescent mode). This reduces energy consumption,but comes at the ‘price’ of reduced responsiveness. To power up a mobile phone, such a

Conceptual Design 127

delay may be acceptable; however, it is unlikely to conform to the specifications requiredfor security- or safety-critical systems. The price paid is therefore the loss to the supplier ofa potential market.The cost of not implementing multiple power modes can be seen in the example of the

mains to DC transformer. The latter is used to provide energy to a host of consumer electronicproducts. As mentioned in subsection 2.4.5, a significant amount of electricity is consumedin Europe by products that the user has ostensibly turned off [141]. Taking the exampleof Germany, a study commissioned by the German Federal Environmental Agency and theFederal Ministry for the Environment reported in 1997 that ‘At least 11% of the electricityconsumed in German households and offices is used by temporarily unused equipmentrunning in standby mode’ [93]. They calculated that this amounted to 20 TWh/annum;assuming an average US cents 20/kWh means this energy will have cost German users 4billion �109� $US/annum. This shows the significant power consumption impact that lowpower devices can have. A technical solution is to use burst-mode transformers.A good example of how to reduce consumption, which may be of service to the IPV

designer, is to use multiple power modes such as with the ARM microprocessor [5] shownin Figure 7.4. Both RUN and IDLE are two in-function modes; SLEEP is otherwise referredto as standby (see words in parantheses in Figure 7.4).For the microprocessor (Figure 7.4), the approach taken implies a penalty for using the

standby mode: the 160ms it takes to return to the in-function state is a time long enoughin its own right to be noticeable to the user. In general, ‘waking’ a microprocessor maycoincide with a higher demand of current than average in-function consumption. Assumingthat the energy saved by switching the chip on and then off standby exceeds leaving it infunction, the designer must decide whether the resulting reduction in quality of service dueto apparent lack of product responsivity is acceptable.The power values in Figure 7.4 can be improved upon today with standby modes some

three factors lower (100 nW) than the SA1100 now available [162]. Such low standby currentsare achieved using multiple modes and are indicative of the trend towards consumptionwhich coincides with what AES can deliver.An aspect of energy-efficient design related to wireless systems that the designer may

consider at the conceptual design phase is where to position the processing (‘intelligence’).This may be at the sensor and/or at the motherstation (central control unit). Assuming

Figure 7.4 StrongARM [5] SA1100 microprocessor power-saving states. The designer can seefrom the standby current that this processor is not well suited to IPV


that the energy limitation is with the (distributed) sensors and not the motherstation, andthat wireless communication is both required and is the most energy-intensive activity,the designer will tend to decentralise some processing. An example of this for an indoortemperature control system is providing the distributed sensors with typical profiles forwhat the temperature should be in each location. Sensor communications then become onlynecessary for correcting those divergences that require action on the part of the motherstationto adjust the temperature. This can contribute to reducing the sensor energy requirements.Such dilemmas, including the standby dilemma, are found not only at the system level,

as in the above examples, but also at all levels of microprocessor design. The issue ofhardware/software partitioning is conceptually reminiscent of the centralisation/distributionof processing in the above example. Essentially, it may be more energy efficient to set someprocesses in hardware, potentially at the cost of reduced flexibility of the overall design,than rely entirely on the processor for all computations.Further microprocessor level energy efficiency techniques can be categorised as hardware

or software. An example with regard to hardware from which a number of energy-savingtechniques may be inferred is the power consumption of a CMOS gate:

PT =PSW +PSC +PSBY (7.6)

where

PT = total power (W)PSW = switching power (W)PSC = short-circuit power (W)PSBY = standby (or leakage) power (W)

Switching power has generally attracted the most attention in processor design and can bemodelled for an individual CMOS gate as follows:

PSW = 05�V 2S fCKCLESW� (7.7)

where

VS = supply voltage (V)fCK = clock frequency (Hz)CL = output load capacitance (F)ESW = switching activity factor (probability of switching)

From equation (7.7), power can be reduced most effectively by a reduction in voltage. Thisis apparent in the drop from 5 to around 1V in gate voltage that has come with the adventof process geometries below 05�m [70]. However, as the switching power of general-usemicroprocessors has decreased, the leakage power, PSBY, has become increasingly important.

Other factors in equation (7.7) can also contribute to reducing hardware energy consump-tion; techniques include voltage scaling, reducing load capacitance and reducing switchingactivity [14].It is also possible to improve energy efficiency at the software level. A compelling

approach is energy-aware compilation, in which the programmed code is analysed for theenergy it demands. This allows specific lines or subroutines to be modified to reduce


overall consumption. Further software techniques to reduce energy consumption may berelated to the choice of algorithm (e.g. MPEG-2), code compression and operating systemsupport [145].Overall, the designer should be aware that a wide range and significant number of

techniques and methodologies exist for minimising the energy consumed by microprocessortechnology. Literature in this area may (unscientifically) use ‘low power’ rather than ‘lowenergy’.

7.3.1.1 Circuit of the energy system

Another important aspect of energy-efficient design is selecting the appropriate IPV energysystem circuit. Therefore, the basic circuits mentioned in subsection 1.4.2 are further analysedhere to support the designer in circuit selection and specification. These are based on typicaldesigns for the functions ‘respond to light’, ‘sense and display’, ‘light, move, compute oramplify’ and ‘communicate (wirelessly)’ and are shown in Figures 7.5 to 7.8 respectively.The variables are as follows:

PV1�2�3�4 = photovoltaic module, characterised in section 4.4 and a model of which isshown in Figure 4.19

i1�2�3�4 = current at that point in the circuitRA1�A2�A3�A4 = resistance of an applicationRB4 = battery internal resistanceC2�4 = capacitanceiR1�R2�R3�R4 = current passing through a resistanceiC2�C4 = current passing through a capacitanceiB3�B4 = current passing through a batteryiZ3�Z4 = current passing through a Zener diodeu1�2�3�4 = voltage across a resistanceuB4 = voltage across a batteryuZ3 = reverse breakdown voltage across an ideal Zener diode

PV1RA1 u1

i1

iR1

Figure 7.5 Equivalent circuit diagram for Figure 1.2 (respond to light function)


RA2 u2c2

i2

iR2

PV2

ic2

Figure 7.6 Equivalent circuit diagram for Figure 1.3 (sense and display function)

RA3

u3

i3

iR3

PV3

iZ3 iB3

Figure 7.7 Equivalent circuit diagram for Figure 1.4 (light or movement function)

RA4

C4RB4u4

i4

iR4

PV4

uB4

iZ4 iB4 iC4

Figure 7.8 Equivalent circuit diagram of Figure 1.5 (wireless communication function)

From Kirchoff’s first law, which states that all current flowing into a node is equal to allcurrent flowing out of the node, the following relationships may be deduced.The general equation for Figure 7.5 is

i1 = iR1 (7.8)

and

iR1 =u1RA1

(7.9)


For Figure 7.6

i2 = iC2 + iR2 (7.10)

and, given

iC2 =C2

du2dt

(7.11)

and

iR2 =u2RA2

(7.12)

it follows that

i2 =C2

du2dt

+ u2RA2

(7.13)

The general equations for Figure 7.7 are

i3 = iZ3 + iB3 + iR3 (7.14)

and

i3 = iZ3 + iB3 +u3RA3

(7.15)

Assuming an ideal Zener diode with a reverse breakdown voltage uZ3, the battery is chargedwhen

u3<uZ3 (7.16)

and in this case

iZ3 = 0 (7.17)

in which case equation (7.15) simplifies to

i3 = iB3 +u3RA3

(7.18)

The general equations for Figure 7.8 are

i4 = iZ4 + iB4 + iC4 + iR4 (7.19)

i4 = iZ4 + iB4 +C4

du4dt

+ u4RA4

(7.20)

i4 = iZ4 +u4 − uB4RB4

+C4

du4dt

+ u4RA4

(7.21)


7.3.1.2 Wireless system design

For wireless applications, the transceiver is likely to be the highest power component. Ensur-ing that this power is at the correct level and that the communications transmit effectively(e.g. no message repeating necessary) is therefore an important energy efficiency concern.The designer will benefit from having details of the buildings in which the products willbe used, such as layout and materials. However, these alone are unlikely to be usable as‘radio signals are subject to attenuation, reflection and the interference and time-dispersioneffects of multipath propagation’ [85]. Therefore, a simulation may be recommendable, usingtechniques such as ray tracing [168, 231].

7.3.2 Maximising Available Energy

At the conceptual design phase, without specifying the exact components, the designershould be aware of a number of guidelines that will support the collection of as much radiantenergy as possible beyond those related to location that were mentioned in subsection 7.2.2.These may concern human interaction, PV technology selection and obstacles.

7.3.2.1 Human interaction

The first of these is to consider human interactions with the product. Whether by accident, outof ambivalence or for deliberate reasons, product users may influence the amount of energyavailable to the product. Where possible, it is therefore advisable to include foolproofingsuch as is shown in Figure 7.9. This wall-mounted device is designed so that the plane of thesolar module is at an angle to the wall. Given that radiant energy in the built environmentusually has a greater vertical component than horizontal component, by the cosine law[equation (3.5)] this module will collect more energy than a module whose plane is vertical(i.e. parallel to the wall).Users may feel that it is in their interest to tamper with products, such as those that are

monitoring their use of services, e.g. hot water. Equally, applications that have home security

Figure 7.9 Foolproofing the inclination of the solar module of a solar-powered thermometer.A typical value for such an angle would be 25�


functions might be tampered with by a potential intruder. The designer should thereforeconsider how to make products resilient to such behaviour. If using an IPV energy system foreither of these examples, it may be necessary to dissimulate (subsection 7.4.1) the module aswell as add software that, for example, may recognise, signal and record incorrect product use.

7.3.2.2 PV technology selection

The results from testing the eight PV technologies in Chapter 5 indicate that IPV applicationsthat receive the majority of radiant energy from daylight of over 10W/m2 should prioritisethe use of crystalline silicon, polycrystalline silicon or CIGS solar cells. The results alsosuggest that amorphous silicon, cadmium telluride and dye cell technologies are betteradapted for applications that generally receive radiant energy of an intensity less than10W/m2 (see subsection 5.2.3).If a majority of this radiant energy is from fluorescent sources, for the examples tested

in subsection 5.2.4 amorphous silicon solar modules showed the greatest performanceimprovement compared with results under a daylight spectrum. The results also indicate thatcrystalline silicon is less well adapted to fluorescent sources, as confirmed elsewhere [173].Thin-film solar modules in general offer the advantage that the voltage required by the

application can be achieved in a single module by adjusting the number of solar cells inseries. While this is electrically feasible using wafer-based technologies such as crystallinesilicon, thin-film technology provides a more convenient single unit.For applications requiring more than 3 V, convenient voltage tuning may also avoid the

necessity for a DC–DC converter, not to mention the associated conversion losses of such adevice.Open-circuit voltage, shown graphically in Figure 4.13 and discussed further in subsec-

tion 4.4.4, is another issue. Typical values at 1000W/m2 are 07± 01V. Of the samplestested in Chapter 5, two groups appear. Amorphous silicon, cadmium telluride and dye cell(the upper group in Figure 5.3) had at least a 0.1V/cell higher open-circuit voltage than thecrystalline, polycrystalline and CIGS samples tested (the lower group in Figure 5.3).Open-circuit voltage is relatively linear with intensity on the logarithmic scale; a heuristic

for all samples is 0.1V/cell per decade of intensity. This may be helpful when the designeronly has data for one intensity level (e.g. 1000W/m2�.

7.3.2.3 Obstacles

All non-transparent objects that lie in a straight line between the radiant energy source andthe PV module may reduce the energy collected. In order to maximise the available energy,the designer should be aware of the impact of obstacles, be they indoors or outdoors. As hasbeen seen in subsection 3.4.1, an obstacle that has an aspect ratio of 1, as viewed from themodule, may reduce the local irradiance by four-fifths (see Figure 7.10).

7.3.3 Variation in Design

It is generally to the designer’s advantage to reduce variation in the many product parametersthat must be considered. For products that must function all year round and are storagelimited, despite the fact that daylight irradiance may have the highest absolute intensities at


Tot

al in

stan

tane

ous

irra

dian

ce(W

/m2 )

Radiometer

Breadth, B Obstacle

Height,H

0 0.5 1 1.50

5

10

15

20

–80%

Aspect Ratio (H/B)

Figure 7.10 Effect of aspect ratio on instantaneous irradiance, based on Figures 3.16 and 3.17

J F M A M J J A S O N D

Month

Cum

ulat

ive

irra

dian

ce(k

Wh

/m2 m

onth

)

Cum

ulat

ive

irra

dian

ce(M

J /m

2 mon

th)

250

600

900

300

0

200

150

100

50

0

Factor of 10

Figure 7.11 Factor of 10 variation in cumulative outdoor irradiance in Payerne, Switzerland, 1998(courtesy of World Radiation Center, Davos)

some times, it may be advisable therefore not to rely on daylight-sourced irradiance. This isbecause daylight radiant energy may vary by a factor of 10 over the year (see Figure 7.11).Furthermore, even if the designer takes the worst case (e.g. December in Figure 7.11), somelocations in certain bad years may receive much less than this amount of energy. The rangeof average radiant energy over 59 locations in Switzerland supports this. As Figure 7.12shows, the very month of December which the designer might choose coincides with thewidest range of average global irradiance! If this is an issue for a particular IPV design, itmay be advisable to prioritise the collection of electrically sourced irradiance.Alternatively, the designer may need to tolerate a certain degradation of service.

7.3.4 Charge Storage Components

Three basic concepts are important with regard to effective charge storage in ambient energysystems. The first is that the designer should be aware of the limitations of certain tech-nologies. For example, nickel cadmium batteries suffer relatively high self-discharge as well


J F M A M J J A S O N DMonth

80%

40%

0%

–40%

–80%

Val

ue r

ange

com

pare

dw

ith a

vera

gae

for

all v

alue

s

Figure 7.12 Range of average global outdoor insolation between 1981 and 2000 for 59 locations inSwitzerland [155]

as being ill adapted to trickle charge owing to the memory effect (see subsection 6.3.2).Other technologies, such as rechargeable alkali manganese (RAM), suffer less from thesedrawbacks (see the self-discharge comparison in Figure 7.13). Further charge storage tech-nological comparisons can be found in section 6.6.Secondly, for batteries in general, the total energy delivered by a battery is related to

the current discharge profile. If a current greater than the rated current of the battery isdischarged, then the ratio of the delivered energy to the energy stored in the battery maybe reduced. The usable battery life may thus also be diminished. This effect is called therate capacity effect. On the other hand, for a battery that is infrequently discharged for briefintervals, it may be possible to improve the ratio of energy delivered to energy stored inthe battery [190]. This may be of service to IPV designers, especially for products that arerelatively rarely in function as described in the second category of subsection 7.2.1.The third useful concept that may also pertain to extending battery life is the use of hybrid

storage, especially for wireless communication applications. An example is the use of a

0

200

400

600

800

1000

1200

1400

1600

1800

0 100 200 300 400

Storage Time (days)

Cap

acity

(m

A h

) NiMH

NiCd

RAMTM

Figure 7.13 Secondary cell self-discharge technology comparison, courtesy of BatteryTechnologies Inc., Canada


supercapacitor and a battery in parallel; the former can deliver brief peaks of power whilethe latter can deliver the majority of the charge storage capacity (see section 6.5).

7.3.5 Functionality

For the case where interaction with the product is required, one or more buttons may beused. It is technologically feasible that one or more ‘smart’ solar cells could fulfil thebutton function, as well as the radiant energy collection. This hybrid idea has not attractedintellectual property right protection, unlike the use of photovoltaic switches [59]. The costeffectiveness of the hybrid approach may be evaluated using a cost–function analysis [224].

7.4 EMBODIMENT

In the embodiment phase, the third phase of Table 7.1, a single product design will beselected from one or more prototypes. As the product takes form, it may be easier to recognisewhether its appearance is acceptable, with respect to both the solar module (subsection 7.4.1)and the product casing (subsection 7.4.2). The economic and production feasibility may alsobe confirmed (see subsections 7.4.3 and 7.4.4 respectively).

7.4.1 PV Module Appearance

Product purchase decisions are made for both rational and irrational reasons, with a generalpropensity for the latter [224]. It is therefore to be expected that the designer will meetconcerns with regard to non-technical issues, as these may strongly affect purchase decisions.Such concerns apply whether converting an existing product to IPV or developing a new one.One such issue is the visual impact of the product, in abstracto or relative to the area in

which it will be used. In IPV there is a potential technical conflict of interest between thePV modules which are ideally in black colours (for radiant energy absorption) and the pastelor white colours often selected for the inside surfaces of buildings, for example (with theaim of reflecting incoming light).Globally, two solution areas may be considered to alleviate issues related to the PV

module. Firstly, the designer may choose a module whose intrinsic properties are appropriate.Secondly, the module can in some way be dissimulated. The consequence of both theseapproaches, as can be expected, is that the effective electrical efficiency of the PV surfacemay be reduced.

7.4.1.1 Select PV module intrinsic properties

The majority of solar cells and modules have a monoplanar form, appear relatively reflective(gloss rather than matt) and have a grey or dark-blue colour. The advent of thin-film PVtechnologies provides designers with a greater range, especially in terms of shape and colour.The shape of thin-film solar modules may take any three-dimensional shape that has arelatively constant cross-section along one axis, as illustrated in Figure 7.14. Figure 7.15indicates the range of colour that may be achieved, depending on PV technology and theproduction processing chosen.

Embodiment 137

Figure 7.14 Proposed solar clothing [257], reproduced with permission of VFH-Technologies SA

Figure 7.15 Comparison of three thin-film PV technologies: the top left sample is cadmiumtelluride (light grey); the three others are amorphous silicon in red and green (top right), black

(bottom left) and purple (bottom right)

7.4.1.2 Dissimulating PV module

A more common approach in IPV product design is to reduce the visual impact of the PVmodule by distracting attention from the PV or concealing it from view. An example ofthe latter is the use of a translucent cache on top of the solar module, as applied to solarwatches, for example. While this is a visually effective approach, it suffers the disadvantageof reduced incident radiation on the PV module. A rule of thumb is a 50% loss of availableenergy on account of such a cache [202].Another form of cache is the use of a transparent lens cover, as shown in Figure 7.16.


Figure 7.16 Solar-powered thermometer with a rectangular transparent window (within a grey ovalcache) covering a five-cell PV module. The square window below the oval is an LCD display

The designer may choose to reduce the visual impact of the PV module by the use ofcolour. This may be achieved by adapting the product housing to the solar module colour,or by including some form of more gradual blending colour, such as a collar around the PVmodule edge of a colour(s) between that of the PV and the casing. An example is the greyoval cache shown in Figure 7.16.A final form of integration is that of the solar module as a feature, such as the company or

product logo. While to the author’s knowledge no examples are available on the market, theidea has been patented [97]. This may be achieved by the selective deposition of thin-filmPV modules in a shape other than the usual stripes, or by overprinting any PV module. Therisk of this solution is that cells of the module may not receive balanced amounts of radiantenergy, which would reduce the effective efficiency of the module.

7.4.2 Product Housing

A second appearance issue is that the product housing may need to be adjusted to accom-modate the solar module. This need can be minimised if sufficient product outer surface isavailable and a flexible thin-film PV module can be adhered to it. If this is not possible,it implies that a window in the casing will be required and that the casing may need to beenlarged in at least one dimension. Both of these changes can reduce the resistance of thecasing to temperature, shock and bending. It is therefore recommended that a mechanicalresistance analysis of proposed modifications be made. With a change in scale, the plasticinjection material properties should also be taken into account, for example to ensure that

Embodiment 139

the molten plastic viscosity is still compatible with the desired moulding. Product housingsolutions that the designer may consider include altering existing parts (e.g. ribs) or addingfurther parts (e.g. glazing).

7.4.2.1 Ribs

Based on a mesh structural analysis, this involves adding extra material to ensure that thecasing is not undermined with respect to shock or bending, especially in the delicate areassuch as the corners of the window.

7.4.2.2 Glazing

The IPV solar module may be deposited on glass and therefore require little protection.However, for issues related to the application such as appearance and structural integrity, itmay be appropriate to use a window of transparent plastic, for example. The ‘glazing’ mayalso protect against dust and dirt, especially for solar modules that are not vertically oriented.

7.4.3 Cost

While cost is important at all stages of design, during the embodiment phase, the exact costs ofthe IPV energy system can be confirmed, such as the example in Figure 7.18. In this case themost expensive function is the charge storage. This is to be expectedwhen abattery is necessary.At this phase the designer may also be able to confirm that the ideal price of the more

expensive components has been found. An indication of this may be gained by using a

Figure 7.17 Example of a window in the housing protecting an LCD and a solar module

Solarmodule

20%

Circuit20%

Battery60%

Figure 7.18 Example of the cost breakdown of an IPV energy system based on a total of $US 5


1.00

10.00

100.00

1000.00

10 100 1000 10000 100000

Capacity, Ah (J)Pric

e,F

S ($

US/

unit)

AgO Li Button NiCd

Figure 7.19 Battery cost function for selected examples of three technologies (AgO= silver oxide,Li= lithium, NiCd=nickel–cadmium); data [79]

cost–function analysis. An example is the relationship between cost and charge storagedevice capacity, Ah (J). The latter can be summarised either graphically (see Figure 7.19) ormathematically [see equation (7.23)], as the example of charge storage shows.

FS = f�Ah� (7.22)

FS =

003Ah + 29 if 13<Ah< 111

10−5Ah + 42 if 112<Ah< 1499

0003Ah + 6 if 1500<Ah< 17000

(7.23)

where

FS = price of a single unit ($US)Ah = capacity of the charge storage device (J)

Cost functions are useful to the designer as they summarise the effect of moving from onetechnology or threshold to another. This can be seen in the example shown in Figure 7.19.Assuming that no devices other than those shown on the graph are feasible, it is advisable toavoid exceeding a capacity of 1500 J as this would require switching to a nickel–cadmiumdevice that would cost at least an order of magnitude more than the lithium devices shownin Figure 7.19. In practice it is unlikely that single-use batteries (AgO and lithium) wouldbe compared with rechargeables (NiCd).Given the importance of correctly specifying the battery, the designer may use models

to determine the optimum device. One such model that indicates technical feasibility of aproduct with respect to charge storage capacity is described in subsection 6.5.2.

7.4.4 Production Quality

An area much influenced by design decisions is manufacturing. One way to integrate thisinto the design process is to predict quality levels before production. Toyota achieve thisby considering three levels of product quality: that of the components (e.g. componentswith compatible tolerances), the assembly process (e.g. statistical process capability) and theoperators (e.g suitably skilled).

Detailed Design 141

In order to support the operator, the designer may also consider assembly error-proofing.The typical top three assembly quality issues can be categorised as use of incorrect parts,omission of parts and incorrect assembly [44]. The designer should therefore minimise therisk of these problems occurring.In general, the designer should ensure that all necessary tolerances can be respected. This

can be represented diagrammatically as in Figure 7.20.Here, the outer circle represents the customer requirements (quality desired). The two

other circles are the product specification (quality specified) and the acceptability of themanufacturing output (quality achieved). Ideal quality may be represented by the coincidenceof all three circles. Reality may look more like the diagram (Figure 7.20). For example,quality specified or achieved may lie outside the quality desired. This represents activitythat does not satisfy customer requirements and such a diagram may support the designer inthe communication of the improvements that should be made.

7.5 DETAILED DESIGN

The last phase of the design flow chart in Table 7.1 is the detailed design. As the designerfinalises the product, issues that may have previously appeared minor may gain importance.Examples of such issues that have been successfully resolved for IPV are described insubsection 7.5.1. The designer may also choose to communicate to the consumer how bestto use the product, especially with regard to maximising radiant energy (seen in subsec-tion 7.3.2). Suggestions on how this may be achieved are covered in subsection 7.5.2.

7.5.1 Alternative Energy Source

A number of noticeably good design solutions have been mentioned already (see Figures 7.9and 7.16). Two others that relate to providing flexibility of energy source to the user arepresented here. This may be necessary as the product user may wish to extend product life,either into environments of insufficient radiant energy (e.g. cellar) or if the ambient energysystem malfunctions.Two solutions inspired by existing IPV products may be considered. Firstly, an optional

battery bay may be added, as in Figure 7.21. Alternatively, a socket for an external powersupply may be provided, as shown in the example from the solar stapler in Figure 7.22.

Quality desired Quality specified

Quality achieved

Figure 7.20 Three quality circles [118] – simplified constraint-based methodology


Figure 7.21 Back of a solar-powered thermometer with the bay for an optional battery (author’sphotograph)

Figure 7.22 External transformer socket on the rear of a solar-powered stapler (author’sphotograph)

It can be seen that the three equivalent circuits shown in Figures 7.6 to 7.8 use at leastone diode. A simple LED may be used, as in the solar alarm clock shown in Figure 7.23.

7.5.2 Information on Ideal Use

Since it is unlikely that most users will appreciate where concentrations of ambient energycan be found, the designer may choose to inform them via instructions, for instance. Anexample of this for the installation of IPV wall-mounted devices under electrical lightsources has been mentioned in subsection 3.4.3. For this same example, Figure 7.24 showsthe kind of instructions that may support the user. As can be seen from this diagram,this advice may be quite intuitive provided a suitable diagram is available. Attemptingto explain this in words may make for a voluminous (and therefore uninviting) usermanual.

Case Studies 143

Figure 7.23 Modified solar alarm clock to show LED used (author’s photograph)

2 m

1 m

Mark of light on wall Good product locations

Fluorescent tubes

Figure 7.24 Suggestion for installation instructions or manual

7.6 CASE STUDIES

In order further to illustrate how the above design phases and chapters may be appliedin practice, two case studies are presented here. The cases are representative of the twocategories defined in subsection 7.2.1, namely low storage (solar calculator, see Figure 7.17)and high storage capacity (solar wireless sensor, see Figure 7.15).

7.6.1 Clarification

The first step in clarification is to determine the ball-park technical feasibility of the proposedproduct as shown in section 7.2. This feasibility depends on the amount of charge storagethat the product will require. For very low charge storage capacity requirement products,such as solar calculators (e.g. less than 10�W h/day or 40mJ/day), or use only in situations


with sufficient radiant energy, inequality (7.4) applies. If this inequality is satisfied, then thestorage capacity can be estimated (see section 6.5); if inequality (7.4) is not satisfied, thedesigner will continue as for a wireless sensor.For greater charge storage capacity such as for wireless sensors, or use with insufficient

instantaneous radiant energy, inequalities (7.5) and (7.3) must be satisfied. Assuming theyare, then the charge storage capacity may be specified as described in section 6.5. The abovecan be summarised in a flow chart (see Figure 7.25).

Key

Storage capacityestimation START

<0.1 mW

>0.1 mW>0.1 h/day

<0.1 h/day

Mean in-function

time?

Mean in-functionpower?

Is inequality(7.4) (PA > PIF)

satisfied?

Is inequality (7.3)(EA > PIF.tIF + PSBY.tSBY)

ok?

Is inequality (7.5)(EA > PSBY.tSBY)

satisfied?

Productunfeasible

withpresent

constraints

Specifychargestoragecapacity

Changeconstraints

?

END

No

No

No

No

Yes

Yes

Yes

Yes

Figure 7.25 IPV product feasibility flow chart

Case Studies 145

As can be seen, if inequality (7.3) or inequality (7.5) is not satisfied, then the proposedproduct is unfeasible with the present constraints unless the latter can be relaxed, for examplewith the methods described in subsection 7.6.2.

7.6.2 Conceptual Design

The probability of product technical feasibility can be increased in a number of ways, whichmay be necessary either to relax initial constraints (see above) or to extend the envelopeof feasible situations. The ways of increasing technical feasibility investigated in this workinclude optimising radiant energy collection and correctly selecting photovoltaic modulesand charge storage.

7.6.2.1 Radiant energy

For the case of the solar calculator, the product must function with the radiant energyavailable at a desk. The solar sensor has more scope for influencing available energy: firstly,the position of the solar module in the built space, such as close to a window, as can beconcluded from subsections 3.3.2 and 3.4.3; also, the importance of orientation with respectthe radiant energy source, as has been shown in subsections 3.4.4 and 3.6.2. Another way toimprove technical feasibility for the solar sensor is to consider the obstacles to the radiantenergy associated with the windows (see subsection 3.4.2). If the solar module collects mainlydaylight, the environment in which the building is found will exert an influence. Geographicallocation (subsection 7.3.3), or nearby buildings, for instance (see subsection 3.4.1), shouldbe taken into account.

7.6.2.2 PV technology

The radiant energy available is a determinant of the appropriate photovoltaic technology,as concluded in section 5.6 and described in subsection 7.3.2 (PV technology section). Theelectrical efficiency can also be improved in a number of ways; the principles are outlinedin section 4.7, while practical solutions are described in section 5.5. Such improvements canserve for both the above-mentioned product cases.

7.6.2.3 Storage selection

Last in this conceptual phase is the selection of charge storage. For the case of the solarcalculator, the storage required is low (see subsection 7.6.1) and self-discharge should not bea limiting factor. For the solar sensor, the requirements are the opposite. Significant storagecapacity is required (e.g. a month without radiant energy), and therefore minimal storagedevice self-discharge is paramount. Techniques for the solar sensor case can be found insubsection 7.3.4 and section 6.6.

7.6.3 Embodiment

Conclusions can be drawn for the two case studies from all four of the subsections insection 7.4.


From subsection 7.4.1 it can be seen that PV module appearance may be more easilyresolved for the solar calculator than the solar sensor. This is because the solar calculatorcolour is likely to be closer to that of the PV module than for the solar sensor. The dissim-ulation techniques described can be of use for the latter, such as the cache in Figure 7.16.With regard to the housing, the solar calculator is likely to be ‘glazed’ as shown in

Figure 7.17. For the solar sensor, structural issues are likely to be more important.As the solar sensor requires relatively high storage capacity, it can be expected to have a

cost breakdown similar to that in Figure 7.18. The solar calculator is likely to have roughlyequal costs for the solar module, capacitor, casing and circuit.

7.6.4 Detailed Design

From the detailed design (section 7.5) it can be seen that the alternative energy source forboth cases taken here could include a battery, such as Figure 7.21. The solar sensor mightalso be used with an external transformer, as in Figure 7.22. User information for the solarcalculator is relatively intuitive as the user receives immediate feedback of lack of radiantenergy, i.e. the display fades. The solar sensor will not benefit from such close attention andshould therefore be installed with care following guidelines such as shown in Figure 7.24.

7.7 CONCLUSION

A number of guidelines specific to ambient energy and IPV have been presented. They arecategorised according to the design process phases [188], although the exact phase in whichthey are found is not a prescription to the designer. While it is not possible to provide anexhaustive methodology for all ambient energy products, many of the parameter relationshipsfor IPV have been explained. Whenever appropriate, values for the expected IPV coefficientsare mentioned. Links to other chapters are regularly provided, and it is hoped that this willencourage the reader to explore the rest of the book.Of all the guidelines, there are three that are especially important.

1. For wireless sensors, the designer should at first be less concerned with absolute appli-cation power than with the average energy consumed [see inequality (7.3)]. Given theimportance of standby current, recently released nW standby processors [46] indicatenew potential for IPV and ambient energy sources (AES).

2. A route to wider ambient energy source application is the use of low-energy designtechniques such as those covered in subsection 7.3.1.

3. Independent of how well the technical design is executed, product success relies heavilyon non-technical issues such as product appearance (see subsection 7.4.1).

7.8 FURTHER READING

Two useful general texts on low-power design methods include [204] and [105]. Morespecific subject areas include [15] which considers the system power and [147] which focuseson memory optimisation.

Acknowledgements

This book could not have been written without the SATW (Zürich) who awarded me ascholarship for eighteen months; Professor Jacques Jacot provided a salary for the remainderof the preparation time. Thank you. I am also deeply indebted to Professor Jacot for playingthe essential role of supervisor. He was ably supported by my steering committee (ProfessorHannes Bleuler and Drs Thomas Maeder and Léon Benmayor). I would also like to thankmy friends, especially those who showed their skills by re-reading my drafts: Craig Mobbs,Mark Rainey and Richard Miguda. I am also grateful to Professor Dowson, to my reviewers,to Sheril Leich, Lynsey Gathercole and the staff at Wiley and Integra.Many contributed as much or more on the personal level than the scientific. My col-

leagues, especially my office mate Mathieu Oulevey, Karine Genoud, Fabrice Dusonchetand Giancarlo Corradini. I will miss you. Equally, I would like to express my sinceregratitude to my parents, Anthony and Anne-Marie, and my brother, Nicholas. Last but notleast, my deepest thanks go to my wife Maya for her guidance and encouragement duringthe writing of this book.

Zürich, Switzerland

Remerciements

Ce livre n’aurait pas vu le jour sans la SATW (Zurich) qui m’a attribué une bourse pendantdix-huit mois ainsi que le professeur Jacques Jacot qui a fourni un salaire pour le reste dela préparation. Merci. Je suis également endetté envers le professeur Jacot d’avoir joué lerôle essentiel de directeur. Il a été habilement secondé par mon comité de parrainage (leprofesseur Hannes Bleuler et les docteurs Thomas Maeder et Léon Benmayor). Je voudraiségalement remercier mes amis, en patriculier ceux qui ont su monter leurs qualités en relisantmes ébauches: Craig Mobbs, Mark Rainey et Richard Miguda. Je suis aussi reconnaissant auprofessor Duncan Dowson, aux experts qui ont lu ma thèse, Sheril Leich, Lynsey Gathercoleet les employés de Wiley et Integra.Plusieurs personnes ont contribué autant au niveau personnel que scientifique. Mes

collègues, en particulier Mathieu Oulevey, Karine Genoud, Fabrice Dusonchet et Giancarlo

xx Acknowledgements

Corradini. Vous me manquerez. J’aimerais exprimer ma gratitude sincère envers mesparents, Anthony et Anne-Marie et mon frère, Nicholas. Enfin, mes remerciements du fonddu cœur à ma femme Maya qui m’a motivé et guidé pendant la préparation de ce livre.

Zürich, Suisse

Abbreviations and Symbols

Author’s note:Greek symbols can be found at the end of this list.

a-Si, aSi amorphous silicona constant of the phenomenological modelaAM� aF value taken by a under AM1.5 or fluorescent source respectivelyas constant of the heuristic modelA− electron acceptorAA standardised 1.5V battery sizeAES ambient energy system(s)AM variable representing the air mass spectrumAM0 air mass zero; the characteristics representing the radiant energy found just

outside the Earth’s atmosphere; ASTM standard E490-00aAM1.5 air mass 1.5; the characteristics representing the radiant energy typically found

at sea level. The basis for STC; ASTM standard G159-98AR aspect ratioA surface area of an object �m2�ACS cross-sectional area �m2�Ads area receiving radiant flux from all directions (m2)Ah charge storage capacity (A h)AO capacitor plate overlap area �m2�b constant of the phenomenological modelbAM� bF value taken by b under AM1.5 or fluorescent source respectivelybs constant of the heuristic modelB12 buffer between machines 1 and 2B distance or breadth from building to obstaclec speed of a photon in a vacuum = 299× 108 m/sCB conduction band (eV)CC correlation coefficientCdTe cadmium tellurideCFL compact fluorescent lamp


152 Abbreviations and Symbols

CIGS copper indium gallium diselenideCIS copper indium diselenideCMOS complementary metal oxide semiconductorCVD chemical vapour depositionC current of charging and discharging (A)CF capacitance (F)CL CMOS output load capacitance (F)C2�C4 capacitance in an equivalent circuit (F)d distance between capacitor plates (m)D+ electron donorDF daylight factorDLC double-layer capacitorDoD depth of discharge (%)Dn�Dp diffusion coefficientseM1 effective machine 1e emissivity constant of a surfaceEA electrical energy availableEg band gap energy (eV)EQ quantum energy (J)ESW CMOS switching activity factorf frequency (Hz)fCK CMOS clock frequency (Hz)FF fill factor (%)FS cost of batteries ($US)GaAs gallium arsenideGaInP gallium indium phosphideGe germaniumGn radiant energy intensity �W/m2�Gn0 radiant energy intensity of 1000W/m2

h battery capacity (A h)hc Planck’s constant = 6625× 10−34 J shMAV extra storage due to the minimum application voltage (A h)hIF in-function storage capacity (A h)hSBY standby consumption storage capacity (A h)hSD storage self-discharge capacity required (A h)hTTC total theoretical storage capacity (A h)Hz HertzH height of a building (m)iB3� iB4 current passing through a battery in an equivalent circuit (A)iC2� iC4 current passing through a capacitance in an equivalent circuit (A)iR1�R2�R3�R4 current passing through a resistance in an equivalent circuit (A)iZ3� iZ4 current passing through a Zener diode in an equivalent circuit (A)i1�2�3�4 current at a point in an equivalent circuit (A)InP indium phosphideIPV indoor photovoltaic or indoor photovoltaicsIRE incident radiant energy

Abbreviations and Symbols 153

I current (A)IA average current available (A)Ids irradiance from all directions �W/m2�ID deterministic current consumption (A)IDiode semiconductor junction diode current (A)Ie radiant energy emitted �W/m2�IIF in-function current (A)IL radiant energy induced/light-generated photocurrent (A)Ils intensity of radiant energy emitted from a line source �W/m2�IM current at the maximum power point (A)Ipa intensity of radiant energy from a source of parallel radiation (W/m2)Ips intensity of radiant energy from a point source �W/m2�IPR parasitic current due to parallel resistance (A)Irls irradiance received from a line source (W/m2)Irpa irradiance received from a source of parallel radiation (W/m2)Irps irradiance received from a point source (W/m2)IS saturation current, also denoted by I0 (A)ISC short-circuit current (A)ISR parasitic current due to series resistance (A)I–V curve current against voltage graphJ Joulek Boltzmann’s constant, parameter of probability density functionK permittivity of capacitor separation material (at a fixed temperature)l wavelength (m)lux photometric corresponding to one lumen per square metre or the equivalent

of 0.0929 foot candlesLASER light amplification by stimulated emission of radiationLCA life cycle analysisLCD liquid crystal displayLED light-emitting diodeLILT low intensity, low temperatureLint light incident inside a buildingLext light incident outside a buildingLn�Lp diffusion lengthsMB megabyte or 1 million bytes (technically 1 048 576 bytes). Not to be confused

with megabits (Mb)mono-Si monocrystalline siliconMPP maximum power pointMtoe million tonnes of oil equivalent (12 700GWh)multi-Si multicrystalline siliconM1�M2 machine 1 and 2 respectivelync index of refraction (complex number)NiCd nickel cadmiumN machine non-availability. The ratio of ‘mean time between two consecutive

in-function phases’ and ‘mean time in function’NA�ND densities of acceptors or donors respectively

154 Abbreviations and Symbols

NC�NV effective densities of states in conduction and valence bandsNDIP dipole non-availabilityN1�N2 non-availability of machine 1 and 2 respectivelyOLED organic light-emitting diodepSi polycrystalline silicon PV technologyPC photochemical (a PV technology)PDA personal digital assistantPEFC polymer electrolyte fuel cellPEMFC proton exchange membrane fuel cellPV photovoltaic or photovoltaicsPV 1�2�3�4 specific photovoltaic modules in equivalent circuitsPA power converted from the radiant flux and available to the ambient energy

application (W)Pdip proportion of time that the dipole runsPDS probability that the application demand will be satisfiedPF maximum power that can be delivered by a capacitorPI incident power (W)PIF in-function power (W)Pr radiated power (W)PSBY standby power (W)PSC short-circuit power (W)PSW switching power (W)PT total power (W)P2 proportion of time that the application is in-functionq electronic charge = 1602× 10−19 coulombr distance between radiant energy emitting surface and receiving surface (m)RI��RI� refractive indexes of materials , and 1 respectivelyR portion of incident radiation reflected (%)RA application resistance �"�RA1�2�3�4�B4 specific application resistances in an equivalent circuit �"�Rds radiant flux from all directions (W)Res effective series resistance �"�RP parallel resistance, also called shunt resistance, RSH�"�RS series resistance �"�sr steradians Stefan–Boltzmann constant = 567× 10−8W/m2 K4

SD spectral distributionSMA shape memory alloySn tinSR spectral responseSTC standard test conditionsS�)� number of incident photons per unit time, area and energyt time (s)tIF� tSBY time in function or on standby respectively (s)TCO transparent conductive oxideTR transmissivity ratio


TRE transmitted radiant energyT temperature (K)uB3� uB4 voltage across a battery in an equivalent circuit (V)uC2� uC4 voltage across a capacitance in an equivalent circuit (V)uR1�R2�R3�R4 voltage across a resistance in an equivalent circuit (V)uZ3 voltage across a Zener diode in an equivalent circuit (V)UPS uninterruptible power supplyU1�U2 rate of production of machine 1 and 2 respectively (part/s)VB valence band (eV)V voltage (V)Vmax maximum voltage that can be delivered by a capacitor (V)VM voltage (V) at the maximum power pointVOC open-circuit voltage (V)VS CMOS supply voltage (V)w wavelength (m)W WattWP ‘watt peak’ (W); watts delivered by a solar module under standard test con-

ditions of 1000W/m2�25 �C and a known light spectrum (e.g. air mass 1.5)x-Si or xSi crystalline siliconx distance from the semiconductor surface (m)ZnO zinc oxide

, ratio of machine 1 in-function duration to machine 2 in-function duration,c absorption coefficient,f constant of the phenomenological model,s constant of the heuristic model1 ratio of machine 2 non-function duration to machine 1 non-function duration1f constant of the phenomenological model1s constant of the heuristic modelB constant defined in battery capacity dipole model9s constant of the heuristic model% electrical efficiency (%)� temperature ��C)�,� �1 angle of radiant energy to normal of surface�0 temperature of 25 �C* extinction coefficient (complex number)&1�&2 in-function duration of machine 1 and 2 respectively/1�/2 non-functioning time of machine 1 and 2 respectively= standard deviation) frequency of incoming radiant energy (Hz)8 distance from the semiconductor surface (m)8s constant of the heuristic model

About the Author

Dr. Julian Randall is a Senior Research Fellow at the ETHZ (Eidgenössische TechnischeHochschule Zürich), Switzerland. His Bachelor of Engineering with honours is from Univer-sity of Wales College, Cardiff and he holds a doctorate from the EPFL (Ecole PolytechniqueFédérale de Lausanne), Switzerland. He is a Chartered Engineer and a member of the IEEand IEEE.Dr. Randall is an active researcher of indoor photovoltaic solar applications, most recently

for a wearable autonomous location tracking system [207]. He is the general chair of theSecond International Forum on Applied Wearable Computing in Zürich, Switzerland.His research interests include, but are not limited to, ambient energy power sources,

context-aware wearable systems, autonomous systems and human–computer interfaces.Any queries should be directed via [email protected].

Conclusion

Ambient energy systems (AES) are expected to be the basis for making a noticeable improve-ment in the quality of life experienced indoors. The key barriers to the greater use of AESby designers fall under three headings: lack of experience, lack of information and costpressures. Taking each of these in turn, the chief contributions of this book are as follows.

1. The designer lacks experience. A good example of this is the appreciation of whatambient energy capital is available. Given that humans are relatively insensitive to theenergy flows around them, detailed characterisation advantageously counters the flawedintuitive judgements that might be made.For the example of photovoltaic solar cells, the radiant energy in the indoor space is

measured and modelled in Chapter 3. The results summary in Table 3.4 indicates thata typical well oriented and located IPV product may receive of the order of 1kWh/m2

month (100mWh/cm2 month) of radiant energy. At least 10 times this value may alsobe achieved, on the window ledge, for example. The most significant parameters that arefound along the ‘photon chain’ between the light source and the PV device are identified.Recommendations are provided to enable the designer to achieve the best performancefrom a proposed IPV application.Product design advice is another area where lack of experience may affect the designer.

Such guidelines are often proprietary, and therefore confidential, or consider only asingle product, so may not be applicable. In the final chapter (Chapter 7), therefore,IPV guidelines and heuristics are provided. These are built on existing product designtheories, the results of the report and further findings. It is concluded that three top designpriorities are:

(a) for higher-power products, such as wireless, the designer should at first be lessconcerned with absolute application power than the average energy consumed;

(b) a route to wider ambient energy source application is the use of low-energy designtechniques;

(c) irrespective of how well the technical design is executed, product success reliesheavily on non-technical issues such as product appearance.


148 Conclusion

2. There is a lack of information available to the designer, or insufficient time to collectit. Furthermore, this lack of information is likely to be most apparent at the beginningof the project, the very time when it can have the most impact on design success (seeChapter 2).One of the main areas of lack of IPV information is with regard to photovoltaic col-

lectors under indoor conditions. The history of PV technology is briefly presented inChapter 4 which shows how PV was developed and what products have been realised.From this history, the potential for improvement can be inferred, such as for wireless(communicating) PV-powered indoor products. Then the technological basis of photo-voltaics is explained, including how solar cells work and the advantages and limitationsof PV. A section of improvements that can be achieved is also provided. Thus fore-armed, the designer is therefore in a better position to select appropriate photovoltaiccomponents, as well as to manage suppliers.In the second chapter on PV characterisation (Chapter 5) the lack of comparable low

radiant energy intensity solar cells is addressed. Results for 21 samples representing eightmaterial technologies tested by the author under two spectrum types and radiant energyin the range 1–1000W/m2 have been presented. Thereby, it has been demonstratedthat a 1 sun efficiency reference cannot be projected reliably down to indoor radiantenergy intensities for all solar cells. Equally, it is shown which samples performed bestunder indoor radiant energy intensity and spectra. Also, the reasons why such solar cellcomparisons are not applicable for indoor PV (IPV) design have been explained. Forexample, efficiency may vary markedly with intensity in the decades 1–100W/m2, thelimiting factor being the parallel current, IP.The lack of appropriate models has also been described and addressed; two models

are shown to correlate well with experimental results in the intensity range of interest forIPV design. These may be of service to IPV researchers and designers alike.

3. The designer is under cost pressure. In order to be able to benefit from the diffusenature of ambient energy for higher-power products (e.g. that communicate wirelessly),the energy collection time must greatly exceed the energy use time (e.g. a ratio of 300:1for IPV). Therefore, energy collection must be complemented with energy storage. Giventhat many AES require charge storage and that this component often represents themajority of the cost of the system, determining the correct capacity is essential. It isshown in Chapter 6 that the storage capacity (directly relatated to cost) is related to theprobabilty of the storage satisfying a desired specification (a factor of the price for whichthe product can be sold). This is achieved by the development of a model presented insection 6.5.All the above elements individually contribute to improving the chances for success

of future ambient energy systems. Together, and combined with the further informationprovided, they form a compelling argument that ambient energy technology deservesgreater attention and will be developed.

FURTHER WORK

The results and conclusions for the main research areas mentioned above are complementedby the following.

Conclusion 149

1. Characterisation of the indoor environment. Ambient energy in the built environmentis available at, but not limited to, other frequencies than the radiant energy chieflyconsidered here. These include ‘mechanical’ vibrations in the 50–200Hz range [215]with a magnitude of vibration from ‘about 0.1 to about 10m/s2’ [217]. A complementto this work would by a study of the built space for such frequencies. By combining theresults of Chapter 3 and this future study, a more general mapping of the ideal locationsfor indoor ambient energy products could be developed.With regard to the simulation of radiant energy indoors, a well known mathematical

non-diffusive billiard ball model could be applied to IPV photon modelling if the energydiffusion created by the imperfect reflectivity of the walls were taken into account. Atpresent, while the billiard ball model has attracted hundreds of journal contributions,none deals with diffusion.

2. Technological selection (photovoltaic and storage). An ideal outcome from the IPVpractitioner’s perspective would be a model based on easily accessible data (such asAM1.5 efficiency) which would provide a prediction of cell performance over the fullrange of intensities tested here. The phenomenological model [equation (5.8)] is valid foronly some of the samples across the full range tested, 08–1000W/m2, namely Solems, BPMillenium and Tessag. In order to extend the validity, more terms are required to modelthat part of the efficiency–intensity curve where the gradient becomes negative. It is alsonecessary to investigate the physical meaning of control parameters a [equation (5.10)]and b [equation (5.11)].Another area requiring further understanding is the physical mechanism that induces RP.

It is thought to be related to intrinsic leakage (junction resistance) and extrinsic leakage(perimeter).The storage capacity model presented in section 6.5 could be improved in a number of

ways, some of which have already been mentioned. One approach by which the modelwould be brought closer to the designer selection would be to combine it with costfunctions for the various IPV system components in order to optimise the surface of PVmaterial and storage capacity with respect to cost and, by corollary, quality of service.

3. Power management. Ideally, batteries would be avoided altogether, and some believethis may be possible in the future, especially for low power range electronic applicationssuch as IPV systems [196].

To Maya

DESIGNING INDOOR SOLARPRODUCTSPhotovoltaic Technologies for AES

Julian F. RandallETHZ, Switzerland

Copyright © 2005 by Julian F. RandallPublished by John Wiley & Sons, Ltd

Email (for orders and customer service enquiries): [email protected] our Home Page on www.wiley.com

All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system ortransmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning orotherwise, except under the terms of the Copyright, Designs and Patents Act 1988 or under the terms ofa licence issued by the Copyright Licensing Agency Ltd, 90 Tottenham Court Road, London W1T 4LP,UK, without the permission in writing of the Publisher. Requests to the Publisher should be addressed tothe Permissions Department, John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester,West Sussex PO19 8SQ, England, or emailed to [email protected], or faxed to �+44� 1243 770620.

Designations used by companies to distinguish their products are often claimed as trademarks.All brand names and product names used in this book are trade names, service marks, trademarks orregistered trademarks of their respective owners. The Publisher is not associated with any product orvendor mentioned in this book.

This publication is designed to provide accurate and authoritative information in regard to the subjectmatter covered. It is sold on the understanding that the Publisher is not engaged in renderingprofessional services. If professional advice or other expert assistance is required, the services of acompetent professional should be sought.

Other Wiley Editorial Offices

John Wiley & Sons Inc., 111 River Street, Hoboken, NJ 07030, USA

Jossey-Bass, 989 Market Street, San Francisco, CA 94103-1741, USA

Wiley-VCH Verlag GmbH, Boschstr. 12, D-69469 Weinheim, Germany

John Wiley & Sons Australia Ltd, 42 McDougall Street, Milton, Queensland 4064, Australia

John Wiley & Sons (Asia) Pte Ltd, 2 Clementi Loop #02-01, Jin Xing Distripark, Singapore 129809

John Wiley & Sons Canada Ltd, 22 Worcester Road, Etobicoke, Ontario, Canada M9W 1L1

Wiley also publishes its books in a variety of electronic formats. Some content that appears in print maynot be available in electronic books.

Library of Congress Cataloging in Publication Data

British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library

ISBN-13 978-0-470-01661-9 (HB)ISBN-10 0-470-01661-2 (HB)

Typeset in 10/12pt Times by Integra Software Services Pvt. Ltd, Pondicherry, IndiaPrinted and bound in Great Britain by Antony Rowe Ltd, Chippenham, WiltshireThis book is printed on acid-free paper responsibly manufactured from sustainable forestryin which at least two trees are planted for each one used for paper production.

Glossary

Azimuth The horizontal angle measured in a clockwise direction in degrees (0–360) fromeither a North or South reference line. Navigators and surveyors use North. Astronomersand meterologists use South. Azimuth is synonymous with (compass) bearing

Charge controller A device that protects the storage device (battery) by managing itscharging and discharging

Collector A photovoltaic cell or module

Compound semiconductor A semiconductor composed of two or more elements, suchas the binary semiconductor cadmium telluride (CdTe)

Conventional cell A crystalline silicon cell, having a typical thickness of 350± 50�m or3–4 times the diameter of a human hair

Daylight factor The ratio of the light intensity at a point indoors with the light intensityvertically upwards outdoors

Diffuse radiation Incident radiant energy received that has been scattered (e.g. by theatmosphere) or reflected (e.g. by an object) 1 or more times before reaching the photovoltaiccollector

Direct radiation Incident radiant energy that, apart from refraction, has travelled in astraight line from source to collector. An example of such radiation is that light whichcreates a shadow

Doping The addition of a trace amount (e.g. a dopant:semiconductor ratio for crystallinesilicon of <10−6) to a semiconductor material. In the case of silicon, group III elements suchas B create electron acceptor holes and are called p-dopants, group V elements such as Pare electron donors and are called n-dopants


158 Glossary

Efficiency The ratio of the electrical output of a photovoltaic cell, or module, to theradiant energy incident on it

Elemental semiconductor A semiconductor in which all atoms, apart from dopants, arethe same, e.g. silicon and germanium

Emissivity, e A measure of material propensity to radiate energy. A surface with e= 1 isan ideal radiator (black body). All practical surfaces have e< 1. In theory, e can be zero

Energy payback time The time necessary for an energy-producing system or device toproduce as much energy as was required for its manufacture and assembly. Typically 2–4years for PV modules outdoors

Extrinsic semiconductor Doped (impurity-added) semiconductor material usually pre-pared from an intrinsic semiconductor where the dopant-induced conductivity dominates

Fill factor The ratio of the maximum power point power to the product of open-circuitvoltage and short-circuit current. Ideally it has a value as close to 1 as possible, achievedby having as orthogonal an I–V curve as possible

Fresnel lens A flat plate of glass or plastic with prismatic structuring which producesa similar effect to a conventional lens. May be used to direct incident radiation moreeffectively onto the collector

Hybrid system An energy system combining two or more sources of energy such asphotovoltaic with a piezo (mechanical to electrical energy) or Peltier (heat to electricalenergy) system; also used for two or more charge storage devices

In-function mode The active state of an electronic device, when it is providing a desiredservice

Insolation The cumulative sunlight reaching a point �W/m2 day� measured verticallyupwards

Intrinsic semiconductor Undoped (no impurity) semiconductor material where thesemiconductivity dominates

IPV Acronym of the adjective Indoor Photovoltaic and the noun Indoor Photovoltaics

I–V curve Current against voltage graph for a photovoltaic cell or module under a rangeof loads from short circuit to open circuit. The resulting shape indicates the cell or moduleperformance, in particular via the intercept of the voltage axis (open-circuit voltage), theintercept of the current axis (short-circuit current) and the maximum power point

LASER Acronym of Light Amplification by Stimulated Emission of Radiation

Load Electrical energy consumed by the application or the photovoltaic system

Maximum power point The position on a particular I–V curve at which the product of Iand V (i.e. the power) is maximum

Monolithic series (inter)connection A means of electrically connecting in a singlestructure the top electrode of a typically thin-film technology cell to the bottom electrode of

Glossary 159

the neighbouring cell without short-circuiting either cell. May be referred to as ‘structured’in PV jargon

Open-circuit voltage, Voc The voltage across the electrodes of a photovoltaic cell ormodule when illuminated but no external current is flowing

Parallel connection A means of increasing the current of a photovoltaic module. This isachieved by electrically connecting the positive electrodes of more than one cell togetherand the negative electrodes of these cells together. Rarely applied in indoor PV applications

Peak power The output of a PV module under standard test conditions (STC). Alsoreferred to as Watts peak, Wp

Phonon A quantised lattice vibration, e.g. sound, in a solid medium. necessary for theindirect band gap material absorption process

Photon A fundamental unit of light energy

Photopic Light seen by a human

Photovoltaic A system that converts light to electricity. The derivation is from phos, theGreek word for light, and volt after Alessandro Volta (1745–1827), the renowned inventorof the battery at the turn of the nineteenth century

Photovoltaic cell A basic photovoltaic diode device made of semiconductor materials.Like a battery, it has two electrodes. If a resistance is placed across these electrodes, itproduces voltage and direct current on exposure to light

Photovoltaic module A number of photovoltaic cells electrically connected and in a formthat can be used in practice. For outdoor use, this implies the cells being assembled on atleast one glass sheet and encapsulated, thus protecting them from the environment

Photovoltaic system A combination of photovoltaic modules with all other necessaryphotovoltaic system components to form an autonomous source of power

Photovoltaic system components Typical components for indoor applications includestorage (secondary batteries) and storage discharge protection to prevent the cells actingas a current drain in darkness (diode). Less common components are the maximum powerpoint tracker (MPPT) and DC–DC charge converter. The cost of these components as partof a photovoltaic system can be referred to as the balance of system (BOS) costs

PV Acronym of the adjective Photovoltaic and the noun Photovoltaic(s)

Quaternary material A compound containing four elements, e.g. copper indium galliumdiselenide (CIGS)

Series connected A means of increasing the voltage of a photovoltaic module achievedin the most simple form by connecting the positive electrode of one cell to the negativeelectrode of another cell. The voltage across the remaining unconnected electrodes will beincreased in this tandem arrangement, while the current will be constant throughout

Short-circuit current, ISC The current (A) flowing when the electrodes of a photovoltaiccell or module under illumination are electrically connected via an external circuit of zeroresistance

160 Glossary

Solar constant The directly incident radiant power from the sun on the Earth’s meanorbit. Outside the atmosphere, approximately equal to 1367± 5W/m2

Spectral response The relative current output of a solar cell under monochromatic lightas a function of wavelength

Standard test conditions (STC) The most common testing conditions for measuringphotovoltaic cell or module performance. Based on IEC standard IEC-904-3 (1989), i.e.1000W/m2�25 �C, AM1.5 (air mass), solar spectrum perpendicular to the cell surface

Standby mode The state of an electronic device that is powered on but is otherwiseinactive (not in function). Synonym of ‘quiescent mode’

Ternary semiconductor A semiconductor based on three elements. An example is abinary semiconductor with the addition of a relatively small amount of a third element, e.g.copper indium diselenide (CIS)

Thin-film cells Photovoltaic cells manufactured by the superposition of photosensitivematerial layers. Typical total thickness of a few micrometres, much less than the diameterof a human hair

This page has been reformatted by Knovel to provide easier navigation.

INDEX

Index Terms Links

A

absorption 28 29 31 61 69

136

actual storage capacity 108

amorphous

silicon 59 73 79 88 90

93

amplify 8 129

application

photovoltaic 3 7 55 102

artificial light sources 30 46 82

aspect ratio 36 38 133

azimuth 27

B

band gap 59 71 73 86

batteries xv xxi 20 96 102

134

battery

primary 19 105 116

secondary 20 97 115 135

black body 28

building

physics 34

Index Terms Links


C

cadmium 98

cadmium telluride 66 72 74 83 90

93 115

capital investment 15 97

case studies 143

cell

battery 95

fuel 2 105

photovoltaic 53

solar 53

charge control 106

charge storage 95 101

collector

photovoltaic 53 148

communication 4 99 130

compound, see semiconductor compound

compute 6

concentration 78

conceptual design 14 122 126 145

conduction band 59 75

contribution xxi

copper indium diselenide (CIS) 66

copper indium gallium diselenide (CIGS) 72 74

copyright 3

crossover point 46

crystalline silicon 59 72 86 133

Index Terms Links


D

daylight factor 34 51

daylighting 39

density

energy 99 117

power 2 99 117

Derwent 3

design

for manufacturing 140

design process 122

designer responsibility 22

detailed design 122 145

dimensioning 92 114

dipole 61 110

display 30

doping, see semiconductor doping

double-layer capacitor 103 105 117

E

effect

memory, see memory effect

efficiency

photovoltaic 56

electrochemical capacitor 103

electrochemical charge storage 101

embodiment 122 136 145

energy

ambient 2

available 10 23 51 132

Index Terms Links


energy (Cont.)

density 2

harvesting 2 52

scavenging 2 52

source 1

energy constraint 107

engineering

design 13 16

equivalent circuit 129

experiments 38 41 42 47 80

F

fill factor 64 85 91

fluorescent 7 30 82

flywheel 100 117

Fritts, Charles 54

front contact 62 69 76

fuel cells 105

in-function mode 109 123

functional analysis 15

G

gallium 61 63 75 81 88

Gedanken experiment 43

glazing/floor ratio 36 51

gold capacitor 103

Index Terms Links


H

heuristic 89 109 114 123 126

133

hybrid 118

I

ideality 77

insolation 135

intellectual property 3 136

irradiance

map 47

solar 54

I–V curve 65

L

life cycle analysis 18

light 30

creation 25

incidence 25

reflection 25

translation 25

line source 26

location

IPV 7 23 44 52

M

market overview 8 43 55 58 98

Index Terms Links


maximum power point 64

memory effect 99 135

microcrystalline silicon 90

mismatch, see spectral mismatch

mobile 1 9

model

battery 114

Markovian chain 110

phenomenological 85

storage capacity 110

technology specific 90

module

photovoltaic 75 79 136

monolithic series connection 57 69 75

motivation xxi

movement 30

multijunction 81

N

natural light sources 29 34

nepro solar LED watch 58 74

net curtain transmissivity 41

nickel cadmium 99 102 115 134 140

O

obstacles 37

open-circuit voltage 67 83 133

orientation 7 48 145

Index Terms Links


P

packaging 101

patents 3 126 138

phonon 62

photochemical 81 84

photometric 10 33 43

photon 24 41 59

photopic 33 42

photovoltaic

application 57 121

cell 53

collector 53

conventional 54

efficiency 64 71 76

history 54

indoor 5

module 55

thin film 75

pioneer IPV products 56

Planck 24 28

p–n junction 60

point source 26

power

density 99 117

product design 15 121

product liability 17

profile

application 49

building 49

Index Terms Links


profile (Cont.)

product use 125

user 125

pseudo capacitance 104

Q

quarternary, see semiconductor quaternary

R

radiant energy spectra 28

radiation

diffuse 32

direct 27 78

indirect 34 47

radiometric 37

Rayleigh (JohnWilliam Strutt) 28

rechargeable 97

recombination 70

recommendations 132

reflection 32 33

refraction 31 32

resistance

parallel 70

series 72

respond to light 30

S

scale model 34

secondary, see semiconductor secondary

Index Terms Links


self discharge 109

semiconductor

compound 63

doping 60

elemental 63

phototonic 59

quaternary 160

secondary 160

ternary 63 160

thickness 63 70 73 157

sensor

application 100

mote xxi 5 116 128 144

sensors 5 34 116

series-interconnected 76

short circuit current 66 77

silicon amorphous, see amorphous silicon

silicon crystalline, see crystalline silicon

simulation 50 132 149

Smith, Willoughby 54

solar constant 27 160

solar simulator 80

Spadini, Paolo 58

specific energy 99 102

specific power 99 102 117

specification

battery 106

circuit 129

product 5 15

spectral mismatch 69 77 88

Index Terms Links


spectral response 72 84

springs 100

Staebler–Wronski 73

standard test conditions 79

standby current 22 108

standby mode 10 107 124 126

Stefan–Boltzmann 28

storage

mechanical 100

storage capacity 114

super capacitors 96 97 102 103

superposition 46

T

taxonomy 5 62

technology overview 66 72 96 97

ternary, see semiconductor ternary

thermal radiation 10 28

thermalisation 69

thickness solar cell 70

thin film 74 75

tools 5

toys 5

transmissivity

net curtain 41

photopic 43

thermal 43

venetian blind 42

transparent conductive oxide 91

traps 70

Index Terms Links


typical values 51 124

U

ultracapacitor 103

V

valence band 59

venetian blind transmissivity 42

vibrational

energy 2 52 62 149

W

Waldram 39

Introduction

MOTIVATION AND OBJECTIVES

It has been estimated that a source energy saving of $US 55 billion per annum in the UnitedStates alone, or the equivalent of 35 million metric tonnes of carbon emissions, could bemade with improved building automation [235]. Wireless sensors would provide the data toa control centre.While sensors in buildings are not new, a major barrier remains: that of powering the

devices. The commonly accepted solutions are either to hard wire the sensor or to changethe batteries on a regular basis. These have the disadvantage that installation (e.g. wiring)and maintenance (e.g. batteries) costs are likely to rival or exceed those of the sensor beinginstalled. In order to offset this cost, especially for sensors that communicate wirelessly,the power could be drawn from ambient energy. Each sensor node could therefore functionautonomously and cost significantly less than today.Autonomous wireless sensors offer the potential for saving global sensor system cost.

With lower costs, greater implementation can be expected with associated benefits includinga reduction in the use of fossil fuels. Another advantage of the low cost of such sensorsis that they could be integrated into building materials, such as ceiling and floor tiles, andinstalled by non-specialists.As the example of sensors shows, using ambient energy systems can be advantageous both

from the economical and the ecological point of view. Against this background, this bookargues that ambient energy systems could and ought to be more widely applied for poweringelectronic devices within the built space than at present. The case for this is made takingthe example of solar (photovoltaic) technologies; some of the findings may be transferredto other ambient energy technologies.

CONTRIBUTIONS

The basis for the case for ambient energy systems in the built space rests on their technicalfeasibilty. Important factors that are investigated include the available energy, the ambi-ent energy system performance under such conditions and the performance of the charge

xxii Introduction

storage required. Taking the example of solar cell technology, the main contributions of thisbook are:

• two forms of a phenomenologically based model for solar cell response – these modelssimulate PV response across a range of radiant energy intensities from 1000 down to1W/m2, for both daylight and electrically sourced spectra;

• further experimental results that are not available elsewhere, such as the characterisationof radiant energy and PV performance indoors;

• a model for capacity specification confidence for charge storage, which describes therelationship between charge storage and the probability that the application will havesufficient charge to function;

• a range of design guidelines covering all design process phases that support the imple-mentation of the book findings in practice.

It is thereby shown under which circumstances such devices would be feasible. By providingmodels validated with experimental results, this work supports future product design efforts.

BOOK STRUCTURE

The topics covered by the chapters of this book are sequentially organized to gather theknowledge required for an overall understanding of the problems related to indoor photo-voltaic (IPV) and ambient energy system (AES) product design.Chapter 1 describes the state of the art in low-power energy sources and in particular solar

(photovoltaic) cells, identifies the areas where information is less complete and presents howthis book proposes to fill these gaps.Chapter 2 examines the IPV product engineering design process, highlighting the impor-

tance of reliable initial guidance and the influence of environmental concerns on the finaldesign.Chapter 3 takes the example of photovoltaic solar cells indoors, and presents a character-

isation of the radiant energy in the built space.Chapter 4 addresses the lack of widely accessible information about photovoltaic (PV)

solar cells, by giving a historical and technological overview of PV and the reasons whysolar cell efficiency is relatively low.Chapter 5 explores the correct design of IPV cells and modules and considers the charac-

teristation of solar cells indoors on the basis of the results of Chapters 3 and 4.Chapter 6 determines the ways in which ambient energy system charge storage require-

ments may be met by balancing the many storage parameters.Chapter 7 considers the many issues necessary for AES or IPV product design from the

practitioner’s perspective and takes the form of guidelines and, where appropriate, heuristics.

Preface

The use of ambient energy systems (AES) for indoor applications is often limited to productsof low power (microwatts, �W). This book defends the case that, under certain conditions,AES based on photovoltaic solar cells may be used for higher-power devices (mW rating).Such devices, which will soon be available [43], would allow fully autonomous sensornodes which have a number of advantages. One is that wiring, which may otherwise be themost expensive element of a sensor network installation, could be avoided. In the case ofbattery-powered sensor networks, battery maintenance could be reduced.The motivation for the book is both financial and ecological, as two examples for the

United States show: it has been estimated that a source energy saving of $US 55 billionper annum, or the equivalent of 35 million metric tonnes of carbon emissions, could bemade by using AES sensors in buildings as a matter of course [235]. Another examplesupports the case that using AES to extend battery life would reduce toxic waste as batteriesrepresent ‘less than one per cent of total municipal solid waste generated’ but ‘account fornearly two-thirds of the lead, ninety percent of the mercury, and over half of the cadmiumfound in the waste’ [150]. A further motivation is the trend towards more energy efficientand comfortable buildings reflected in the Swiss Minergie standard [3]; the latter would befurthered by the use of building sensors.The main avenues of this engineering design research are:

• power management: ensuring that the application and AES consume minimal energy;• characterising the influence of the environment on the ambient energy available;• technological selection: with respect to the efficiency of conversion of ambient energy

into charge and the storage of this charge;• design methodologies that should be applied during all design process phases.

Following these avenues, the case of solar (photovoltaic) cells is considered. The maincontributions are:

(a) two forms of a phenomenological model of solar cell response under indoor radiantenergy conditions;

(b) the validation of (a) with extensive experimental results which are necessary to theengineering designer but were hitherto unavailable;

(c) a model of the service efficiency of AES with respect to charge storage capacity;

xvi Preface

(d) a characterisation of the radiant energy collected by indoor solar cells;(e) a range of design guidelines and heuristics that complement existing product design

methodologies for indoor AES.

Overall, this book identifies the technological barriers that at present hinder the further useof ambient energy sources. The results obtained, however, prove that solutions can be foundby following the proposed design guidelines; hence further achievements can be expected.

Préface

L’usage de systèmes d’alimentation utilisant l’énergie ambiante (SAES) pour des applicationsd’intérieur se limite souvent aux produits de basse puissance (microwatts, mW). Ce travaildéfend la thèse que, sous certaines conditions, les SAES basés sur des photopiles (solaires)pourraient être employés pour des dispositifs de puissance plus élevée (milliWatt, mW).De tels dispositifs, qui seront bientôt disponibles [43], permettraient, par exemple, la misesur pied de réseaux de capteurs entièrement autonomes, évitant ainsi le câblage, qui estsouvent l’élément le plus cher de telles installations. Dans le cas de réseaux de capteursalimentés par des batteries, c’est la maintenance de ces dernières qui s’en trouverait réduite.Ce livre est motivé par des considérations tant économiques qu’écologiques, comme

l’illustrent deux exemples provenant des États-Unis. Dans le premier exemple [235], il estestimé qu’une réduction de la consommation d’énergie équivalente à 55 milliards US$ parannée, c’est-à-dire l’équivalent de 35 millions de tonnes d’émissions à base de carbone,serait envisageable en utilisant systématiquement des capteurs SAES dans les bâtiments.Le deuxième exemple [150] démontre que l’augmentation de la durée de vie des batteriesqu’apporterait l’usage des SAES réduirait sensiblement la quantité de certains déchets tox-iques. En effet, bien que les piles ne représentent que moins d’un pour-cent des déchetssolides municipaux, elles représentent à elles seules presque les deux-tiers du plomb, lesquatre-vingt-dix pour cent du mercure, et plus de la moitié du cadmium trouvé dans lesdéchets. Une motivation supplémentaire est la tendance vers des bâtiments plus confortablesqui se reflète dans la norme suisse Minergie [3]; le succès de ce dernier pourrait-être amélioréen utilisant des capteurs.Les principaux axes de cette recherche sont:

• la gestion de la puissance: visant à assurer une consommation minuscule d’énergie.• la caractérisation de l’influence de l’environnement sur l’énergie ambiante disponible.• le choix des technologies en tenant compte du rendement de la conversion de l’énergie

ambiante en charge ainsi que des performances du stockage de charge.• des méthodologies qui s’appliquent à toutes les phases de la conception.

L’étude des cellules (photovoltaïques) solaires selon cette approche (d’axes) à donné lesprincipales contributions suivantes:

(a) deux formes d’un modèle phénoménologique de la réponse à l’énergie radiative dephotopiles à l’intérieur de bâtiments.

(b) la validation de a) par d’amples résultats expérimentaux, indispensables à la conceptiond’équipements SAES mais qui n’étaient jusqu’à présent pas disponibles

(c) un modèle du taux de service du SAES en fonction de la capacité de stockage de charge

Preface xvii

(d) une caractérisation de l’énergie radiative captée par les photopiles solaires à l’intérieurde bâtiment.

(e) une gamme de directives et d’heuristiques de conception de SAES, qui fournissent uncomplément aux méthodologies existantes.

De façon générale, ce livre identifie les barrières technologiques qui entravent actuellementl’utilisation des sources d’énergies ambiantes. Les résultats obtenus prouvent toutefois quedes solutions existent si on suit les directives de conception proposées; il en découle que defuturs réalisations seront bientôt sur le marché.

References

[1] ABB, Renewable Energy: Status and Prospects 1998, Internal ABB report.[2] E. Alsema, Environmental Aspects of Solar Cell Modules, Report 96074, Department of Science,

Technology and Society, Utrecht University, Utrecht.[3] Verein Minergie (AMI) Association Minergie, Minergie Standard, Register: 2 Reglement Stand:

July 2002 HB2-2 from http://www.minergie.ch/download/standards.pdf on 03.05.01.[4] D. Anderson, J. Bishop and E. Dunlop, Energy Rating of Photovoltaic Modules, Proceedings

of the 16th European Photovoltaic Solar Energy Conference and Exhibition, 1–5 May 2000,Glasgow.

[5] ARM, Details of ARM Products, http://www.arm.com.[6] S. Arrhenius, Nobel Prize for Physics 1921 (and others), http://mirror.nobel.ki.se/laureates/

physics-1921-press.html.[7] ASTM International, Standard Solar Constant and Zero Air Mass Solar Spectral Irradiance

Tables, E490-00a Standard.[8] E. Baruh, La salle BM2.135: Environment Propice Pour Accueillir Des Applications Photo-

Voltaïques ‘Indoor’? EPFL internal project, July 2002, Lausanne, Switzerland.[9] P. Barrade, H. Ravokatrasolofo and S. Pittet, Energy storage device by means of supercapacitors,

Internal poster, EPFL, Lausanne, Switzerland.[10] P. Barrade and A. Rufer, Supercapacitors as Energy Buffers: A Solution for Elevators and for

Electric Buses Supply, PCC-OSAKA 2002: Proceedings of the Power Conversion Conference,2002, Osaka, Vols I–III, pp. 1160–1165.

[11] P. Barrade, S. Pittet and A. Rufer, Series Connection of Supercapacitors, with an Active Devicefor Equalizing the Voltages, IPEC 2000: Proceedings of International Power Electronics Con-ference, 3–7 April, Tokyo, Japan.

[12] BATREC website, Batteries, http://www.batrec.ch/english/batteries.htm on 22 September 2002.[13] Y. Bellouard, Personal Communication, Discussion with Dr Yves Bellouard, EPFL.[14] L. Benini, System-Level Power Optimization, Course entitled ‘Advanced Digital System Design’,

Lausanne, Switzerland, October 2002.[15] L. Benini and G. de Micheli,Dynamic Power Management, Boston, MA (etc.): Kluwer Academic

Publishers, 1998, ISBN 0-7923-8086-X.[16] R. Bertrand, A. Schneider, M. Hammer and S. Michaud, SOLERO P/L Power Design, Technical

note document SOLERO-2100-10, ESTEC Contract 15776/01/NL/PA, p. 17.


162 References

[17] H. J. Bergveld, W. S. Kruijt and P. H. L. Notten, Electronic-network Modelling of RechargeableNiCd Cells and Its Application to the Design of Battery Management Systems, Journal of PowerSources, 77, 143–158, 1999.

[18] BIBLING-Elektronische Bibliothek für Ingenieure, Image of VDI 2221 in German, downloadedhttp://www.ub.ruhr-uni-bochum.de/Projekte/Bibling/Ingenieur.htm on 29.09.02.

[19] R. W. Birkmire, Compound Polycrystalline Solar Cells: Recent Progress and Y2K Perspective,Solar Energy Materials and Solar Cells, 65 (1–4), 17–28, January 2001.

[20] M. A. de Blas, J. L. Torres, E. Prieto and A. Garcia, Selecting a Suitable Model for CharacterizingPhotovoltaic Devices, Renewable Energy, 25, 371–380, 2002.

[21] D. Bloch, 200 ans d’électrochimie: sources d’énergie pour objets mobiles, de la pile Volta auxpiles à combustible miniatures, Atelier Microtechnique, Arc et Senans, 29–31 October 2002.

[22] J. D. Booker, M. Raines and K. G. Swift, Designing Capable and Reliable Products, Oxford:Butterworth-Heinemann, 2001, ISBN 0-7506-5076-1.

[23] G. Bopp, H. Gabler, et al., A Systematic Effort to Define Evaluation and Performance Param-eters and Criteria for Lead–Acid Batteries in PV Systems, Proceedings of the 13th EuropeanPhotovoltaic Solar Energy Conference, 23–27 October 1995, Nice, France.

[24] M. Born and E. Wolf, Principles of Optics, Cambridge University Press, 1999, p. 46, ISBN0 521 642221.

[25] H. Brezet and C. Van Hemel, Ecodesign: A Promising Approach to Sustainable Production andConsumption, United Nations Publication UNEP, Paris, 1997, ISBN 92-807-1631-X.

[26] K. Bücher, True module rating, Proceedings of the 13th European Photovoltaic Solar EnergyConference, 23–27 October 1995, Nice, France, pp. 2097–2103.

[27] K. Bücher (Fraunhofer ISE), Site Dependence of the Energy Collection of PV Modules, SolarEnergy Materials and Solar Cells, 47, 85–94, 1997.

[28] K. Bücher, G. Kleiss and D. Bätzner, RRCModule Energy Rating: A Module Survey, Proceedingsof the 26th Institute of Electrical and Electronics Engineers, Inc., Photovoltaics SpecialistsConference, 30 September–3 October 1997, Anaheim, CA, pp. 1187–91.

[29] M. Burgelman and A. Niemegeers, Calculation of CIS and CdTe Module Efficiencies, SolarEnergy Materials and Solar Cells, 51, 129, 1998.

[30] M. Burgelman and A. Niemegeers, Influence of Illumination Conditions on the Design of Thin-Film Modules, Solar Energy Materials and Solar Cells, 57, 85–95, 1999.

[31] S. Burnside, S. Winkel, et al., Deposition and Characterization of Screen-printed Porous Multi-layer Thick Film Structures from Semiconducting and Conducting Nanomaterials for Use inPhotoVoltaic Devices, Journal of Materials Science – Materials in Electronics, 11 (4), 355–362,June 2000.

[32] M. Burri, Alimentation solaire d’une souris sans fil, University of Neuchatel internal project(travail de semestre) Winter 2000–2001.

[33] BUWAL, Bundesamt für Umwelt, Wald und Landschaft, Separately collected waste forrecycling; quantities and rates, downloaded from http://www.umwelt-schweiz.ch/imperial/md/content/abfall/separatsammlung_e.pdf on 03.06.05.

[34] M. Cagalj, J.-P. Hubaux and C. Enz, Minimum-Energy Broadcast in All-Wireless Networks:NP-Completeness and Distribution Issues, Proceedings of ACM MobiCom 2002, Atlanta, GA,September 2002.

[35] California Energy Commission, Public Interest Energy Research, 2001 annual report. PIERprogram project summaries. Appendix A.

[36] F. Callias, Energie solaire récupérée au poignet, CSEM technical report 188.[37] M. Camani, N. Cereghetti, D. Chianese and S. Rezzonico, Comparion and Behaviour of PV

Modules, Proceedings of the 2nd World Conference Exhibition on Photovoltaic Solar EnergyConversion, 6–10 July 1998, Vienna, Austria.

[38] A. Canto, Bye-bye lithium (micro fuel cells by Robert Hockaday), SVM October 2000,http://svm.vunet.fr.

[39] COM (2001) 31 final, Environment 2010: Our Future, Our Choice, EC communication2001/0029 COD.

[40] CEN (European Committtee for Standardization), Environmental management – Life cycleAssessment – Principles and Framework, EN ISO 14040: 1997.

References 163

[41] B. Charny, Motorola Steps on the Gas for Cell Phones, published 02.10.01, downloaded fromhttp://news.com.com/2100-1033-273818.html?legacy=cnet on 03.06.05.

[42] D. M. Chapin, C. S. Fuller and G. L. Pearson, A new p–n junction photocell for converting solarradiation into electrical power, Journal of Applied Physics, 25, 676–677, 1954.

[43] A. Chandrakasan, R. Amirtharajah, J. Goodman, W. Rabiner, Trends in Low Power DigitalSignal Processing, Proceedings of the 1998 Institute of Electrical and Electronics Engineers,Inc., International Symposium on Circuits and Systems (ISCAS) ’98, 31 May–3 June 1998,Vol. 4, pp. 604–607.

[44] B. Cheldein, Computational Producibility Evaluation in Product Development Seminar, Seminarat EPFL, Lausanne, Switzerland, 01.11.02.

[45] D. Chianese, N. Cereghetti, S. Rezzonico and G. Tavaglini, 18 Types of PV Modules under theLens, Proceedings of the 16th European Photovoltaic Solar Energy Conference and Exhibition,1–5 May 2000, Glasgow.

[46] Chipcon, CC1000 RF Transceiver, Promotional information.[47] CIE, Lighting of Indoor Work Places, Draft International Standard ISO/DIS 8995 CIE DS

008.3/E.[48] CIE, Spatial Distribution of Daylight – CIE Standard Overcast and Clear Sky, ISO 15469

CIE S 003.[49] CIT Ekologik, Diagram of Phases of an LCA, downloaded from http://www.lcait.com/img/

LCA_ISO14040.gif on 021111.[50] J. F. Cole, Battery energy-storage systems – an emerging market for lead/acid batteries, Journal

of Power Sources, 53, 239–243, 1995.[51] R. Compagnon, D. Raydan, Irradiance and Illumination Distributions in Urban Areas, Pro-

ceedings of the 17th International Conference on Passive and Low Energy Architecture (PLEA)2000, Cambridge, UK.

[52] R. Compagnon, Measured cumulative solar irradiance for a location, Personal communication.[53] R. Compagnon, Potentiel de captage du rayonnement solaire en milieu urbain, Proceedings of

the 2001 Conférence Internationale Energie Solaire et Bâtiment (CISBAT).[54] R. Compagnon, Simulation, Advanced Radiance Training Course, Fribourg, Switzerland.[55] R. Compagnon, Simulations numériques de systèmes d’éclairage naturel à pénétration latérale,

EPFL thesis 1193, Lausanne, Switzerland.[56] Communication from the Commission, Energy for the Future: Renewable Sources of Energy,

White Paper for a Community Strategy and Action Plan COM(97)599 final (26/11/1997).[57] A.-S. Corbeau, Les applications portables, downloaded http://www.annso.freesurf.fr/

portable.html on 22.10.02.[58] F. de Coulon and M. Jufer, Introduction à l’électrotechnique, Traité d’Électricité, Vol. I, 1981,

Presses Polytechniques Romandes, Lausanne.[59] Derwent Innovations Index, Patents mentioning photovoltaic switch: JP2002127178-A,

US5969581-A, DE29622049-U and CH685216-A5, Search via ISI Web of Knowledge on 10April 2003.

[60] DG-TREN: European Commission Directorate – General Energy and Transport, Code of conducton efficiency of external power supplies, downloaded from http://energyefficiency.jrc.cec.eu.int/pdf/COC_PSfinal_v6june2000.pdf on 23.10.02.

[61] US Department of Energy, Energy use of some typical home appliances, fromhttp://www.eere.energy.gov/consumerinfo/refbriefs/ec7.html on 05.05.03.

[62] DTI, Quality Assurance Programme 1992–1996: Federal Ministry of Research and Technology,germany, Translation 0219-92, DTI, London, UK.

[63] D. Dubois and J.-P. Forestier, Productivité et en-cour moyen d’un ensemble de deux machinesséparées par une zone de stockage, RAIRO Autom. Syst. Analysis and Control, 16, 105–132.

[64] W. Durisch, K. Lam and J. Close, Behaviour of copper indium diselenide module under realoperating conditions, Proceedings of World Renewable Energy Conference VII 2002, Cologne.

[65] V. Dyakonov, I. Riedel, C. Deibel, J. Parisi, C. Brabec, N. S. Sariciftci and J. Hummelen,Electronic Properties of Polymer–Fullerene Solar Cells Studied with Light-Induced ElectronSpin Resonance and Admittance Spectroscopy, Proceedings of the 17th European PhotovoltaicSolar Energy Conference and Exhibition, October 2001, Munich.

164 References

[66] Ecolife Network, Ecodesign Guide – Environmentally Improved Product Design Case Studies ofthe European Electrical and Electronics Industry, downloaded from http://www.ihrt.tuwien.ac.at/sat/base/ecolife/index.html on 021111.

[67] ECR (European Commission Regulation), ENVIRONMENT Eco-label; revised Community eco-label award scheme, Regulation (EC) 1980/2000 of the European Parliament and of the Councilof 17 July 2000 (Official Journal L 237, 21.09.2000) and Regulation 880/1992 (a voluntaryinstrument).

[68] J. A. Eikelboom and M. J. Jansen, Characterisation of PV Modules of New Generation,Energieonderzoek Centrum Nederland (ECN), Postbus 1, 1755 ZG Petten report ECN-C-00-067,June 2000.

[69] N. H. Eklund and P. R. Boyce, The Development of a Reliable, Valid, and Simple OfficeLighting Survey, Journal of the Illuminating Engineering Society, Summer 1996.

[70] E. Macii, Power Analysis and Low-Power Design, Advanced Digital System Design, Lausanne,Switzerland.

[71] Unknown, Batterien and Supercaps, Energie and Umwelt: Forschung am Paul Scherrer Institut.[72] Environmental Protection Agency Envirowise Website, Fact Sheet: Disposal of Alkaline Bat-

teries, http://es.epa.gov/techinfo/facts/proact5.html on 22 September 2002.[73] Joint Stock Company ESMA, FAQ and library, downloaded from http://www.esmacap.com/

FAQ/?lang=English on 021217.[74] EU_Stand-by Initiative, List of participants of the code of conduct on efficiency of external

power supplies for the European actions to Improve the energy efficiency of electrical equipmentwhile either OFF or in standby, downloaded from http://energyefficiency.jrc.cec.eu.int/html/s_b-ParticipantsCoC.htm on 23.10.02.

[75] EurObserv’ER, A 29% remarkable growth, Systèmes solaires 136-2000-1.[76] W. J. Fabrycky, Modeling and Indirect Experimentation in System Design Evaluation, Journal

of the International Council on Systems Engineering (INCOSE), 1 (1), 133–144, July–September1994.

[77] A. L. Fahrenbruch and R. H. Bube, Fundamentals of Solar Cells: Photovoltaic Solar EnergyConversion, Academic Press, Inc., ISBN 0-12-247680-8.

[78] P. Fairley, Solar on the Cheap, Technology Review – MITs Magazine of Innovation, 48–53,January/February 2002.

[79] Farnell, Component Catalogue, Vols 1 and 2, February 2002.[80] H. Field, Solar cell spectral response measurement errors related to spectral band width and

chopped light waveform, 26th Institute of Electrical and Electronics Engineers, Inc., PhotovoltaicSpecialists Conference, 29 September–3 October, 1997, Anaheim, CA.

[81] F. Figge and C. Butz, Solar energy: A Key Energy Source for the 21st Century? SarasinSustainable Investment Outlook.

[82] S. Fink, W. Lannggassner and U. Wagner, Ganzheitliche energetische Bilanzierung von bat-teriebetriebenen Kleingeräten, Elektrizitätswirtschaft, 97 (13), 1998.

[83] B. Flipsen, G. Timmers, R. van den Hoed and I. Leeflang, In Search for Application Field ofFuel Cells, Proceedings of Electronics Goes Green 2000+, Vol. 1, Technical Lectures, 11–13September 2000, Berlin, Germany, pp. 719–725.

[84] S. Forrest, P. Burrows and M. Thompson, The Dawn of Organic Electronics, Institute ofElectrical and Electronics Engineers, Inc., Spectrum, August 2000.

[85] S. J. Fortune, D. M. Gay, B. W. Kernighan, O. Landron, R. A. Valenzuela and M. H. Wright,WISE Design of Indoor Wireless Systems: Practical Computation and Optimization, Instituteof Electrical and Electronics Engineers, Inc., Computational Science and Engineering, Spring1995, Vol. 2, No. 1, pp. 58–68.

[86] Franklin Associates Ltd, Resource and environmental profile analysis of a manufactureredapparel product. Life cycle analysis (LCA): woman’s knit polyester blouse, Prepared forAmerican Fiber Manufacturers Association, downloaded from http://www.fibersource.com/f-tutor/LCA-Chapter %201.htm on 11.11.02.

[87] M. J. French, Conceptual Design for Engineers, London (etc.): Springer, 1999, ISBN 1-85233-027-9.

[88] C. Fröhlich, Observations of Irradiance Variations, Space Science Reviews, 94 (1), 15–24,November 2000.

References 165

[89] C. Fröhlich, Homogenised Solar Irradiance, downloaded from ftp://ftp.pmodwrc.ch/pub/data/irradiance/composite/DataPlots/comp_d30_60_0409_vg.pdf on 26.09.04.

[90] V. M. Fthenakis and P. D. Moskowitz, Thin-Film Photovoltaic Cells: Health and EnvironmentalIssues in Their Manufacture, Use and Disposal, Progress in Photovoltaics, 3 (5), 295–306,September–October 1995.

[91] P. Fullana, Life Cycle Assessment of a Mobile Phone, Presentation by Randa Group S.A. forTelefónica Servicios Móviles S.A., July 2000.

[92] C. Gemmer and M. Schubert, Solar Cell Performance Under Different Illumination Conditions,Proceedings of the 17th European Photovoltaic Solar Energy Conference and Exhibition,October 2001, Munich.

[93] German Federal Environment Agency report, Waste while standing by, downloaded fromstandby.lbl.gov/articles/Germany.html on 14.11.02.

[94] G. Giroult-Matlakowski, U. Theden and A. W. Blakers, Photovoltaic Market Analysis, WorldConference and Exhibition on PV Solar Energy Conversion, Vienna, 1998.

[95] K. K. Govil and K. K. Aggarwal, New method for maximum power consideration in solar cellswith and without resistive losses, Electronics Letters, 17 (1), 8 January 1981.

[96] E. L. Grant and R. S. Leavenworth, Statistical Quality Control, McGraw-Hill InternationalEditions, ISBN 0-07-024117-1.

[97] T. Grayson, Photovoltaic power pack for portable equipment, e.g radios – has connector extend-ing from rechargeable battery terminals, mounting hinges, clips or pouch and photovoltaic cellin series with diode, World Patent WO 95/12220, p. 11, lines 34–35.

[98] L. N. Green and E. Bonollo, The Development of a Suite of Design Methods Appropriate forTeaching Product Design, Global J. of Engng Educ., 6 (1), 2002.

[99] M. A. Green, Solar Cells – Operating Principles, Technology and System Applications, PrenticeHall, NJ, ISBN 0-13-82270.

[100] S. Guha, J. Yang, A. Banerjee and T. Glatfelter, Amorphous Silicon Alloy Solar Cells for SpaceApplications, Proceedings of the 2nd World Conference and Exhibition on Photovoltaic SolarEnergy Conversion, 1998, Vienna, Austria, pp. 3609–3613.

[101] R. Hahn and J. Müller, Future Power Supplies for Portable Electronics and Their Environmen-tal Issues, Proceedings of Electronics Goes Green 2000+, Vol. 1, Technical Lectures, 11–13September 2000, Berlin, Germany, pp. 727–734.

[102] T. L. Halaczek and H. D. Radecke, Batterien und Ladekonzepte: Lithium-lonen, Nickel–Cadmium, Nickel-Metall-Hybrid-Akkus, Power-Kondensator, Ladetechnik Herstellerübersicht,Franzis’ Verlag GmbH, ISBN 3-7723-4603-0.

[103] J. Harrison, Investigation of polycarbonate as a suitable ‘greenhouse’ material forthe solar cooker, Florida Solar Energy Center, November 2001, downloaded fromhttp://www.fsec.ucf.edu/Solar/PROJECTS/SolarCooker/cooker.htm on 15 August 2002.

[104] M. Hatcher, An eye for an eye, Opto and Laser Europe, Issue 78, October 2000.[105] P. J. M. Havinga and G. J. M. Smith, Design Techniques for Low-Power Systems, Journal of

Systems Architecture, 46 (1), 1 January 2000.[106] R. Hebner, J. Beno and A. Walls, Flywheel Batteries Come Around Again, Institute of Electrical

and Electronics Engineers, Inc., Spectrum, April 2002.[107] D. Hessman, Energy-wave vector graph for a direct band gap semiconductor, downloaded on

02.04.02 from www.ftf.lth.se/FTF/Person/HessmanD/Thesis/Introduction.htm.[108] R. Hill, Photovoltaic Market Analysis, Solar Cells and Their Applications, Chapter 19, p. 417,

ISBN 0-471-57420-1.[109] W. Hively, Reinventing the wheel (Jack Bitterly), Discover, August 1996.[110] R. G. Hockaday, Microfuel Cells (TM) for Portable Electronics, Manhattan Scientifics, Inc.,

presentation San Diego, 2000.[111] M.-O. Hongler,Modélisation des flux de production. Performances et pilotage des installations,

Cours d’Assemblage du Professeur Jacques Jacot, EPFL, Switzerland.[112] R. G. Hopkinson, Architectural Physics Lighting, Her Majesty’s Stationery Office, UK.[113] H. J. Hovel, Semiconductors and Semimetals, Vol. 11 – Solar Cells, Academic Press, Inc.,

ISBN 0-12-752111-9.[114] B. Iannotta, Batteries not included (Jack Bitterly), New Scientist, 11 January 1997.

166 References

[115] International Energy Agency, Renewables in global energy supply, an IEA fact sheet, page 1,downloaded from http://www.iea.org/leaflet.pdf on 14 July 2003.

[116] IEC Standard, IEC-904-3, International Electrotechnical Commission, Geneva, Switzerland.[117] International Energy Agency, Non-hydro renewable energy ‘will grow faster than any other pri-

mary energy source, at an average rate of 3.3% per year for the next 30 years’, http://www.re-focus.net/news/021002_1/.

[118] J. Jacot, Le management de la qualité, Chapter 3, Cours d’Industrialisation du ProfesseurJacques Jacot, EPFL, Switzerland.

[119] A. J. Jansen and A. L. N. Stevels, Human Power, a Sustainable Option for Electronics,ISEE-1999. Proceedings of the 1999 IEEE International symposium on electronics and theenvironment.

[120] M. Johnson, Percent of lit commercial floor space lit by incandescent, fluorescent and com-pact fluorescent lighting, US Energy Information Administration, 1995 Commercial EnergyBuildings Survey.

[121] S. Kalpakam, Finite Dam with Alternating Linear Input and Output, J. Math. Phys. Sci. 11 (5),443–460, 1977.

[122] H. Kameno, et al., Basic Design of 1 kW h class flywheel energy storage system, Proceedingsof the 8th International Symposium on Magnetic Bearings ISMB-8, 26–28 August 2002, Mito,Japan, pp. 575–580.

[123] N. H. Karam, J. H. Erner, R. R. King, M. Haddad, L. Cai, D. E. Joslin, D. D. Krut, M.Takahashi, J. W. Eldredge, W. Nishikawa, B. T. Cavicchi and D. R. Lillington, High EfficiencyGalnP2/GaAs/Ge Dual and Triple Junction Solar Cells for Space Applications, Proceedingsof the 2nd World Conference and Exhibition on Photovoltaic Solar Energy Conversion, 1998,Vienna, Austria.

[124] Kassensturtz of 8th October, SF1, Die Alte tuts noch lange, downloaded fromhttp://www.sfdrs.ch/system/frames/highlights/kassensturz/index.html on 10.10.02.

[125] K. R. Kefauver, G. C. Walsh, L. P. Hromada Jr, J. A. Kirk and R. B. Zmood, A 50 Wh OpenCore High-Speed Flywheel, Proceedings of 1999 Intersociety Energy Conversion EngineeringConference, downloaded from http://www.enme.umd.edu/ice_lab/papers/ps/iecec9∼1.PDF on22.10.02.

[126] S. Keller, M. Wagner, S. Scheibenstock, W. Jooss, A. Kress, G. Hahn, P. Fath and E. Bucher,Simple Mini Module Fabrication Schemes for High Voltage Silicon Solar Cells, 16th EPVSECE,1–5 May 2000.

[127] H. Keppner, J. Meier, P. Torres and D. Fischer, Microcrystalline Silicon and MicromorphTandem Solar Cells, Appl. Phys. A, 69, 169–177, 1999.

[128] A. King, Portable Power: Advanced Rechargeable Lithium Batteries, MRS Bulletin, 27 (8),August 2002.

[129] D. L. King, M. A. Quintana, J. A. Kratochvil, D. E. Ellibee and B. R. Hansen, PhotovoltaicModule Performance and Durability Following Long-Term Field Exposure, Prog. Photovolt.Res. Appl., 8, 241–256, 2000.

[130] Z. Kristl and A. Krainer, Energy Evaluation of Urban Structure and Dimensioning of BuildingSite Using Iso-shadow Method, Solar Energy, 70 (1), 23–34, 2001.

[131] M. A. Kroon, R. A. C. M. M. van Swaaij and J. W. Metselaar, Miniature High-Voltage a-Si:HSolar Cell for Integration in Microelectronic Devices, 16th EPVSECE, 1–5 May 2000.

[132] K. Lahiri, A. Raghunathan, S. Dey and D. Panigrahi, Battery-Driven System Design: A NewFrontier in Low Power Design, Proceedings of the ASP-DAC/Int. Conf. VLSI Design, January2002, Bangalore, pp. 261–267.

[133] A. Lal, Y. Bilbao, R. M. León, H. Guo, H. Li, S. Santanam, R. Yao, J. Blanchard and D.Henderson, A Nuclear Microbattery for MEMS Devices, Proceedings of the 9th InternationalConference on Nuclear Engineering, April 2001, nice.

[134] D. S. Landes, L’heure qu’il est – les horloges, la mesure du temps et la formation du monde,Editions Gallimard.

[135] R. Lankey and F. C. McMichael, Life-Cycle Methods for Comparing Primary and RechargeableBatteries, Environ. Sci. Technol., 34, 2299–2304, 2000.

References 167

[136] R. Lankey, Materials Management and Recycling for Nickel–cadmium batteries, PhD, Depart-ment of Civil and Environmental Engineering, Carnegie Mellon University, Pittsburgh, PA,15213.

[137] R. Lankey and F. C. McMichael, Life-Cycle Environment Comparison of Primary and Sec-ondary Batteries, Proceedings of the 1999 Institute of Electrical and Electronics Engineers,Inc., International Symposium on Electronics and the Environment.

[138] R. Latta, Percent of households using incandescent, fluorescent and compact fluorescentlighting 1993, US Energy Information Administration, 1993 Residential Energy ConsumptionSurvey.

[139] A. Luque, Coupling Light to Solar Cells, Advances in Solar Energy, Vol. 8, Chapter 14, editedby Morton Prince.

[140] T. L. Macdonald, Fuel cells for portable devices, EDN Power Supplement, September 2002.[141] P.-A. Magnin, Afin que ‘off’ soit vraiment ‘off’, Énergie Environnement – Magazine Officiel

d’Information de la Confédération et des Cantons Romands 9, Spring/Summer 2002.[142] A. Mahadavi, et al., A Two-way Interference Approach to Daylight Simulation, Journal of the

Illumination Engineering Society, 66–74, Winter 1993.[143] S. L. Mak, Where Are Construction Materials Headed, Building Innovation and Construction

Technology, 8, August 1999.[144] G. Marsh, RE storage – the missing link (data from Electricity Storage Association

http://www.electricitystorage.org/), Renewable Energy Focus, 38–41, March/April 2002.[145] P. Marwedel, Energy aware compilation, Course entitled ‘Advanced Digital System Design’,

Lausanne, Switzerland, October 2002.[146] A. Marsh and C. Raines (Square One research and the Welsh School of Architecture at

Cardiff University), A comparison of the spectral transmissions of a range of different glasstypes, downloaded on 22.01.03 from http://www.squ1.com/index.php?http://www.squ1.com/concepts/glass- properties.html.

[147] A. Macii, L. Benini and M. Poncino, Memory design techniques for low energy embeddedsystems, Kluwer Academic Publishers, ISBN 0-7923-7690-0.

[148] J. McGinnes, P. Corry and P. Proctor, Amorphous-Semiconductor Switching in Melanins,Science, 183 (4127), 853–855, 1974.

[149] T. J. McMahon, T. S. Basso and S. R. Rummel, Cell shunt resistance and photovoltaicmodule performance, Proceedings of 25th Institute of Electrical and Electronics Engineers,Inc., Photovoltaics Specialists Conference, 13–17 May 1996, Washington, DC, pp. 1291–1294.

[150] F. C. McMichael and C. Henderson, Recycling Batteries, Institute of Electrical and ElectronicsEngineers, Inc., Spectrum, February 1998, pp. 35–42.

[151] J. Merten and J. Andreu, Clear separation of seasonal effects on the performance of amorphoussilicon solar modules by outdoor I/V measurements, Solar Energy Materials and Solar Cells,52, 11–25, 1998.

[152] J. Merten, J. M. Asensi, C. Voz, A. V. Shah, R. Platz and J. Andreu, Improved EquivalentCircuit and Analytical Model for Amorphous Silicon Solar Cells and Modules, Institute ofElectrical and Electronics Engineers, Inc., Transactions on Electron Devices, 45 (2), February1998.

[153] J. Merten, et al., Variable Illumination Measurements of Microcrystalline Silicon Solar Cells,Proceedings of the 14th European Photovoltaic Solar Energy Conference and Exhibition,Barcelona, Spain, 1997, pp. 1424–1427.

[154] D. H. Meadows, et al., The Limits to Growth: A Report for the Club of Rome’s Project on thePredicament of Mankind, London: Potomac, ISBN 0-85644-008-6.

[155] MeteoSchweiz, Average global insulation data for Switzerland, available onhttp://www.meteosuisse.ch/fr/Climat/Climatactuel/IndexClimatactuel.shtml 11.03.03.

[156] Meteotest, Meteonorm 4.0 software, http://www.meteonorm.com.[157] E. L. Meyer and E. van Dyk, Monitoring ISC�VOC and the Performance Parameters of Photo-

Voltaic Modules, Proceedings of the 17th European Photovoltaic Solar Energy Conference andExhibition, October 2001, Munich.

[158] E. L. Meyer, The Effect of Low Irradiance and Cell Shunt Resistance on Photovoltaic ModulePerformance, Master of Science thesis, University of Port Elizabeth, South Africa, Chapter 3,pp. 15–30.

168 References

[159] S. Michaud, Lampe de poche solaire à LED, EPFL internal student project (Projet de semestre).[160] L. Michel, Méthode expérimentale d’évaluation des performances lumineuses de bâtiments,

EPFL thesis 2042, pp. 12–14.[161] P. E. Michelli, Medieval stained glass, downloaded from http://www.ariadne.org/studio/

michelli/sgmedieval.html on 10.11.02.[162] Microchip, nanoWatt Technology Products, from http://www.microchip.com/1010/suppdoc/

design/lowpwr/sub/index.htm on 05.05.03.[163] P. D. Moskowitz, N. Bernholc, V. M. Fthenakis, R. M. Pardi, H. Steinberger and W. Thumm,

Environmental, Health, and Safety Issues Related to the Production and Use of CadmiumTelluride Photovoltaic Modules, Advances in Solar Energy, Vol. 10, Chapter 4, pp. 211–246.

[164] H. J. Möller, Semiconductors for Solar Cells, Norwood, MA: Artech House, 1993, ISBN0-89006-574-8.

[165] C. Moreau, Pas de piles à combustibles dans les portables avant 2005, Industries et Techniques,(825), April 2001.

[166] Cl. Morel, Y. Golay and D. Lymberis, Lumière naturelle et energie – etudes typologiques,Final report of OFEN Project 53708 (Annexe 1) EPFL, CH-1015 Lausanne, Switzerland, 1994.

[167] S. Michaud, Conception pour l’environement – ecodesign d’un jouet, Cours Science-Technique-Société (STS) ‘Development durable’, Spring 2000.

[168] S. P. Morgan, Prediction of indoor wireless coverage by leaky coaxial cable using ray tracing,Institute of Electrical and Electronics Engineers, Inc., Trans. Vehicular Technology, 48 (6),2005, November 1999 (also in Proceedings of VTC’98).

[169] P. D. Moskowitz, An Overview of Environmental, Health and Safety Issues in the PhotovoltaicIndustry, Solar Cells and Their Applications (ed. L.D. Partain), New York: Wiley, 1995,Chapter 18, pp. 391–415.

[170] D. Moyer, Window and flat glass for historical archaeologists, downloaded fromhttp://www.digitalpresence.com/histarch/glass.html.

[171] T. M. McAloone, Industrial Application of Environmentally Conscious Design, London: Profes-sional Engineering Publishing, 2000, ISBN 1-86058-239-7.

[172] U. Muntwyler, et al., Report on the first phase of the IEA implementing agreement for hybridand electric vehicle technology and programmes – 1993–1999, Chapter 5: Batteries and Super-Conductors, available from www.ieahev.org/pdf/chapter5.pdf on 6.11.02.

[173] H. Nakajima, Solar Array Enrich Consumer Products, JEE, 26–29, June 1979.[174] NEC Laboratories, NEC uses carbon nanotubes to develop a tiny fuel cell for mobile applica-

tions, http://www.labs.nec.co.jp/Eng/Topics/data/r010830/ on 02.05.03.[175] NEC/Tokin, Double-layered capacitors ‘Supercapacitor’, Tokin product catalogue http://

www.nectokin.net/now/english/product/pdf_dl/SuperCapacitors.pdf.[176] Nedo, Spectra of daylight and spectral response of solar cells representing 5 PV technologies,

downloaded 02.04.02 from www.nedo.go.jp/…/eng/solar1/solar_cell/thin_film_cell.htm.[177] R. Nelson, Probability, Stochastic Processes and Queueing Theory: The Mathematics of Com-

puter Performance Modeling, New York: Springer-Verlag, 1995, ISBN 0-387-94452-4.[178] Unknown, The 2000 Nobel Prize for Chemistry, http://www.nobel.se/chemistry/laureates/2000/

public.html on 28.3.02.[179] G. A. Norris and P. Notten, Current Availability of LCI Databases in the World, downloaded

from http://www.sylvatica.com/Database%20summary%20May%202002.pdf 22.10.02.[180] P. Nostell, Preparation and optical characterisation of antireflection coatings and reflection

materials for solar energy systems, PhD Solid State Physics, Uppsala University, 2000, p. 10.[181] Official Journal L307, 08/11/1991 P. 0034 - 0036, Council Decision 91/565/EEC of 29 October

1991 concerning the promotion of energy efficiency in the Community (SAVE programme),downloaded from http://www.environment.fgov.be/Root/tasks/atmosphere/klim/pub/wet/eu/91-565-EEG_en.htm on 23.10.02.

[182] L. Oiva, A. Middendorf and K.-H. Zuber, Case Study on the Environmental Impacts of MobilePhone, Proceedings of Electronics Goes Green 2000+, Vol. 1, Technical Lectures, 11–13September 2000, Berlin, germany, pp. 749–752.

[183] H. Oman, Electrochemical Storage Device Configuration, MRS Bulletin, 24 (11), 33–39,November 1999.

References 169

[184] K. Omura, A. Hanahusa, T. Arita, H. Higuchi, T. Aramoto, T. Nishio, S. Sibutani, S. Kumazawaand M. Murozono, Recent Technical Advances in Thin-Film Cds/CdTe Solar Cells, Proceedingsof World Renewable Energy Conference, 1996, pp. 405–409.

[185] A. Oritz-Conde, et al., Exact Analytical Solutions of the Forward Non-ideal Diode Equationwith Series and Shunt Parasitic Resistance, Solid-State Electronics, 44, 1861–8164, 2000.

[186] A. Oritz-Conde, et al., Modeling Real Junctions by a Series Combination of Two Ideal Diodeswith Parallel Resistance and Its Parameter Extraction, Solid-State Electronics, 45, 223–228,2001.

[187] Osram, Lampes Fluorescentes 4.02, Programme d’éclairage 2000 Osram (commercialcatalogue).

[188] G. Pahl and W. Beitz, Engineering Design – A Systematic Approach, Berlin (etc.): Springer,1998, ISBN 3-540-19917-9.

[189] W. Palz, Solar Electricity, UNESCO, Butterworths, 1978.[190] D. Panigrahi, C. Chiasserini, S. Dey, R. Rao, A. Raghunathan and K. Lahiri, Battery Life

Estimation of Mobile Embedded Systems (rate-capacity effect), Proceedings of the InternationalConference on VLSI Design, January 2001, pp. 55–63.

[191] A. G. Pandolfo, A. M. Vassallo and G. L. Paul, Practical Aspects and Hurdles in the Devel-opment of Low-Cost High-Performance Supercapacitors, 7th International Seminar on DoubleLayer Capacitors and Similar Energy Storage Devices, 8–10 December 1997, Deerfield Beach,FL, 1997.

[192] D. H. Park and J. G. Zeikus, Electricity Generation in Microbial Fuel Cells Using Neutral Redas an Electronophore, Appl. Envir. Microbiol., 66, 1292–1297, 2000.

[193] B. Paule, Eléments d’éclairagisme, RAVEL, Office fédérale des questions conjoncturelles,Switzerland.

[194] B. Paule, Eclairage des bureaux, RAVEL, Office fédérale des questions conjoncturelles,Switzerland.

[195] PDE: Products Designed for the Environment, Mid-rate travel charger H H H uses less than0.12 watts of power in the standby mode, downloaded from http://www.motorola.com/EHS/environment/products/products.html on 23.10.02.

[196] N. Peguiron, Personal communication, Interview with Director of LOGMA SA.[197] J. Perlin, Evolution of PV Technology 1839–1973, downloaded from http://www.

siemenssolar.com/deutsch/09d.htm.[198] J. Perlin, From Space to Earth: The Story of Solar Electricity, Ann Arbor, MI: aatec Publica-

tions, ISBN 0-937948-14-4.[199] H. Petterson, Solar Cells in Electronic Consumer Products, Proceedings of Electronics

Goes Green 2000+, Vol. 1, Technical Lectures, 11–13 September 2000, Berlin, Germany,pp. 713–718.

[200] H. Pettersson and T. Gruszecki, Long-Term Stability of Low-Power Dye-Sensitised Solar CellsPrepared by Industrial Methods, Solar Energy Materials and Solar Cells, 70 (2), 203–212,December 2001.

[201] PPG Glass, History of glass, downloaded from http://www.ppg.com/gls_ppgglass/architect/history.htm on 20.11.02.

[202] 50% transmissivity of solar watch cache, Private conversation.[203] G. Prophet, Supercaps for Supercaches, EDN Europe, January 2003.[204] J. M. Rabaey and M. Pedram, Low Power Design Methodologies, Kluwer Academic Publishers,

ISBN 0-7923-9630-8.[205] B. Ram, A. Stevels, H. Griese, A. Middendorf, J. Müller, N. F. Nissen and H. Reichl, Envir-

onmental Performance of Mobile Products, ISEE – 1999. Proceedings of the 1999 IEEEInternational Symposium on electronics and the environment, 1999.

[206] J. F. Randall, C. Droz, M. Goetz, A. Shah and J. Jacot, Comparison of 6 photovoltaic materialsover 4 orders of magnitude of intensity, Proceedings of the 17th European Photovoltaic SolarEnergy Conference and Exhibition, October 2001, Munich.

[207] J. Randall, O. Amft and G. Tröster, Towards Lux Trace: Using Solar Cells to Measure DistanceIndoors, Proceedings of Location and Context awareness (LoCA 2005 springer LNCS) incooperation with Pervasive 2005, 12–13 May 2005, Oberpfaffenhofen (near Munich), Germany.

170 References

[208] G. Rebitzer, D. Hunkeler and O. Jolliet, Simplifying LCA: A Comparison of Methods for ProductDesign, Proceedings of the Society of Environmental Toxicology and Chemistry (SETAC)Europe Annual Meeting, 12–16 May 2002, Vienna.

[209] N. Reiter, Diode-based Energy Source? Scavenge Circuits! http://www.escribe.com/science/freenrg/m10443.html on 02.05.03.

[210] M. Reinhold, U. Hartmann, M. Badenius and H. Hauer, A Solar Powered Letter Scale with45microW Power Dissipation, Proceedings of the 1999 Institute of Electrical and ElectronicsEngineers, Inc., International ASIC/SOC Conference, September 1999, Washington, DC.

[211] I. Riedel, V. Dyakonov, J. Parisi, C. Brabec, N. S. Sariciftci and J. Hummelen, PolymerFullerene Networks for Photovoltaic Applications, Proceedings of the 17th European Photo-voltaic Solar Energy Conference and Exhibition, October 2001, Munich.

[212] J. Robertson and M. J. Rutter, Band Diagram of Diamond and Diamond-Like Carbon Surfaces,Diamond and Related Materials, 7 (2–5), 620–625, February 1998.

[213] W. Roth and J. Schmid, Photovoltaically Supplied Devices of Low and Intermediate PowerRange, Proceedings of the 8th EC Photovoltaic Solar Energy Conference, May 1998, Florence,pp. 263–269.

[214] W. Roth and A. Steinhüser, Photovoltaische Energieversorgung von Geräten undKleinsystemen, OTTI Technologie-Kolleg; Frauenhofer ISE.

[215] S. Roundy, Energy Scavenging for Wireless Sensor Nodes with a Focus on Vibration-to-Electricity Conversion, Presentation 19 February 2003 University of California, Berkeley, CA.

[216] J.-M. Rouiller, Labélisation de l’énergie électrique d’origine renouvelable, Forum-EPFL-Industrie, 3 October 2002.

[217] S. Roundy, Energy Scavenging for Wireless Sensor Nodes with a Focus on Vibration toElectricity Conversion, PhD thesis, University of California, Berkeley, CA.

[218] S. Roundy, P. K. Wright and K. S. J. Pister, Micro-electrostatic vibration-to-electricity con-verters, The American Society of Mechanical Engineers International Mechanical EngineeringCongress and Exposition, 17–22 November 2002, New Orleans, LA.

[219] S. Roundy, P. K. Wright and J. Rabaey, A Study of Low Level Vibrations as a Power Sourcefor Wireless Sensor Nodes, Computer Communications, Autumn 2002, in print.

[220] M. Rubin, K. Von Rottkay and R. Powles, Window Optics, Solar Energy, 62 (3), 149–161,1998.

[221] A. Rufer and P. Barrade, A Supercapacitor-Based Energy-Storage System for Elevators withSoft Commutated Interface, Institute of Electrical and Electronics Engineers, Inc., T. Ind. Appl.,38 (5), 1151–1159, September–October 2002.

[222] A.-Ch. Rufer and H. Ravokatrasolofo, Static Converter for Complementary Energy Storagewith Battery and Supercapacitor, Power Conversion, June 1999, Proceedings 161.

[223] D. Ruffieux, A miniature powering solar cell, Towards an autonomous millipede walkingmicrosystem. Thèses en microtechnique (Université de Neuchâtel) 1551, 2000.

[224] P. Ryser and J. Jacot, Gestion et évaluation de projets, Industrialisation course, Chapters 2 and3, given at EPFL.

[225] J. Sam, Constraint Consistency Techniques for Continuous Domains, Thesis 1423, EPFL,Lausanne, Switzerland.

[226] SANYO Energy (Europe) Corporate GmbH, Coin type rechargeable lithium battery (ML-series)where nominal voltage is 3V and (constant voltage) charge is 31±015V, Specification sheetsML614, ML621, ML1220, ML2016 and ML 2430.

[227] D. U. Sauera, M. Bächlerc, G. Boppa, W. Höheb, J. Mittermeib, P. Sprauc, B. Willerd andM. Wollnyd, Analysis of the Performance Parameters of Lead/acid Batteries in PhotovoltaicSystems, Journal of Power Sources, 64 (1–2), 197–201, January 1997.

[228] J.-L. Scartezzini, L. Michel, C. Roecker and R. Rhyner, Laboratoire de lumiere naturelle,Laboratoire de Lumiere Naturelle. Programme interdisciplinaire LUMEN. Projet OFEN. TheManual. EPFL, CH-1015 Lausanne, Switzerland, 1994.

[229] H. Schmidhuber and C. Hebling, First Experiences and Measurements with a Solar PoweredPersonal Digital Assistant (PDA), Proceedings of the 17th European Photovoltaic Solar EnergyConference and Exhibition, October 2001, Munich.

[230] L. Schmidt-Mende, et al., Self-organized Discotic Liquid Crystals for High-Efficiency OrganicPhotovoltaics, Science, 283, 1119–1122, 10 August 2001.

References 171

[231] S. Y. Seidel and T. S. Rappaport, Site Specific Propagation Prediction for Wireless In-buildingPersonal Communication System Design, Institute of Electrical and Electronics Engineers, Inc.,Transactions on Vehicular Technology, 43 (4), 879–891, November 1994.

[232] N. S. Shenck and J. A. Paradiso, Energy Scavenging with Shoe-mounted Piezoelectrics, Micro,Institute of Electrical and Electronics Engineers, Inc., 21 (3), 30–42, May/June 2001.

[233] H. Shirakawa, E. J. Lewis, A. G. MacDiarmid, C. K. Chiang and A. J. Heeger, Synthesisof Electrically Conducting Organic Polymers: Halogen Derivatives of Polyacetylene, �CH�x,J. Chem. Soc., Chem. Commun., 578–580, 1977.

[234] W. Siemens, On the electromotive action of illuminated selenium, discovered by Mr Fritz ofNew York, The Electrical World, 1885 and http://www.siemenssolar.com/deutsch/09c.htm on19.3.02.

[235] J. L. da Silva Jr, J. Shamberger, et al., Design Methodology for PicoRadio Networks, Proceed-ings of Design, Automation and Test in Europe Conference and Exhibition, 13–16 March 2001,Munich, Germany: ICM/Neue Messe, 2001, pp. 314–323.

[236] A. Slawsby, Taking Charge: Trends in Mobile Device Power Consumption, IDC bulletin 27514,June 2002.

[237] W. Smith, Effect of Light on Selenium during the Passage of an Electric Current, Nature, 20February 1873, p. 303.

[238] SolData Instruments, CIE (Commission International d’Eclairage) photopic, downloaded fromwww.soldata.dk/def-sid1.htm 18.08.02.

[239] D. J. Spiers and A. D. Rasinkoski, Predicting the Service Lifetime Of Lead/acid Batteries inPhotovoltaic Systems, Journal of Power Sources, 53, 245–253, 1995.

[240] D. L. Staebler and C. R. Wronski, Reversible Conductivity Changes in Discharge-ProducedAmorphous Si, Applied Physics Letters, 31 (4), 292–294, 15 August 1977.

[241] L. Steiner, Scellement, remplissage et assemblage des cellules solaires à base de colorant,EPFL internal student project (projet de semestre).

[242] G. Strobl, et al., High efficiency Si and GaAs solar cells for LILT applications, Photovoltaicgenerators, energy storage. Paris European Space Agency, 1995.

[243] R. W. Sturgis, Tort Cost Trends: An International Perspective, Tillinghast Corporation,Seattle, WA.

[244] S. M. Sze, Semiconductor Devices: Physics and Technology, New York: Wiley, 2001, ISBN0-471-33372-7.

[245] D. E. Tarrant and R. R. Gay, Commercialization of CIS-based thin-film PV, NREL/SR-520-28597 Phase 1 Annual Technical Report, August 1998–August 1999.

[246] A. N. Tiwari and H. Zogg, Des couches minces pour de l’électricité solaire, da Revue Poly-technique, No. 1635, March 2000.

[247] J.-F. Tricaud, Atlas énergétique du rayonnement solaire pour la France, Laboratoired’énergétique solaire du CNRS, Edition Pyc, ISBN 2-85330-033-1.

[248] TRIPS: Agreement on trade-related aspects of intellectual property rights, Article 33: Term ofProtection: ‘twenty years counted from the filing date’, PART II – Standards concerning theavailability, scope and use of intellectual property rights.

[249] O. Ulleberg, Stand-alone Power Systems for the Future: Optimal Design, Operation andControl of Solar-Hydrogen Energy Systems, PhD dissertation, Department of Thermal Energyand Hydropower, Norwegian University of Science and Technology, Trondheim.

[250] US Department of Energy National Center for Photovoltaics, PV roadmap: introducing the pho-tovoltaic industry roadmap H H H full speed ahead, http://www.nrel.gov/ncpv/intro_roadmap.htmlon 19.4.02.

[251] UNEP GRID Geneva from International Road Federation 1997, Number of motor vehicles,downloaded from http://www.unep.org/geo2000/english/i13.htm 09.10.02.

[252] Vanderbilt University news release posted 14 April 2001, Turning diamond film into solarcells, http://spaceflightnow.com/news/n0104/14diamondarrays/.

[253] N. J. C. M. van der Borg and E. J. Wiggelinkhuizen, Irradiation in the Built Environment,Proceedings of the 17th European Photovoltaic Solar Energy Conference and Exhibition,October 2001, Munich, VB2.44.

[254] VDI-Richtlinie 2221, Methodik zum Entwickeln und Konstruieren technischer Systeme undProdukte, Düsseldorf: VDI-Verlag, 1993.

172 References

[255] VDI Society for Development, VDI 2221:1993-05 systematic approach to the developmentand design of technical systems and products, Design and Marketing (VDI-EKV), Düsseldorf.Berlin: Beuth Verlag GmbH.

[256] N. Vela, F. Fabero, J. Cuenca and F. Chenlo, Accelerated Overcharge Test Procedure for aQuick Assessment of Lifetime of PV Batteries, Proceedings of the 16th European PhotovoltaicSolar Energy Conference and Exhibition, 1–5 May 2000, Glasgow.

[257] VHF Technologies, Solar Surfer Image, downloaded from http://www.flexcell.ch/products/helios.html on 030321.

[258] C. Vincent, Lithium Batteries, Institution of Electrical Engineers, UK, Review, March 1999.[259] W. Vouets, Nouvelles tendances pour la production d’énergie électrique et essor de la pro-

duction décentralisée, Forum – EPFL – Industrie, 3 October 2002.[260] G. J. Ward and F. M. Rubinstein, A New Technique for Computer Simulation of Illuminated

Spaces, Journal of the Illumination Engineering Society, Winter 1988, pp. 80–91.[261] G. J. Ward, F. M. Rubinstein and R. D. Clear, A Ray Tracing Solution for Diffuse Interreflection,

Computer Graphics, 22 (4), August 1988.[262] J. Watson, Light emission from solids – black body spectrum, downloaded from

http://vcs.abdn.ac.uk/ENGINEERING/lasers/solids.html on 10.11.02.[263] C. A. Webber and R. E. Brown, Savings potential of ENERGY STAR voluntary labeling pro-

grams, Proceedings of the 1998 American Council for an Energy Efficient Economy SummerStudy on Energy Efficiency in Buildings, LBNL-41972.

[264] Webster’s Dictionary, Ambient: (Am"bi*ent), n. Something that surrounds or invests; as, airH H H being a perpetual ambient. Sir H. Wotton, downloaded from http://www.bibliomania.com/2/3/257/119 2/21589/1/frameset.html on 19.04.03.

[265] J. Wijngaard, The effect of interstage buffer storage on the output of two unreliable productionunits in series, with different production rates, AIIE Transactions, 11 (1), 42–47, 1979.

[266] M. Williamson, Watts in Space, Institution of Electrical Engineers, UK, Review, January 1999,pp. 19–23.

[267] M. Wolf, Historical Development of Solar Cells, Proceedings of the 25th Power Sources Symp.,23–25 May 1972, pp. 120–124.

[268] S. Wyss, Comparaison des cellules ‘Grätzel’ avec les cellules photovoltaïque conventionelles,Cours Science-Technique-Société (STS) ‘Development durable’, Spring 2000.

[269] M. T. Xuan, Personal communication, Discussion with Dr Mai Tu Xuan, EPFL, Switzerland.[270] M. Yamaguchi and A. Luque, High Efficiency and High Concentration in Photovoltaics, Insti-

tute of Electrical and Electronics Engineers, Inc., Transactions on Electron Devices, 46 (10),October 1999.

[271] Yoram Kaufman, Solar radiation and climate experiment (SORCE), downloaded fromhttp://earthobservatory.nasa.gov/Library/SORCE/ on 10.11.02.

[272] R. Zallen, Course sources (solid state) pn junction, downloaded from www.phys.vt.edu/∼rzallen/ on 07.05.02.

[273] Z. Zhao, M. Vinson, T. Neumuller, J. E. McEntyre, F. Fortunato, A. T. Hunt and G. Ganguly,Transparent conducting ZnO:Al films via CCVD for amorphous silicon solar cells, Proceed-ings of 29th Institute of Electrical and Electronics Engineers, Inc., Photovoltaic SpecialistsConference, 20–24 May 2002, New Orleans, LA.

[274] L. Zubieta, R. Bonert and F. Dawson, Considerations in the design of energy storage systemsusing double-layer capacitors, IPEC 2000: Proceedings of International Power ElectronicsConference, 3–7 April 2000, Tokyo, Japan.

Contents

About the Author xiii

Preface xv

Acknowledgements xix

Introduction xxi

1 State of the Art 11.1 Introduction 11.2 Low-power Energy Sources 11.3 Intellectual Property Rights 3

1.3.1 Methodology 31.3.2 Results 31.3.3 Conclusion 5

1.4 IPV Taxonomies 51.4.1 Product Use Taxonomy 51.4.2 Function (or Circuit) Taxonomy 51.4.3 Radiant Energy Application Taxonomy 71.4.4 Mean Energy Taxonomy 10

1.5 IPV Gaps in Knowledge 101.5.1 Radiant Energy Available 101.5.2 PV Solar Cells 101.5.3 Charge Storage 111.5.4 Energy Source Guidelines 111.5.5 Applications 11

1.6 Conclusion 11

viii Contents

2 Engineering Design 132.1 Introduction 132.2 Defining Design 132.3 Trends in Engineering Design 162.4 Life Cycle Methods 18

2.4.1 Introduction 182.4.2 Definition 182.4.3 LCA for IPV Designers 192.4.4 LCA in IPV Design 202.4.5 Designer Responsibility 222.4.6 Summary 22

2.5 Conclusion 22

3 Radiant Energy Indoors 233.1 Introduction 233.2 Physics of Buildings 24

3.2.1 Radiant Energy 243.2.2 Radiant Energy Spectra 283.2.3 Basic Optical Parameters 31

3.3 Photometric Characterisation 333.3.1 Characterisation Methodology Issues 343.3.2 Daylight Factor 343.3.3 Nearby Obstacle Aspect Ratio 363.3.4 Glazing/Floor Ratio 363.3.5 Lighting Recommendations 37

3.4 Radiometric Characterisations 373.4.1 Obstacles 373.4.2 Window Transmission 403.4.3 IPV Location 443.4.4 IPV Cell Orientation 483.4.5 Further Profiles: Buildings, Users and Applications 49

3.5 Computer Simulation 503.6 Discussion 50

3.6.1 Summary of Parameters 513.6.2 IPV Designer Recommendations 51

3.7 Conclusion 523.8 Future Work 523.9 Further Reading 52

4 Fundamentals of Solar Cells 534.1 Introduction 534.2 Brief History of Solar Collectors and PV 54

4.2.1 The ‘Selenium Years’ 544.2.2 The ‘Silicon Years’ 544.2.3 History of IPV 564.2.4 IPV Today 57

Contents ix

4.3 Photonic Semiconductors 594.3.1 What is a Semiconductor? 594.3.2 Photonic Semiconductor Properties 604.3.3 Solar Cell Categories 62

4.4 Photovoltaic Technology 644.4.1 Electrical Efficiency Calculation 644.4.2 Fill Factor 644.4.3 Short-circuit Current 664.4.4 Open-circuit Voltage 674.4.5 Power Curve 68

4.5 Suboptimal Solar Cell Efficiency 684.5.1 Optical Issues 694.5.2 Material Quality Issues 704.5.3 Parasitic Resistance 704.5.4 Efficiency Losses Summary 71

4.6 IPV Material Technologies 724.6.1 Spectral Response 724.6.2 Amorphous Silicon Thin Film 734.6.3 Polycrystalline Thin Film 744.6.4 Conventional Silicon Cell 744.6.5 Other PV Technologies 754.6.6 Thin-film Modules 75

4.7 Efficiency Improvements 764.7.1 Current and ISC 774.7.2 Voltage and VOC 774.7.3 Fill Factor 774.7.4 Concentration 78

4.8 Conclusion 784.9 Further Reading 78

5 Solar Cells for Indoor Use 795.1 Introduction 795.2 Technology Performance at Indoor Light Levels 80

5.2.1 Experimental Procedure 805.2.2 Results 825.2.3 Efficiency with Intensity 835.2.4 Spectral Response 84

5.3 Indoor Light Level Model Presentation 855.3.1 Phenomenological Model 855.3.2 Heuristic Model 895.3.3 Technology-specific Models 90

5.4 Discussion 905.5 Designing PV Modules for Indoor Use 91

5.5.1 Transparent Conductive Oxide 915.5.2 Dimensioning of the Module 92

x Contents

5.6 Conclusion 935.7 Further Work 93

6 Indoor Ambient Energy Charge Storage 956.1 Introduction 956.2 Trends in Charge Storage 96

6.2.1 Electrical Storage Parameters 976.2.2 The Market 986.2.3 The Waste 98

6.3 Charge Storage Technology 996.3.1 Introduction 996.3.2 Technology Characterisation 996.3.3 Mechanical Storage Technology 1006.3.4 Electrochemical Storage Technology 1016.3.5 Batteries 1026.3.6 Double-layer Capacitors (DLCs) 1036.3.7 Fuel Cells 105

6.4 Charge Storage Parameters 1056.5 Determining the Storage Capacity 107

6.5.1 Relatively Stable Factors 1086.5.2 Non-deterministic Parameters 1096.5.3 Heuristic of Total Storage Capacity 114

6.6 Electrochemical Storage Technologies 1156.6.1 Double-layer Capacitors (DLCs) 1156.6.2 Secondary Batteries (Accumulators) 115

6.7 Conclusion 1196.8 Future Work 1196.9 Further Reading 120

7 Ambient Energy Power Source Design 1217.1 Introduction 121

7.1.1 Chapter Goals 1217.1.2 Chapter Structure 1217.1.3 Guideline Caveats 122

7.2 Clarification 1227.2.1 Realistic Functions 1237.2.2 Realistic Use Pattern 1257.2.3 Prior Art 126

7.3 Conceptual Design 1267.3.1 Energy-efficient Design and Responsiveness 1267.3.2 Maximising Available Energy 1327.3.3 Variation in Design 1337.3.4 Charge Storage Components 1347.3.5 Functionality 136

7.4 Embodiment 1367.4.1 PV Module Appearance 136

Contents xi

7.4.2 Product Housing 1387.4.3 Cost 1397.4.4 Production Quality 140

7.5 Detailed Design 1417.5.1 Alternative Energy Source 1417.5.2 Information on Ideal Use 142

7.6 Case Studies 1437.6.1 Clarification 1437.6.2 Conceptual Design 1457.6.3 Embodiment 1457.6.4 Detailed Design 146

7.7 Conclusion 1467.8 Further Reading 146

Conclusion 147


Glossary 157

References 161

Index 173

Documents

the-eye.eu · 1 State of the Art 1.1 INTRODUCTION This chapter is structured intwo parts. Firstly, the available knowledge with regard to low-powerenergysourcesisexamined(section1.2).Thisreview