escuela tÉcnica superior de ingenierÍa y sistemas...

UNIVERSIDAD POLITÉCNICA DE MADRID

ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA Y

SISTEMAS DE TELECOMUNICACIÓN

CHARACTERISATION OF THE OPERATION &

MAINTENANCE PHASE IN PV RURAL

ELECTRIFICATION PROGRAMMES

THESIS

AUTHOR: LUIS MIGUEL CARRASCO MORENO

DIRECTOR: LUIS NARVARTE FERNÁNDEZ MADRID, MAY 2015

Dedicado a mis padres,

mi hermana,

mi abuela,

a toda mi familia,

mis amigos

y cómo no, a Adelita.

iii

ACKNOWLEDGEMENTS

"[...] el olmo ya seco de la ermita debe su único verdor a la hiedra que le abraza, pero ella a su vez sólo gracias al viejo tronco

logra crecer hacia el sol."

José Luis Sampedro (La Sonrisa Etrusca)

Escribió Galdós que la experiencia es una llama que no alumbra sino quemando. Creo que en

mi vida me he chamuscado bastante, pero no lo he hecho solo y por eso tengo que agradecer a

muchas personas todo lo que de ellas he aprendido trabajando codo con codo hasta llegar aquí,

empezando por Luis Narvarte, mi tutor y director de tesis, alma mater de este trabajo, excelente

persona y amigo, quien me animó a emprenderme en esto de investigar y quien siempre ha estado

disponible para escuchar, pensar y resolver. A Eduardo Lorenzo, por su experta mirada desde lo alto

que tanto ha servido para enderezar mis torcidos renglones. A todos mis compañer@s del grupo de

sistemas fotovoltaicos del IES, que forman entre tod@s el más cordial ambiente de camaradería de

trabajo que he conocido. A tod@s mis compañer@s de la extinta Isofoton en España y Marruecos

con los que trabajé y aprendí muchas más cosas además de fotovoltaica. Y a much@s más, que

aunque no mencionados, fueron fuente de iluminación.

Agradezco a Isofoton Maroc s.a.r.l. por su colaboración al poner los enormes cimientos en los

que se ha basado el trabajo experimental de esta tesis.

Un último reconocimiento a la Universidad Politécnica de Madrid por su ayuda al financiar

parte de los estudios llevados a cabo en esta tesis con el proyecto 35_FOTOVOLT perteneciente a la

XI Convocatoria de Acciones de Cooperación Universitaria para el Desarrollo.

v

ABSTRACT

With 1,300 million people worldwide deprived of access to electricity (mostly in rural environments),

photovoltaic solar energy has proven to be a cost‐effective solution and the only hope for electrifying

the most remote inhabitants of the planet, where conventional electric grids do not reach because

they are unaffordable. Almost all countries in the world have had some kind of rural photovoltaic

electrification programme during the past 40 years, mainly the poorer countries, where through

different organizational models, millions of solar home systems (small photovoltaic systems for

domestic use) have been installed. During this long period, many barriers have been overcome, such

as quality enhancement, cost reduction, the optimization of designing and sizing, financial

availability, etc. Thanks to this, decentralized rural electrification has recently experienced a change

of scale characterized by new programmes with thousands of solar home systems and long

maintenance periods. Many of these large programmes are being developed with limited success, as

they have generally been based on assumptions that do not correspond to reality, compromising the

economic return that allows long term activity. In this scenario a new challenge emerges, which

approaches the sustainability of large programmes. It is argued that the main cause of unprofitability

is the unexpected high cost of the operation and maintenance of the solar systems. In fact, the lack

of a paradigm in decentralized rural services has led to many private companies to carry out

decentralized electrification programmes blindly. Issues such as the operation and maintenance cost

structure or the reliability of the solar home system components have still not been characterized.

This situation does not allow optimized maintenance structure to be designed to assure the

sustainability and profitability of the operation and maintenance service.

This PhD thesis aims to respond to these needs. Several studies have been carried out based on a real

and large photovoltaic rural electrification programme carried out in Morocco with more than 13,000

solar home systems. An in‐depth reliability assessment has been made from a 5‐year maintenance

database with more than 80,000 maintenance inputs. The results have allowed us to establish the

real reliability functions, the failure rate and the main time to failure of the main components of the

system, reporting these findings for the first time in the field of rural electrification.

Both in‐field experiments on the capacity degradation of batteries and power degradation of

photovoltaic modules have been carried out. During the experiments both samples of batteries and

modules were operating under real conditions integrated into the solar home systems of the

Moroccan programme. In the case of the batteries, the results have enabled us to obtain a proposal

of definition of death of batteries in rural electrification.

A cost assessment of the Moroccan experience based on a 5‐year accounting database has been

carried out to characterize the cost structure of the programme. The results have allowed the major

costs of the photovoltaic electrification to be defined. The overall cost ratio per installed system has

been calculated together with the necessary fees that users would have to pay to make the

operation and maintenance affordable.

Finally, a mathematical optimization model has been proposed to design maintenance structures

based on the previous study results. The tool has been applied to the Moroccan programme with the

aim of validating the model.

vii

ACRONYMS

AECID: Agencia Españolda de Cooperación para el Desarrollo

AEG: Allgemeine Elektrizitäts Gesellschaft

CC: Charge Controller

CFL: Compact Fluorescent Lamp

CM: Corrective Maintenance

DOD: Depth Of Discharge

EC: European Communities

ECU: European Currency Unit

EDP: Energy Demonstration Programme

ESCO: Energy Service Company

EVA: Ethylene‐Vinyl‐Acetate

GEF: Global Environmental Facility

HW: Hardware

IEA: International Energy Agency

IEC: International Electrotechnical Commission

IES‐UPM: Instituto de Energía Solar ‐ Universidad Politécnica de Madrid

LC: Low power Consumption light lamps

LED: Light‐Emitting Diode

LEDC: Less Economically Developed Countries

MAD: ISO code for the Moroccan currency (dirham)

MDG: Millennium Development Goals

MNRE: Ministry of New and Renewable Energy of India

MPPT: Maximum Power Point Tracker

MTTF: Mean Time To Failure

NGO: Non‐Governmental‐Organizations

O&M: Operation and Maintenance

OEI: Organización de Estados Iberoamericanos

ONEE: Office National de l'Electricité et l'Eau (Morocco)

OW: Orgware

pdf: probability density distribution

PERG: Programme d'Electrification Rurale Globale (Morocco)

PLANER: Plan Nacional de Electrificación Rural (Spain)

PM: Preventive Maintenance

viii

PPER: Programme Pilote d'Electrification Rurale (Morocco)

ppp: public‐private‐partnership

PV: Photovoltaic

PVPS‐IEA: Photovoltaic Power Systems Programme ‐ IEA

PVRE: Photovoltaic Rural Electrification

PWM: Pulse‐Width Modulation (charge controller)

REA: Rural Electrification Administration

REDP: Renewable Energy Development Project

SE4ALL: Sustainable Energy for All

SGA: Société Générale Agricole

SHS: Solar Home Systems

SLI: Start‐Lighting‐Ignition (Battery)

SOC: State Of Charge

Solar‐PERG: Photovoltaic PERG programme

Solar‐PERGISO: Solar‐PERG carried out by the private company ISOFOTON

SW: Software

UN: United Nations

UNDP: United Nations Development Programme

USAID: United States Agency for International Development

UTSfSHS: Universal Technical Standard for Solar Home Systems

VAT: Value Added Tax

VRLA: Valve‐Regulated Lead‐Acid (Battery)

WB: World Bank

Wp: Watt peak

ix

SUMMARY

1 INTRODUCTION ......................................................................................................... 2

1.1 THE GLOBAL ACCESS TO ELECTRICITY ................................................................................... 3

1.2 THE ORIGINS OF RURAL ELECTRIFICATION .......................................................................... 10

1.3 REVIEW OF THE DEVELOPMENT OF THE PHOTOVOLTAIC RURAL ELECTRIFICATION .......... 17

1.4 OBJECTIVES OF THE THESIS ................................................................................................. 35

1.5 METHODOLOGY OF THE WORK ........................................................................................... 35

1.6 THESIS STRUCTURE ............................................................................................................. 35

2 THE MOROCCAN PV RURAL ELECTRIFICATION PROGRAMME .................................. 38

2.1 INTRODUCTION ................................................................................................................... 38

2.2 THE PERG PROGRAMME ..................................................................................................... 38

2.3 THE SOLAR‐PERG ORIGIN, DEVELOPMENT AND FEATURES ................................................ 40

2.4 THE ISOFOTON‐PERG PROGRAMME ................................................................................... 43

2.5 SOME COMMENTS ABOUT THE SOLAR PERG DEVELOPMENT ............................................ 50

2.6 THE ISOFOTON‐PERG DATABASE ........................................................................................ 50

3 RELIABILITY ASSESSMENT OF SHS COMPONENTS .................................................... 54

3.1 INTRODUCTION ................................................................................................................... 54

3.2 RELIABILITY ANALYSIS ......................................................................................................... 54

3.3 ANALYSIS OF THE RESULTS .................................................................................................. 60

3.4 APPLICATION EXAMPLE ...................................................................................................... 65

3.5 CONCLUSIONS ..................................................................................................................... 66

4 IN‐THE‐FIELD ASSESSMENT OF BATTERIES AND PV MODULE RELIABILITY IN THE PERG

PROGRAMME ................................................................................................................. 70

4.1 INTRODUCTION ................................................................................................................... 70

4.2 IN‐FIELD BATTERY TESTING ................................................................................................. 71

4.3 IN‐THE‐FIELD PV‐MODULE TESTING .................................................................................... 81

4.4 CONCLUSIONS ..................................................................................................................... 84

5 CHARACTERIZATION OF THE OPERATIONAL & MAINTENANCE COSTS ...................... 88

5.1 INTRODUCTION ................................................................................................................... 88

5.2 COST ANALYSIS .................................................................................................................... 88

5.3 SENSITIVITY ANALYSIS ......................................................................................................... 94

5.4 INFLUENCE OF THE SHS SPATIAL DENSITY ........................................................................... 96

5.5 APPLICATION EXAMPLE ...................................................................................................... 97

5.6 CONCLUSIONS ..................................................................................................................... 99

6 DESIGN OF DECENTRALIZED MAINTENANCE STRUCTURES IN PHOTOVOLTAIC RURAL

ELECTRIFICATION ...........................................................................................................102

x

6.1 INTRODUCTION ................................................................................................................. 102

6.2 BASELINE DATA ................................................................................................................. 103

6.3 METHODOLOGY ................................................................................................................ 103

6.4 MODEL APPLICATION ........................................................................................................ 111

6.5 CONCLUSIONS ................................................................................................................... 115

7 CONCLUSIONS AND FUTURE RESEARCH ..................................................................118

7.1 CONCLUSIONS ................................................................................................................... 118

7.2 FUTURE LINES OF RESEARCH ............................................................................................. 121

PUBLICATIONS GENERATED DURING THIS PHD ..............................................................124

BIBLIOGRAPHY ..............................................................................................................128

CHAPTER 1

INTRODUCTION

2

1 INTRODUCTION

Beyond the reasons that justify the right of every human to have access to modern sources of

energy, the importance of electricity as energy vector, from the first application of the late

nineteenth century to today, lies in the fact that it is easy to transport and simple to operate.

Nowadays there are still 1,300 million people deprived of electricity, 85% of them in remote rural

areas where electrification encounters problems such as high economic investments, low

profitability or difficulty of operation, among others. In these cases, decentralised electrification by

means of solar home systems (SHS) has aimed to be a technical and cost‐effective solution for over

40 years in many countries of the world. Currently, large‐scale electrification programmes with

thousand of SHSs are established in remote and impoverished regions, whose results, in terms of

sustainability, are in doubt. These attempts at electrification are frequently based on assumptions,

such as electricity consumption, device reliability, operating costs, rural spending habits, etc, which

bear little resemblance to reality. The consequences are the long term economic instability of the

programmes, the failure of private operators and the abandonment of SHSs, which has happened in

many initiatives developed in recent decades.

This work presents a study based on a real and large photovoltaic rural electrification (PVRE)

programme, taking advantage of the excellent opportunity that the author took advantage of

whilst, for five years, being part of the management team of the company that operated that

programme, having full access to the detailed maintenance data, failure of the SHS components,

unit costs, management structure, activity organization, etc, during that period. The study provides

the chance, for the first time, to contrast the real data of decentralised electrification with the

classic assumptions, by means of the SHS's reliability statistic research, the characterization of the

actual costs in the operation and maintenance (O&M) phase and the study of the application of the

results in the formulation of PVRE programmes.

This chapter introduces the detailed historical evolution of rural electrification, in general and the

photovoltaic rural technology, in particular, which nowadays has culminated in the implementation

of large PVRE programmes. First, it focuses on the problem of access to electricity and discusses the

difficulties that it faces. Then, a review of the rural electrification origins throughout the 20th

century is presented to show that barriers and solutions at the beginning of rural electrification are

similar to the current challenges. Finally, an historical review of photovoltaic rural technology

evolution shows that the three dimensions that integrate it (hardware, software and orgware) have

unequally evolved to the present day, which gives rise to a still non‐mature technology.

The chapter concludes with the main objectives of the thesis, a brief explanation of the

methodology of the work and the description of the document structure.

Chapter 1: Introduction

3

1.1 THEGLOBALACCESSTOELECTRICITY

1.1.1 Currentstatus:1,300millionpeoplewithoutelectricity

Nowadays, the lack of access to electricity affects to 1,300 million people worldwide, 20% of the

world’s population. This figure, published by the International Energy Agency (IEA) in 2012 ‐ World

Energy Outlook Report, [1] ‐ gives us an overall idea of a problem to be solved globally, similar to

other issues such as hunger, access to clean water, sanitation, etc. According to the IEA, this figure

has decreased since 1990, from 2,000 million people, to 1,300 million in 2010. Not only that, in just

in 8 years (2002 ‐ 2010), it has been reduced from 1,623 million to 1,267, a gap of new 356 million

people with access to electricity (more than the population of the United States of America).

However, these figures are just estimations (as recognized by the IEA), since the lack of access to

electricity is something specific to the marginal and rural areas of the less economically developed

countries (LEDC), where the inaccuracy of the population census, also affected by the double and

opposing effect of population growth and migration to urban areas, precludes any accurate

estimations [2].

1.1.2 TheIEAexpectsthatuniversalaccesstoelectricitywillbeachievedinpartwithSolarHomeSystems

From the perspective of reducing the world population without access to electricity, in 2010 the

United Nations (UN) launched the Sustainable Energy for All (SE4ALL) initiative to "achieve universal

energy access, improve energy efficiency, and increase the use of renewable energy" [3] (It must be

remembered that the UN for many years did not include action on energy poverty in the

Millennium Development Goals).

The 2011 World Energy Outlook report [4] published by the IEA estimated the necessary

investment for electricity universal access, between 2010 and 2030, at US$ 640 billion (this

requirement is small when compared to overall energy‐related infrastructure investment,

equivalent to around 3% of the total). The report suggests that 70% of the required infrastructure

would consist of off‐grid systems: mini‐grids (65% of this share) and stand‐alone off‐grid solutions

(the remaining 35%), that is, solar home systems (SHS), small hydro systems, and others (wind and

biogas). We estimate that nowadays, the SHSs represent 95% of the stand alone system installed

worldwide. So, the IEA foresees an investment of around US$ 150 billion for SHSs to reach universal

access to electricity before 2030. If photovoltaic (PV) systems were sized to meet housing

consumption between 250 and 500 kWh/year [5], the required SHS power would be 180 ‐ 365

watts peak (Wp). Taking into account a unit cost for the installed SHSs of between US$ 6 ‐ 8 /Wp1, it

would correspond to installing more than 50 million SHSs, giving access to electricity to 250 million

people.

1.1.3 Whythelackofelectricityisaproblem

Access to electricity is not considered a universal fundamental right of people [6]. However, there is

a unanimous opinion that electrical supply is a priority factor which is urgent to resolve. Therefore,

in the last decade there have been numerous initiatives to address the problem, such us the Global

1 It includes equipment, transportation, installation of the SHSs and 10% of overhead expenses.

4

Environmental Facility ‐ GEF [7], the Millennium Development Goals [8] and the Sustainable Energy

for ALL (SE4ALL) from the UN [3], "Luz para Todos" in Brazil [9], Power for all in India [10] or "Luces

para aprender" from the Organization of Ibero‐American States (OEI) [11], among others.

Apart from the extended consideration in Western culture that human enrichment as a society, in

economic, social, political and cultural aspects, is necessarily linked to infrastructure development

[12, 13], perhaps, the best arguments that justify the need for access to electricity are included in

the Millennium Development Goals (MDG). In this resolution, adopted by the UN in 2000, although

there are no specific MDGs relating to energy, it has been recognized that MDGs cannot be met

without affordable, accessible and reliable energy services (Table 1):

Table 1: Importance of electrical access for achieving the MDGs

Goal 1:

Eradicate extreme poverty

Access to modern electricity increases household incomes through economic development and reduces the burden of time‐consuming domestic labour. Electricity supply enables poor households to engage in activities that generate income by providing lighting that extends the working day and by powering machines that increase output.

Goals 2 and 3:

Achieve universal primary education and promote gender equality and empowerment of women

For poor people everywhere, access to electricity frees time for education; time that would otherwise be spent collecting traditional fuels, fetching water, processing food or in other physical work. Access to electricity contributes to the empowerment of women. Increasing access to energy brings major benefits for women and girls; in health, education, and productive activities.

Goals 4, 5 and 6:

Reduce child and maternal mortality and reduce diseases

Electricity helps improve health by powering equipment for pumping and treating water; it enable health clinics to refrigerate vaccines, operate and sterilize medical equipment, and provide lighting. It allows the use of modern tools of mass communication needed to fight the spread of HIV/AIDS and other preventable diseases. Access to electricity helps attract and retain health and social workers in rural areas by improving living conditions.

Goal 7:

Ensure environmental sustainability

Energy use and production affect in local, regional and global environments. The environmental damage and its harmful effects can be reduced by increasing energy efficiency, introducing modern technologies for energy production and using renewable energy.

Goal 8:

Develop a global partnership for development

The World Summit for Sustainable Development called for partnerships between public entities, development agencies, civil society and the private sector to support sustainable development, including the delivery of affordable, reliable and environmentally sustainable energy services.

However, these arguments, usually very common in the literature, should be treated with caution,

since, very often, the availability of electricity does not directly involve any development [14, 15].


5

An example that refutes this attribution is the Moroccan Global Rural Electrification Programme

(PERG in French acronym), which will be widely discussed in this work. This electrification

programme focused on providing electricity mainly to housing and not to farmlands, which are the

places where electrification could have some impact on the development of the local economy. The

case of Tizi n'Ait Amer, a small village of just 700 inhabitants in the south of Morocco, is illustrative.

It got access to electricity 10 years ago and every dwelling is connected to the grid. However, no

new economical activities have been developed since the electrification of the village. The only

hope of carrying out new activities has been the extension of agricultural lands, which has recently

become possible thanks to the installation of a photovoltaic water pumping system to irrigate the

new crops, as the wells are 400 meters from the village and the grid does not reach it2. This

example shows that giving access to dwellings is not enough for economical development. Rural

electrification must be more ambitious if new economical activities are to be implemented.

From a different point of view, most modern societies are economically based on the so‐called

"consumer economy", thus it is not surprising that private corporations, financial institutions and

public administrations are interested in the extension of the economy to the rural population,

focusing in the fact that access to electricity contributes to the acceleration of that process

(consider, however, the existing criticisms of the dominant current model of economic growth, but

this subject is far from the arguments addressed in this thesis).

Beyond corporate or market interests, the extension of the access to electricity is currently in the

hands of the people themselves, that even knowing about the electricity, they still live without it

and therefore they demand it. To a greater or lesser extent, modern standards of living have spread

to the most remote areas of the planet, and so electric lighting, television and mobile phones are

currently perceived as basic needs in the rural areas of impoverished countries. The introduction of

these everyday uses requires the availability of electricity. It can be said that after more than a

century of electrification, the current demand for electricity is global.

1.1.4 Blockingfactors

Despite the efforts made to enhance the conditions for people in rural environments, the fact is

that the access to electricity rates are still very low in some regions of the planet (Sub‐Saharan

Africa and South Asia constitutes 95% of the world population without access to electricity).

The evolution of the rate of access to electricity is affected by several factors:

a) Positive factors that increase the electrification rate:

‐ Migration from rural to urban areas

‐ Rural electrification

‐ Maturity, quality and cost reduction of new technologies

b) Negative factors that reduce the electrification rate:

‐ The high birth rates in rural areas of impoverished countries

2 Own sources. The PV pump installed in Tizi n'Ait Amer belongs to a project financed by the Spanish International Cooperation (AECID) and the Universidad Politécnica de Madrid (UPM)

6

‐ The increased costs of conventional technologies

These factors, which could be quantified, depend on other more unpredictable and difficult

weighting factors, such as political will, armed conflict, famine, natural disasters, etc.

Considering the last 3 decades (1980 ‐ 2010), an analysis of the factors involved in global access to

electricity could be carried out just by assigning an indicator to each factor (Table 2).

Table 2: Factors that quantitatively affect the evolution of global access to electricity

Factor Indicator 1980 2010

Migration from rural to urban areas

Rural population (nº of people living in rural environment)

2,675,822,000 (61% of the population)

3,320,679,000 (48% of the population)

Birth World population (nº of people)

4,413,536,000 6,861,918,000 (increased by

55%) Rural electrification programmes

People without access to electricity

2,000,000,000 (45%)

1,300,000,000 (20%)

Maturity and reduced costs of new technologies

Photovoltaic systems costs ($/Wp of the photovoltaic module)

$12 $0.8 (93% reduction)

Increased costs of conventional technologies

Crude oil prices (US$/ barrel) [16]

Jan. 1970 (Before 1970s oil crisis)

US$ 21.00

July 2010 US$ 82.25

1.1.4.1 Theincreaseintheruralpopulation

On the one hand, in spite of the strong migration impact towards the cities (in 2007 there was the

historical phenomenon that, for the first time, the world population changed from mainly rural to

urban), the high global birth rate has meant that in 3 decades the world’s rural population has

increased by 25% (more than 600 million people).

On the other hand, although the rural electrification programmes have contributed to increasing

the rate of access to electricity, it is not known precisely what was this rate in the 1980s, but it can

be estimated that the overall number of people without access to electricity remained constant

during that decade at 2,000 million, which means 45% of the population [17]. If the figure was

reduced to 1,300 million in 2010, it means that the rate of access to electricity is still higher than

the growth rate of the rural population, which is a very encouraging fact (see Figure 1) on the

evolution of the global access to electricity, especially in Asia, where the ratio of people without

power is declining rapidly (China gave access to electricity to more than 700 million people

between 1980 and 2000 [18], and the country's electrification rate currently exceeds 99% [5]). Sub‐

Saharan Africa, however, remains as the only region of the world where the number of people

without access to electricity is increasing.


7

Figure 1: Global evolution of population, rural population and lack of access to electricity until 2013. World Bank [19]

1.1.4.2 Conventionalelectrificationisbecomingmoreexpensive

The current high costs of conventional rural electrification systems are affected, among other

factors, by the increased prices of fossil fuels. For example, in US, the cost of electricity for

residential use has doubled in three decades. In Europe, between 2002 and 2013, the cost of

electricity for households has gone up by 61%.3

Among the less electrified regions of the world, Sub‐Saharan Africa has the most expensive

electricity tariff in the world, on average between US$ 0.13 ‐ $0.14/kWh (in comparison, electricity

tariffs in Latin America, Eastern Europe and East Asia are around US$ 0.08/kWh.) [5], which lie well

below the true cost of production, which on average is US$ 0.18/kWh, preventing any return in

capital, thus threatening the long‐term sustainability of the utilities in the region [20].

If, in addition, we consider the investment needed to provide access to electricity to rural

communities, it should be noted that the infrastructure costs for conventional electrification

(extensions of electricity grids mainly through the medium and low voltage lines) has increased

considerably. These lines use raw materials such as iron and copper, whose market prices have

increased 2 and 5 fold respectively from 1980 to now [21]. The average cost of a medium voltage

line is around €6,000/km (case of 11 kV; cost of medium voltage transformers or operation and

maintenance not included [22, 23, 24, 25]) and its impact on the energy costs can be estimated at

€2.5c/kWh/km [26].

At the same time, during the last 40 years, the silicon flat‐plate photovoltaic industry (that

represents more than 90% of the global photovoltaic market) has reduced its costs by 93%, so in

the sunniest countries, such as the Mediterranean area, or most of the African continent, it is now

3 Note, however, that the integration of renewable energy sources into the European energy mix has also affected the increase in tariffs.

‐

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

1950 1960 1970 1980 1990 2000 2010 2020

Million of peo

ple World

Rural WorldWithout electricity access

7,125

3,336

1,285

8

feasible to produce solar electricity at a cost around4 €8c/kWh , which could lead to the possibility

of a medium term change of paradigm.

On other matters, from 1980 to 2010, electrical power consumption in the world has increased by

more than 150%, at a rate nearly 5% per year (Figure 2). However, the human population has

grown at a rate of 1.8% per year, which indicates that electricity consumption per capita has

increased almost 3 fold in 30 years. Although the development of the industry carries a lot of

weight in these results, it is also obvious that home electricity consumption is growing. This fact

suggests that providing access to electricity leads not only to an increase in the power required to

meet the new connections, but also that this power needs to be gradually increased according to

trends in household consumption.

Figure 2: Evolution of the World’s electricity capacity, generation, consumption and losses [27]

Within the great figures of the world electricity generation, it is worth mentioning the importance

of the distribution energy losses (see Figure 2), which represent 8 ‐ 9% of the electricity generated

every year worldwide. This means annual losses of 1,800∙106 MWh, enough to supply electricity to

a country of more than 300 million inhabitants with the European standard consumption of

electricity (5.4 MWh/person/year [28]). Not only that. Taking into account the minimum electrical

consumption to guarantee basic life conditions (1 MWh/person/year [29]), the figure would

become 1,800 million people; and considering the average consumption per capita in Africa (0.5

MWh/person/year), this figure would rise to 3,600 million people, almost 3 times the world’s

population without access to electricity.

4 It concerns large photovoltaic (PV) power plants. Taking into account a power degradation rate of 1%/year for PV modules and a lifetime of 25 years, 1 kWp PV power could produce around 44,500 kWh for 25 years (solar radiation = 5.5 kWh/m2/day). At current PV power plant investment prices (€1.5 /Wp), a performance ratio PR = 0.75 and O&M costs corresponding to 3% yearly of the investment cost, the produced solar energy

cost would be €7.8c/kWh.

0

1,000

2,000

3,000

4,000

5,000

6,000

0

5,000

10,000

15,000

20,000

25,000

1975 1980 1985 1990 1995 2000 2005 2010 2015

Total Electricity Net Consumption (Billion Kilowatthours)

Electricity Distribution Losses (Billion Kilowatthours)

Total Electricity Installed Capacity (Million Kilowatts)

109 kWh 106 kW


9

1.1.4.3 Politicalandsocialfactors

In the current global energy scenario, with a declining growth rate of the world’s rural population

and viable alternatives to conventional electrification, we can estimate that technical and

economical aspects are not the only cause impeding access to electricity. The development of rural

electrification also depends on other factors such as political will, social acceptance, subsidies and

agricultural development policies, among others.

It is socially accepted that renewable energies, especially photovoltaic technology, are "expensive"

and have low reliability compared to conventional technologies, so they would require a great deal

of investment for implementation and the power supply could not be guaranteed. But, in 2013

subsidies to conventional energies, such as petroleum, reached US$550 billion all around the world,

4 times higher than the amount dedicated to renewable energies [30]; or, in the same context, as

regards rural electrification, the World Bank (WB) argues that subsidies for grid electrification are

significantly greater than those for off‐grid electrification [31].

As regards rural development in impoverished countries, the lack of structure in the agricultural

sector also contributes to impeding access to electricity, since the agricultural policies require

investments in infrastructures to be made in the agricultural economy and dignify the peasant's

lives. Thus, it is very unlikely that a country without agricultural policies will be able to allow the

rural population to get access to electricity. As will be set out below, rural electrification in almost

all Western countries in the mid‐20st century was developed in parallel with agriculture with the

aim of modernizing the countryside and increasing agricultural production ratios.

Finally, it must be taken into account that rural electrification especially addresses a particular part

of society, the peasants. They have been historically constituted as an independent economy

characterized by the fact that the peasantry has always supported itself. The peasant community is

the most aware class with regard to its economy, which determines the decisions that they take

daily. The difference between a peasant and other society member is that the former knows

perfectly what he obtains from his work: he produces what he needs to live and the rest of the

production can be a surplus value when sold on. On the other hand, a worker from the "standard"

society never knows the real value of the product of his work. Thus, is important to realize that

giving access to electricity to rural people means an incursion from the macro‐economy into the

peasant economy, with all the difficulties involved (resistance to change). For example, rural

inhabitants from countries like Morocco are not familiarized with public services such as electricity,

and it is difficult to admit concepts like the payment of monthly fees or the long term contracts, or

contractual rights and obligations.

To better understand the phenomenon of contrast in the development of rural electrification,

which prevails both in the effort to electrify and the problem of electrification, there is nothing

better than referring back at the origins of rural electrification in the Western countries carried out

during the twentieth century.

10

1.2 THEORIGINSOFRURALELECTRIFICATION

1.2.1 Theappealofelectricity

1.2.1.1 Startofthemarketingofelectricity.The1881ParisExposition

Electricity and its applications have fascinated humans since the beginning of the industry over 130

years ago. When today a peasant family without access to electricity in an impoverished country

finally gets access to it, the ability to marvel at the optimum quality of electric lighting in addition to

the possibility of using appliances like TV or mobile phone must be very similar to that experienced

by our ancestors in the late nineteenth century.

It may be argued that the commercial inception of the electrical industry began with the

International Exhibition in Paris in 1881, exclusively dedicated to electricity, that brought together

many of the inventors and industrialists from the emerging sector at the time to exhibit their

creations and show them to the world (Figure 3). It was the closest thing to what we now

understand as an industrial exhibition. It was attended by over 600,000 visitors and had over a

thousand exhibitors (including Thomas A. Edison, Joseph W. Swan, Zénobe T. Gramme, A. Graham

Bell, William Thomson, etc), 19 of whom came from Spain [32].

Figure 3: Overview of the International Exhibition of Electricity, Paris 1881 (appeared in Nature, 1881 second quarter)

Inside the Palais de l'Industrie, which then occupied the place where now stands the Grand Palais

des Champs Elisées, the latter built to host 1900 Universal Exhibition, all kinds of inventions for

electric power generation, transmission and application were exhibited, from a lighthouse, boats

and even an airship driven by electric motors, submarine cables, telegraphy apparatuses,

electrochemical batteries, electric stoves, large magneto‐electric machines, microphones, trams,

etc.


11

But surely, there were two applications which caused more excitement: the phone and lighting. On

the latter, the Spanish magazine "La Ilustración Española y Americana" published in reference to

the Paris Exhibition the following [32]:

"On the bottom left, there are all the known generators: steam, gas, or by means of batteries. Further, a series of powerful gas engines or steam, which set in motion the dynamo‐magneto‐electric machines of Gramme, Lontin, Siemens and Meritens, which send torrents of electricity to the lamps of various systems, which shine splendidly inside the Palace with the most brilliant clarity that human industry has ever produced and with the astonished gaze of man has ever seen."

Thus, electric lighting was the first application of electricity that amazed humanity and became the

engine of development and expansion, thus making other means of artificial lighting practically

inconceivable.

1.2.1.2 Firstpublicsupplyofelectricityinaruralsetting:Godalming1881

Coinciding with the 1881 Exhibition in Paris, and one year before that the famous electric power

plant of Pearl Street in New York (September, 1882) was inaugurated, it took place in September,

1881 in Godalming (England) the first experience in the rural supply of electricity on record, built to

provide street lighting for the town, and replacing the existing gas‐lighting system. In the last

quarter of the nineteenth century, electricity was perceived by society within the realm of the

"scientific". The fact that it was applied in a small town of only 2,000 inhabitants caused a huge

interest around the country. The power generation genius system and the welcome given by not

only the local and surrounding population, but also by the press, because of the good quality of

lighting [33], started the paradigm of what electricity would mean for humanity throughout the

coming century. However, the enormous expectation of the pioneering system and its initial

success had to deal with its technical immaturity and despite the enthusiasm of its promoters, the

private company of electricians Calder & Barnet, eventually abandoned its contract with the Town

of Godalming, which in turn was taken over by Siemens and after numerous problems, causing

continuous and repeated outages, Godalming went back to gas lighting only 2 and a half years after

the start of the new experience. Electricity would come back to Godalming in 1904 and this time

would be forever.

1.2.1.3 Theurbandevelopmentoftheelectrification

Electrification applied to lighting was really confined to the big cities, whose beginnings were

marked by the fierce competition against gas‐lighting, but the rapid popularity of electricity and its

great reception brought about its rapid expansion.

The first urban experiences of using electricity did not go beyond being mere exhibitions. For

example the lighting of Puerta del Sol in Madrid in 1875, or in 1878, to mark the engagement

between King Alfonso XII and his cousin Maria de las Mercedes (who was only 17 years old. She

would die of typhus just five months later, giving rise to the famous legend of the love between

them and the traditional songs that have survived in popular heritage), or other more extravagant

events, like the first night bullfight with not very good results in 1879, which "La Ilustración

Española y Americana" would outline [32]:

"If the shadows of our grandparents hold bullfight functions in the Otherworld, they should be

very similar, because what we saw was a show of silhouettes."

12

In the late nineteenth century, European cities were equipped with a gas lighting service operated

by private companies. The pioneers of electrification were also private companies, and after

electricity superseded gas lighting, many gas companies turned to electricity. Thus emerged a

network of companies that obtained concessions (from municipalities) to illuminate streets or even

whole neighbourhoods. The companies employed steam engines and alternators installed

wherever they could (rented basements, cellars, etc) to power the street lights. Very soon,

theatres, cafes, public buildings, and later dwellings, would also be electrified which led to complex

commercial competition between the numerous electric companies (Figure 4), generating a price

war in order to win customers. Electrical distribution was born, therefore, as a totally private and

decentralized system.

Figure 4: Electricity sales advertisement appeared in an early 20th century newspaper from Barcelona

1.2.1.4 Theworld'slargestindustryemerges:theelectricalindustry

The development of the electricity supply industry was possible thanks to private equity, closely

linked to the European industry. In the case of Spain, the first electric company, also founded in

1881, was the "Sociedad Española de Electricidad", with a company's share capital of 20 million

pesetas, and created by D. Tomás Dalmau, who owned an "optics and physics" shop in Barcelona,

and who had previously introduced the Gramme machine in Spain in 1873, which subsequently

obtained a license for manufacturing.

The "Sociedad Española de Electricidad" installed a multitude of electrical supply equipment for

public and interior lighting in many cities in Spain, especially Barcelona and its surroundings, even

overseas (Cuba and the Philippines) and navy warships. The representative of the company in

Madrid, who was also a partner, the engineer and inventor Artilleryman Colonel Isodoro Cabanyes,

had already equipped his atelier with electricity in 1881 for lighting and motive power. He was

responsible for many of the first electrical project demonstration in Spain. It is worth mentioning

that Cabanyes would work some years later on the use of solar energy for decentralized rural

applications in the field of agricultural irrigation, firstly through a "solar reflector system" (Figure 5)

and afterward with the "solar air engine" [34].

The company was taken over in 1894 by the German company Allgemeine Elektrizitäts Gesellschaft

(AEG) who founded the "Compañía Barcelonesa de Electricidad" in the same year [35, 36, 37, 38,

39].


13

Figure 5: Cabanyes's solar reflector. It appeared in 1890 in the magazine La Gaceta Industrial [34]

Electricity generation, initially produced by means of the steam engine, made the leap to

hydroelectricity, which meant a reduction in the costs of production and consequently electricity

tariffs, initiating the development of large electrical distribution networks.

This new situation led to the need to make major investments in the construction of dams and

reservoirs, artificial waterfalls, high voltage distribution lines, etc. However, the enormous

investments necessary could not be covered by the limited national electric companies, nor even

the public administration, so since the very early days, the electricity industry in Spain, which in the

1930s was the most important in terms of investment, exceeding that of the rail and mining

industries, needed the intervention of international investment holdings to meet the costs of the

rapid development of the electricity sector. In the early 1930s all European utilities were already in

the hands of roughly 20 companies, thus shaping what would later become the paradigm of

centralized electrification [36].

1.2.2 Thebeginningsofruralelectrificationanditsproblemofprofitability

After the introduction of use of electricity in the cities, the Spanish countryside showed little

interest in the new technology. However, the public administration considered electricity as the

panacea for the 3 major rural problems of the Spanish post‐civil war years [40]: unemployment,

poverty and the consequent rural exodus. However, access to electricity in the countryside had to

face two major obstacles: "the enormous cost of setting up the transmission and distribution of

electricity" and the lack of interest of the rural population towards technological innovation. The

first problem was solved through subsidies and as regards the latter, Luis González Abela in his

book "La Electrificación Rural, Problema Nacional " published in 1942 described the problem in thus

[40]:

"... there is only one way to overcome it, which is a very active advertising through pamphlets, daily and technical press, radio, cinema and whatever means possible, which will highlight the transcendental benefits that would result giving access to electricity to our honoured peasants, because there is no reason for them to be second‐class citizens and because they did not commit any offense in having born in the countryside ... "

14

An example of these transcendental benefits was cited in the Congress of Rural Electrification in

1948, held at the School of Industrial Engineers of Madrid [41], in which the importance of using

radio receivers for the Spanish peasant was mentioned:

"... [the peasant] isolation is broken in this way. He belongs to the great human family. He can cultivate his Spirit, increase his knowledge, participate in the national life and enjoy the artistic beauties of music whenever he wants. Not enough can ever be said about the benefits of radio in the life of an isolated peasant. "

Much less documented than urban electrification, rural electrification was carried out in parallel

with the urban, but with a different approach and significant limitations. On the one hand, the

existence of small waterfalls that were used in the flour mills, saw mills, foundries, etc, were

exploited by means of small generators (dynamos) to provide electricity to small towns. Again, the

origin of the electrification system, like the urban one, was absolutely decentralized. From that

mentioned at the 1948 Congress of Rural Electrification, the following is extracted [41]:

"The typical electric mill that is used in many towns and all of its electrical industry is known; it is a completely logical solution, which adequately meets the needs of these people. It is enough to have a small water flow, provided by any ravine that goes to a canal that carries the water to a small pond. At the foot of it, a turbine with a dynamo and engine is installed, achieving a power of 5 to 20 CV; the latter serves to supply electricity to several towns. During the daytime it works as mill, and at night, the dynamo supplies lighting to the town. This is the reality for a large area of the country, and as long as the Spanish countryside does not change its habits, what nowadays seems to be difficult, the National power distribution networks and the rural electrification will be superfluous."

They were small companies including municipalities and agricultural cooperatives which were

commissioned in the early decades to deal with these matters. The technical and productive

limitations of the electrical rural generators, the distribution losses (voltage drops) and the gradual

increase in loads (users added more lighting points, or appliances every year), caused the electric

service to be of very poor quality, with frequent power outages and failures of the generator or

even in the distribution network. From the aforementioned 1948 publication, the following was

cited [41]:

"... the technical solution for creating small local power plants or, at most, at a regional level, installed in waterfalls that are built ad hoc or even using already existing mill and sawmill facilities, is usually not effective, unless, even within modesty of the installations, their energy power far exceeds that required for the loads."

At the time, the notion of critical mass of users that would allow to a company to manage an

electrical network with an economic return was already mentioned [41]:

"... the towns where electrical lighting has not yet arrived, not only will not give profits but losses, even if the facilities were freely outsourced to the nearest distributor, as the expected revenue would be 75‐100 pesetas per month on average at current tariffs, because towns have between 15 and 40 neighbours, most of them with poor access, and therefore the operation of electrical services is very expensive."

"The solution must be sought permitting the rural distributors to apply an adapted tariff throughout its region. In this way, while tariffs remain moderate for the entire electricity rates, the distributors can increase it to get the real rural electrification in the area that they


15

manage. So these new rates shall be applied to the rural market in which villages with up to 2,000 subscribers must be included. "

Another singularity of the rural electrification, which directly affects the problem of the

profitability should also be noted: the collection of the user fees. Given that the peasant and his

family spend most of the daylight hours in agricultural activities, it is most likely that collectors,

when they visit their customers will not find anyone at home, so the already high cost of moving

around remote regions is increased as they have to return repeatedly. In this regard, another

extract from the 1948 Congress is shown [41]:

"Collection of receipts.‐ currently, they are charged at home, which is very expensive because the collector does not always find all the neighbours at home, so he is bound to make several trips, and very likely he may not be able to complete the collection."

As a result of this historical evidence, it can be argued that some of the problems that rural

electrification had to face in the first half of the twentieth century were based on the lack of

profitability for the utilities, due to the high costs of infrastructure (network extensions), no return

on investment (very low consumption of electricity) and insurmountable operation and

maintenance tasks (remote and dispersed customers and difficulty in managing users' fee

collection). As will be seen below, these problems have remained to date.

1.2.3 Ruralelectrificationtomodernizeagriculture

In 1932, during the Second Spanish Republic, the Instituto de Ingenieros Civiles (now known as the

Instituto de la Ingeniería de España [42]), organized a series of conferences on rural electrification

dedicated to electrical energy applied to agriculture, where in a somewhat visionary way it

addressed the tilling of the land by means of electric machines, in addition to the "electroculture of

crops" (direct application of electricity to the crops to influence their development). The focus of

the conferences was the French experience, which had already almost 40,000 electrified towns and

used the "electric‐tiller" (Figure 6) for agricultural work in France, its Protectorates and Colonies

[43]:

"... the Gas Lebón Company, in Algeria [...] had decided to give a subsidy of 300,000 Francs to private farmers and agricultural cooperatives that purchased electric‐tillers of more than 100 H.P."

Figure 6: Electric‐tiller with cable winch, owned by the Société Générale Agricole (SGA). Photo from [43]

16

It is known that later, during the second half of the twentieth century, the engine of development

of rural electrification were policies focused on agricultural modernization, carried out in the

European post‐war as a means of activating the European economy.

"... to turn electrification into a profitable activity, it must cover electric‐tilling, harvesting, threshing and other available operations using electric motors ..." [41]

Thus, the idea was to extend the grids, at the time fed by large hydraulic and thermal power plants,

toward farms with the aim of increasing crop yields through the use of new electrical equipment.

However, the private companies, which had flourished within urban electrification, did not perceive

the same business opportunity in rural electrification that had it had seen in the cities, for the

aforementioned reasons.

1.2.4 Publicsubsidiesforruralelectrification

Then government intervention was required through incentives for both the utilities and the rural

users in order to make the rural electrification attractive to them. In most Western countries rural

electrification was achieved through grants and loans provided to the electricity companies to

ensure a return on the investment, and carrying out awareness campaigns addressed to the rural

population to ensure a minimal electric power consumption.

For example, the US created a rural electrification agency (the Rural Electrification Administration ‐

REA) with the aim of funding the utilities that were electrifying the rural areas [44, 45]. In the

1930s, the US administration launched a promotional campaign aimed at encouraging the peasants

to use electricity (at the time they were reluctant to pay for an electric service that never had

needed before) for different domestic appliances and machinery for agriculture and livestock farm

work (Figure 7).

Figure 7: Two of the advertisements that the REA agency used for electrification promotion in the 1930s to increase awareness among the rural population on the benefits of electricity.


17

Thanks to this campaign, an electrification rate close to 100% in US was attempted in few decades,

which contributed to popularizing the use of domestic appliances, such as television, oven, iron,

bread machine, vacuum cleaner, etc, which would later be exported all over the world. It had the

same impact on agriculture, and the consequent employment of sophisticated electrical power

tools.

1.2.4.1 Publicsubsidies:TheSpanishPLANER

In 1974, in Spain, more than 900,000 rural people still lacked access to the public service electricity

lines (over 6% of rural population). Giving access to electricity to that remote population meant a

huge investment and negative profitability because of the wide dispersion and low purchasing

power of the population. The 1973 National Electrical census indicated that while the density of

subscribers in urban areas was 116.68 per km2, in rural areas it was 11.42 per km2. Moreover, while

the mean urban consumption was 6,244 kWh/year (per dwelling), the rural rate was 885 kWh/year,

i.e. the rural household consumption was 7 times lower than the urban one and the dispersion of

the dwellings was 10 times higher, what meant that the rural electrification costs were 70 times

higher than the urban costs [46].

Although most of the electricity companies in Spain were private, the Spanish government

launched the rural electrification plan, PLANER in Spanish acronym, [47] with the aim of providing

access to the non‐electrified rural population, upgrading rural power grids and contributing to the

increase in agricultural and rural electricity consumption. The programme was carried out between

1976 and 1989. Just from 1982 to 1989 [48], the amount of these subsidies reached 32 billion

pesetas (more than €700 million at current rates [49]).

In parallel to the modernization and extension of the conventional power grids,the first experiences

in decentralized electrification was carried out in the 1980s by means of renewable energies,

promoted by the National Institute for Reform and Development (IRYDA in Spanish acronym) within

the PLANER programme. Around 3 million ECU (European Currency Unit) were dedicated between

1982 and 1985 (€4.6 million at current rates, applying inflation rate) to install more than 2,200

photovoltaic systems [50] in dwellings from decentralized areas.

1.3 REVIEW OF THE DEVELOPMENT OF THE PHOTOVOLTAIC RURALELECTRIFICATION

1.3.1 Introduction

During the second half of the nineteenth century, the rising cost of coal led to the exploration of

other alternatives to replace the coal in industrial applications where thermal processes intervene.

That was how the French professor M. Augustin Mouchot developed his solar thermal system, later

perfected by the engineer Frank Shuman in US in the early twentieth century [51] (see Figure 8).

After the First World War, oil prices dropped dramatically, putting an end to the new global energy

paradigm based on this fossil fuel while technological initiatives based on solar energy were

abandoned.

18

Figure 8: Left: 1878 Universal exhibition in Paris. First parabolic trough solar collector developed by Mouchot in 1866; right: First solar‐generating plant set up in 1913 in Egypt at Maadi by Frank Shuman

The use of solar energy was absolutely forgotten for 6 decades until the 1970s, when the oil crises

of 1973 and 1979 shook the entire energy sector. Then the emerging photovoltaic technology, at

the time restricted to aerospace since in the 50s, Bell laboratories in US developed the first

photovoltaic cells, making the jump to terrestrial applications. This coincided with the first steps in

the manufacture of silicon cells at a much lower cost than existed to date (in 1971, the price of

silicon photovoltaic cells for the aerospace industry was $100/Wp [51]).

Since then, the use of photovoltaics was conceived as a possible solution to electrification in

remote areas. On the one hand, the solar resource is available, to a greater or lesser extent,

everywhere in the World and on the other hand, the photovoltaic module is an element of high

reliability and long life, which makes it ideal for use in isolated areas.

Despite these two great qualities, there have been other factors that have played against the

supposed "idealism" of the photovoltaic technology, such as high costs or low reliability of the

other system components. These negative factors have been evolving during the 40 years of PV

history thanks to the efforts of industry, researchers, installers and especially the users, who

throughout the world have been the great laboratory of the decentralized PV electrification.

1.3.2 TheSolarHomeSysteminPhotovoltaicRuralElectrification

Although the global PV market is currently shared by around 99% dedicated to the grid‐connection

and only 1% (see Figure 9) to off‐grid applications, the use of PV technology in stand‐alone systems

was, until 2000, the most extended application, mainly to provide electricity (lighting and small

appliances) to rural homes through the so‐called solar home systems (SHS). The PV rural

electrification is currently growing annually at a rate greater than 20% [52]. For example, the off‐

grid PV systems power installed in 2013 may have been more5 than 600 MW (with 500 MW

installed in China alone) [53].

5 The author have not found any source reporting reliable data about global off‐grid PV markets


19

Figure 9: Evolution of the off‐grid and grid‐connected global market. The worldwide cumulated PV installed power at the end of 2014 was 177 GWp

The solar home system has been the most used concept for mass electrification of houses in

remote areas, versus the centralized PV systems (pure or hybrid power‐plants) or commonly so‐

called mini‐grids (Figure 10).

Figure 10: Left: Village electrified by SHSs; Right: PV off‐grid power plant (both in Morocco)

The idea in favour of SHS argues that PV users invariably consume more electricity when they are

not personally responsible for the system. This concern is linked to the capacity and size of the

systems, to which the operation and maintenance factor could be added. The management of

collective structures (need of local organizations, agreements, etc) seems to be more difficult than

individual systems. However, SHS has also been imposed versus the mini‐grids for the following

reasons:

0

20

40

60

80

100

120

140

160

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

% Grid‐Connected % Off‐Grid Cumuled GWp GWp

20

Standardization. The same design can be used in different homes or applications of similar ranks, which makes it easier for engineers, developers and installers;

Geographical spread. SHS can be applied in both dense and sparse populations. Mini‐grids are justified only in geographically dense villages;

Local availability of spare parts. SHS components are more standardized than those of mini‐grid power plants, so it is easier to find spare parts locally in countries where PVRE is developed, such as electrochemical batteries, regulators or light bulbs adapted to the SHSs.

An SHS is typically made up of (Figure 11) a small PV generator (35 – 100 Wp), a charge controller,

an electrochemical lead‐acid battery, several lamps and DC plugs to connect low loads, such as TVs,

radios or mobile phone chargers. These systems are usually set up to a 12 Vdc output [54].

Figure 11: Left: Solar Home System electric scheme, right: PV module of the SHS on the dwelling roof

Even though photovoltaic technology applied to rural electrification has reached a solid maturity

after 40 years of development, it still faces several problems, some of which are dealt with in this

thesis. These problems involve not only the body of the technology itself, the SHS (what we are

going to call hardware), but they mainly affect the management of decentralized services in rural

electrification (known as orgware).

To understand this issue we consider the two approaches presented below:

1.3.3 SHS:electrificationsystemordomesticappliance

Taking into account the millions of SHSs that are installed in the world, it can be said that they

consist of a standardized assembly of basic components (generator, charge controller, battery and

loads). The user, in accordance with his economic resources, can purchase an SHS, and even install

it himself in exchange for an equipment warranty. This is something very similar to buying a

domestic appliance.

To refer to an SHS as an electrification facility, similar to the conventional power grid, it must satisfy

certain requirements, which make it equivalent to the electric power grid.

The electrical service from the conventional power grids is managed by large companies that

ensure the supply through a strong system of generation, transmission, distribution and O&M. The

resources of these companies range from sophisticated media and control management to

departments with specialized technical staff, mobility and transport capabilities, intervention


21

protocols, etc. A similar deployment of resources is used for commercial issues, for example to

ensure the collection of fees to end users by means of precise energy meters, switches to which

only the companies can access, direct debit payments, billing departments, etc.

In PVRE, it is difficult to obtain these sophisticated, large and effective management tools, perhaps

due to the limited size of most of the PVRE programs, when compared with the grid, which does

not apparently justify the necessary investment. While, in general, some PVRE programs

demonstrate meticulous care in terms of the quality of the devices, they pay little attention to the

management mechanisms that must ensure the operation and maintenance of the SHSs. So it can

guarantee the quality of the PV system but not its sustainability.

As a response to this problem, many electrification experiences have considered PVRE as

something further from a service notion and closer to a domestic appliance. Thus, the figure of the

service manager is replaced by the figure of the sales and guarantee manager. This model is a copy

of the common domestic appliance market, which has the peculiarity that it has been

institutionalized within the rural electrification field.

As an example for purposes of illustration, PVRE can be compared to bicycle hire services that exist

in many European cities. The purpose of this service is to provide mobility to citizens by means of

bicycles. The bikes are apparently similar to those that we have at home, but they have certain

special features, such as the automatic identification codes for tracking, parking anchorage devices,

etc, which make them different and adapted to a management system. The user rides the bike just

like a normal one, but in parallel to a registration system, subscriptions, card payments, etc. Behind

it there is a complex (and usually expensive) management system that allows the concessionaire to

carry out the O&M of bicycles and renting facilities, and to collect the leasing charges with

guarantees (obviously the correct use of bicycles and the collection of fees is not left to the good

faith of users).

To date it has been usual in PVRE for, even in programmes configured as electric service, the SHS to

be set up in a similar way to the bicycle that we have at home, in accordance with the

aforementioned example. Thus, the O&M managers of these systems do not have any tool to

manage the service offered to their customers and there is no choice but to trust in the honesty of

thousands of SHS's users.

The result of this fact is the well‐known dilemma about whether an SHS is a domestic appliance or,

on the other hand, an electrification system comparable to the conventional one [55]. If the

tendency is to achieve the universality of the access to electricity rights, the SHS cannot be a simple

appliance purchased by the user from any dealer. If the SHS is a real electric supply system, its set

up cannot be simplified to the minimum required components, and in the same way as the public

service of bicycle renting, it will need hardware (the SHS) adapted to the management system

(orgware) to provide the necessary tools to administrate the O&M and allow the user to benefit

from a service with the same guarantees given by conventional electrification.

1.3.4 PVREastechnologicalsystem

As regards the photovoltaic rural technology, understood as a system [56], from a holistic point of

view it consists of three dimensions (Figure 12):

22

The hardware (HW), that refers to the system material body: the SHS, its components, quality,

lifetime, reliability, cost, etc.

The software (SW) is about the use of the system by the user: the consumption, the time of use of

each appliance, the signals of the charge controller and reaction of the user, etc.

The orgware (OW) is the organization model of the rural electrification programme, which provides

the electricity service to the dwellings. In this regard it is taken into account on the one hand,

whether the programme is developed through subsidies, credits, cash sales or a fee for service,

among others. On the other hand, the orgware dimension deals with programme management,

from marketing and installation of the SHSs, to the "after sales" service and the operation and

maintenance.

Figure 12: Hardware, software and orgware interactions in the photovoltaic rural technology system

This scheme, proposed and analyzed for technological systems by the Ukrainian Gennady M.

Dobrov [57] in the late 1970s, has certain peculiarities concerning the 3‐dimension interaction. One

of them is that, traditionally in technical innovation, more attention has been paid (and more

resources dedicated) to the HW and SW than the OW. This negatively affects the technological

system’s sustainability. The orgware, defined by Dobrov as "a set of organizational arrangements

specially designed and integrated using human, institutional and technical factors to support the

appropriate interaction of the technology and the external systems", plays a key role in photovoltaic

rural technology, which has been underestimated throughout PVRE history and currently still

suffers significant deficiencies.

The element that perhaps has evolved most in the PV rural system has been the hardware, both in

the quality of the SHS devices, and adaptation of international standards, and recently, in the

dramatic reduction in market cost.

•SHS components•Quality•Prices•Reliability•Installation

•User SHS know‐how•Consumption

•SHS interface•User manual

•Financementmodel (subsidies, credits, cash sales, fee for service, etc)

•Normes, tenders, engineering•ESCO: marketing, installation, O&M, fee collection•Internal skills and training•Management structure

•O&M management and costs

•Datalogger•Monitoring•Prepayment system•Technical standard•Spare parts

•Enquiries•User skills•Fee payment•O&M fees•Maintenance service

HARDWARE

ORGWARE

SOFTWARE


23

Second, the development of the software dates back to the beginnings of PVRE, when the task of

accommodating the needs and abilities of users to the management and operation of the PV

systems was the first requirement for the successful implementation of this technology. This has

remained until today, constantly adapting to new hardware advancements.

As regards the orgware, despite its developmental delay in PVRE, some of the factors that integrate

it have reached a high degree of maturity. Several management and organizational models have

been well described in the literature and applied in the field, especially since the 1990s, and they

have been studied in depth by recognized international organizations such as the World Bank [17,

58, 59, 60, 61, 62, 63, 64] or the International Energy Agency [65, 66, 67]. However, the orgware

has had several weak points during the development of PVRE, as will be discussed below.

1.3.5 EvolutionoftheHW,SWandOWinPVRE

1.3.5.1 The1960sand1970s. Hardwaredevelopment:reliabilityandcost‐effectivenessindecentralizedruralelectrification

The first terrestrial experiences of PV technology date back to the 1960s when Japan began to use

it in maritime applications (light beacons, communications, etc) [68]. Paradoxically, oil companies

such as Exxon, Texaco and Shell, among others, pioneered the use of photovoltaic solar energy.

These companies had equipped their platforms in the Gulf of Mexico with lighted beacons, which

were fed from non‐rechargeable batteries which were frequently replaced, at an operating cost of

about US$2,100 per replaced battery. In the 1970s, these companies decided to change these

accumulators for rechargeable batteries with a photovoltaic generator, thus reducing the operating

costs by 95%.

It was in 1968, in Niger, when PVRE started formally, through the installation of a system to feed a

television in the Gondel school, close to Niamey [69]. The experience was expanded to other

schools until 1977, after installing 123 PV systems. They were made up of a 282 watt peak (Wp)

photovoltaic power generator, a 40 ampere‐hour (Ah) and 32 nominal volt (V) battery, and a charge

controller to feed a television receiver of 32 W. The cost of these systems was US$3,100 per school

in 1975, with an estimating price of US$0.12/hour of television, which meant US$3.75/kWh.

Despite this enormous cost and considering that the lifetime of the PV system was 10 years (PV

manufacturers at the time gave 5‐year warranties), the solution was 4 times cheaper than the

option of using high‐capacity alkaline cells, for which the TV receivers were originally designed.

In the 1970s, Father Verspieren in Mali [70], and his organization Mali Aqua Viva [71], instigated

the first photovoltaic pumping systems programme for extracting water from wells, in order to try

to solve the disastrous situation of thousands of people affected by the severe drought that

suffered the Sahel region in those years. The use of PV pumps by Verspieren was the result of years

of bad experiences with hand pumps and diesel generators because of their low reliability and high

O&M costs. Mali Aqua Viva carried out the installation of 16 PV pumping systems (reaching a total

power of 21.8 kW) between 1975 and 1980, which was one of the first milestones of PVRE to

consolidate this technology as a cost effective and reliable alternative to diesel generators and

hand pumps.

24

1.3.5.2 FirstpromotionandR&DprogramstoreducethecostsofPV

The oil crisis was the trigger for the first political incentives for industry and research into

photovoltaic technology and its application in rural electrification. As PV was still a technology with

high manufacturing costs, a first researching phase focused on the cost reduction was necessary.

Some of these initiatives were as follows:

In 1975 the Commission of the European Communities financed the first R&D program in the field of non‐nuclear and non‐fossil fuel energies. It devoted US$6.4 million to photovoltaic conversion [72], with the aim of studying and enhancing the photovoltaic cells, to later evaluate them in several 5 kWp prototype systems [73].

In 1976, the Department of Science and Technology of the Government of India launched their Solar Cell Programme Plan with the aim of researching and developing different projects in areas such as the "development of conventional type single crystal silicon solar cells" to get 7‐9% PV conversion efficiencies. The programme was motivated by the low rate of access to electricity in India (less than 10% of the rural inhabitants). They took PV energy into account as a technological and cost‐effective solution, as alternative to the electric grid extensions. At the time a cost of US$60 billion was estimated to mass electrify 100% of the population at 1 kWp per dwelling [74].

In the mid 1970s Mexico had a rural electrification rate of 35% (10.7 million people without access to electricity). The Centro de Investigación y de Estudios Avanzados of the Instituto Politécnico Nacional carried out some projects for PV terrestrial applications against the background of the rural electrification problem. In 1976 two pilot projects were established: "the demonstration project in educational TV" and "PV for two rural telephone stations", both in the East of the Mexican Rocky Mountains, using PV modules of 7 Wp and 12‐15 V with 10% efficiency [75].

In Japan, also in the 1970s, the "Sunshine Project" had the final goal of reducing the cost of PV by a factor of 100 through 5 fields of research: silicon ribbon crystals, silicon thin film, new types of solar cells, II‐IV compound semiconductors and fundamental research [76].

In the same vein, the UK Department of Industry (DIn), the Science Research Council and the European Economic Community (EEC), started research work in the field of the PV cells in the 1970s a with the aim of reducing the manufacturing costs by half (less than £8/Wp) by means of the development of new manufacturing processes of photovoltaic cells [77].

By the mid‐1970s the first pilot projects began, which aimed to direct PV technology applications

towards decentralized electrification, integrating the software dimension while its purpose was to

electrify remote rural populations:

In 1976 in the USA, the "PV Stand‐Alone Application Project", led by the NASA Lewis Research Centre and the United States Agency for International Development (USAID), developed "universal" stand alone PV systems in order to open up a new market for rural electrification in developing countries for domestic lighting applications, water pumping, grain mills, etc. [78, 79, 80, 81].

In 1978 The United Nations Development Programme (UNDP) and the World Bank (WB) launched the UNDP funded GLO/78/004 project to develop small‐scale pumping systems for water supply and irrigation applications in developing countries [82], including field trials of systems in Mali, the Philippines and Sudan.


25

In some forums, such as the "Solar Energy for Development" conference in 1979, Varese (Italy) [83] the need for subsidies began to be discussed, with emphasis to the fact that giving PV systems to the users as a "gift" was something quite inappropriate.

In 1979, during the "Interregional Symposium on Solar Energy for Development" in Japan, the first NASA pilot experiences in the USA and Africa were set out [68]. The key factors related to the reduction in manufacturing costs were shown (between 1975 and 1978 the market price of the PV module had been reduced from US$35/Wp to US$13/Wp and it was expected that it would reach US$0,61/Wp by 1986), together with the high reliability of the PV systems, which would turn this technology into a competitive alternative versus power grids in developing countries in the short‐term.

1.3.5.3 The1980s.EvolutionaryconsolidationofHWandSW

After the first initiatives emerged in the 1970s and glimpsed the enormous potential of PV and the

brilliant expectations of cost reduction in the short and medium term, the 1980s saw the boom in

PV applications in rural electrification. The 1980s was the decade of the demonstration projects and

the beginning of the development and evolution of the 3 dimensions of the PV rural electrification

system together. Some of these projects are listed below:

Energy Demonstration Programme (EDP). The Commission of the European Communities (EC‐Commission DGXII), in addition to carrying out the first EC projects in the field of energy co‐operation overseas (Cameroon, Comoros, Syria, Jordan, Mali, Senegal, etc) [84], it launched the Energy Demonstration Programme (EDP) (15.3 million ECU [85]) in the 1980s and the THERMIE programme in the 1990s [86], from which 145 pilot projects were successfully established in European countries, installing a power capacity of up to 5.7 MW. 61 of these projects were devoted to rural electrification. The purpose of the initiatives was to encourage the photovoltaic market, reducing costs and building up a know‐how on key issues such as the quality and performance of the systems (hardware), the assessment of usage patterns and interaction with the user (software) and the funding models in rural electrification (orgware). As regards the latter, the successful experience of the French Pyrenees electrification, in which the users created local associations for financing, operating and maintaining their PV systems proved interesting [87]. Another significant demonstration project was the PV rural electrification of dwellings in Sierra de Segura (Spain), a remote location in the South of the country [88]. It involved different entities such as the regional utility, two research institutes and a PV module provider. Some of the lessons learned were that PV technology was perceived by the inhabitants as a limited electric source and they feared that getting SHS would be an obstacle to reaching the grid in the future. In addition, they were aware of the need for maintenance and were willing to pay a periodic fee to assure the energy supply [89].

4‐year Energy R&D Programme. Belonging to the EDP programme, 15 pilot projects were carried out with power of between 30 and 300 kWp, reaching almost 1 MWp. It was carried out in decentralized locations within the European Community, most of them on isolated islands such as French Guyana, Crete, Terschelling Island in Holland, Corsica, Giglio in Italy, Pellworm in Germany (with a PV power plant of 300 kWp), or Kythnos in Greece (with a 100 kWp hybrid PV‐diesel power plant) [90]. The evaluation of these projects gave rise to the publication of the first IEC (International Electrotechnical Commission) quality standards dedicated to terrestrial PV technology through the TC‐82 Committee [91].

The Navajo Engineering and Construction Authority ‐ Indian Health Service Home Photovoltaic Power Systems Project provided solar home systems to 191 dwellings on the Navajo Reserve, in the Southwest of the USA. ARCO Solar, the supplier and installer, guaranteed the SHSs to the users for a period of 3 years [92].

26

An agreement between Spanish Cooperation and Senegal in 1985 was responsible for the electrification of the Village of Noto (Senegal), which had 400 inhabitants, by means of SHSs and an 8 kW PV power plant [93].

USAID developed a cooperation solar project in Egypt to check the performance of PV

technology applied to different usages, such as water pumping, an ice‐machine, water

desalination, village stand‐alone systems and grid‐connection. The aim was to demonstrate

the economic viability and open the Egyptian market to this technology [94].

Colombia was highlighted through a boom in PV technology applied in rural electrification, with more than 6,000 SHS installed in the mid‐80s of the decade and over 10,000 telecommunications systems [95].

France began on its own PV programme in the 1980s for industrial development in order to manufacture PV modules locally reducing costs and creating a national PV market [96]. Several PVRE projects were carried out in some of its islands, for example Polinesia [97], New Caledonia, Guadeloupe and the Marie‐Galante Islands [98], even in regions from the South of France [87] by means of the EC‐Commission DGXVII programme (the Solar Energy Demonstration Projects), which also financed the provision of PV systems to electrify mountain refuges in the Alps and Pyrenees mountains [99].

Some demonstration projects were also carried out in Spain, for example the 100 kW power plant in San Agustín de Guadalix (close to Madrid) [100], 10 kW in Caravaca (Murcia), in addition to other CEC DG‐XVII projects. However, the development of a spontaneous and non‐demonstration PV market was quite remarkable. It was the largest in the world during that decade, after the United States, with more than 8,000 SHSs installed between 1980 and 1986, which represented 35% of the more than 1.6 MW installed in total in the country [50, 101, 102]. The other PV facilities were professional applications such as telecommunications (20%) or holiday homes located in the countryside (40%). The Spanish experience was studied in depth by the Insitituto de Energía Solar from the Universidad Politécnica de Madrid (IES‐UPM). For the first time the consumption profiles and usage habits of the SHSs were characterized, which contributed to the software evolution providing a valuable feedback [103, 104]. One of the most revealing conclusions was that, in general, electricity consumption in PVRE was low, which has been a common trend in many other experiences analyzed later in other countries.

The software approach was also studied in depth by researches in Chile [105], in rural communities in the North of the country (the Camarones commune), where several aspects of the software were characterized such as the different consumption habits, the available information of the users with respect the use of the system, its maintenance and procurement of spare parts, the degree of satisfaction in using SHSs, the adaptability to the electrical applications, etc. These studies conclude that social aspects must be considered before formulating and sizing the SHSs to adapt them to real needs and get wide acceptance by the users.

The importance that the software dimension gained within the technological system was

mentioned in different studies, which took into account the sociological factor [106]. Some of the

failures in the first pilot projects were justified by technical problems, for example the low quality

of SHS components, limited size of the PV system or high prices. Instead, it was common that the

real problems came from the ignorance of certain sociological factors:

even if the technology impact assessments were frequently carried out after the installation, the cases in which the studies were done before the implementation of the projects were few;


27

without the previous sociological study, it was difficult to size the system adapted to the real needs of the future users correctly;

these deficiencies in the implementation of PV rural projects caused conflicts regarding the existing social structure, so many of these projects were doomed to failure.

Some referent literature at the time approached the need to consider the maintenance and

reliability of the systems as the most important criteria when formulating PVRE projects in

developing countries [107, 108, 109]. The efficiency and reliability enhancement of SHSs, as well as

the need for technically qualified staff to carry out the maintenance services were some of the

most important issues to ensure the success of the projects. However, this key factor (the orgware)

was not paid enough attention during the coming decades.

1.3.5.4 The1990sand2000.Theorgwareevolution

During the 1990s the photovoltaic rural technology reached some maturity that crystallized in new

programs and projects, mainly in impoverished countries. Many of these electrification experiences

have been the subject of studies that have highlighted the different financing models applied in

each case. The studies published by the World Bank were outstanding, such as the "Best practices

for photovoltaic household electrification programs" [17], which, based on several real experiences

in Indonesia, Sri Lanka, the Philippines and the Dominican Republic, released the first

recommendations about the implementation of financing and management models in PVRE

programs (OW), which in theory would ensure their success.

The outstanding attention given during the 1990s and 2000s to the management models and the

orgware in general has been reflected in some studies such as the Photovoltaic Power Systems

Programme (PVPS‐IEA) [53], on the nature of the institutional and financial framework, the

implementation and financing mechanisms used and the level of capacity building and quality

assurance that affect the success of PV deployment in 16 different programs carried out in 16

countries from Asia, Africa, the Pacific, South America, Middle East [66].

The results of many of these studies agreed on the generalized lack of available information on the

performance of SHSs and projects. Some experts had realized at the time that "independent

evaluations of PV projects are scarce (and also difficult to perform), so, despite some interesting

exceptions [...], data related to technical problems on realities in the field have rarely reached the

available literature" [110].

As regards this concern, the Netherlands Energy Research Foundation was very conclusive and it

made a large assessment based on experiences from 32 countries in Africa, the Pacific, South

America and Asia. The study was articulated on the basis of the HW, SW and OW [111]. The main

conclusion of the study was the marked lack of feedback of the programs carried out and its

disastrous impact on the learning curve of the technological system.

As regards the SHS, in parallel to the proliferation of rural electrification programs, tens of technical

standards emerged, most of them for local specific projects. They served to define the technical

requirements of these programs. In order to standardize the technical criteria in one standard rule

that might serve in any PVRE programme, the Thermie B: SUP‐995‐96 initiative, carried out by the

IES‐UPM and based on already existing standards made ad hoc for real experiences in countries

28

such as Bolivia, India, Brazil, Kenya, Indonesia, etc, [112] resulted in an international quality

standard for SHSs, which is still used in PVRE programs worldwide [113].

Due to this concern about the quality of SHSs, an ambitious in‐field study of batteries working in a

large PVRE programme was carried out for the first time, with 40,000 SHSs in Mexico, with the aim

of assessing the performance of previously installed photovoltaic (PV) rural electrification systems

[114]. 555 batteries were analyzed within their first 18 months of operation. Despite the

maintenance of the systems being devolved to the users, the study demonstrated that the state of

the batteries was satisfactory (only 4% of the batteries showed malfunctioning or high

degradation).

The beginning of the century led to the assessment of many PVRE experiences carried out at the

time. The most significant programs were reviewed, from which significant conclusions were

extracted on the financing and management models that were implemented (OW). More

international technical standards for SHSs were published (HW), and the user perception (SW) was

analyzed in depth [115]. It was thus preparing for the jump to large PVRE programs.

The historical overview of the PVRE evolution shows that the orgware development during the

1990s and the beginning of 21th century was mainly devoted to the role of the financing models.

The change in scale of PVRE meant the need to mobilize large sums of money to finance the

decentralized electrification. When mobilizing funds, large financial institutions needed a good

definition and structuring of business models, which had just been defined and studied thoroughly,

with support, for example, from the WB.

In this way , the development of large PVRE programs was possible in the first decade of the 21st

century, for example in Senegal (10,000 SHS [116], where there is currently an additional

concessions programme which aims to electrify the rural regions of the country by means of 18,000

SHSs [117, 118] ), Ghana (10,000 SHSs [119]), South Africa (30,000 SHSs [120]) the PERG in

Morocco (51,000 SHSs [121]), Sri Lanka (20,000 SHSs [122, 123]), Nepal (110,000 SHSs [122]), India

(the Remote Village Electrification Programme has equipped more than 10,000 villages by means of

SHS and PV power plants as of 2013 [124]. Kenya and Tanzania (more than 320,000 SHSs have been

installed in Kenya and more than 40,000 in Tanzania as of 2013 [125, 126]), China (from 2002 to

2007, more than 400,000 solar home systems were sold in north‐western China under the US$316

million World Bank/Global Environment Facility‐supported Renewable Energy Development Project

(REDP) [127]). Also in Latin America some large programs have seen the light [119], such as in

Bolivia (60,000 SHSs), Honduras (5,000 SHSs), Peru (10,000 SHSs), Dominica Republic (3,000 SHSs)

or Argentina (30,000 SHSs).

But among all these experiences, it highlights the case of Bangladesh [128, 129, 130, 131]. From

mid‐2000 to May 2013 over 2,000,000 SHSs were installed [132], in a power range of between 30

Wp and 75 Wp per SHS. In 2013 the installation rate was 1,000 SHSs/day, therefore at the end of

2014 the operational SHSs could have got around 3,000,000 SHSs [133]. This programme is

currently the most important in the world in terms of volume. It has been developed under a credit

sales model. The keys to its success are found in the correct balance of the HW, SW and OW

dimensions, highlighting an economically affordable HW for users which is also sized to meet their

needs; an efficient microcredit funding system which is well supported by solid institutions; and an

O&M structure based on the creation of local employment through the training of technicians.


29

Therefore it is evident that the "financial model" factor has been developed in depth leading to a

worldwide phenomenon of investments, loans, financial products, etc. that has allowed the leap in

scale of PVRE programs in terms of size, which is manifested by the millions of SHSs installed in less

than 15 years worldwide (some studies estimate more than 5 million [134]).

1.3.6 StateoftheartoftheHW,SWandOWinPVRE

Throughout the history of PVRE, rural photovoltaic technology has evolved in the three dimensions

(HW, SW and OW) as summarized below:

1. The SHS (Hardware)

‐ PV modules. The most used PV technology at the beginning of PVRE was amorphous silicon (a‐Si), which was rapidly replaced by crystalline silicon (c‐Si), up to now. Other technologies such as thin‐film, CdTe, etc [135] have not still found an application niche in PVRE. The c‐Si PV module has reduced its cost by 95% in four decades, has increased its power efficiency by 2.5 (in 2014 an efficiency of 25% had been reached [136]), and the quality of the assembly components have been optimized (front glass, EVA‐tedlar encapsulate 6 , electrical connections, module insulation, etc) to obtain a long lifetime product (more than 25 years) and low degradation rates (PV manufacturers guarantee currently 80% of the PV module nominal power up to 20 years). The cost of the PV cells has been reduced from US$100/Wp in the early 1970s to less than US$0.36 today (Figure 13). The photovoltaic industry has also contributed to this reduction by lowering the production process costs, from metallurgical silicon to the assembly and lamination of the PV modules, in addition to the enhancement of its efficiency.

Figure 13: The price evolution of Silicon PV cell (in US$ per watt). Source Bloomberg, New Energy Finance & pv.energytrend.com

6 The Ethylene‐vinyl acetate (EVA) and the Polyvinyl fluoride (PVF, known by the trade mark Tedlar®) are two polymers used for PV module encapsulation

30

‐ SHS components (Balance Of System):

i. The solar battery. The technology of electrochemical lead‐acid batteries has not significantly changed in 4 decades. The deep‐cycle lead‐acid battery, used in stationary applications such as telecommunications since the 1970s, has been the most adapted battery for PV systems. However, the usual budget constraints linked to the local unavailability of deep‐cycle batteries in many developing countries has extended the use of start‐lighting‐ignition (SLI) batteries in PVRE programs due to its low cost.

In the field of PVRE the most significant progress in storage technology has been the use of VRLA (valve‐regulated lead‐acid) batteries, known as "free maintenance", which have a gel‐electrolyte, thus they do not require the periodic filling of water to keep the electrolyte. This aspect makes the VRLA very attractive in decentralized electrification as it represents a significant lowering of the maintenance costs.

The lead‐acid battery, with over 100 years of existence, has very little chance nowadays to reduce its cost. Rather, the lead, as major component of the battery, is a metal which is becoming more expensive every year. The hope of reducing the cost of the SHS batteries is the development of new technologies, such as lithium‐ion, but this solution seems to be still far from becoming an economically viable alternative.

ii. Charge controllers. The evolution of this device has basically been an improvement in the control algorithms and adaptation to the SHS‐user system. On the one hand, the electronic progress has led to PWM control (Pulse‐Width Modulation), the voltage thresholds in charging and discharging considering the temperature and the setup of algorithms presumably adapted to the charge and discharge curves of the batteries. In recent years the maximum power point tracker (MPPT) technology [137] has been imposed, especially in medium‐sized controllers. On the other hand the user interface is becoming more "friendly" through the implementation of LEDs and acoustic signals or digital displays from which information about the battery status, circuit faults, etc. can easily be obtained.

iii. Lamps. Lighting is the main load in SHSs. Even if lamps are not a part of the energy generator, they become part of the system. The importance of lamps is that the performance of the SHS depends on the quality of its loads. Typically SHSs run at 12 VDC, so the lamps must be adapted to that voltage, and their consumption very low, according to the size of the SHS [138]. The lamps have evolved from early fluorescent tubes to the later compact fluorescent lamps (CFL) and the current LED lamps. This development has led to more reliable lamps, larger lifetime, cost reduction and better quality of lighting [139].

‐ Quality: PVGAP (Global Approval Program for Photovoltaics), IEC (International Electrotechnical Commission) or Thermie‐B are some of the initiatives carried out throughout the three decades, which have generated a wide set of technical norms and quality standards as regards PVRE [134, 140, 141]. The IEC has published since the founding of the TC‐82 technical committee for solar photovoltaic energy systems in 1981, the largest collection of technical standards relating to both the photovoltaic technology and PV systems and components. Specifically, the IEC TS 62257 series [142] is devoted to PVRE and aims to provide different players (such as project implementers, contractors, supervisors, installers, etc.) the foundations for the setting up of renewable energy and hybrid systems with AC voltage below 500 V, DC voltage below 50 V and power below 50 kVA.

The role that these standards have played in PVRE has been called into question by different authors [110] due to their effectiveness which could often not be demonstrated, among other things because the decentralized nature of PVRE has not allowed the required feedback from these in‐field experiences to be obtained, which would be necessary in a


31

correct process of developing standards. Furthermore, the behaviour of rural users is difficult to standardize, and difficult to simulate in the laboratories.

By the mid‐90s, under the Thermie‐B program, The Universal Technical Standard for Solar Home Systems (UTSfSHS) was released as the result of an ambitious survey of experiences in the field, resulting in a set of standards and recommendations that are still applied today in PVRE programs (Peru, Bolivia, Morocco, Tunisia, among others)7. The UTSfSHS guarantees the quality and proper operation of the SHS components as well as proposing easy and cheap quality control procedures to check the SHS performance "in situ", with no need for prestigious laboratories far from the locations where PVRE programs are developed.

‐ Installation: as with the SHS components, their installations and connections have to fulfil several quality criteria to ensure an optimal performance of the system. These criteria are compiled in international standards such as the aforementioned UTSfSHS. This issue was one of the first concerns that were addressed in terms of quality at the beginning of the development of SHSs and has been the object of several publications [143].

‐ Reliability: it has been indicated that the PV module is a very reliable device, its failure rate is very low and its lifetime very long. However, the other components of the system have not enjoyed the same features. For example, it is said that a battery has a lifetime of 3 to 10 years, according to some factors such as its quality, kind of technology or even its cost. The same applies to the charge controller or lamps. But, in rural electrification there is another factor that affects the reliability of SHS: the user. In a PVRE programme with thousands of SHSs, the users' behaviour as regards SHS will largely determine the possibility of having more or less failures depending on the good or bad use of the system.

To date, there are no studies reporting the reliability of SHSs working under real conditions and considering the intervention of the user, so this component of the hardware is one of the outstanding pending issues in photovoltaic rural technology.

2. Use of the SHS (software).

The trend over the 40 years life of PVRE as regards software has been:

The user should be familiar with both the operation and the limitations of the SHS

The SHS interface should be as clear as possible so that the user is able to diagnose the state of operation of the system

The user must know the procedure to follow in case of failure (warning the ESCO)

The few reported experiences show that consumption is in general lower than expected but they gradually grow as users become more familiar with the system and get new appliances

The size of the SHSs is a controversial matter since the needs perceived by the users rarely match those considered by engineers and promoters. For example, lighting is often considered a priority for rural inhabitants, while they may prefer to have access to TV.

3. The management of the rural electrification programs (Orgware)

The good use of SHS by the users and their correct operation and maintenance are aspects linked to how the PVRE programs are designed and configured (orgware). Some of the first experiences devolved the maintenance responsibility to the user, giving very poor results. Other

7 Own sources.

32

programs have persisted in this practice by supplementing with a period of warranty service that could be of 2 years or more.

In the so‐called fee for service model, in which specialized technicians are responsible for the maintenance in exchange for a fee charged to the user, she/he only receives training on the proper use of the system. The warranty and maintenance falls to the concessionaire energy service company (ESCO).

The wide range of organizational models [144] that have been used to develop hundreds of PVRE programs worldwide in the last 4 decades is due to the presumed well‐intentioned will of their promoters in order to adapt the programs to the social and economic realities of the rural communities. The first experiences were pilot projects, as has been mentioned above, many of which remained at that stage and few gave rise to large‐scale projects. Non‐governmental‐organizations (NGOs) implemented programs in which the SHS was given to the user as a gift, after a technical training addressed to the user in order to be responsible for the maintenance. The experience led to some authors to advise against this practice due to the maintenance not being carried out properly and the SHS not being appreciated enough (simply for being free) by the user and then the SHS had the risk of being abandoned.

The cash sales model was implemented in the 1990s in Kenya, Tanzania and Uganda [145, 126, 146, 147, 148] with great success in terms of the number of SHSs sold. The model has been replicated in other countries with varying degrees of success. This model treats the SHS as a domestic appliance. To run the model it is necessary for people to have a high purchasing power, which is not common in most rural regions without access to electricity in the world. So, this model is geographically limited. As mentioned above, with the goal of increasing rural electrification ratios by means of access to lower‐income households, a new innovation has been introduced: the pico‐system. These systems are currently being commercialized in countries where the cash sales model has been successful [149], reaching a target of customers with lower resources.

Based on the low income of rural inhabitants, this concern has been addressed by micro‐credit schemes. Micro‐credits are usually used for small business activities, but they are granted for SHSs in countries where favourable social and cultural conditions ensure the repayment of loans even if using an SHS is not a profitable activity. This model is running with enormous success in India and Bangladesh, 2 countries with the largest PVRE programs in the World. The success when extending this scheme to other countries has not been demonstrated yet.

Finally, the fee for service model is generally based on subsidizing much of the cost of the SHS and then the user pays the rest. A specialized energy service company (ESCO) who installs the SHSs will be responsible for their maintenance for 10 years or more, and receives a fee from the user (usually monthly). This model would be the one closest to the notion of a public utility service.

In addition to these factors relating to financial issues of the orgware, there are other questions linked to the operation of the photovoltaic rural technological system. The sustainability of the programs requires a number of factors which are still in an underdeveloped state, and which have been the cause of failure of most them:

Monitoring: A 10‐year O&M activity needs a constant feedback in all aspects: technical, economical, etc.

Maintenance structure: to date the cost and the necessary size of the operation and maintenance structure for providing electricity in decentralized rural environments is still not known. This can be considered one of most significant technical challenges in PVRE which has been responsible for the lack of economic sustainability in most PVRE programs


33

Spare parts: the reliability of the SHS components are not known when they operate in real conditions. Thus the forecasting of spare parts through the programme lifetime is not available. This concern affects the design of the maintenance structure and the profitability of the programme

Fee collection: as most of the programs carried out to date have not considered a hardware adapted to the management of non‐payment cases, the ESCO has to devote many resources and time to collecting the fees, with long, frequent and recurrent displacements, even to the dwellings, where it is often not easy to meet the customers or gain to access to the houses. In response to this problem some experiences have integrated a control system of payments into the hardware, such as prepayment cards [126, 150, 151], or the "pay as you go" strategy [152, 153], among others. These innovations have had limited impact and effectiveness to date. In addition PVRE tends to minimize the investment costs, however these innovations tend to increase the capital costs.

1.3.7 ThechallengeofPVRE

After this assessment and a review of 40 years of PV rural technology development the following

conclusions can be extracted:

‐ Progress in the technology has been affected by the gap in the development of the three dimensions. Both the HW and the SW have co‐evolved since the early 1970s and it was not until the 1990s that the OW started to play an important role in PVRE. Some authors argue that when introducing new technologies into human communities, the OW and SW should go ahead of the HW in order to "prepare" the receiver for the new technology and ensure that he will adopt it successfully [154]. In PVRE the opposite has happened.

‐ Despite the integration of the OW into the technological system, the implementation of this dimension has been arduous. A widespread case in PVRE has been the lack of promotion and awareness addressed to the future users, which is generally a source of problems related to the bad use of the SHS and consequently the cause of failures and non‐payments of the fees to the ESCOs.

‐ The poor integration of the rural PV system elements between the 3 dimensions is the cause of the lack of sustainability of PVRE programs. Another illustrative example is the economic imbalances of the ESCOs in many PVRE programs, due to the low fees paid by the users. The "fee", which can be collected monthly, is intended to cover the maintenance costs. It is an orgware element, since it determines the economic viability of the operation and maintenance service. If the quota is set incorrectly, the system will have a financial imbalance that would wreck the programme.

‐ Finally, the lack of feedback or monitoring leads to ignorance of the effectiveness of the

measures taken for the programme development. If there is no feedback during the O&M

phase, the replacement rates of the SHS components are unknown, as well as the lifetime of

the systems, the degree of satisfaction of the users, the economic viability of the O&M, etc.

As the funding mechanisms for PVRE are well defined and have a good monitoring, the same

care during the operation phase would be desirable. But this element has implementation

problems in remote areas with difficult access and poor communications. Only a very well

organized structure with sufficient and adequate resources to develop the activity over a

long period of time can ensure the necessary feedback for the system.

34

Thus it could be argued that photovoltaic rural technology has reached a remarkable maturity in

the HW and SW dimensions, except HW adaptation to the notion of a public utility service and the

lack of understanding of the real reliability of the SHS components. The OW, however, has

progressed well in some of its factors, such as the financing models, but it has still serious

limitations in matters relating to the sustainability of PVRE programs, firstly due to the lack of

feedback, and secondly, because there is still no paradigm concerning the design of specific

maintenance structures for decentralized environments. The lack of both the reliability of the SHS

working under real conditions and the O&M costs left without tools for those responsible for the

programme when designing the maintenance structures.

So far, both the public body responsible for the formulation of electrification programs and private

companies charged with implementing them, have designed the maintenance structures based on

theoretical estimates, such as those assuming that the annual maintenance cost corresponds to 1%

‐ 3% of the capital cost (investment) [155, 156, 157], or fitting the maintenance costs to the

remuneration obtained from the user fees, which were previously established according to the

economic possibilities of the users or calculated from what was previously spent using conventional

lighting systems, such as candles, kerosene or others.

This situation reflects the lack of a paradigm on the O&M in PVRE programs. Many of the programs

carried out so far imposed, apparently, low fees, such as South Africa [120] (monthly fee: 6

US$/SHS∙month), Zambia [158] (12 US$/SHS∙month), Peru [159] (5 – 8 US$/SHS∙month), Tunisia

[115] (5.2 US$/SHS∙month) or the Pacific Islands [17] (5 US$/SHS∙month). Note that the dispersion

of the fee's can also be intended as an indicator on the immaturity of the paradigm. The experience

has proved that many programs have experienced economic imbalances in the operation phase as

the user fees did not cover the real costs of the O&M, after which the ESCO abandoned the

programme, leaving thousands of SHSs without a maintenance service [17, 120, 158, 160, 161].

Therefore, the O&M is, in general, a more expensive activity than expected; moreover, the fees

paid by the users for the maintenance service are generally lower than necessary. These two facts

lead to an unsustainable financial situation of the PVRE programs.

Thus, the current challenge is to ensure the sustainability in PVRE programs, introducing the notion

of public service in the decentralized rural environment in order to achieve the paradigm in which

the conditions that make the global access to electricity affordable will be defined.

This PhD thesis will address some of the identified problems for PVRE sustainability, as shown in

Table 3:

Table 3: PV rural system factors discussed in this work

HARDWARE Reliability of the SHSs working in real conditions.

SOFTWARE O&M tariff sizing adapted to the real maintenance costs.

ORGWARE Design of maintenance structures for decentralised environments. Characterization of the O&M cost structure.


35

1.4 OBJECTIVESOFTHETHESIS

The general objective of this thesis is to contribute to the technical and economic characterization

of large PVRE programs, including the O&M phase to ensure their sustainability.

This objective is broken down into the following specific objectives:

‐ Determining the real reliability parameters of the SHS components working under real conditions in which the "user" is involved

‐ Characterizing the actual operation and maintenance cost of a real PVRE programme

‐ Studying the battery and PV module degradation when working under real conditions within a PVRE programme

‐ Establishing the design criteria to optimize the maintenance structures in PVRE programs.

1.5 METHODOLOGYOFTHEWORK

To achieve the proposed objectives a wide experimental work based on the data collected during

the 5 years of development of a real PVRE programme carried out in Morocco with more than

13,000 SHSs installed and managed by an ESCO in a vast region of more than 200,000 km2 has been

carried out. A great effort has been dedicated to the assessment of a database with more than

80,000 inputs related to the technical data of the maintenance phase, in order to carry out a

comprehensive reliability study of the SHSs.

Furthermore, both in‐the‐field experiments have been conducted in several SHSs belonging to the

Moroccan PVRE programme in order to study the power degradation of a PV modules sample after

6 years of operation and the capacity degradation of a sample of batteries which have been tracked

for 18 months. Both have been working in real SHSs of the programme. In the case of the battery

assessment, several electrical operation parameters have been collected weekly and the battery

capacities have been measured every 6 months.

The second line of work has addressed the characterization of the management and O&M cost

during the same 5‐year period, based on the accounting records from the same ESCO.

Finally, a mathematical model has been developed based on a mixed integer linear optimization

with the aims of designing maintenance structures adapted to the special criteria concerning the

PVRE activity, illustrating its functionality through several examples from the Moroccan

programme.

1.6 THESISSTRUCTURE

The thesis is structured as follows:

Chapter 2: A brief description to the Moroccan electrification programme, known as PERG, because

all the experimental work of this thesis has been articulated around this programme.

Chapter 3: The description of the procedure carried out for the reliability study of the SHS

components, from the debugging of the maintenance database to the study results and application

examples.

36

Chapter 4: The results of the in‐the‐field experiments related to the degradation of PV modules and

batteries working under real conditions.

Chapter 5: The characterization of the real costs corresponding to the first 5 years of development

of the PERG programme and to the forecast for 10 years of O&M.

Chapter 6: The modelling process of the decentralised maintenance structures, based on a mixed

integer linear optimization, as well as its application to several regions of the Moroccan

programme.

Chapter 7: Summary of the most significant contributions of this thesis in terms of publications and

some possible future directions of research.

Bibliography: references and bibliography referred in this work.

CHAPTER 2

THE MOROCCAN PV RURAL

ELECTRIFICATION PROGRAMME

38

2 THEMOROCCANPVRURALELECTRIFICATIONPROGRAMME

2.1 INTRODUCTION

The research work of this PhD is based on a real PVRE experience. The PVRE programme chosen has

been one of the concessions of the Moroccan Rural Electrification Global Programme (PERG in the

French acronym) which installed more than 13,000 SHSs in 3 years, disseminated in around 200,000

km2 in the South of the country. This programme belongs to a global initiative managed by the

Moroccan national utility to provide access to electricity by means of grid extensions and stand

alone systems to almost 100% of the rural inhabitants.

The PERG programme is paradigmatic because, despite being formulated in a careful manner in

terms of financing and organizational schemes, supported by previous studies in social,

demographic, technical and economical issues, the Solar‐PERG concessionaries (ESCOs)

experienced serious financial problems when carrying out their work during the maintenance

phase. These problems have led to most of the ESCOs leaving their concessions because of financial

imbalances.

This chapter describes the Solar‐PERG concession assigned to the private ESCO ISOFOTON (it will be

referred to as Solar‐PERGISO). Being a public‐private‐partnership (ppp) between the ESCO and the

national utility, the development rules of the programme are set out, as are the objectives of the

concession and the real achievements and main barriers encountered during the implementation

and operation.

In the operative frame, ISOFOTON was responsible for selling the SHSs directly to the final users

through a fee for service mechanism, as well as its supply and installation within the 3 first years of

the programme (from 2006 to 2008). The installed SHSs were subject to a 10‐year maintenance

period. The ESCO must guarantee the systems during this period and must repair or replace

damaged components. The ESCO was also responsible for collecting the monthly fees that the users

must pay for the maintenance service.

The maintenance and cost databases in which almost all the experimental work of this thesis has

been based are described and explained at the end of the chapter.

2.2 THEPERGPROGRAMME

In 1995, more than 82% of the Moroccan rural population lived still without access to electricity,

that is, around 10 million people. The Government of Morocco declared universal rural access to

electricity as a political priority and in 1996 began supporting the implementation of the Rural

Electrification Global Programme. After this great effort, the electrification rate in Morocco is

currently 98.5%.

Chapter 2: The Moroccan PV Rural Electrification Programme

39

During this period (1996 ‐ 2013), the PERG programme gave access to electricity to 2,027,120 rural

households in 37,099 towns by means of electric grid extensions and 51,559 more by means of

solar home systems in 3,663 towns (the Solar‐PERG) [162]. These achievements have been made

through the construction of 42,900 km of Medium Voltage lines, 111,700 km of Low Voltage lines

and 21,650 MV/LV transformers (a power of more than 1,500 MVA). Moreover, for the off‐grid

electrification, 51,559 solar home systems have been installed with a mean peak power of 60

Wp/SHS (more than 3 MWp installed).

The PERG was promoted by the ONEE (Water and Electricity National Office), the National Utility,

who needed the financing of more than €2.27 billion to complete the programme, from which,

around €1 billion came from international institutions. However, it is worth mentioning that local

authorities have greatly contributed to financing the programme, although these figures rarely

appear in the published summaries.

The ONEE established the financial terms that would apply in the PERG development. In the case of

the grid electrification, the financing model determined that the users paid 2,500 dirhams (MAD)

for the subscription (grid connection)8, the local municipality contributed 2,085 MAD/connection

and the remaining cost would be covered by the utility (the ONEE) until a fixed maximum threshold

per household, set out in Table 4, that varied during the four consecutive phases of the

programme.

Table 4: Maximum threshold cost for household grid connection

Period PERG phase Maximum threshold cost per

household (MAD)

1996 ‐ 2002 1 & 2 10,000

2002 ‐ 2004 3 14,000

2004 ‐ 2006 4 ‐ 1st phase 20,000

2006 ‐ 2016 4 ‐ 2nd phase 27,000

The PERG rules dictate that when the grid connection cost to electrify a rural household exceeded

the mentioned maximum thresholds, the house would be electrified by solar energy, unless the

local authority covered the additional costs.

At the beginning it was estimated that 150,000 households would be equipped with SHSs [163]. At

the end of the programme, just 51,559 homes had been electrified by means of PV systems, and

not only that, most of these solar households are currently being connected to the grid. It can be

deduced that more than 100,000 households have been connected to the electrical grid exceeding

the threshold cost, reaching amounts that sometimes exceed 90,000 MAD/connection. Overall, the

global cost of the PERG may have been close to €3.5 billion (taking into account the local

authorities financing), in addition to €0.86 billion corresponding to the user subscriptions and the

municipality contributions, what means an average PERG cost of €2,000/household.

8 1 dirham (MAD) = 0.09272 euro (€) exchange rate at 04/12/2014

40

2.3 THESOLAR‐PERGORIGIN,DEVELOPMENTANDFEATURES

The technical definition and financial arrangements of the Solar‐PERG were based on the evaluation

of a previous programme, called PPER (Rural Electrification Pilot Programme), which was the first

attempt at establishing a programme in the field of off‐grid rural electrification in Morocco. It

electrified 1,500 dwellings in 30 different villages from 3 provinces between 1996 and 2001. After

that, the Solar‐PERG was established as a fee‐for‐service model in which the management of the

projects would be delegated to private energy service companies (ESCO) in a public‐private‐

partnership with the ONEE [164, 165]. The concessions to the ESCOs were made through several

tenders covering different regions of Moroccan territory. There were 4 different tenders between

2002 and 2005, whose awards are detailed in Table 5 and Figure 15:

Table 5: Solar‐PERG concessions and achievements

TENDER ESCO SHSs

OBJECTIVE SHSs

INSTALLED9 PROVINCES

OFF‐PERG Various ‐ 5,387 2002 ‐ PERG I TEMASOL 16,000 16,000 (100%) 4

2003 ‐ PERG II SUNLIGHT POWER MAROC 12,000 6,513 (54%) 3 BP SOLAR MAROC 4,000 950 (24%) 1

2004 ‐ PERG III TEMASOL 37,000 7,609 (21%) 20

2005 ‐ PERG IV ISOFOTON 34,500 13,600 (39%) 13 TEMASOL 5,500 1,500 (27%) 5

TOTAL Solar‐PERG 109,000 46,172 (42%) 46

TOTAL Solar‐PERG + others 51,559

The installation process was carried out according to the chronological scheme showed in Figure

14:

Figure 14: Solar PERG chronological scheme for the annual SHS installed

9 Some of the figures are approximate.

1500385 976 1308 1218

5070

83229533

5248

26911658

50

3929

45395132

0

2000

4000

6000

8000

10000

12000

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

PERG ISOFOTON

PERG other ESCOs

Before PERG

Number of SHS installed


41

Figure 15: Distribution of the Solar‐PERG concessions between the different ESCOs

Figure 16: Solar‐PERGISO local agencies distribution. The different colours indicate the covered area by each local agency

Morocco has a surface of 446,550 km2 (we consider the international recognition of Morocco

without the Western Sahara territory), so the mean geographical density of SHSs belonging to

Solar‐PERG corresponds to 0.1 SHS/km2. This extremely low geographical density of systems has

been one of the barriers that the ESCOs had to face.

42

In the operative frame, the ESCOs were responsible for the marketing, sales, installation and the

operation & maintenance of the SHSs in accordance with the following principles:

The SHS cost is subsidized 90% by the utility. The remaining 10% is paid by the user through

a fee collected by the ESCO.

There were different kinds of SHSs (Table 6), depending on PV power and service charges,

whose choice corresponded to the users.

The user signs a subscription contract with the ONEE, and a second contract with the ESCO

and the ONEE for the O&M service.

The delay in installing the SHS is 2 weeks from signing the contract and paying the

subscription fee.

The SHS ownership remains with the ONEE utility.

The ONEE inspects the new installed SHSs. If favourable, the ESCO will issue an invoice.

Table 6: Different SHSs offered in the Solar‐PERG programme

TENDER PV power Services User connection fee in € and (MAD), taxes

included

Monthly O&M fee in € and (MAD), taxes

included

PERG I&II

50 Wp 4 lamps + 1 socket (12 V)

69 (700) 6 (65)


167 (1,800) 9 (96)


287 (3,100) 12 (129)

PERG III&IV


83 (900) 6 (65)

200 Wp 4 lamps + 3 socket (12 V) + 1 fridge

370 (4,000) 14 (150)

Figure 17 PERG III&IV standard 75 Wp SHS scheme and components description


43

Despite the variety of SHSs, almost 100% of the SHSs installed in phases I and II were 50 Wp power

systems and 75 Wp for the phases III and IV (the smaller systems), which means that, in general,

users prefer low costs systems.

The O&M contract establishes the user and ESCO obligations for a period of 10 years, which are

detailed in Table 7:

Table 7: Clauses of the O&M contract for the user and the ESCO

USER ESCO

Respect of the SHS use rules

Payment of the O&M fee every month

Payment of surcharges when there is payment delay

Contact the ESCO when failure or malfunction of the SHS

Allow entry to the house to the ESCO technicians

Train the user to use the SHS

Collect the O&M fees

Visit each SHS every 6 months for preventive maintenance

Repair the SHS failures or malfunctioning within 48 hours after user claim

Pick up (dismantling) the SHS after 3 months of non‐payment of the O&M fee

It is worth mentioning that the ESCO BP‐SOLAR, concessioner of 4,000 SHSs in 1 province

(Chichaoua), that finally installed less than 1,000 SHSs, used a prepayment system for collecting the

monthly maintenance fees, based on a rechargeable electronic card that users had to take to the

ESCO staff, and after recharging, to insert it into the charge controller. There was no feedback

about the performance of the prepayment system since BP‐SOLAR was the first ESCO to abandon

the programme because of economical imbalances.

2.4 THEISOFOTON‐PERGPROGRAMME

The study of this PhD is based on the experience of the ESCO called ISOFOTON, marked in green in

Figure 14, which installed 13,600 SHSs between 2006 and 2008 in 13 different provinces, as shown

in Figure 15.

The Solar‐PERGISO region is geo‐morphologically characterized to be located, partly, in a

mountainous area (50% of the region is covered by 3 mountain ranges: the Medium Atlas, the

Grand Atlas and the Antiatlas), and partly by wide desert areas (Table 8). These features give rise to

difficult accessibility in the region and a wide dispersion of villages and households.

Table 8: Geomorphological distribution of the region

PERG‐ISOFOTON

surface Mountains Hills Desert Plain

km2 214,531 109,636 2,990 79,148 22,757

% 100% 51.1% 1.4% 36.9% 10.6%

The extremely low geographical density of SHSs was a remarkable obstacle for the ISOFOTON‐PERG

development. The average density of SHSs in the region was 0.068 SHSs/km2, which means that

there is 1 SHS per 14.7 km2. There are just two provinces, Al Kalaa des Sraghna and Beni Mellal,

44

whose geographical density is more than 0.4 SHSs/km2. In the vast majority of the region, the

density is less than 0.1 SHS/km2, coinciding with the most mountainous and desert areas.

ISOFOTON deployed a structure to manage the programme for each of its two phases:

Marketing, sales and installation phase (3 years)

O&M phase (10 years)

This management structure was made up of the headquarters and warehouse in Casablanca and 9

local agencies dispersed throughout the Solar‐PERG region. The staff was made up of executive

managers, accountants and administrative employees in the headquarters in addition to

commercial and technical staff in the local agencies.

Table 9 and Figure 16 summarize the distribution of the 13,600 SHSs in the different provinces and

indicate the surface and the geographical density of the systems. The local structure for the O&M

phase was made up of 9 local agencies, 43 employees and 19 vehicles. It should be noted that,

although 13,600 SHS were installed during the 3 first years of the programme, after the installation

phase, that figure was reduced to 13,449 due to the cancellation of some contracts because of lack

of payment. Thus, the O&M phase was developed for that latter figure.

Table 9: Summary of the O&M structure location (Local Agencies) in the different provinces of the PERG programme. The provinces’ areas and the installed SHSs are also indicated.

Provinces Province Capital

Area km2

Number of SHSs

Density (SHS/km2)

Local Agency location

TOTAL ‐ 214,531 13,600 0.063 9

Ben Slimane Ben Slimane 2,760 889 0.311 Ben Slimane

Errachidia Errachidia 59,585 968 0.016 Errich

Beni Mellal Beni Mellal 6,638 2,754 0.410 El

Ksiba

Azilal Azilal 9,800 1,817 0.185 Azilal

Al Haouz – Marrakech

Taghnaout 7,883 868 0.109 Ait Ourir

Al Kalaa des Sraghnas


10,070 4,445 0.437 Ben Guerir

Ouarzazate – Zagora

Ouarzazate 55,298 845 0.015 Ouarzazate

Taroudant Taroudant 16,500 697 0.042 Taroudant

Tiznit – Guelmim – Assa‐Zag

Tiznit 45,997 317 0.007 Tiznit

The choice of the location of the local agencies by the ESCO management staff was down to logistic

and administrative reasons. This is why some of the agencies were located in the capital of the

provinces, close to the ONEE’s bureaus, banks and regional government offices. Nevertheless, other

agencies were based in rural community centres to be closer to the PERG users, such as in

Errachidia, Al Kalaa des Sraghnas, Al Haouz or Beni Mellal.


45

The local agency serves both as the executive office and warehouse. One administrative employee

is in charge of the administrative tasks and remains permanently in the agency. The SHS' spare

parts are stocked in the warehouse, where the O&M technicians pick them up for the maintenance

and return the failed components from the in‐the‐field SHSs. The O&M teams are made up of two

technicians and one vehicle.

2.4.1 THEISOFOTON‐PERGSHS

The SHSs installed by ISOFOTON were mainly the 75 Wp power system kind. Just 30 of the 13,600

systems were of the 200 Wp typology. The 75 Wp SHS components are summarized in Table 10.

Table 10: 75W Solar‐PERGISO SHS component description

Qty Description Picture Qty Description Picture

1

75 Wp monocrystalline photovoltaic module

1 15 A PWM charge controller

1 150 Ah C20 modified SLI battery (12 V)

3 7WDC LC lamps

1 11WDC LC lamp

The SHS was sized for an autonomy of 5 days (tender condition), which means that considering 150

Ah as nominal capacity (Cnom), and a designed depth of discharge DOD = 40%, the design load was Ld

= 12 Ah/day, which corresponds approximately to the use of all the charges for 2 hours every day.

The SHS kit comprised a schematic manual for the user. It was a poster in Arabic installed close to

the charge controller attached to the indoor wall (Figure 18).

2.4.2 THEISOFOTON‐PERGPHASES

2.4.2.1 MARKETINGANDSALES

The commercial strategy to sell the SHSs was based on three approaches: the collective marketing;

the door to door sales; and the institutional approach.

In the first case, the collective marketing was carried out in the local markets (the so‐called souks)

organized once each week in the Centre of the Rural Communities10. A commercial team, by means

10 A Rural Community is a small region that comprises several villages and has an Administrative Center. Each Moroccan Province is made up of several Rural Communities, or Prefectures for the urban cities.

46

of a prototype of the SHSs (Figure 19), promoted the Solar‐PERG programme and tried to get

subscription contracts.

The door to door consisted of the commercial teams visiting the rural dwellings in order to get to

know to the future customers. This sales method required a great deal of effort to move to the

dwellings, which supposed a high economic cost. The O&M technicians used to do the door to door

sales when they visited the villages for the maintenance actions.

The institutional approach was based on the contact with the local authorities to get the necessary

support for the Solar‐PERG promotion (Figure 21).

These three commercial approaches were dependent on each other. For example, it is very usual in

the rural communities of Morocco for decisions not to be made at the individual level, but within

the collective sphere. In general, a family does not decide to install an SHS if the other neighbours

decide not to do the same thing. For this reason, the door to door was effective only when all the

inhabitants were previously convinced by the officers from the municipal or regional governments.

Even then, the Solar‐PERG programme had a generally negative response from the rural population

as they wanted to be electrified by the electrical grid, and not by means of solar systems. So, the

support of the local authorities was very important, in those cases in which these authorities were

not against the solar electrification.

However, despite the big efforts made by the ESCO in terms of commercial activity, the number of

SHSs installed remained well below the initial target (13,600 against 34,500 SHSs). Moreover, the

PV electrification was considered by the population as a temporary solution, waiting for the

definitive grid electrification, which currently is being carried out. This great gap between the

objective and the real achievement can be considered as a paradigmatic case regarding the

weakness of the assumptions on which the PVRE programmes are based.

2.4.2.2 INSTALLATION

After the subscription of a new customer and the signing of the contract, the ESCO delivered the

new contracts to the provincial office of the ONEE to be signed by the utility. In few days the ESCO

picked up the signed contracts from the ONEE and had a delay of 15 days to make the installations.

The installations were planned by the ISOFOTON local agency (LA) manager and supervised by the

headquarters staff. The SHS kits were stored in the local agency. Each morning, the technician

teams departed from the LA with a vehicle together with the solar kits to be installed and the list of

new customers.

The installation follows the ONEE reference prescriptions in terms of quantity and quality of the

components (Figure 22). In fact, in each province, an ONEE technical team went every month to the

rural villages for the commissioning of the new installed SHSs (Figure 20).

The installers, after finishing the installation, trained the users, often women and children, in the

use of the SHS, the meaning of the charge controller indication lights, and the procedures for

paying the O&M monthly fees and how to warn the ESCO when the SHS fails or malfunctions.


47

Finally, the SHS kit also included a record sheet in which the ESCO must register the maintenance

visits, indicating the date of the visit and the incidents that occurred.

2.4.2.3 O&M

The O&M period of a SHS began when the installation was accomplished. During the first 3 years,

ISOFOTON deployed a structure to cover all the activities: commercial, installation and O&M. When

the installation phase was finished, the structure was adjusted to just the O&M activity.

The maintenance comprised both the preventive and the corrective maintenance (PM and CM).

The preventive maintenance (PM) consists of checking the SHS performance. The battery must be

filled with distilled water if necessary, the PV module cleaned, the electric connections checked and

failed lamps replaced. The PM is a deterministic process as each SHS must be checked every 6

months, according to the Solar‐PERG rules. The maintenance technicians recorded the date of the

visit on the maintenance record sheet.

The corrective maintenance (CM) is carried out after prior notification by the user when she/he

notices a problem in the SHS operation. There was a free phone number, but they usually

contacted the local agency or the technicians directly when they visited the weekly markets. Then,

the technicians went to the dwelling, within 48 hours, to repair the system or replace the failed

components and return to start the SHS. The corrective visits were also recorded on the

maintenance record sheet.

The fee collection was carried out during the maintenance visits, in the local agency, or mainly in

the weekly markets organized in the centres of the rural communities (r), the souks. These

traditional souks attract most of the surrounding rural population and are the best places to meet

the SHS users, collect the monthly fees and replace failed lamps that users bring from their

dwellings. Note that lamps are the only SHS components that can be replaced in the souks.

The maintenance technicians are in charge of collecting the fees in the souks every week. They

spend the morning at the souk and the afternoon is dedicated to the maintenance activity in the

surrounding villages.

In some regions, there was a remarkable problem with the delay in payments, which habitually

obliged to the ESCO staff to go to the villages, dismantle some systems and even to take legal

action against the customers that did not allow to entry to the houses to take away the system.

2.4.2.4 MANAGEMENT

The management refers to the organization that the ESCO uses to manage the programme. It

involves the headquarters staff and resources, the flow of information between the customers‐local

agencies‐headquarter, the project and economic balances, forecasting, ONEE communication,

reports, etc.

All the staff of the ESCO ISOFOTON corresponding to the headquarters represented about 20% of

the total and evolved in the first 5 years from 14 people between 2006‐2008 to 11 and 8 in the

years 2009 and 2010 respectively.

48

The ESCO developed an ad hoc software tool to manage the programme that consisted of an

extensive database in which all data corresponding to user identification, dates of contract

signature, installation, fee payment, maintenance, stock, etc, were recorded.

Figure 18: User manual. It consists of a schematic poster including information on the SHS operation, warnings and use recommendations


49

Figure 19: Marketing of the Solar‐PERGISO programme carried out by the ISOFOTON staff in a souk (local market) in the Zagora province. The SHS prototype was used to show the system and explain its

operation

Figure 22: Outdoor and indoor installation stages of a SHS in the Errachidia province

Figure 20: ONEE inspector Figure 21: Meeting of the management staff of

ISOFOTON with the Assa‐Zag province authorities

50

2.5 SOMECOMMENTSABOUTTHESOLARPERGDEVELOPMENT

Apart from the success of the PERG programme in terms of achieved objectives, the Solar‐PERG did

not achieve the initial goals (only 42% of SHS were installed) and the private stakeholders involved

had to face to some extrinsic difficulties:

The expected objectives as regards the number of SHSs that must be installed were not

achieved because of lack of potential customers.

The reduction in the number of SHSs had a negative impact in the O&M management

whose costs were significantly increased.

The PERG programme developed by means of grid connections became strong competition

for the Solar‐PERG as people preferred the power grid instead of the SHSs and local

authorities supported it, in terms of financing.

The public‐private‐partnership (ppp) on which the fee for service model was based,

devolved the Solar‐PERG promotion tasks to the ESCOs, whose corporate images were very

weak, as regards the rural population. They may have known the ONEE (the National

Utility) but they showed distrust towards the ESCOs.

The SHS cost increased dramatically during the period 2006‐2008 due to prices of many

raw materials increasing considerably, such as the silicon (the main component of the PV

modules), the lead (the main component of the batteries) and the copper (the main

component of the electric wires).

The Solar‐PERG contract signed between the ESCOs and the ONEE was very unfavourable

for the ESCOs, due, for example, to the lack of price revision or the economic

compensation clauses.

Although these problems have been specific of the PERG programme and they cannot be

generalized to other experiences, it is worth mentioning that these kinds of problems are intrinsic

to the public–private‐partnership projects and they may happen in any stage of the programme.

As a result of the aforementioned comments, most of the ESCOs have been forced to leave the

programme due to financial imbalances in the management of the O&M phase. In a few regions,

some local technicians and subcontractors continue to operate the Solar‐PERG customers and their

SHSs. In many others they have been completely abandoned.

Nowadays most of the dwellings electrified by the Solar‐PERG are being connected to the national

power grid, thus, it is predictable that in a short time, the grid will cover almost 100% of the

Moroccan countryside and the SHSs will become obsolete.

2.6 THEISOFOTON‐PERGDATABASE

The experimental work of this thesis has been based on the data taken from the Solar‐PERGISO, in

particular from the maintenance and cost databases that the ESCO drew up during the first five

years of the programme development. These databases were established through real data

extracted from the day‐to‐day activity as feedback for the management of the programme.


51

The advantage that the author was part of the ISOFOTON management staff during these five

years, has allowed, on the one hand, full access to the databases and on the other, to have the

necessary know‐how and capabilities to properly interpret the collected data for further analysis.

2.6.1 THEMAINTENANCEDATABASE

It was made up of all kind of maintenance incidents such as failures of the SHS components, repairs,

device replacements or preventive actions. Every maintenance activity recorded had the date of

the activity, the complete SHS location data (personal data of the customer, contract number,

name and code of the village, rural community and province), a description of the maintenance

activity and all possible devices repaired or replaced (detailing the serial number of the new and

replaced devices) and the name of the technicians that carried out the action.

The ESCO used an ad‐hoc management software elaborated by CEGID [166] to manage all the

Solar‐PERGISO data. The maintenance database was integrated into the software functions.

Between the programmes began at the end of 2005 and November 2010 the maintenance

database had cumulated more than 80,000 maintenance inputs out of the 13,600 SHSs installed.

The compiling of the database was made in accordance with the following description:

The maintenance technicians visiting the SHSs, either because of the preventive

maintenance schedule or after the user failure warnings (corrective maintenance). During

the visit, the technicians made an inspection of the SHS operation and diagnosed the

possible faults. All this information is written in a maintenance datasheet as shown in

Figure 23. So, after every maintenance visit, a maintenance datasheet is generated.

The maintenance datasheets are transmitted to the headquarters administration to record

the maintenance actions onto the database.

The corrective inputs are checked and validated with another database: the spare‐part

stocks management, which the ESCO details in parallel within the overall management of

the programme.

2.6.2 THECOSTSDATABASE

It comes from the economic and financial management of the ESCO during the same period. From

these data the ESCO carried out the accounting of the company, so it is a complete and detailed

database compiled from the real costs of the the Solar‐PERGISO.

52

Figure 23: Maintenance datasheet model used by the ISOFOTON technicians after maintenance activities

CHAPTER 3

RELIABILITY ASSESSMENT OF SHS

COMPONENTS

54

3 RELIABILITYASSESSMENTOFSHSCOMPONENTS

3.1 INTRODUCTION

The reliability of the SHS components is a key factor in the design of the sustainable maintenance

structures of the PVRE programmes. However, perhaps because of the intrinsic difficulties in

gathering data in decentralised frameworks, available literature does not offer real data, thus

impeding the qualification of the reliability. Moreover, manufacturers do not usually provide any

information on failure rates or mean time to failure of the devices, only lifetime estimations, for

example the number of switch on/off cycles that a lamp can resist, the characteristic life in cyclic

use of batteries, or the annual power loss rate of the PV modules. However this kind of information

is based mainly on laboratory experiments and cannot be transferred to a programme of thousands

of devices in operation and, most importantly, where other factors are involved in their

performance, for example the user factor.

In fact, what is relevant in PVRE programmes with thousands of SHSs and long maintenance periods

is to know how the SHS components fail and how frequently they do so. This information is useful

for spare parts forecasting which is one of the pillars that allows optimized maintenance structures

to be designed. To cover this deficiency, this work has carried out a reliability study based on the

maintenance data obtained from the 5 years of operation of the Moroccan PERG programme.

The source of the data corresponds to the ISOFOTON ESCO, that has more than 80,000 corrective

and preventive maintenance data inputs, collected between October 2005 and June 2010.

The questions about which of the SHS’s components break down together with the frequency will

be answered by analyzing the time distribution of failures of every SHS component, using concepts

of reliability engineering and by determining the reliability function R(t), the failure rate λ(t) and the

Mean Time To Failure (MTTF).

3.2 RELIABILITYANALYSIS

3.2.1 SHScomponentsandfailureclassification

The description of the SHS components is summarized in Chapter 2.4.1 and Table 10.

3.2.1.1 Photovoltaicmodule

The PV module corresponds to a 36 series‐associated monocrystalline cells with a 12 VDC nominal

rated voltage. These modules usually have a very high reliability index and generally few problems

are associated with them. Failures in PV modules can be linked, generally, to a poorly‐made

electrical contact or a flaw in the by‐pass diode [167]. It must be noted that the module problems

are frequently related to a difference between the peak power value shown on its label and that

attained in experimental practice. In any case, this problem does not involve a failure in the

operating system. The expected faults in PV modules are as follows:

Chapter 3: Reliability Assessment of SHS Components

55

Diode failures. This failure can cause a malfunction in the solar module thus cutting off the

operation of some or all PV cells. If a PV module does not work for few days, the battery will not be

charged and the electricity supply will not be available. After some days of an inoperative solar

module, the battery self‐discharge can give rise to a deep discharge and it will be irreversibly

damaged. This battery failure is known in reliability engineering as a secondary failure, as it is

caused by the failure of another component (main failure). This kind of PV module failure is usually

repaired by replacing the damaged diode; hence, it is a repairable failure.

Broken PV module. Hail, heavy storms, vandalism, etc, can cause PV module breakage and,

therefore, it is a catastrophic failure.

Thermographic defects. There are sometimes hotspots in the solar cells, and even in the

interconnections between cells (bus), as a result of defects in the manufacturing process, which can

be detected by infrared thermographic cameras. The affected cell becomes reverse biased and

dissipates power in the form of heat [168, 169], which can destroy the PV module.

3.2.1.2 Chargecontroller

The charge controller uses a Pulse Width Modulation (PWM) system to regulate the charge and

discharge of the battery from the PV module toward the loads. Its maximum admissible current is

15A during charge and discharge. This device protects the battery by keeping it working within a

prescribed voltage range for an optimal use. It also shows the state of charge (SOC) of the battery

by means of LEDs located at the front of the device, which shows the SOC battery level. The LEDs

also warn of system malfunctioning, usually by showing a red LED when a deep battery discharge,

overload or a short circuit occurs. The charge controller is an electronic device, so its faults occur

randomly [167]. This means that the failure rate is constant with time. This behaviour is usually

described by exponential distributions of failures, which will be explained later. The failures in this

device will be non‐repairable (catastrophic failures).

3.2.1.3 Battery

The battery is a 12V modified Starting Lighting Ignition (SLI) lead‐acid battery which is

manufactured in Morocco and its technical report shows a capacity of C20 = 150 Ah (at 1.8V

minimum discharge voltage per cell and at 20ºC).

The lead‐acid battery lifetime is limited by its ageing effects, leading to decreasing capacity and

decreasing efficiency, giving rise to higher inner resistance or even to total breakdowns. Sudden

failures (catastrophic) can occur in batteries but they are less significant than failures through

continuously ongoing processes. The main causes of ageing are: anodic corrosion, sulphatation,

degradation of the separator, growth of dendrites and loss of inner surface in the negative

electrode [170]. Then, the ageing effect causes degradation failures. According to the standard IEC

60896‐11:2002(E) [171], the lifecycle of the battery is considered before its residual capacity drops

below 80% of its nominal capacity. However, in the real case of field trials, a residual capacity of

less than 80% may be a satisfactory battery performance for some users. Some studies indicate that

in an SHS with 3 – 4 days of theoretical autonomy, battery degradation will become noticed by the

user (in the sense of less availability of energy) when the useful ampere‐hour (Ah) have decreased

up to one‐third of the nominal battery capacity [172]. The user behaviour with regard to the SHS

operation will be closely linked to the battery cycle life.

56

3.2.1.4 Lowpowerconsumptionlamps

The SHS includes three 7WDC lamps and one 11WDC lamp. These devices are made up of a

fluorescent tube and electronic ballast. The life time for lamps is usually measured as the maximum

number of cycles (switch on/off) that lamps can resist. The Universal Technical Standard for Solar

Home Systems [173] has fixed this resistance to at least 5,000 cycles. It may be noted that, unlike

the rest of components, lamps have a discrete operation. They do not work continuously, but by

cycling. However, we do not consider this difference for the reliability analysis and we will use the

time variable instead of the cycle parameter. The expected failure types in lamps are, on the one

hand, the random electronic failures in the ballast, and on the other hand, the ageing of the

fluorescent tube.

3.2.1.5 Electricalkit

The SHS electrical installation is made up, in addition to the devices described above, of wires, light

switches, connection box, bulb sockets, a DC plug, etc. These components may also be the cause of

system failures as a result of manufacturing defects or installation mistakes. However, we do not

consider these failures because of a lack of data in the database, but we are aware of the

significance of what this SHS failure causes and on the need to study it in more detail.

3.2.2 Thesourceofdata

The database has been updated daily from the corrective and preventive maintenance activities

carried out by the ESCO technicians. It includes details on the failures of batteries, PV modules,

charge controllers and low power consumption (LC) light lamps for every SHS. These failures are in

most cases catastrophic failures, but we must take into account the following considerations:

The PV module failures resulting from diode breakdowns are considered repaired failures,

and not catastrophic failures.

The batteries replaced by ageing effects will be treated as catastrophic failures.

Before applying the reliability analysis, it was necessary to arrange the database by carrying out a

data debugging to remove the invalid inputs, such as mistakes in dates, erroneous PV identification

codes, or non‐representative samples resulting from insufficient data. The database’s figures, after

debugging, are shown in Table 11. It is worth noting that the number of SHSs considered in the

study is still high (7,595 after debugging).


57

Table 11: Recap of the maintenance database after debugging

Database Data inputs after debugging

Number of SHSs 7,595

Corrective & preventive Maintenance data 44,070

Failures

Batteries 714

PV modules 20

Charge controllers 433

7 W lamps 2,337

11 W lamps 755

We have worked with a very large sample in which there are 44,070 maintenance inputs related to

failures, as well as survivors (devices that have not failed during the period considered).

3.2.3 Reliabilityconcepts

The operating life of each component can be determined based on the failure database. After 5

years of operation, there are some components which have failed (failure data) and others which

have survived (the so‐called suspended or right censored data). The failure and suspended data can

be used to determine a probability density distribution (pdf) of failures, hereinafter referred to as

failure distribution f(t), from which we obtain the cumulative distribution function F(t) [174, 175]:

t

dttftF0

)()( (1)

F(t) is the probability that a component will fail before time=t. On the other hand, the reliability

function R(t) can be defined as the probability of a component surviving for a time interval. It is

given by the complementary expression of the cumulative distribution function F(t):

)(1)( tFtR (2)

The failure rate λ(t) is the frequency with which a system or component fails. Its function

represents the conditional probability of failure in the interval (t , t + dt) of that component, given

that there was no failure before time ≤ t. It can be expressed as number of components failing per

time unit. Its mathematical expression is:

)(

)()(

tR

tft (3)

58

Finally, the Mean Time to Failure (MTTF) can be defined as the expected value of time until failure.

It measures the average time between failures with the assumption that the failed system is not

repaired.

0

)( dttftMTTF (4)

3.2.4 Distributionfit

One of the most successful models used in reliability engineering is the Weibull distribution

because of its versatility in fitting many different failure models. The Weibull probability density

function is shown in Table 12.

Depending of the shape parameter β value, the trend of λ(t) will be decreasing for β<1; constant for

β=1; and increasing for β>1. Once f(t) is known, the other reliability functions ‐ λ(t), R(t), and MTTF ‐

can be obtained as shown in Table 12, where:

)

11(

is the Gamma function )(x for )

11(

x

When β comes close to 1, the reliability distribution R(t) approaches an exponential function and

the failure rate becomes constant (λ(t)= λ). This distribution is usually a good fit for electronic

devices, which follows a random model of failures independently of time. The exponential reliability

functions are also detailed in Table 12, where γ is the location parameter and means that the

failure distribution begins at t = γ. Note that if γ=0, then MTTF=1/λ, and R(t) for t=MTTF is:

368.0)( 1 etR . This means that the survival probability for t=MTTF and γ=0 is 36.8% [176].

On the other hand, if Weibull’s scale parameter β≈3.5, the failure model approaches the Normal

distribution. It means that there is a dominant failure mechanism, for example ageing, even if other

mechanisms intervene in the causes of the failure. The Normal functions appear in Table 12, where

θ represents the mean and σ is the standard deviation. In this case, the survival probability for t =

MTTF is 50%, and the mean θ will correspond to MTTF [175].


59

Table 12: Expressions of the failure distribution f(t), the failure rate λ(t), the reliability function R(t) and the mean time to failure MTTF for the Weibull, the Exponential and the Normal distributions. Note that

Exponential and Normal are specific cases of the Weibull distribution

Function Weibull (α, β, ϒ)

Exponential (α=1/λ, β=1, ϒ)

Normal (θ, σ)

f(t)

)(

1)()(t

ettf )()( tetf

2)(2

1

2

1)(

t

etf

λ(t) 1)()(

tt

dte

e

tt t

t

0

)(2

1

)(2

1

2

2

2

11

2

1

)(

R(t)

)(

)(t

etR )()( tetR dtetR

t t

0

)(2

1 2

2

11)(

MTTF )1

1(

MTTF

1MTTF dtetMTTF

tt 2)(2

1

0 2

1

Parameters

α: scale parameter

β : shape parameter

γ : location parameter

λ : failure rate

γ : location

parameter

θ : mean

σ : standard deviation

3.2.5 Failuredistributionfit

In order to get the best fit for an experimental failure distribution, the failure and suspended data

need to be put in order and the cumulative probability calculated. The accuracy of this distribution

can be improved by calculating the median ranking of cumulative percentages. This approximation

can be given by the Bénard estimation [174]:

4,0

3,0

n

iri

(5)

Where ri is the median rank for each failure, i is the ith order failure value and n is the sample size.

The median ranks are an estimate of the unreliability for each failure.

After the ranking process, the data will be ready to be plotted looking for the best fit. We will

determine how well the data fit an assumed distribution by using some of the many statistical

indices that measure the goodness of fit. One of these methods is the least square test. The

goodness of fit as derived by this method is called the correlation coefficient (ρ). The closer the

value ρ is to 1, the better the fit [174].

60

Once the reliability distribution is defined, the characteristic parameters can be obtained for every

distribution, and then, the reliability equations will be solved.

3.3 ANALYSISOFTHERESULTS

3.3.1 Distributionfitting

Different fit distributions have been tried for each SHS’s component, obtaining the following

outcomes:

i. The charge controller has the best fit with the one parameter (γ=0) exponential

distribution. Figure 24 shows the exponential fit. The two external lines mark the

confidence bounds of 95%.

ii. LC lamps have the best fit with a two‐parameter exponential distribution. Figure 24 shows

the 7WDC LC lamp fit, with a confidence level of 95%. The result is that the failure rate is

λ=5.96%/year. Note that the straight line does not start in coordinate (0, 100), but begins at

a time before zero (γ= ‐0.294 years). This fact can be interpreted because of the first points

appearing at the beginning of the failure distribution on the plot paper follow a different

distribution, which may be due to an early failure period (β<1). However, its impact on the

fitting result is imperceptible, since the correlation coefficient shows a very high value close

to 1. As regards the 11WDC lamp, it shows the same behaviour as the 7WDC lamp. It has a

location parameter γ=‐0.288 years and the failure rate figure is 5.97%/year. The two‐

parameter exponential fit in lamps, in contrast to the one‐parameter exponential charge

controller fit, can be explained by the fact that the failures in lamps present a residual

infant mortality. As shown in the graph, this period is very short, which indicates an

acceptable quality control made by the manufacturer.

Figure 24: Left: Failure distribution for the charge controller; Right: Failure distribution for 7WDC LC lamp. Both devices fit an exponential function (logarithmic scale on the y‐axis). In the case of the charge

controller, the exponential distribution is a one‐parameter function. For lamps, the distribution fits a two‐parameter exponential model. Both show a very good correlation coefficient (ρ) and a low uncertainty for

95% confidence bounds.

iii. The battery has the best fit using the normal distribution. This is coherent with the fact that

the main cause of failure is ageing. The correlation coefficient is perceptibly lower than the

precedent ones, but its figure remains very close to 1. However, there is a range of initial

failures which do not fit the normal distribution. In Figure 25, where F(t) is represented


61

instead of R(t), we can see that the failures until t~1.4 years, out of the Normal fit straight

line, have a different tendency. By trying a mixed Weibull distribution, shown in Figure 25,

we successfully fit the distribution until t=1.4 years with a shape parameter β figure of less

than 1. This means that the failure rate during this period decreases over time, which is to

say, the failures are the result of an infant mortality. In what follows, we will maintain the

Normal distribution as the best fit, although the first year behaviour has forced us to carry

out a more detailed analysis which is presented in Chapter 4.

Figure 25: Left: Normal fit of battery failure distribution. The Normal probability plot on the y‐axis represents the cumulative distribution function F(t); Right: Weibull fit. A slope change can be appreciated in the fit distribution line in t≈1.4 years. The shape parameter β in this stretch line is <1, hence, the failure rate has a decreasing tendency. It is important to note that the failure distribution does not fit the Normal function exactly in the first period until 1.4 years. After that, the Normal has a high goodness of fit and the correlation coefficient (ρ) presents a figure very close to 1 for the adjustment of all of failures altogether

iv. Finally, the PV module's reliability evaluation has been achieved from the data detailed in

Table 13:

Table 13: PV module failure figures: Left: figures from the maintenance database; Right: some failure information gathered from the maintenance technicians

PV failures declared in the database

Number of failures

PV failures declared by the maintenance technicians

Number of failures

Replacement but unidentified failures

20 Diode failures 3

Broken PV modules without replacement

5 Breakage of the module through natural causes

9

Unidentified failures. No replacement

15 Breakage of the module through human causes (vandalism)

1

Sum 40 Hot spots at the junctions between cells

6

62

There have been 20 replacements of PV modules resulting from catastrophic failures. In addition,

another 20 failures have been declared even though they have not been replaced. Otherwise, after

an interview with some of the ESCO maintenance technicians, we found that other different types

of PV failures have occurred. The diode failures and hot spots in the cell bus were usually repaired

“informally” by the technicians in the field (they access the cell bus through the tedlar layer, and

then they weld it). The broken PV modules were not replaced, sometimes the users’ request a

termination of the contract, perhaps because the users do not want to assume the obligation of

paying for a new PV module when the breakage was the result of human misuse.

After the evaluation of these figures, we have concluded that there is not enough information to

characterize PV module failures (we only have 40 failure inputs). The evaluation of the failure rate

for the PV modules will need more details on the causes of the failure and an in‐the‐field peak

power degradation and thermographic analysis. An in‐depth analysis on this is presented in Chapter

4.

3.3.2 Reliabilityfunctionsofcomponentsandsystem

The parameters (σ, θ) for Normal fit and (γ, λ) for Exponential distribution have been calculated and

are shown in Table 14. We see elevated goodness of fit coefficients (ρ) for each of the fit

distributions proposed.

Table 14: Parameters of normal and exponential reliability functions, correlation coefficients and MTTF of the SHS components with 95% confidence bounds

Parameter BATTERY CHARGE

CONTROLLER 7W LAMP 11W LAMP

Reliability

Normal

θ (years) 5.46 ‐ ‐ ‐

σ (years) 2.27 ‐ ‐ ‐

Exponential

λ (% / year) ‐ 3.67 5.96 5.97

γ (year) ‐ 0 ‐0.294 ‐0.288

Goodness of fit Correlation

coefficient (ρ) 0.9762 0.9973 0.9939 0.9954

MTTF (years).

95% confidence bound

5.5

± 3.4%

27.2

± 9.5%

16.5

± 4.0%

16.5

± 7.0%

The R(t) and λ(t) functions have been determined from the parameters in Table 14 and equations

from Table 12. We can see in Figure 27 that lamps and charge controller reliabilities are greater

than battery reliability after the 2nd and 4th operating years respectively. Table 14 shows that the


63

two lamps have an identical failure rate; (λ≈5.9%/year) being a higher value than the charge

controller (3.67%/year). The battery failure rate, however, increases over time, typical behaviour

when an ageing process is predominant.

The reliability of the system has been calculated according to the following series model diagram

(Figure 26):

Figure 26: Block diagram representing the SHS series reliability model made up of 7 independent components. It is assumed that lamps work in a series model, because if just one lamp fails, one of the

household rooms will not have lighting, hence the system will have failed

It is assumed that the SHS is made up of 7 independent components. The system fails when one of

the components fails. The main goal of an SHS is to provide energy to the loads (lamps and small

household appliances). If the PV module or charge controller does not work, the battery will

become damaged through lack of charge or/and deep discharge. If the battery capacity crashes,

there will be no energy available to feed the charges, hence, the system fails. As regards the lamps,

they are not working in series within the system, but we have assumed that when one of the lamps

fails, because of the lack of service in the room where this lamp worked, then the system fails.

Usually, when 1 lamp fails, a wide area of the dwelling will have an absence of lighting, and we have

considered this fault as an overall failure of the system. Hence the proposed SHS model operates

according to a series of system lamps.

As regards PV module reliability, it is considered to be close to 1 [177] in order to calculate the SHS

reliability function as the product of the individual component probabilities of survival [178]:

RSHS(t) = ΠRi(t) = RPV(t) RBAT(t) RCC(t) R7WL(t) R7WL(t) R7WL(t) R11WL(t) (6)

The RSHS(t) function is plotted in Figure 27. It shows an exponential tendency and a very negative

steep slope resulting from the battery function effect. On the other hand, the SHS series failure rate

λSHS(t) function is the sum of the individual component failure rate, as shown in Figure 28:

λSHS(t) = ΣRi(t) = λPV(t) + λBAT(t) + λCC(t) + λ7WL(t) + λ7WL(t) + λ7WL(t) + λ11WL(t) (7)

PV module

RPV(t)

11WDC Lamp

R11WL(t)

7WDC Lamp

R7WL(t)

Pb‐acid Battery

RBat(t)

Charge Controller

RCC(t)

7WDC Lamp

R7WL(t)

7WDC Lamp

R7WL(t)

64

Figure 27: Charge controller, battery, 7 & 11 W lamps, and series system (SHS) reliability functions

Figure 28: Charge controller, battery, 7 & 11 W lamps and series SHS failure rates

Finally, the MTTF figures for every component are shown in Table 14. This parameter has been

calculated by taking into account a 95% confidence boundary [179]. Note that the battery is the

component with the lowest MTTF value (5.5 years ±3.4%) which is in accordance with its reliability

and failure rate functions. On the other hand, the most reliable component is the charge controller

(27.25 years ±9.5%). Finally, the lamps have a very similar MTTF value between them (16.5 years

±4% for 7W lamps and 16.5 years ±7% for 11W lamps). It is important to note that the MTTF

concept has a different meaning according to the distribution model chosen. In the case of the


65

battery, whose failure distribution fits a Normal model, 5.5 years is the time in which 50% of

batteries will have failed. In the case of the charge controller, the MTTF figure means that 36.8% of

devices will have survived after 27.25 years. The case of the lamps, whose distribution model is a

two‐parameter exponential function, is similar to that of the charge controller, except that the

location parameter γ intervenes in the MTTF mathematical expression in accordance with the

equation from Table 12. Therefore 36.8% lamps will fail before 16.5 years for both 7 and 11WDC

lamps.

As regards the accuracy of the calculated MTTF, it can be observed that the confidence bounds are

very close (less than 10%), therefore, their values have a low uncertainty.

Some authors [180] have studied the mean lifetime of charge controllers and solar batteries based

on standard reliability level electronic parts belonging to charge controllers. These studies have

concluded with MTTF results of between 30 – 40 years for charge controllers and 6 – 10 years for

solar batteries. These theoretical results, based in exponential distributions, are significantly higher

than the experimental results shown in this work, but they may serve as a comparative reference

for the MTTF ranges achieved in our study.

3.4 APPLICATIONEXAMPLE

The reliability, and therefore the accumulated cost, of a PVRE programme is directly linked to the

material’s repairs and replacements when some of the devices fail. The ESCO is forced to draw up a

forecast for the maintenance period in order to calculate the optimal stock of spare parts, the

number of technical teams and how many vehicles will be necessary, among other things. Then, an

application example of the resulting reliability functions is proposed in order to determine the

annual stock of spare parts required for a hypothetical 100,000 SHS programme with a 10‐year

maintenance period. The results, in terms of number of units of the spare‐part stock per year, is

shown in Table 15:

Table 15: Maintenance period forecast for a 100,000 SHS PVRE programme with 10 years of maintenance. Figures represent the device‘s units required per year

Year 1 2 3 4 5 6 7 8 9 10

Charge Controllers

3,670 per year

7W Lamps 17,880 per year

11W Lamps 5,970 per year

Batteries 2,700 5,910 10,834 16,704 21,677 23,497 21,801 18,82 16,618 16,855

As shown in Table 15, charge controllers and lamps have a constant failure rate, unlike the battery’s

spare parts, which changes every year according to the normal distribution of failures. The annual

evolution of the quantities of spare parts is represented in Figure 29:

66

Figure 29: Annual evolution of the spare‐part requirements for a 100,000 SHS PVRE Programme with 10 years of maintenance

3.5 CONCLUSIONS

Through this reliability study of the Solar‐PERGISO programme, the failure distribution of every SHS

component has been evaluated in order to obtain its reliability function R(t), its failure rate λ(t) and

its MTTF. The results achieved show the following conclusions:

Charge controller and light lamp failure distribution can be established by an Exponential

function. The 2 types of lamps (7 and 11 WDC) show an identical behaviour as regards their

reliability function and failure rate and MTTF parameters.

Battery failure distribution is better established by a Normal function. Although the failures

in the first 1.4 years do not fit the normal model exactly, the whole failure distribution has a

high goodness of fit and it presents a correlation coefficient close to 1. In these first 1.4

years, failure distribution fits a Weibull model with a scale parameter β of less than 1;

hence the failure rate in this period has a decreasing tendency, corresponding to infant

mortality causes.

As regards the PV modules analysis, the maintenance database does not give enough

information about the failure mechanisms of this component. The failures found were: 3

damaged diodes, 10 frontal glass breakages and 6 hotpots at the junctions between cells.

The resulting R(t) and λ(t) functions have been shown for each component and for the

whole system. It is important to note that battery is the main limiting factor as regards the

reliability of the system, since its reliability figures are much lower than those of the lamps

or charge controller.

The MTTF results show, on the other hand, that the charge controller is the most reliable

component (about 27.2 years ±9.5%) and the least reliable is the battery (5.5 years ±3.4%),

while 7WDC and 11WDC LC lamps have a similar MTTF value (16.5 years ±4% for 7WDC lamps

and ±7% for 11WDC lamps). In these results we realize again that the battery has a low

MTTF value as opposed to those of the other components.

‐

5

10

15

20

25

0 1 2 3 4 5 6 7 8 9 10

Battery Charge Controller7W Lamp 11W Lamp

Nº of spare parts / year x 103


67

These results allow the maintenance structure in PVRE programmes to be characterized. The

calculation of the spare‐parts stock over a period of 10 years for a 100,000 SHS programme has

been shown as an example. It can also be very helpful when a quality improvement process needs

to be carried out. This study shows that an improvement in quality of the battery could increase the

system’s reliability if the battery performance reaches that of the lamps or charge controller.

Finally, the peculiarities of the study results regarding the low reliability of the batteries and the

few failures of the PV modules have been the cause of a greater in‐depth analysis of these two

components, which is detailed in the next Chapter.

CHAPTER 4

IN-THE-FIELD ASSESSMENT OF

BATTERIES AND PV MODULES

RELIABILITY IN THE PERG PROGRAMME

70

4 IN‐THE‐FIELDASSESSMENTOFBATTERIESANDPVMODULERELIABILITYINTHEPERGPROGRAMME

4.1 INTRODUCTION

The reliability study presented in Chapter 3 has shown that lamps and charge controllers have a

constant yearly failure rate of 6.0% and 3.7% respectively. Corresponding MTTF values are 27 and

17 years. However, the battery, affected by ageing factors, presented a variable failure rate and its

mean time to failure (MTTF) was 5.5 years.

A further analysis of the failure database showed that it only included the completely dead

batteries that required replacement by new ones. International technical standards consider that

lead‐acid batteries are dead when the remaining capacity drops below 80% of their initial capacity.

Nevertheless, in PVRE, batteries operate even when the remaining capacity is much lower than

80%, namely until the user perceives that the battery must be replaced. Consequently, further

studies to analyse the capacity degradation of the batteries and to establish their real lifetime and

reliability are required.

On the other hand, PV modules were not analysed in the previous reliability study because of the

very low failure data found in the database: just 20 failures out of 13,600 installed PV modules. This

low failure rate did not allow a proper reliability study to be carried out, but it indicated that PV

modules have a very high reliability as regards the other components.

This Chapter describes two in‐field assessments carried out in the Moroccan Solar‐PERGISO

programme to determine both, the capacity deterioration of the batteries and the quality status of

the PV modules when operating under real conditions in a sample of 41 SHSs from 6 provinces

(Figure 30). In 32 of these dwellings a datalogger has been installed to record the battery’s

operating parameters.

As already mentioned, the Solar‐PERGISO region is located partly in a mountainous area (50% of the

region is covered by mountains), and partly by wide desert areas. Climate varies between the

different areas, but the continental climate predominates with very hot summers and cold winters

(it becomes very dry and extremely warm during the long summer, especially in the southern

lowlands). In the mountainous regions rainfall can reach up to 350 mm per year and snow also falls

in winter. Precipitation in the south can reach 100 mm per year. Horizontal solar irradiation varies

between 4.7 and more than 5.5 kWh/m2/day as yearly average. This allows batteries and PV

modules to be tested under very different climatological conditions.

Chapter 4: In‐Field Assessment of Batteries and PV Modules Reliability in the PERG Programme

71

Figure 30: Moroccan PERG region divided into 12 provinces (shaded area). The table indicates the battery, datalogger and PV module sample distribution in the different provinces.

4.2 IN‐FIELDBATTERYTESTING

The battery used in the Solar‐PERGISO is a flat plate, lead‐acid, SLI (starting‐lighting‐ignition)

technology produced in Morocco (data sheet in Table 16). Note that this is the lowest cost battery

technology used in PV stand‐alone applications, and in this case, it reached a cost of 100 €/SHS

(0.06 €/Wh). On the other hand, the SHS's battery was sized for an autonomy of 5 days (tender

condition), which means that considering 150Ah as nominal capacity (Cnom), and a designed depth

of discharge DOD = 40%, it can be deduced that the design load was Ld = 12 Ah/day.

Table 16: Battery data sheet [181]

Storage capacity 150 Ah C20, at 20 °C and 10.8 V (1.8 V per cell) as final discharging voltage

Autonomy 5 days (DOD=40% and daily load Ld=12 Ah)

Electrolyte density 1.26 g/cm3 at 20 °C

The charge controller (CC) is a PHOCOS CML‐15 model, which has a pulse width modulation (PWM)

control including a temperature compensation algorithm. It protects the battery against deep

discharge cutting the load consumption when the battery reaches 11.4 V (at 25°C), as required by

Province BatteriesData‐

logger

PV

modules

Beni Mellal 14 12 0

Azilal 8 8 6

Marrakech 2 2 0

Al Haouz 2 1 0

Ouarzazate 11 9 23

Zagora 3 0 12

Total 40 32 41

72

the Solar‐PERGISO technical specifications imposed by the programme promoter (the ONEE). Other

features can be found in the manufacturer’s datasheet [182].

In order to analyse battery ageing rates, we have substituted existing batteries with new ones at 40

SHSs already in operation for more than 5 years. In this way, energy consumption habits must be

stabilized. 32 of these SHSs have also been equipped with the PHOCOS CX20 model charge

controllers [182], capable of recording past daily energy consumption values.

Several capacity tests have been carried out in these new batteries throughout their operating life

as detailed in Table 17.

Table 17: Different battery capacity tests achieved during their operation under real conditions

Test 1.1 1.2 2 3 4

Month 0

(as delivered by the manufacturer)

0 (after initial full charge)

6 12 18

Tests 2, 3 and 4 have been carried out by temporally collecting the batteries from the households

and putting them together on a testing stand (Figure 31) for around one week, during which the

SHS keeps working through a spare battery. Initial full charging was accomplished by means of

largely available domestic battery chargers, leaving the batteries for at least 12 hours in flotation

after the end of the charge. Then, each battery was discharged through two 50W halogen lamps.

This way, the discharging current, ID, was close to C20 (ID ≈ 2x50 W / 12 V = 8.3A). Discharging was

maintained until the battery voltage VB reached 10.8V. ID, VB, and ambient temperature was

manually recorded every 30 minutes. So, the accuracy of the measurement is about ± 4 Ah. As the

temperature affects the results, they have been corrected 0.5%/°C. So, we estimate an accuracy of

the result at about 6%. The measurements of the electrolyte density were also taken both before

and after carrying out the tests for each battery.

Finally, one of the most degraded batteries was opened up after Test 4, and the plates extracted for

direct visual inspection.

Figure 31: Left: Battery test stand for capacity testing. Batteries are connected to a resistive load that consists of two halogen lamps of 50W each; Right picking up the battery from one of the dwellings of the

sample


73

Figure 32 shows the voltage evolution of one of the batteries in Test 1.2. Note that the voltage

drops slowly during the discharging process but it falls quickly when voltage is close to 11.4 V,

which means that the DOD when the batteries reach the CC low voltage disconnection is practically

100%. This reveals that the protection threshold was incorrect and it was not protecting the

batteries, although, as will see below, this does not seem very relevant, since excessive discharging

rates are not common.

Figure 32: Voltage and energy curves of one of the battery discharging tests. The crosses above the voltage line represent the different readings taken during the test. The dashed line above the light grey area

represents the capacity extracted from the battery when the charge controller disconnects the battery at 11.4V. The dashed line above the thin and darker area shows the remaining capacity of the battery

between 11.4 and 10.8 V.

4.2.1 Capacityresults

Figure 33 shows the battery capacity histograms for the 40 batteries as delivered (Test 1.1) and

after the initial charge (Test 1.2). Figure 34 describes the time evolution of the capacity once the

batteries are in operation (Test 1.2, 2, 3 and 4). Table 18 summarizes the corresponding mean and

standard deviation values.

0

20

40

60

80

100

120

140

160

6.0

7.0

8.0

9.0

10.0

11.0

12.0

13.0

14.0

0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00

Voltage(V)

Voltage (V) Energy (Ah)

Time (h)

Energy (Ah)

11.4 V10.8 V

74

Figure 33: Capacity distribution of batteries from tests 1.1 (after delivery by the manufacturer) and 1.2 (after initial full charge). Note that characteristic capacity distribution of test 1.1 is broader than that of

test 1.2 and its mean value is shifted to the left

Figure 34: Capacity representation of the entire battery sample on the Tests number 1.2, 2, 3 and 4. Note that from Test 2, the average capacity is situated below 80% as regards the nominal capacity

0%

20%

40%

60%

80%

100%

120%Measured Capacity / Nominal Capacity

Test 1.2 Test 4Test 3Test 2

18 months

40%

05

10

15

45%

50%

55%

60%

65%

70%

75%

80%

85%

90%

95%

100%

105%

Test 1.1

Test 1.2

M d i / i l i

Number of bateries


75

Table 18: Corresponding capacity (C20) mean (μ) and standard deviation (σ) for the different tests

TEST μC20test/C20nominal (%) μ C20 (Ah) σ C20 (Ah)

1.1 84 126 15

1.2 92 138 11

2 76 114 16

3 65 97 25

4 62 94 25

The following comments apply:

Low mean and large spread capacity values of delivered batteries suggest, respectively, insufficient

‘formation’ of the plates and poor manufacturing quality. It is worth remembering that

incompletely formed plates contain not only proper active materials (lead dioxide on the positive

electrodes and sponge lead on the negative electrodes) but also remnants of other materials (PbO,

PbSO4) which lead to initial capacities that are below the nominal values. Moreover, these

remnants induce anomalous crystallization around them. Thus, this initial poor quality acts as a

seed for accelerated degradation. This quality problem is also seen in the broader dispersion of the

measured electrolyte density for the total battery set (see Figure 35).

Figure 35: Average electrolyte density distribution from the six cells of each battery in Test 1.2. Measurements were taken before (grey colour) and after (black) the capacity tests. The high dispersion

could be the result of poor manufacturing quality

0

2

4

6

8

10

12

1085

1095

1105

1115

1125

1135

1145

1155

1165

1175

1185

1195

1205

1215

1225

1235

1245

1255

1265

1275

1285

1295

1305

1315

After test

Before Test

Density (g/l)

Frecuency

76

The initial charge further helps the formation of the plates, leading to an increase in battery

capacity. However, half the specimen still shows capacities below 95% of the nominal value, which

is the recommended threshold for SHS battery acceptance (see the Universal Technical Standard

for Solar Home Systems [112]). Large capacity spread still persists.

Not surprisingly, subsequent observed degradation rates, with the batteries already in operation,

are very high. An experimental model of capacity degradation has been derived simply by

calculating the time that each battery is in use to reach the states corresponding to 80%, 70%, 60%,

and 50% of its nominal capacity. For each capacity state, there are some batteries that have

reached that state (it could be defined as "failures") and others that do not reach that state

("survivors") [174]. For instance, there are 21 batteries that reached the degradation of 70% of the

nominal capacity (0.7∙Cnom) at some point during the 18 months of the testing period ("failures"),

and 15 batteries that never reached this degradation ("survivors”).

Figure 36 shows that the unreliability function F(t) of the batteries that got 70% of their initial

capacity fits a Normal distribution. Furthermore, the mean value (or mean time to failure ‐ MTTF) of

each one of the Normal fit distributions at different capacities have been calculated (MTTF80%,

MTTF70%, MTTF60% and MTTF50%). Based on these averages, an experimental degradation function

has been extracted relating the MTTF and the capacity degradation (Figure 37).

Figure 36: Normal fit of the battery’s unreliability F(t), that is to say, the distribution of batteries that reach 70% of their initial capacity as observed in the in‐the‐field battery testing. The two contour lines indicate the 95% confidence boundary of the fit distribution. The correlation coefficient (ρ) shows how well the

linear regression model fits the data. In this case is very close to 1, indicating a good fit.

Failure at Cmeasured / Cinitial = 70%ρ = 0.93

UnreliabilityF(t) (%

)


77

Figure 37: Capacity tendency degradation of batteries extracted from the different capacity tests. When operation time reaches the MTTF = 5.5 years, the remaining capacity of the battery is just 18%. The two

contour lines indicate the 95% confidence boundary of the degradation function

The question of battery lifetime deserves further comment. Worldwide accepted standards [183,

184] indicate that batteries reach the end of their lifetime when the remaining capacity is below

80% of their initial capacity. In adhering to this figure, the degradation function derived here

implies that the SHS batteries concerned would have to be replaced before 1 year of operation.

However, in reality, batteries survive until the user complains about frequent blackouts. The Solar‐

PERGISO experience shows that, on average, this happens according to a MTTF = 5.5 years of

operation. As shown by the previously derived degradation model which corresponds to a

remaining capacity of the battery of around 18% of the nominal value. That is, international battery

standards and SHS user behaviour differ by factors between 4 and 6, when the remaining capacity

of the battery and lifetime are considered respectively. We think that the explanation of this rather

astonishing result relies on the fact that real energy consumptions are well below the standard

design values . Thus, most users are happy with its SHS even if it performs substantially below

specifications.

At this point, it is important to clarify that the capacity that remains in the battery when it has

0.18∙Cini does not necessarily match the available capacity when the battery operates within the

SHS, as evidenced below. Batteries at high capacity degradation levels have unexpected behaviours

due to the elevated internal resistance when current is flowing in the charge and discharge states,

and then, the charge controller’s algorithms are malfunctioning. That is the case when PV module

charges a degraded battery: the voltage in the terminals rises quickly because of a substantial

increase in the internal resistance, and it is interpreted by the charge controller as final charge

voltage, and it starts the floating charge mode even if the battery is not fully charged. Thus, the

charge of the battery by the ensemble PV module – CC does not reach the full charge and then, the

available capacity for the user is even lower than this 18% of Cini.

Finally, one of the most degraded batteries of the sample (it reached a capacity of less than

0.3∙Cnom) was inspected in the laboratory in order to check the actual state of the plates after

0%

20%

40%

60%

80%

100%

0 1 2 3 4 5 6 7 8 9 10

MTTF (years)

Capacity / nominal capacity (%)

MTTF = 5.5 years

Possible definition of death of battery: 18 % of nominal capacity

78

carrying out the Test 4. The electrolyte of the battery was completely removed and then the

container was opened and the positive and negative plates of the cells extracted.

The inspection showed that most of the positive active mass was lost (shedding effect) and was

deposited at the bottom of the cells (see Figure 38). The low physical resistance of the plates to the

mechanical loads occurred during the cycling operation can be the cause of the loss of active mass,

thus bringing about the dramatic decrease in battery capacity and its early death.

Figure 38: Loss of active mass in the positive plates in one of the cells of the battery

4.2.2 Observationsonenergyconsumption

After the analysis of the recorded data from the dataloggers it can be seen that:

Mean daily energy consumption is μload = 5.2 Ah/day, and the standard deviation σload= 4.18

Ah/day, which indicates that there is no standard load profile, however the real loads are

on average lower than the load of design (12 Ah/day).

Despite that, some users are affected by blackouts, associated with the low voltage

protection. This may happen on days in which energy consumption is exceptionally large. In

fact, disconnections at the most affected houses are probably related to consumption

surges during the weekend: 1/7 = 14%, which fits with the batteries that have suffered a

greater number of disconnections (see Figure 40(b)).

Dataloggers indicate that the average battery’s depth of discharge (DOD) ranges from

between 30% and 50%.

The state of charge (SOC) recorded by the dataloggers after sunset is, on average, 70%, while the

percentage of days in which the dataloggers indicate full charge is 74%. The results shown in Table

19 and the histogram in Figure 39 indicate that even if most batteries reach full charge every day,

the mean SOC after sunset is not 100%. Assuming that in general the loads start from the afternoon

onwards, it could be deduced that the charge controllers algorithms that are used to calculate the

SOC and the full charge state of the batteries are distorted by the battery degradation and do not

work properly.


79

Table 19: Statistical results of the SOC in the morning and in the afternoon and days in which batteries are fully charged

Figure 39: Distribution of days in which the batteries reach the full charge as a percentage

Given that the charge cycles of the batteries are characterized by low DOD (normal operation), with

sporadic high DOD and load disconnections (exceptional operation), it can be established that the

capacity degradation is related to both the capacity throughput [185], that is, the quantity of

energy that passes through the battery, and with the number of disconnections due to low battery

voltage that, according to the previous analysis, is not frequent but can dramatically affect the

health of the battery.

In principle, with other circumstances being equal (solar radiation, ambient temperature, charge

controller algorithms, etc.) the higher the energy consumption, the larger the battery degradation.

In fact, this can be observed in reality, despite the anomalous behaviour of the highly degraded

batteries analysed here. Figure 40 shows a) the average daily load (Ah) of each battery during the

18 months of operation; b) the percentage of days with at least one disconnection of the load due

to low battery voltage; and c) the remaining capacity of the batteries after 18 months of operation

ranked from least to most. The eight batteries in dark are those whose remaining capacity are

equal to or lower than 50% of the Cnom

SOC morning

(%) SOC afternoon

(%) Battery full

charged days (%)

Mean (μ) 66 70 74

Median 67 71 83 Standard Deviation (σ) 11 9 25

Mode 61 71 99

0%

5%

10%

15%

20%

25%

30%

35%

20% 30% 40% 50% 60% 70% 80% 90% 100%

Days with full charged battery(%)

Battery frecuency

θ = 74% σ = 25%

80

Figure 40: Bars in the same vertical position correspond to the same battery; a) Average daily load(Ah); b) Percentage of days with low voltage disconnections; (c) Remaining capacity of batteries after 18 months of

operation measured in Test 4. In dark, the 8 most degraded batteries (less than 50% capacity)

The data represented in Figure 40 (a), (b) and (c) can be merged in a mathematical linear

combination such as z=Ax+By, where x represents the percentage of days of low battery

disconnections, y the average daily load (in percentage as well) and z the capacity degradation. A

and B are two parameters calculated for the best linear regression (A = 0.42; B = 0.58) (see Figure

41).

0

5

10

15

a) Daily average load energy (Ah)

0%

5%

10%

15%

20%

b) Days of low battery disconnections (%)

0%

20%

40%

60%

80%

100%

30 24 09 22 36 16 03 02 15 40 21 01 13 06 28 10 11 39 08 38 23 04 29 12 37 14 26 27 31

Code of each battery from the sample

c) Capacity Test 4 (CT4/Cnom)


81

Figure 41: Relationship between the degradation of the batteries (CT4 / Cnom) and the combination of both the percentage of days of low battery disconnections (x) and the average daily load percentage (y) with

respect to a normalized load of 20 Ah. A and B are two parameters calculated for the best linear regression when compared with battery degradation. The results show that A = 0.42 and B=0.58

These results suggest that battery's degradation is affected by the combination of both the normal

and the exceptional operation roughly in the same proportion according to the values of A and B.

The goodness of fit is supported by a R2 = 0.494, which is a representative value when social

behaviours are involved, as is the case in SHS where the users are not only consumers, but also

managers of the systems.

4.3 IN‐THE‐FIELDPV‐MODULETESTING

The PV module model is an ISOFOTON IS‐80S, consisting of 36 monocrystalline silicon cells and, at

the moment of the testing, they had been working for a period of 6.15 years on average.

The testing method consisted of in‐the‐field measuring of the I‐V curve of the PV modules by means

of a capacitive load [186]. This I‐V curve has been extrapolated to standard test conditions (STC) to

extract the peak power of the PV modules. Then, this actual peak power has been compared with

the power of the flash report provided by the manufacturer as a part of the quality control carried

out at the end of the manufacturing process.

During the test, the on‐plane effective irradiance, Geff, and cell temperature, Tc [187], have been

measured from a coplanar reference PV module [188, 189], of the same model as that of the

sample. Thus, uncertainty due to thermal, angular and spectral responses is not an issue. As shown

in Figure 42, after cleaning both modules and waiting for cell temperature stabilization (both

modules must reach similar temperature values), the capacitive load test is carried out when the

effective irradiance is above 800 W/m2 [189] in order to minimize uncertainty.

Linear regression:R² = 0,494

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

0% 20% 40% 60% 80% 100% 120%

Ax+By

Capacity in Test 4 / nominal capacity

82

Figure 42: Outdoor measurement equipment. The PV module below is one of the sample modules, and the upper one is the reference module. Both are stabilized in temperature before testing and they are installed in the same plane. The calibrated cell is used to check that the effective irradiance Geff is greater than 800

W/m2 when carrying out the test

The main statistics of the results are summarized in Table 20 and Figure 43, where ΔIm* is the

variation of the maximum current with respect to that of the flash report and ΔPm* is the maximum

power variation with regard to the same report.

Table 20: Statistics of the PV module degradation, in STC conditions, related to the data from the manufacturer flash report after 6.15 years of operation

ΔIm*(%) ΔPm*(%)

Mean (μ) ‐4.8 ‐6.7

Median ‐5.0 ‐6.7

Standard deviation (σ) ‐1.6 ‐2.0

The test shows a mean reduction of PM* in 6.7%, and the standard deviation σ = 2.0%.

Reference PV module

Sample PV module

Calibrated cell

Capacitive load &

oscilloscope box


83

Figure 43: Distribution of the measured loss of power of the 41 PV modules after 6.15 years of operation

We can assume that the mean annual power (PM*) degradation corresponds to 1.1%. However, PV

modules can experience a premature degradation after the first hours of exposure to solar

radiation. The possible early degradation is included in this yearly rate, since degradation in the first

days of operation was never measured. Considering an early degradation in terms of maximum

power PM* between 0% and 4% [190], we can conclude that the mean yearly degradation goes

from 0.4% to 1.1%, which is consequent with other reported experiences [191, 192].

Inspection of hotspots has been carried out with a thermographic camera (model Infracam FLIR

E60). The results show that only one PV module presented temperature variations (ΔT) between

cells higher than 20 °C. That PV module had a ΔT of around 30°C, reaching a hot‐spot temperature

of 86°C (see Figure 44).

Figure 44: Thermal image of the rear side of a PV module showing a hot cell 30°C higher than the other surrounding cells of the module

0

1

2

3

4

5

6

7

8

9

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0

Freq

uen

cy (nºmodules)

Loss of power (%)

Mean = 6.7%Standard deviation = 2.0%

≈30 °C

84

It has been observed, as well, some visual defects in a few PV modules, such as browning in the

encapsulation ethylene‐vinyl‐acetate (EVA) layer [193, 194], or defects in small areas of the anti‐

reflective coating in one cell [169], but because of its low frequency they have no statistical

relevance in this study.

4.4 CONCLUSIONS

In order to analyse the battery degradation process, a sample of 40 batteries installed in the Solar‐

PERGISO has been periodically tested over a period of 18 operation months. The main results are:

i. On average, the capacity drops to 60% of the initial capacity after eighteen months of

operation. In less than six months the batteries have reached a remaining capacity of less

than 80% of the nominal capacity, the threshold usually used to define the battery lifetime,

which is not the case for PVRE users.

In spite of this fast degradation, past studies based on the O&M database show that

batteries keep working for much longer (MTTF=5.5 years) until the users perceive that the

SHS does not satisfy their needs and ask for the replacement of the battery.

Correlating these two results, a model has been proposed to calculate the remaining

capacity that leads to user dissatisfaction, which is 18% of the initial capacity for this

particular programme.

ii. Based on current recorded data, it has been determined that there is no a standard load

profile, but loads are in general very low as regards the storage capacity of the batteries.

The mean load consumption has been μload=5.2 Ah/day and the standard deviation σload=4.2

Ah/day. It must be noted that the design load was much higher (12 Ah/day), which could

suggest that battery degradation would have been more accelerated if both design and real

loads had been closer. This leads to an operation normally based on a high SOC, with

sporadic events of low battery disconnections. A model that correlates these two types of

operation with the battery capacity degradation has been proposed.

iii. The reasons for this fast degradation of the battery capacity could be as follows:

Bad manufacturing quality of batteries. The in‐the‐field testing has exhibited a rapid

degradation of batteries even working at low loads. Moreover, they have shown a wide

dispersion in terms of capacity and density of electrolyte. Ultimately, the visual inspection

of the plates from one of the most degraded batteries has shown the dramatic loss of

active mass in the positive plate which suggests a very low physical resistance of the plates

The behaviour of degraded batteries (with capacities of less than 0.8∙Cnom) is difficult to

predict, which makes it very complex to adapt the algorithms of the charge controllers.

The threshold of the Solar‐PERGISO charge controller, to protect the battery against over

discharge, allows a DOD close to 100%, which is completely incorrect, although this does

not seem to be very relevant, since excessive loads are not common


85

The results and conclusions reported in this Chapter could lead to the question of the convenience

of using local manufactured batteries in rural electrification programmes. It is assumed that

manufacturing quality depends only on the quality process and raw materials used and not on the

country, but it is also true that quality involves an increase in prices, and therefore, this must be

adapted to the local markets. However, successful experiences have taken place in Brazil and

Bolivia where local batteries have been used in PVRE programmes (according to author's own

sources). Thus, the use of local batteries in general must not be excluded, but compliance with the

minimum quality standards should be tested.

As regards the PV modules, a set of 41 mono‐crystalline modules have been tested in the field to

assess their state of quality, resulting in a mean power degradation rate of between 0.4% and 1.1%

per year, which matches other experiences reported in the bibliography. Defects related to

hotspots, broken diodes, browning and other defects are almost negligible. Hence, silicon PV

modules show good reliability behaviour, especially when compared with the other SHS devices,

such as batteries, lamps or charge controllers.

CHAPTER 5

CHARACTERIZATION OF THE

OPERATIONAL & MAINTENANCE COSTS

88

5 CHARACTERIZATIONOFTHEOPERATIONAL&MAINTENANCECOSTS

5.1 INTRODUCTION

As mentioned above, most PV rural electrification projects based on solar home systems (SHSs) and

fee for service concept [50, 114] have failed because real operation and maintenance (O&M) costs

have been larger than initially expected [120, 160, 158]. Fees lower than the real cost produces

serious financial imbalances, making O&M unfeasible and leading to the abandonment of SHSs by

the ESCOs.

In fact, despite many programmes and subsequent evaluations having been carried out since the

1970s [17, 66, 115, 195, 196, 197, 198, 199, 200], real O&M costs are hardly reported at the

available literature [201], making difficult the task of designing new SHSs programmes with

appropriate maintenance and management structures.

This Chapter presents an assessment and evaluation of the operation and maintenance costs of the

Solar‐PERGISO programme, with the aim of characterizing its cost structure. Based on the data

extracted from the 5‐year operational costs of the ESCO, the programme has been analyzed to take

out the most relevant economical aspects involved in the O&M phase as well as the comparative

appraisal between the 3 main activities: installation, O&M and management.

Remember that the Solar‐PERGISO rules provided that the ESCO must guarantee the SHS for a period

of 10 years, during which it must repair or replace the damaged components. It is also responsible

for collecting the monthly fees that the users must pay for the maintenance service. These fees are

established by the ONEE utility at 4.9 €/month/SHS (excluding VAT), that it is equivalent to 59.0

€/year/SHS. This fee will remain unchanged over the 10 years of maintenance service.

5.2 COSTANALYSIS

We will distinguish 3 main activities (see Figure 45): installation, O&M and general management.

The installation refers all the work and activities required to install the SHSs, as well as the purchase

of equipment and the marketing. The operation and maintenance of the systems requires the

technical maintenance of the SHSs (including spare parts) and the collection of user’s fees.

The general management (ESCO headquarters, management staff, etc) is linked to the others, so it

can be considered as an indirect cost. However, the management has been taken into account as

an independent activity in this study.

The different costs, such as taxes, banking fees, insurances, staff training, office supplies,

contingences fees, financial expenses, transports, customs, etc, are included in these 3 activities

according to their involvement in each one.

Chapter 5: Characterization of the Operational & Maintenance costs

89

Figure 45: Solar‐PERGISO programme activities

It has been necessary to assign every unit cost from the accounting database to the different

activity levels as represented in Figure 45. The cost allocation of the first 5 years of the Solar‐

PERGISO programme according to the different activities classification are shown in Figure 46 and

Table 21: from 2006 to 2008 the activity was mainly devoted to the installation of the SHSs, and

2009 ‐ 2010 to the operation and maintenance service. Obviously, O&M was also carried out from

2006 to 2008 for already installed SHSs.

The total expenses in these 5 years reached the amount of €12.5 million, distributed as shown in

Figure 46. Note that we have considered every expense involved in the development of the

programme, taking into account all operative and financial costs before amortization.

Figure 46: Distribution of actual costs in the first 5 years of programme

0 €

100 €

200 €

300 €

400 €

500 €

600 €

700 €

800 €

900 €

2006 2007 2008 2009 2010

Equipements

Installation

Marketing

Fees collection

Maintenance & Spare parts

General management

Year

€/SHS

End of the installation phase

90

Table 21: Solar‐PERGISO cost balance (units in €/SHS). Note the elevated costs of the O&M activities in 2009 and 2010, where just the cost of collecting the monthly fees represents almost half of the fee collected

Item 2006 2007 2008 2009 2010

Equipment 478 461 454 ‐ ‐

Installation 87 84 81 ‐ ‐

Marketing 94 80 64 ‐ ‐

Fee collection 12 10 8 26 24

Maintenance & Spare parts 15 15 16 33 40

General management 112 54 36 23 18

Total (€/SHS) 798 704 659 82 82

Given the failure rate of the SHS components in this programme (see Chapter 3), and considering

that the total number of SHSs in the O&M phase (13,449) remains practically fixed, it is reasonable

to suppose that the O&M costs in 2010 are representative of the following years. Therefore, we can

calculate the whole cost of the programme after the 10‐year period of O&M just by making the

hypothesis that the yearly discount rate will be similar to the inflation rate in the coming years.

In this way, the Solar‐PERGISO, whose completion year will be in 2018, should have an overall cost of

€21.2 million, which when referring to all the installed systems is €1,574/SHS.

We can see in Figure 47 that the installation costs are similar to the O&M costs for the whole

programme period, and that the major costs are the initial equipment (29.4%) in the installation

phase, and the maintenance activity including spare parts (26.5%).

Figure 47: Overall cost distribution of the Solar‐PERGISO programme for 13,449 SHSs and 1.008 MWp installed (considering 75 Wc/SHS). Percentage distribution of costs for each group of activities

O&M Management9.4%

Maintenance26.5%

Fee Collection14.9%

Equipments29.4%

Marketing5.0%

Installation5.7%

Installation Management

9.1%

Before set‐up

Management (18.5%)

Installation (40.1%)

O&M (41.4%)

After set‐up


91

5.2.1 Installation

We must consider that the equipment costs, during the installation phase, was influenced by two

important facts: 1) the prices of the PV modules went up considerably as compared to the previous

years, reaching more than €3.5/Wp (€262/SHS) in the 2006‐2008 period; 2) the market price of lead

shot up reaching limits greater than 4 times its cost in 2005, which directly affected the battery

costs, in which the cost of the lead is 50% , reaching €0.75/Ah (or €1.51/Wp). So, the purchase price

of the PV kits was around €6.2/Wp (€465/SHS). Note that, at the photovoltaic market current prices

of 2015 (around €0.8/Wp for "small size" PV modules), the cost of this PV kit would be around

€3.7/Wp (€277/SHS). However, this dramatic reduction in PV module cost means just an 11.6%

decrease in the cost of the programme, showing that the cost of PV module is not the key point in

decentralized rural electrification.

5.2.2 O&M

Maintenance includes both spare parts and the maintenance structure. Its cost represents 26.5% of

the overall programme cost, as observed in Figure 47, and it can be expressed as an annual 9.01%

of the equipment investment (€41.7/SHS∙year). This figure is so far from other figures found in

some publications [155, 156, 157], where the maintenance cost is estimated at 1 – 3% per year of

the investment in equipment.

Costs involved directly in the maintenance are shown in Figure 48. We see that the cost of the

spare parts is the most important (€22.9/SHS∙year), especially battery, which represents

€19.25/SHS∙year (according to MTTF = 5.5 years [18]), followed by lamps: €2.93/SHS∙year (MTTF =

16.5 years).

Figure 48: Maintenance cost structure (note that fee collection cost is not included)

The second largest cost of the maintenance is the maintenance structure (€14SHS∙year). This

structure refers to the direct costs of this activity related to staff, offices, stores, vehicles, fuel,

€0.95

€3.86

Battery €19.25

Charge C. €0.61

Lamps €2.93

D. water €0.11

General Maintenance

Structure €3.19

Local Maintenance

Structure €10.81

‐ €

5 €

10 €

15 €

20 €

25 €

Equipments (spare parts)

Maintenance structure

Transports Other costs

€/SHS/year Total maintenance cost:

€41.7/SHS/year

54.9% 9.3%2.3%33.6%

% of the total maintenance cost

92

telephone, etc. These costs are divided between the local structures close to the SHSs, as local

agencies, technical staff, etc; and the general structure located in the ESCO headquarters. Over 75%

of costs are attributed to the local maintenance structure (€10.81/SHS∙year), versus the general

maintenance structure (€3.19/SHS∙year). This data confirms again that the decentralized character

is a very important factor in a PVRE programme: for example, the cost of fuel is as high as

€5.5/SHS∙year, and the annual distance per SHS travelled by the O&M vehicles is 57 km/SHS∙year.

Distilled water (€0.11/SHS∙year) is associated to the open lead battery used in this programme.

However, the cost of the maintenance linked to the distilled water for the batteries is 30% of the

total cost of the battery maintenance (the part of the overall maintenance cost intended for the

battery, which includes the spare parts, a part of maintenance structure, transport and other costs),

and represents 6% of the overall cost of the programme (€9.5/SHS∙year). This activity requires the

SHSs to go at least once a year, and recommendable once every 6 months. This opens the door to

evaluating the convenience of using “free maintenance” batteries, which do not require this

practice.

Within the maintenance activity, the cost of transport refers to the transport of spare parts to

supply the local stores, plus the import of goods which are acquired from outside of the country.

Under "other costs" we refer to the documentation (maintenance forms), financial expenses and

other contingencies.

Another of the major costs of the programme is the fee collection activity (14.9% of the total cost –

or €23.4/SHS∙year ‐ as shown in Figure 47). The method of carrying out this activity is based on the

daily presence of the ESCO staff in the main rural communities from each region, coinciding with

the days in which the local markets (souks) are organized, and demands a great effort in human

capital and mobility, which justifies its high cost.

In addition there is another circumstance which makes the activity even more expensive: the

management of non‐payment. When the user does not pay the monthly fee for more than 3

months, the ESCO agents must go to the user’s dwelling in order to collect the arrears, or even to

rescind the subscription contract and remove the PV system. This activity is very delicate and it

further increases the cost of the fee collection work. After the end of the installation phase, in

2009, the ESCO rescinded 151 customer contracts because of non‐payment of monthly fees.

Considering all of the O&M costs over the whole duration of the project (2006 – 2018), we can get

the average yearly cost of the O&M: €76.03/year∙SHS, just €59.09/year∙SHS being the yearly user

fees. Making a cash‐flow balance for a 5% constant annual discount rate, and given a ideal

collection rate of 100%, we will reach a total deficit of €3.4 million for the O&M activity over the

duration of the whole project.

In order to evaluate the accumulative costs of the main SHS components during the whole duration

of the programme, Figure 49 shows the component cost of installation as well as their replacement

when they fail (initial installation + spare parts).


93

Figure 49: Detail of the general costs during the whole duration of the programme. The equipment costs are made up both of the initial installation and the spare parts of the maintenance activity. Figures in

brackets detail the cost per SHS.

From Figure 49 it is shown that the most important cost of the programme is not the PV module,

but the battery and the general management (both 18.5%). The PV cost of the module, of course, is

significant, but it accounts for just 15.5% of the overall cost. Considering 2015 PV market prices, the

PV module would mean only 4.36% of the cost of the programme.

5.2.3 Management

The relevance of the general management activity within the cost structure (18.5% of the overall

cost as shown in Figure 47 and Figure 49) must be highlighted. The ESCO had designed the

management structure for a 34,500 SHSs programme as initially planned. The lack of potential

customers and the electric grid expansion achieved by the ONEE, dramatically limited the number

of SHSs installed. This fact suggests that the size of the PVRE programmes must be big enough to

support the high cost of decentralized management structures.

5.2.4 Energycost

Finally, we can reach a figure of the electricity cost by calculating the electricity available for the

users for the whole duration of the project. Taking into account that the daily average radiation on

the tilted surface is around 5.5 kWh/d/m2 and considering the performance loses (75%), we obtain

an available overall production of 15.19 GWh for the 13,449 SHSs over 10 years, leading to a cost of

€1.39/kWh ( not so far from the figures of other programmes, e.g. India: US$0.65 – 1.35/kWh [202],

Zambia: US$1.6 – 2.1/kWh [203] or some African countries: US$0.45 – 1.30/kWh [20]). On the

other hand, the tariffs for on‐grid electricity in Morocco goes from €0.083/kWh to €0.137/kWh

18.5%

11,9%

14.9%

5.0%

5,7%

5.5%

2.7%

1.8%

18.5%

15.5%

0% 5% 10% 15% 20%

General management

O&M (without s. parts)

Fees Collection

Marketing

Installation

Miscellaneous

Lamps

Charge controller

Battery

PV moduleINITIAL INSTALLATION + SPARE PARTS (€243/SHS)

(€292/SHS)

(€28/SHS)

(€43/SHS)

(€86/SHS)

(€91/SHS)

(€79/SHS)

(€234/SHS)

(€188/SHS)

(€291/SHS)

94

(2015 prices [121]) according to the quantity consumed, having an initial subscription fee of €227

and a monthly fixed fee of €0.75 [162].

5.3 SENSITIVITYANALYSIS

Figure 50 shows that both, the variation in the battery and PV module costs, have the most impact

on the cost of the total programme. This analysis has been carried out by varying just the cost of

the components, without changing any other parameter. It is remarkable that battery affects

slightly more than the PV module. An 80% variation in the cost of the battery means a 15%

variation in the overall cost of the programme . In the case of lamps and charge controllers, the

variation in costs has much less impact. Note that the fuel for the ESCO vehicles has more influence

than lamps and charge controllers in the sensitivity analysis of costs.

Figure 50: Sensitivity analysis of the SHS component costs and fuel belonging the Solar‐PERGISO programme

Figure 51 represents the cost variation of the programme regarding the variation in the MTTF value

in lamps, batteries and charge controllers, without changing their costs. Obviously, the tendency is

that by increasing reliability, the overall cost of the programme decreases. We can see that battery

reliability has more influence than lamps or charge controllers on the overall cost of the

programme: a 30% increase in battery MTTF leads to a reduction in the programme cost of 5.5%.

‐20%

‐15%

‐10%

‐5%

0%

5%

10%

15%

20%

25%

‐90% ‐60% ‐30% 0% 30% 60% 90% 120%

BATTERY

PV MODULE

FUEL

LAMPS

CHARGE CONTROLLER

∆G

loba

l PE

RG

Cos

t

∆ components cost


95

Figure 51: Sensitivity analysis regarding the reliability variation of the SHS components

Increasing reliability means increasing the cost of the kit components. Therefore, the impact on the

overall cost depends on the device’s cost/reliability relationship.

We can see in Figure 52 the cost/MTTF relationship of each device for keeping the programme’s

overall cost unchanged. This means, for example, if the battery MTTF decreases by 60% regarding

the MTTF figure, and its cost varies by ‐77.5%, the programme’s overall cost would not change.

These cost/MTTF relationships of the SHS components will help us to assess whether a reliability

variation can positively or negatively vary the programme’s overall cost depending on the

component costs that were considered.

Figure 52: Relationship between cost and MTTF in order to keep the overall cost of the programme constant when the MTTF varies. Example: the red dot in the upper right corner indicates a battery price 40% higher than the original, whose MTTF would be 80% longer, which means that the overall cost of the

programme would be lower than by using the original battery

‐120%

‐100%

‐80%

‐60%

‐40%

‐20%

0%

20%

40%

60%

80%

100%

‐100% ‐80% ‐60% ‐40% ‐20% 0% 20% 40% 60% 80% 100% 120%

∆ component MTTF

∆co

mpo

nent

cos

t

BATTERY

CHARGE CONTROLLER

LAMPS

‐15%

‐10%

‐5%

0%

5%

10%

15%

20%

‐90% ‐60% ‐30% 0% 30% 60% 90% 120%

∆

∆G

loba

l PE

R C

ost

BATTERY

CHARGE CONTROLLER

LAMPS

96

For example, it is evident that a higher quality battery will be more expensive, but in addition it

implies a longer life cycle, a greater reliability, and a lower maintenance cost. If the battery used at

the beginning in the Solar‐PERGISO is replaced by a different one with a price 40% higher than the

original, and its MTTF is 80% longer, the overall cost of the programme will be less than by using

the original battery, as shown in Figure 52.

5.4 INFLUENCEOFTHESHSSPATIALDENSITY

The spatial density of SHSs will largely determine the design of the maintenance structure, and will

directly affect some unit costs, such as staff, fuel, vehicles, transport of spare parts, the number of

local agencies and stores, etc.

A management cost analysis of the Solar‐PERGISO has been carried out regarding each local agency,

taking into account the number of SHSs managed by each agency, in addition to the agency staff

and goods, the surface of the local region and its geomorphologic features.

Figure 53 shows that decreasing the SHS density, the management structure cost increases fitting

an exponential function. This is due to the fact that general management costs increase rapidly

when the SHS’s density is very low. On the one hand fixed costs are spread among a smaller

number of SHSs, and on the other hand, the dispersion and inaccessibility of the systems increases

the variable costs.

Figure 53: Management local structure cost per SHS as regards the SHS density for each region. The geomorphologic features have been considered

From this cost/density distribution model, a sensitivity analysis can be carried out to know how the

SHS density affects the overall cost of the programme.

Figure 54 shows the variation of the overall cost of the programme when the SHS density changes

around the density figure. It fits an exponential function. The fact that this variation does not fit a

y = 16.291x‐0.226

R² = 0.8397

0 €

10 €

20 €

30 €

40 €

50 €

60 €

0.000 0.100 0.200 0.300 0.400 0.500

Annual real cost of the local structure / SHS (€/SHS/year). Data 2009 ‐ 2010

SHS density (SHS/km2)

€/SHS/year


97

linear function has a great significance on the decentralized rural electrification and will be a major

factor when designing a PVRE programme.

Figure 54: Sensitivity analysis of the PERG according to the variation in the SHS density

5.5 APPLICATIONEXAMPLE

An application exercise is presented here to illustrate the influence of the results of this Chapter in

the design of PVRE programmes. We will imagine a hypothetical programme with 3 different

characteristics as regards the Solar‐PERGISO.

1) The cost of PV modules is reduced by half.

2) The number of SHSs does not vary, but the programme’s surface will be 10 times smaller than the Solar‐PERGISO area.

3) The battery MTTF will be double that of the current one and its cost will be 75% higher than indicated in the Solar‐PERGISO. It is an open battery that needs distilled water.

Table 22 summarizes the features of the Solar‐PERGISO and the example:

Table 22: Distinguishing features of Solar‐PERGISO and application example programmes

Case Parameter Solar‐PERGISO Application example

1 PV module cost (€/Wp) 3.35 1.675

2 SHS density (SHS/km2) 0.068 0.68

3 Battery MTTF (years) 5.46 10.92

Battery cost (€/Wp) 1.31 2.29

‐10%

‐5%

0%

5%

10%

15%

0.000 0.100 0.200 0.300 0.400 0.500

SHS density (SHS/km2)

0.068 SHS/km2

∆G

loba

l PV

RE

Cos

t

98

The overall programme costs resulting from applying the modification of each of the parameters

(cases 1 to 3) and all of them together (case 4), are summarized in Table 23:

Table 23: Resulting figures (in €/SHS) after applying the 4‐case features. The percentages indicate the cost variation as regards the original costs

Case

Total cost (€/SHS) Installation cost (€/SHS) O&M cost (€/SHS) Management cost (€/SHS)

Example Δ cost /

PERG Example

Δ cost /

PERG Example

Δ cost /

PERG Example

Δ cost /

PERG

PERG 1,574 0.0% 631 0.0% 651 0.0% 291 0.0%

1 1,452 ‐7.7% 509 ‐19.3% 651 0.0% 291 0.1%

2 1,434 ‐8.9% 603 ‐4.6% 541 ‐16.9% 290 ‐0.3%

3 1,472 ‐6.5% 706 11.8% 472 ‐27.5% 294 0.9%

4 1,211 ‐23.1% 555 ‐12.0% 361 ‐44.5% 294 1.0%

It can be shown that case 4 achieves a reduction of 23.1% of the original overall cost. The increasing

of the SHS density (case 2) mainly affects the O&M, reducing its cost by 16.9%. The reduction in the

PV module cost (case 1) just affects the installation activity, and the new battery (case 3) increases

the installation cost by 11.82% but reduces the O&M by 27.50%. The management cost does not

vary more than 1.0%.

For case 4, the distribution of costs is shown in Figure 55. We highlight the following aspects:

The major cost of the programme is not the battery (18.3%) nor the PV module (10%), but the management (24.3%). The halving of the cost of the PV module and the use of a battery with a cost/reliability relationship as we have chosen, is the reason why the cost of the programme is reduced. Even though the installation cost increases, the O&M cost decreases because the number of spare parts, mainly the battery, has been greatly reduced.

We also see that the management cost changes from 18.5% to 24.3%. Management is practically a fixed cost, and its reduction depends on increasing the number of SHSs in the programme. By keeping the number of systems in this example and reducing the maintenance costs, the relative cost of management will increase. This fact implies that it is necessary to improve the design of the maintenance and management structures, which will be the objective of the following Chapter, and to implement extensive rural electrification programmes.


99

Figure 55: Details of the overall costs of the application example as regards the SHS component costs

The cost of energy in this application example has been reduced from €1.3/kWh to €1.0/kWh, and

the maintenance cost is reduced from €76.0/year∙SHS to €45.4/year∙SHS. The example has an

economic profitability of €1.8MM over the programme life.

5.6 CONCLUSIONS

The Solar‐PERGISO programme has been analyzed in order to identify the different costs that have

been involved in its development. Based on the real data costs of the first 5 years of the

programme we have obtained the overall cost for 10 years of maintenance, detailing each activity.

The overall programme cost reaches €21/Wp (or €1,574/SHS), where 40.1% of this cost

corresponds to the installation phase (€631/SHS); 41.4% to the O&M activity (€652/SHS) and 18.5%

to the general management (€291/SHS). The main conclusions reached in this paper are as follows:

Around 50% of the overall cost is invested during the installation phase, and the other 50% during the O&M period.

The O&M of the systems (maintenance, spare parts and fee collection) reaches €76.0/SHS∙year. This figure is further higher than usually considered in photovoltaic maintenance and it is not covered by user’s fees, causing unsustainable financial balances.

The PV module represents just 15.5% in the overall cost (€243/SHS), versus 18.5% of the battery (€292/SHS), so the battery has to be considered as the most expensive component in decentralized PV rural electrification.

The maintenance of the open lead batteries leads to the frequent addition of distilled water, whose overall cost means 6% of the overall cost of the programme (€9.5/SHS∙year).

The SHS dispersion and inaccessibility plays an important role in the cost structure, given by a mean SHS density of 0.068 SHS/km2. This feature affects the maintenance cost. For example, the cost of fuel in this programme represents €5.5/SHS∙year and the annual rate of distance per SHS travelled by the O&M vehicles reaches 57 km/SHS∙year.

24,3%

8,7%

14,2%

4,4%

7,2%

7,1%

3,5%

2,3%

18,3%

10,0%

0% 5% 10% 15% 20% 25% 30%

General management

O&M (w/o spare

parts)

Fees Collection

Marketing

Installation

Miscellaneous

Lamps

Charge controller

Battery

PV module

INITIAL SH

S INVESTEMEN

T + SPARE PARTS

% of the global programme cost

100

The energy delivered by the SHSs, expressed as available electricity for customers has been calculated, reaching a cost of €1.3/kWh.

Finally, we have shown an application example that demonstrates how by taking into account the

results of this analysis, the design of the PVRE programmes can be improved. Therefore, this study

opens the door to the creation of a design tool for costs to formulate future PVRE programmes.

CHAPTER 6

DESIGN OF DECENTRALIZED

MAINTENANCE STRUCTURES IN

PHOTOVOLTAIC RURAL

ELECTRIFICATION

102

6 DESIGNOFDECENTRALIZEDMAINTENANCESTRUCTURESINPHOTOVOLTAICRURAL

ELECTRIFICATION

6.1 INTRODUCTION

It is usual when designing a PVRE programme that sizing the SHS, according to local or international

standards, is often one of the issues most concentrated on, followed by the establishment of the

financial terms. As regards the latter, the incomes from the users play a decisive role when

establishing the fees to be paid for the maintenance service. As has been already mentioned, fees

are usually calculated according to what users pay for traditional lighting (candles, kerosene,

batteries recharging, etc), which is around US$ 3‐10 /month [17, 204, 205], an amount that in many

cases is not enough to cover the ESCO O&M costs.

As it has been shown, the main cause of high maintenance costs is the decentralisation factor that

has been often ignored by the promoters of electrification. In fact, decentralization makes the O&M

costs unknown and uncertain (travelling to repair an SHS can be ten times more expensive than the

repair itself), depending on aspects such as the geographical density of the SHSs, their reliability,

road access, the different local costs (personnel, vehicles, fuel, taxes, etc).

In this regard, the present Chapter aims to develop an ad‐hoc tool for the design of maintenance

structures in decentralised areas where PVRE programmes are going to be implemented, looking

for the minimization of the O&M costs. A mathematical optimization tool, modelled with GAMS

[206], one of the most powerful algebraic modelling languages, and solved with the CPLEX

optimiser [207], has been developed and applied to the Solar‐PERGISO programme in order to

compare the results.

The required input data to create the model has been defined together with several variables and

restrictions. An objective function, whose optimization criteria is the minimum cost of the O&M

activity, is calculated, obtaining the following outputs:

‐ composition of the maintenance structure (number of technicians and vehicles),

‐ location and quantity of local agencies (local maintenance headquarters),

‐ scheduling of the preventive maintenance and fee collection,

‐ the overall cost of the activity.

A prototype model has been implemented and validated in two provinces of the Solar‐PERGISO

region to show the usefulness of the tool when the data on the O&M costs and the reliability of the

SHS is available.

Chapter 6: Design of Decentralised Maintenance Structures in Photovoltaic Rural Electrification

103

6.2 BASELINEDATA

Throughout the contract signed by the three parties (ESCO, ONEE and user), the ESCO committed

itself to maintaining the systems for a period of 10 years, in accordance with the following terms:

The ESCO has to repair or replace every defective or malfunctioning SHS component, within

48 hours after being notified by the user (corrective maintenance).

The ESCO must visit every SHS at least once each 6 months to check the state of the SHS

and to fill the battery with distilled water (preventive maintenance).

The user has to pay the ESCO a monthly fee of €4.9 (taxes not included) corresponding to

the maintenance service.

The ESCO has to maintain a minimum O&M local structure to ensure the aforementioned

service.

The local O&M structure deployed by the ESCO was presented in Chapter 2 and it is shown here for

reasons of clarity (Table 24). In all, the local structure was made up of 9 local agencies, 43

employees and 19 vehicles.

Table 24: Summary of the O&M structure location (local agencies) in the different provinces of the Solar‐PERGISO programme. The provinces’ areas and the SHSs installed are also indicated.

Provinces Province Capital

Area km2

Number of SHSs

Density (SHS/km2)

Local Agency location

TOTAL ‐ 214,531 13,452 0,062 9

Ben Slimane Ben Slimane 2,760 857 0,311 Ben Slimane

Errachidia Errachidia 59,585 959 0,016 Errich

Beni Mellal Beni Mellal 6,638 2,723 0,410 Ksiba

Azilal Azilal 9,800 1,809 0,185 Azilal

Al Haouz – Marrakech

Taghnaout 7,883 862 0,109 Ait Ourir



10,070 4,396 0,437 Ben Guerir

Ouarzazate – Zagora

Ouarzazate 55,298 841 0,015 Ouarzazate

Taroudant Taroudant 16,500 689 0,042 Taroudant

Tiznit – Guelmim – Assa‐Zag

Tiznit 45,997 316 0.007 Tiznit

The provinces are administratively divided in several rural communities (R), comprising a rural

centre (r) and a number of dispersed villages (V).

6.3 METHODOLOGY

The steps to build and run the modelling tool are described here. It is based on a mixed integer

linear optimisation model. The input of the tool is made up of a set of parameters with which

certain associated variables are calculated. An objective function is optimised according to a set of

constraints that must be satisfied. Figure 56 shows the schematic running of the tool:

104

Figure 56: Modelling tool scheme. The inputs describe the situation in the field and the outputs are the results of the modelling tool according to the optimization function and the set of restrictions.

The model will consider at least one local agency in the province located in the capital or in one of

the rural community centres (variable rBL ). The first objective that the model must meet is the

location of the local agency (agencies). Each rural community will be assigned to a local agency

(variable , 'r rBA ) in such a way that the SHSs in the rural community will only be attended to by

technicians from the associated agency.

The second element to be determined will be the number of vehicles associated to each local

agency. They are required for the O&M technicians to get to the villages and souks and one vehicle

has been assigned for every two technicians. Each local agency will hold at least one vehicle

(variable rNCR ).

The model is developed taking into account what has been described in Chapter 6.2 and the

following work dynamic:

On the workday, the O&M teams depart early in the morning from the local agency in the

vehicle with two possible destinations: a) the rural dwellings for maintenance work; b) one

of the souks for fee collection (and maintenance in the community of the souk, if

necessary). Figure 57 summarizes the workday scheme. If option a), the team will go to the

rural dwellings according to a forecasted planning whether PM or CM for repairs. If option

b), the team will move to the souk. There, at least one of the technicians will remain and

the other one can carry out maintenance tasks in the villages of the rural community, if

necessary. When the work of the day is ended, they return to the local agency.

OPTIMIZATION FUNCTION

and

RESTRICTIONS

OUTPUTS

‐ Local agencies: quantity and location

‐ O&M structure: number of technicians and vehicles

‐ Schedule of preventive maintenance visits and fees collection

‐ Global cost

INPUTS

‐ Number of SHSs and location

‐ Road network (distances)

‐ Main Rural Centers location

‐ Number of villages and location

‐ Time needed for displacements

‐ Time for maintenance intervention

‐ Time devoted for fee collection

‐ Local markets (souks) schedule and location


105

Figure 57: O&M teams workday scheme

The technician’s daily work must comprise the travelling time (departure from the local

agency and return) in addition to the maintenance action time. In no case it should exceed

the workday.

As regards the souks, not all of them will be visited for fee collection, but just those where

PERG users usually frequent (defined by parameter ,r db ). The time devoted to the souk

work is expressed by parameter ,r dtb .

The time necessary to undertake a maintenance activity (tas), either preventive or

corrective (since arriving to the dwelling until the departure) is also defined. Travel and

time for maintenance activities allow the daily work of the technicians to be calculated,

which is not permitted to exceed the workday (dailyt ).

6.3.1 Inputparametersandassociatedvariables

The optimisation model is designed to be used in a particular province that consists of a network of

nodes representing the capital of the province and several rural communities (R), which includes

some small villages (V). Around each village, there are few dispersed dwellings where a number of

SHSs are installed.

The model will run for a given planning period or "planning horizon" (D). To avoid problem

dimensions, this planning period will not be a complete year, but the period in which at least one

visit to each rural community must be carried out. Because of the level of service quality committed

to, every SHS must be visited for preventive maintenance twice a year. So, each rural community

should be visited at least once every six months. However, being realistic, as there is corrective

maintenances and fees must be collected, the assumption will be that each rural community has to

be visited every two months. Then, the proposed planning period will be two months,

corresponding to 56 working days.

MORNING AFTERNOON EVENING

LOCAL AGENCY

Departure of O&M teams

SOUK

VILLAGES LOCAL AGENCY

PM and CM Maintenance

Fee collection

Return of O&M teams

106

1,...,RR Rural communities in the considered province.

1,...,VV Villages in each rural community.

1,...,DD Days to consider in the simulation horizon time.

The transport network consists of paved roads that link the main towns (almost all the rural

community centres), and tracks to reach the small villages. Distances and mean speed according to

the road type are used to calculate the travel time. Figure 58 shows a schematic example of one of

the Solar‐PERGISO provinces.

Figure 58: Example of a nodes network. Each black spot represents a rural community centre. Segments outline the paved roads linking the centres

The inputs parameters are described in Table 25:

Table 25: Input parameters

rnv

Number of villages in the rural community rR .

,v ra

Element of a 0‐1 matrix such that it is equal to 1 if village vV belongs to the rural community rR and 0 otherwise.

vnshs

Number of SHSs in the village vV .

, 'r rdist

Distance between the rural centres , 'r r R .

, 'r rtd

Time needed for travelling between the rural centres , 'r r R .

'rdist

Average of distances from the rural centre of the rural community rR to each of its villages vV .

'rtd

Average time for travelling within the same rural community rR .

tas Time needed to attend to a solar home system.


107

,r db

Element of a 0‐1 matrix such that it is equal to 1 if there is a market (souk) in the rural community rR on day d D and 0 otherwise.

,r dtb

Time devoted to the market (souk) located in the rural community rR on day d D .

numc Maximum number of vehicles available to be distributed among the local agencies.

dailyt Length of technician’s workday.

rcmc

Cost of locating a local agency rR .

rcnc

Cost of assigning a team (vehicle and two technicians) to the possible local agency located in the rural community rR .

ctr Travelling cost per unit of distance.

From the previous setting, parameters described in Table 26 are obtained:

Table 26: Parameters obtained according to the inputs introduced

vnsys

Number of SHSs to be attended to during the planning horizon (proportional part of

the total SHSs to be visited), computed as 2365v v

Dnsys nshs

''rdist

Average of distance assigned to a SHS in the rural community rR , computed as

| 1

'''

vr

r rr

vv a

dist nvdist

nsys

V

''rtd

Average time devoted to attend to an SHS in the rural community rR , (including

travelling time) computed as

| 1

'''

vr

rr

vv a

td nvtd tas

nsys

V

After setting the input parameters, the model variables are defined according to Table 27:

Table 27: Model variables. They represent the decisions to be taken

rBL 0‐1 variable which equals 1 if a local agency is located in the rural community rR and 0, otherwise.

, 'r rBA 0‐1 variable which equals 1 if a possible local agency located in rR attends to the rural community 'r R and 0, otherwise.

rNCR

Number of vehicles assigned to the possible local agency located in the rural community rR .

, ',r r dBR 0‐1 variable which equals 1 if the rural community 'r R is visited from the possible local agency located in the rural community rR on day d D and 0, otherwise.

',r dNSR Number of solar home systems in the rural community 'r R attended to day d D .

, ',r r dNCT Number of vehicles travelling to the rural community 'r R from a possible local

108

agency located in the rural community rR on day d D .

, ',r r dTM Time spent in the rural community 'r R coming from the possible local agency located in the rural community rR on day d D .

6.3.2 Modelassumptions

6.3.2.1 Travelscheme

For reasons of simplicity, it has been assumed that to reach a village (V) it is always necessary to

pass through the rural centre, which is also supposed to be placed in the geographical centre of the

rural community. Data on the exact location of the different villages (nνr) are not available, they are

assumed to be dispersed throughout the rural community around the geographical centre at an

average distance (dist'r) (see example in Figure 59). Moreover, the model will consider that

travelling from a village (νr) to another within the same rural community (ν'r) has to be done by

passing through the centre.

Figure 59: Assumed rural communities and villages distribution

Furthermore, it will be assumed that the travelling to rural communities from the local agencies are

round‐trips, i.e., not configuring routes. The main reason to do so is that when computing costs

because of corrective maintenance, it is very plausible that each rural community must be visited

directly at least once within the planning period.

6.3.2.2 Correctivemaintenance

As a corrective maintenance is a stochastic event, it cannot be planned in advance. However,

thanks to the SHS reliability study (see Chapter 3), the average percentage of failed SHS

components per month is known. The optimisation model plans only the preventive maintenance

visits, and incorporates the corrective actions as part of the preventive maintenance just by adding

to the corresponding schedule the expected number of corrective actions to be carried out within

the planning period. The optimisation model provides a daily schedule that considers the corrective

Rural Community Centre.

Circles represent thedist’r distance to thevillages(nvr).


109

maintenance by forcing at least one visit to each village every two months within the planning

period. For example, if a corrective maintenance is required in a certain village, due to the

established level of service quality, it must be carried out within 48 hours, and all preventive

maintenance scheduled in this village for the following days will be carried out the day of the

corrective one, and other tasks should be rescheduled.

6.3.2.3 Criteriaandobjectivefunction

The model is developed under the criterion, firstly, of giving a certain level of service quality of the

maintenance. Previous hypotheses are assumed in order to ensure that with the resources and cost

obtained it will be possible to carry out the maintenance operations with the level of quality

assumed by the contract. So, these hypotheses and the service conditions will be included in the

model as hard constraints (constraints that must be adhered to by any feasible solution). Then, cost

will be the criterion to be minimised, assuming a fixed level of service. Total cost is made up of the

cost of locating local agencies, the cost of teams (vehicles + technicians) and the cost of travelling:

Costs related to the location of local agencies, such as office renting, human resources and

other operational costs (it includes a fixed cost per province and a variable cost per number

of local agencies established):

r rr

cmc BL

R

Costs related to the number of teams assigned to each local agency (it includes the annual

fixed costs of the vehicle as insurance and maintenance, and the salaries and some

expenses of the two technicians):

r rr

cnc NCR

R

Costs related to travelling (fuel), made up of travel covered from the local agencies to rural

communities and from rural communities to villages:

, ' , ', ' ',, ' '

2 2 ''r r r r d r r dr r d r d

ctr dist NCT ctr dist NSR

R D R D

Then, the function to be optimised is as follows:

, ' , ', ' ',, ' '

min[ 2 2 '' ]r r r r r r r r d r r dr r r r d r d

cmc BL cnc NCR ctr dist NCT ctr dist NSR

R R R D R D

6.3.3 Constraintdefinition

The constraints defined must guarantee that the service is provided with a given level of quality and

the schedule is robust enough to be adapted in real operations. Besides the assumptions included

at the beginning of this section, it is considered that two preventive maintenance calls are made to

each SHS per year. A proportional number of SHSs will be considered in the planning period model

(in our case, two months D=56).

So, the main constraints included in the model are as follows:

1) At least one local agency must be located in the province:

110

1rr

BL

R

2) Each rural community 'r R must be visited from one and only one local agency.

, ' 1, 'r rr

BA r

R

R

3) A rural community 'r R is assigned to a possible local agency only if this is established, and,

if that is the case, it could be attended to day d D .

, ', , ' , , ' , Dr r d r r rBR BA BL r r d R

4) If there is a souk on day d D , in the rural community 'r R , then it must be visited.

, ', ', , ' ,r r d r dr

BR b r d

R

R D

5) All the SHSs included in a rural community 'r R and contemplated in the planning horizon

must be visited:

, , ' , 'r d v r vd v

NSR a nsys r

D V

R

6) On day d D , the SHSs of a rural community 'r R can be visited only if that rural

community is visited on that day

, , ' , ','

, , Dr d v r v r r dv r R

NSR a nsys BR r d

V

R

7) Vehicles can be assigned to a rural community if there is a local agency located in that rural

community.

,r rNCR numcBL r R

8) The number of vehicles used for travelling from a local agency to a rural community 'r R on

day d D is limited by the number of vehicles assigned to the local agency.

, ', , , ' ,r r d rNCT NCR r r d R D

9) The time spent in the rural community r' ϵ R on day d ϵ D is the maximum from among the

time spent in the souk and the time spent doing SHSs maintenance. For rural communities where a local agency is located, it is not necessary to take into account the souk time since

users are attended to directly by the local agency staff.

, ', , , ',(1 ) (1 ), , ' ,r r d r d r r r dTM tb BL M BR r r d R D

, ', ', , ','' (1 ), , ' ,r r d r r d r r dTM td NSR M BR r r d R D

10) The time spent in the rural community 'r R by all the vehicles displaced there, plus the

travelling time to the rural community of all the vehicles (round‐trip) must be less than the

workday of the technicians.

, ', , ' , ', , ',2 , , ' ,r r d r r r r d r r dTM td NCT dailyt NCT r r d R D


111

11) The total time spent on day d D by the teams assigned to a local agency is at most the

length of their workday.

, ' , ', , ','

, ,r r r r d r r d rr

td NCT TM dailytNCR r d

R

R D

Together with these constraints, a set of conditions to improve the convergence of the optimisation

algorithms is included. The model is implemented in GAMS and solved with the CPLEX optimiser.

6.4 MODELAPPLICATION

To apply the proposed model, two real cases in the provinces of Azilal and Al Kalaa des Sraghnas

from the Moroccan Solar‐PERGISO programme are set out. The results will be compared to the real

maintenance structure deployed by the ESCO and the real costs associated.

6.4.1 Example1:Azilal

From Table 28 to Table 30 the input parameters in the case of Azilal are summarized.

Table 28: Rural community parameters of the Azilal province

(R) Rural communities vnsys 'rdist (km)'rtd

(min)rnv ,r db

(Souk)

0 Azilal (Capital) 22 0 0 0 1(Thursday)

1 Afourer 124 4.1 8 2 0

2 Agoudi N'lkhair 361 7.5 15 9 0

3 Ait Abbas 22 8.3 17 6 1(Friday)

4 Ait Bououli 80 11.1 22 5 0

5 Ait Mazigh 1 6.9 14 3 1(Monday)

6 Ait M'hamed 8 13.0 26 1 0

7 Ait Taguella 43 6.0 12 5 0

8 Ait Tamlil 3 12.8 26 12 0

9 Anergui 10 10.5 21 2 0

10 Beni Hassan 23 6.1 12 1 0

11 Bin El Ouidane 80 6.8 14 1 0

12 Bni A'yat 19 6.4 13 8 0

13 Bzou 3 8.2 16 1 0

14 Foum Jamaa 8 5.3 11 2 0

15 Imlil 55 5.4 11 2 0

16 Isseksi 10 7.8 16 3 1(Friday)

17 Moulay Aissa Ben Driss 71 6.9 14 2 0

18 Ouaouizeght 18 5.3 11 3 1(Wednesday)

19 Ouaoula 1 8.1 16 6 1(Wednesday)

20 Rfala 5 7.5 15 1 0

21 Tabante 66 10.3 21 1 0

22 Tabaroucht 90 5.9 12 2 0

23 Tamda Noumarcid 2 7.3 15 6 1(Thursday)

112

24 Tanante 16 6.9 14 1 1(Tuesday)

25 Taounza 8 5.9 12 3 0

26 Tidili Fatouaka 349 5.0 10 2 0

27 Tilougguite 9 12.0 24 12 1(Saturday)

28 Timoulilte 302 4.1 8 3 0

29 Zaouiat ahansal 22 16.7 33 11 1(Monday)

Table 29: Time parameters for Azilal

Time needed to attend to a solar home system ( tas ) 20 minutes

Time devoted to the market located in the rural community ( ,r dtb ) 4 hours

Maximum time allowed in a workday (dailyt ) 9 hours

Table 30: Eligible costs based on real expenses in Azilal. Associated costs means other secondary expenses linked to the main concept (i.e. telephone and fax costs of the local agency)

Concept (comprising associated costs) Annual unit cost (€)

Local agency 3,521

Agency head 6,589

Administrative employee 3,068

O&M technician 4,295

Vehicle (type VAN) 3,462

Vehicle (type 4x4) 6,402

Vehicle fuel consumption (€/100 km)11 5.31

Other inputs are the distances (distr,r' expressed in km) and travel times (tdr,r'. expressed in minutes)

between the different rural communities (R).

6.4.1.1 Results:

Table 31 summarizes the results of the model optimization:

11 Average fuel prices between 2006 ‐ 2010 in Morocco


113

Table 31: Optimization model results compared to the actual data for Azilal

Output parameters Model results ESCO real data

Local agency location ( rBL ) R=0 (1 local agency located in

AZILAL) 1 local agency located in

AZILAL

Vehicles ( rNCR ) 3 3

O&M technicians 4 4

Local structure costs (€/year) 48,011 48,011

Travel costs (fuel) (€/year) 7,257 8,687

O&M annual cost (€) 55,268 56,698

The results show that both the model solution and ESCO real data are similar (O&M annual costs

differ by 2.5%, which is not significant), what means that the ESCO probably deployed an optimized

structure in this province.

As regards the corrective maintenance, the model leads to a result through which, after assigning

all the preventive maintenance activities, there will be around 283 hours per year available for

corrective maintenance activities. According to the previous reliability study, during the 10 years of

maintenance, the highest number of corrective maintenance activities will happen during the sixth

year, in which the failure rate of the batteries reaches its maximum (see Table 32). If lamps are

replaced in the souks, then the number of corrective maintenance activities over the sixth year

corresponds to 397 batteries + 62 charge controllers = 459 CM actions. Whether the time dedicated

to the maintenance is 20', around 153 hours will be necessary, less than the available 283 hours,

according to the model results. For the other years during the maintenance period, the availability

for CM will be even larger.

Table 32: failure forecasting for the SHS components according to the reliability study. During the 6th year the failure rate of batteries reaches its maximum. Batteries and charge controllers are replaced in the user

dwelling while lamps are replaced in the souk.

Year 1 2 3 4 5 6 7 8 9 10

Batteries 8 62 186 286 369 397 361 305 272 278

Charge controllers 12 39 62 62 62 62 62 62 51 24

TOTAL 20 101 248 348 431 459 423 367 323 302

6.4.2 Example2:AlKalaadesSraghnas

In the case of the province of Al Kalaa des Sraghnas, with 55 rural communities and 4,396 SHSs, the

results of the model optimization are shown in Table 33, taking into account the same time

parameters defined in Table 29:

114

Table 33: Optimization model results compared to the actual data for Al Kalaa des Sraghnas

Output parameters Model results ESCO real data

Local agency location ( rBL ) R=18 (1 local agency located in

JAAFRA) 1 local agency located in

BEN GUERIR

Vehicles ( rNCR ) 4 5

O&M technicians 6 8

Local structure costs (€/year) 59,170 66,875

Travel costs (fuel) (€/year) 13,413 23,280

O&M annual cost (€) 72,583 90,155

The results of the model optimization indicate a lower annual maintenance cost of 19.5% as regards

the real ESCO data, due to the reduction of 1 maintenance team and the optimization of

maintenance travelling, in addition, the local agency has been relocated (Figure 60).

Figure 60: Al Kalaa des Sraghnas province map. The subdivisions show the different rural communities. The arrow indicates the movement of the local agency from Ben Guerir (ESCO location) to Jaafra (model

location)

As regards the CM, there will be around 346 hours per year available for corrective maintenance

activities. The highest number of corrective maintenance activities will also happen during the sixth

year (see Table 34). The number of corrective maintenance activities over the sixth year

corresponds to 1,076 batteries + 167 charge controllers = 1,243 CM actions. Around 414 hours will


115

be necessary, more than the available time, according to the model results (346 hours are enough

to make 1,040 CM actions). As shown in Table 34, only the 5th, 6th and 7th years have more than

1,040 CM to carry out, thus the maintenance structure calculated by the model will be able for the

remaining maintenance period. Just in these 3 critical years, it will be necessary to extend the

number of maintenance teams from 3 to 4. That means that in these 3 years the costs of two

technicians and 1 vehicle must be added to the cost structure, which has been considered within

the resulting costs shown in Table 33.

Table 34: failure forecasting for the SHS components according to the reliability study in Al Kalaa des Sraghnas. In years 5th to 7th the failure rate of batteries reaches its maximum. Batteries and charge

controllers are replaced in the user dwelling. Lamps are replaced in the souk.

Year 1 2 3 4 5 6 7 8 9 10

Batteries 31 158 507 777 1002 1076 985 836 745 755

Charge controllers 43 98 167 167 167 167 167 167 125 70

Total 74 256 674 944 1169 1243 1152 1003 870 825

6.5 CONCLUSIONS

A model based on a mixed integer linear optimisation has been introduced with the aim of

estimating costs and designing maintenance structures for PV rural electrification programmes.

Constraints included in the model ensure a determined operability and level of service quality for

maintenance. The objective function defined must minimise the associated costs of:

‐ Local agencies

‐ Vehicles and technicians

‐ Maintenance and fee collection displacements

Based on the Solar‐PERGISO, the usefulness of the tool has been illustrated using two of the

provinces belonging to the programme.

The results of the application show in one case that the model solution matches the real ESCO data

in that province (Azilal), which suggests that the ESCO optimized the maintenance structure and

costs in that case. In the other province (Al Kalaa des Sraghnas), the model has optimized the

maintenance structure reducing by 1 the number of maintenance teams (2 technicians and 1

vehicle) for 7 of the 10 years of the maintenance period; it has relocated the local agency to a

better location and it has optimized maintenance travel. Thus, the cost of the O&M in this province

has been reduced by 19.5%, which suggests that the ESCO had deployed and oversized structure.

Throughout these results it has been illustrated that it is possible to design O&M structures in PVRE

programmes on the basis of the knowledge of the reliability parameters of the SHSs and the

operating costs of the programme.

CHAPTER 7

CONCLUSIONS AND FUTURE RESEARCH

118

7 CONCLUSIONSANDFUTURERESEARCH

7.1 CONCLUSIONS

PV technology plays an important role in the solution of the overall lack of access to energy. It is

expected that 250 million people will get access to electricity by means of solar home systems

before 2030.

Large PV rural electrification programmes currently taking place in many countries have shown

problems of sustainability in the maintenance phase as O&M costs are systematically larger than

expected, leading the private operators (ESCOs) to leave the programmes because of negative

financial balances. The cause of these concerns is that PVRE programmes are generally formulated

on the basis of assumptions that are not in line with reality.

This PhD has studied a real PVRE programme developed in Morocco with more than 13,000 SHSs

with the aim of analyzing the real factors involved in operation and maintenance, such as the

reliability of the SHSs operating under real conditions and the actual costs of the O&M phase. The

excellent opportunity for the author to have had full access to the detailed data of maintenance

and costs of the Moroccan programme developed by the ESCO ISOFOTON (the Solar‐PERGISO) for

five years, provides the opportunity, for the first time, to contrast the real data of the PVRE with

classic assumptions.

The work has been structured in two axes: the reliability study of the SHSs and the cost

characterization of the O&M phase.

Through the reliability study, the failure distribution of every SHS component has been evaluated

on the basis of the real failures extracted from the maintenance database that the ESCO drew up

during the first 5 years of operation of the programme. The following conclusions of the study are

highlighted:

Charge controllers and lamps have showed a failure distribution established by exponential

functions, as applied to electronic devices, in which failures happen randomly. The 2 types

of lamp (7 and 11WDC) have shown an identical behaviour as regards their reliability

function, failure rate and MTTF parameters.

The battery failure distribution has been defined by a Normal function, which is more

adapted to devices affected by aging factors. Although the failures in the first 1.4 years do

not fit the Normal model exactly, the whole failure distribution has shown a great goodness

of fit and presents a correlation coefficient close to 1. In these first 1.4 years, the failure

distribution fits a Weibull model with a scale parameter β of less than 1, which means that

the failures are the result of infant mortality.

As regards the PV modules analysis, the Solar‐PERGISO maintenance database does not give

enough information on the failure mechanisms of this component. The failures found were:

Chapter 7: Conclusions and Future Research

119

3 damaged diodes, 10 frontal glass breakages and 6 hotpots at the junctions between cells.

This information has not been enough to apply the reliability study but it can be argued

that the PV module failure data prove that they are very reliable compared to the other

components, with a minimal impact on the SHS series reliability.

The MTTF results have proved that the charge controller is the most reliable component

(about 27.2 years ±9.5%), 7W and 11W LC lamps have a similar MTTF value (16.5 years ±4%

for 7W lamps and ±7% for 11W lamps) and the least reliable component is the battery,

whose MTTF = 5.5 years ±3.4%.

As battery is the main limiting factor as regards the reliability of the system, and PV module

reliability has not been possible to asses, two in‐field degradation tests of both batteries and PV

modules have been carried out to analyze the behaviour of these components operating under real

conditions, in which user behaviour is involved in the degradation processes.

A sample of 40 batteries installed in the Solar‐PERGISO has been periodically tested over a period of

18 months in order to analyse the capacity degradation process of the battery. The main results are

as follows:

On average, the capacity drops to 60% of the initial capacity after eighteen months of

operation. In less than six months the batteries have reached a remaining capacity of less

than 80% of the nominal capacity, the threshold usually used to define the battery lifetime,

which is not the case for the Solar‐PVREISO users.

In spite of this fast degradation, reliability studies show that batteries keep working for

much longer (MTTF=5.5 years) until the users perceive that the SHS does not satisfy their

needs and ask for the replacement of the battery.

Correlating these two results, a model has been proposed to calculate the remaining

capacity that leads to user dissatisfaction, which is 18% of the initial capacity for this

particular programme.

Based on current (A) recorded data, it has been determined that there is no a standard load

profile, but loads are in general very low as regards the storage capacity of the batteries.

The mean load consumption has been μload=5.2 Ah/day with a standard deviation σload=4.2

Ah/day. It must be noted that the design load was much higher (12 Ah/day), which could

suggest that battery degradation would have been more accelerated if both design and real

loads had been closer. This leads to an operation normally based on a high SOC, with

sporadic events of low battery disconnections. A model that correlates these two sorts of

operation with the battery capacity degradation has been proposed.

The reasons for this fast degradation of the battery capacity could be as follows:

Bad manufacturing quality of batteries. The in‐the‐field testing has shown a rapid

degradation of batteries even working with low loads. Moreover, they have shown wide

dispersion in terms of capacity and density of the electrolyte. Ultimately, the visual

inspection of the plates from one of the most degraded batteries has shown the dramatic

120

loss of active mass in the positive plate which suggests a very low physical resistance of the

plates.

The behaviour of degraded batteries (with capacities of less than 0.8∙Cnom) is difficult to

predict, which makes it very complex to adapt the algorithms of the charge controllers.

The threshold of the charge controller to protect the battery against over discharge allows

a DOD close to 100%, which is completely incorrect, although this does not seem to be very

relevant, since excessive loads are not common.

As regards the PV modules, a set of 41 mono‐crystalline modules have been tested in the field to

assess their state of quality, resulting in a mean power degradation rate of between 0.4% and 1.1%

per year, which matches other experiences reported in the bibliography. Defects related to

hotspots, broken diodes, browning and other defects are almost negligible. Hence, silicon PV

modules show good reliability behaviour, especially when compared with the other SHS devices.

As regards the second pillar of the thesis, a characterisation of the Moroccan Solar‐PERGISO costs

has been achieved. Based on the real data costs of the first 5 years of the programme, the overall

cost for 10 years of maintenance has been obtained, detailing each activity (installation, O&M and

management). The overall programme cost reached €1,574/SHS , where 40.1% of this cost

corresponds to the installation phase (€631/SHS); 41.4% to the O&M activity (€652/SHS) and 18.5%

to the general management (€291/SHS). The main conclusions extracted are as follows:

Around 50% of the overall cost is invested during the installation phase, and the other 50% in the O&M period.

The O&M of the systems (maintenance, spare parts and fee collection) reaches €76/SHS∙year . This figure is further higher than usually considered in photovoltaic maintenance and it is not covered by user fees, causing unsustainable financial balances. In the case of the Solar‐PERGISO the annual fee was €59/SHS , significantly lower than the O&M unit costs.

The PV module represents just 15.5% in the overall cost (€243/SHS), versus 18.5% of the battery (€292/SHS), so battery has to be considered as the most expensive component in decentralized PV rural electrification.

The maintenance of the open lead batteries leads to add distilled water frequently, whose global cost means 6% of the overall programme cost (€9.5/SHS∙year).

The SHS dispersion and inaccessibility plays an important role in the cost structure, given by a mean SHS density of 0.068 SHS/km2. This feature affects the cost of maintenance. For example, the cost of fuel in this programme represents €5.5/SHS∙year.

The energy delivered by the SHSs has been calculated, expressed as available electricity for customers, reaching a cost of €1.3/kWh .

This PhD work has concluded with the proposal of a mathematical tool based on a mixed integer

linear optimization created with the aim of designing maintenance structures for PV rural

electrification programmes. The proposed model comprises a set of parameters and variables used

to define an objective function, which depends on a set of constraints to be satisfied. The objective

function must minimize the associated costs of:

‐ Local agencies

‐ Vehicles and technicians

Chapter 7: Conclusions and Future Research

121

‐ Maintenance and fee collection displacements

Based on the Solar‐PERGISO, the usefulness of the tool has been illustrated by applying it in two of

the provinces belonging to the programme.

In one case the model has come up with a solution that matches the real ESCO data in that province

in terms of the size of the structure and costs. In the other case, the model has optimized the

maintenance structure reducing by 1 the number of maintenance teams, has relocated the local

agency and has optimized the maintenance displacements. Thus, the cost of the O&M in this

province has been reduced by 19.5%.

These results illustrate that it is possible to design O&M structures in PVRE programmes based on

the availability of the reliability parameters of the SHSs and the operating costs of the programmes.

7.2 FUTURELINESOFRESEARCH

This research work has tried to open the door to the design of maintenance structures for

decentralized rural electrification. The criteria evaluated in this study as regards the reliability of

the SHSs and the characterization of the O&M costs, both on the basis of the Moroccan PVRE

programme has been useful to understand that the sustainability foundations that PVRE promoters

currently show weaknesses. Thus it is necessary to extend this analysis to other experiences with

the aim of universalising the results and being able to propose a paradigm of decentralized rural

electrification. In this line of work, the following areas of research are proposed:

The creation and application of friendly tools able to gather the data relating the reliability

and cost for the new programmes that will be developed. The goal is to make the collection

of data easier in order to extend the feedback in PVRE.

The development, application and analysis of results of new fee collecting systems, such as

the prepayment cards or the "pay‐as‐you‐go" model, with the aim of adapting the user

payment to the new technologies, already assumed for the rural population (for example

the payment through the mobile phone), in addition to the reduction in the management

cost of fee collection in PVRE programmes.

To establish new patterns for PVRE sizing taking into account that batteries degrade until

ranges of 18% of the initial capacity. Moreover, new in‐the‐field studies of batteries

working under real conditions could be very useful to extend the results of this PhD. The

definition of a standardized cycling model of batteries specifically for PVRE could be useful

for the improvement of quality and reliability.

Finally, as well as international standards have been developed to ensure the quality of the

SHSs and the good practises in the installation phase, it would be desirable the

development of an O&M standard for PVRE to guaranty the sustainability of long period

maintenance phases.

PUBLICATIONS GENERATED DURING

THIS PhD

124

PUBLICATIONSGENERATEDDURINGTHISPhD

Papers in international journals:

Carrasco LM, Narvarte L, Peral A, Vázquez M. Reliability of a 13,000‐SHS photovoltaic rural electrification programme. Prog. Photovolt: Res. Appl. 2013, 21. p.1136–1145 [Impact Factor: 9.696; Q1]

Carrasco LM, Narvarte L, Lorenzo E. Operational costs of A 13,000 solar home systems rural electrification programme, Renewable and Sustainable Energy Reviews, April 2013, Volume 20. p. 1‐7 [Impact Factor: 5.510; Q1]

Carrasco LM, Narvarte L, Martínez‐Moreno F, Moretón R. In‐field assessment of batteries and PV modules in a large photovoltaic rural electrification programme, Energy, 1 October 2014, Volume 75. p. 281‐288 [Impact factor: 4.159; Q1]

Gago Calderón A, Narvarte L, Carrasco LM, Serón Barba J. LED bulbs technical specification and testing procedure for solar home systems, Renewable and Sustainable Energy Reviews, January 2015, Volume 41. p. 506‐520 [Impact Factor: 5.510; Q1]

Carrasco LM, Narvarte L. Electrificación rural fotovoltaica "Solar Home Systems". Fiabilidad y costes de mantenimiento. "Era Solar", v. null (n. 173); 2013 p. 26‐29, pp.. ISSN 0212‐4157.

Papers in international congresses:

Guest lecture:

Carrasco LM, Narvarte L. Reliability issues and cost structure in the 13,000 SHS rural electrification programme in Morocco. 3rd Symposium Small PV Applications ‐ Rural Electrification and Commercial Use, June 2013, Ulm, Germany, p. 270‐275

Oral:

Carrasco LM, Narvarte L. Maintenance characterization of a 13,000 SHS PV program in Morocco. 2nd Symposium Small PV Applications ‐ Rural Electrification and Commercial Use, June 2011, Ulm, Germany, p. 46‐51

Carrasco LM, Narvarte L. Fiabilidad y costes de mantenimiento en un programa de electrificación rural de 13.000 solar home systems. XV Congreso Ibérico y X Iberoamericano de Energía Solar. Vigo, Galicia, Junio 2012, Vigo, España.

Carrasco LM, Narvarte L. Operational costs and reliability in a large rural electrification programme based on solar home systems. 28th European Photovoltaic Solar Energy Conference and Exhibition (EU PVSEC) 2013 Proceedings, p. 3831 ‐ 3835. Paris, France.

Carrasco LM, L. Narvarte, A. Peral, Reliability Analysis of a 13,000 SHS PV Rural Electrification Programme in Morocco, 26th European Photovoltaic Solar Energy Conference and Exhibition (EU PVSEC) 2011 Proceedings, p.4044 ‐ 4047. Hamburg, Germany

Publications generated during the PhD

125

Gago Calderón A, Narvarte L, Carrasco LM, y Díaz Martínez M. LED bulb lamps technical specification and testing procedure for solar home systems. 3rd Symposium Small PV‐ Applications Rural Electrification and Commercial Use, June 2013, Ulm, Germany.

BIBLIOGRAPHY

128

BIBLIOGRAPHY

[1] IEA, “World Energy Outlook,” Paris, 2012.

[2] C. N. Doll and S. Pachauri, “Estimating rural populations without access to electricity in developing countries through night‐time light satellite imagery,” Energy Policy, vol. 38, no. 10, pp. 5661 ‐ 5670, October 2010.

[3] [Online]. Available: http://www.se4all.org/. [Accessed October 2014].


[5] IEA, “Africa Energy Outlook ‐ A Focus on Energy Prospects in Sub‐Saharan Africa,” Paris, 2014.

[6] S. Tully, “The Human Right to Access Electricity,” The Electricity Journal, vol. 19, no. 3, pp. 30 ‐ 39, April 2006.

[7] [Online]. Available: http://www.thegef.org/gef/. [Accessed November 2014].

[8] UN Energy, “The energy challenge for achieving the Millennium Development Goals,” New York, 2005.

[9] [Online]. Available: http://luzparatodos.mme.gov.br/luzparatodos/asp/. [Accessed October 2014].

[10] Banerjee, G. Sudeshna, D. Barnes, B. Singh, K. Mayer and H. Samad, “Power for All: Electricity Access Challenge in India,” Washington, DC, 2013.

[11] [Online]. Available: http://lucesparaaprender.org/web/. [Accessed November 2014].

[12] P. Cook, “Rural Electrification and Rural Development,” in Rural Electrification Through Decentralised Off‐grid Systems in Developing Countries, S. Bhattacharyya, Ed., Springer London, 2013, pp. 13 ‐ 38.

[13] P. Cook, “Infrastructure, rural electrification and development,” vol. 15, no. 3, pp. 304 ‐ 313, 2011.

[14] J. Chontanawat, L. C. Hunt and R. Pierse, “Does energy consumption cause economic growth?: Evidence from a systematic study of over 100 countries,” Journal of Policy Modeling, vol. 30, no. 2, pp. 209 ‐ 220, 2008.

[15] C. de Gouvello and D. Laurent, “Maximizing the productive uses of electricity to increase the impact of rural electrification programs,” 2008.

[16] [Online]. Available: http://www.macrotrends.net. [Accessed February 2015].

Bibliography

129

[17] A. Cabraal, M. Cosgrove‐Davies and L. Schaeffer, “Best practices for photovoltaic household electrification programs,” Washington, D.C., 1996.


[19] [Online]. Available: http://www.worldbank.org/. [Accessed December 2014].

[20] C. Heuraux, L'Électricité au Coeur des Défis Africans. Manuel sur l'électrification en Afrique, Paris: Karthala, 2010.

[21] [Online]. Available: http://www.lme.com/. [Accessed November 2014].

[22] S. Mahapatra and S. Dasappa, “Rural electrification: Optimising the choice between decentralised renewable energy sources and grid extension,” Energy for Sustainable Development, vol. 16, p. 146 – 154, 2012.

[23] L. Parshall, D. Pillai, S. Mohan, A. Sanoh and V. Modi, “National electricity planning in settings with low pre‐existing grid coverage: Development of a spatial model and case study of Kenya,” Energy Policy, vol. 37, no. 6, pp. 2395 ‐ 2410, June 2009.

[24] K. Q. Nguyen, “Alternatives to grid extension for rural electrification: Decentralized renewable energy technologies in Vietnam,” Energy Policy, vol. 35, no. 4, pp. 2579 ‐ 2589, April 2007.

[25] A. Inversin, “Reducing the cost of grid extension for rural electrification,” 2000.

[26] S. Szabó, T. K.Bódis a and M. Moner‐Girona, “Sustainable energy planning: Leapfrogging the energy poverty gap in Africa,” RenewableandSustainableEnergyReviews, vol. 28, pp. 500 ‐509, 2013.

[27] [Online]. Available: http://www.eia.gov/. [Accessed October 2014].

[28] International Energy Agency, “Climate & Electriciy Annual Data and Analysis,” OECE/IEA, Paris, 2011.

[29] M. G. Pereira, M. A. V. Freitas and N. F. d. Silva, “Rural electrification and energy poverty: Empirical evidences from Brazil,” Renewable and Sustainable Energy Reviews, no. 14, p. 1229 – 1240, 2010.


[31] R. A. Cabraal, “Experiences and lessons from 15 years of World Bank support for photovoltaics for off‐grid electrification,” in 2nd International Conference on the Developments in Renewable Energy Technology (ICDRET), Dhaka, 2012.

[32] J. M. García de la Infanta, Primeros Pasos de la Luz Eléctrica en Madrid y otros Acontecimientos, Madrid: Fondo Natural, 1986.

[33] F. Haveron, The Brillant Ray ‐ How the Electric Light was brought to Godalming in 1881, The

130

Godalming Electricity Centenary Celebrations Commitee, 1981.

[34] J. Sánchez Miñana, “El Artillero Isidoro Cabanyes (1843 ‐ 1915): Una Vida de Inventos,” in Quaderns d'Història de l'Enginyeria, vol. XIV, Barcelona, ETSEIB, 2014, pp. 83 ‐ 154.

[35] M. Arroyo and G. Nahm, “La Sociedad Española de Electricidad y los inicios de la industria eléctrica en Cataluña,” in Las Tres Chimeneas. Implantación industrial, cambio tecnológico y transformación de un espacio urbano barcelonés, Barcelona, Fecsa, 1994, pp. 25 ‐ 51.

[36] F. F. Sintes Olives and F. Vidal Burdils, La Industria Eléctrica en España. Estudio Económico ‐Legal de la Producción y Consumo de Electricidad y de Material Eléctrico, Barcelona: Montaner y Simon, S. A., 1933.

[37] J. Maluquer de Motes, “Los pioneros de la segunda revolución industrial en España: la Sociedad Española de Electricidad (1881‐1894),” Revista de Historia Industrial, no. 2, pp. 121 ‐ 142, 1992.

[38] G. Núñez Romero‐Balmas, “Empresas de producción y distribución de electricidad en España (1878‐1953,” Revista de Historia Industrial, no. 7, pp. 39 ‐ 80, 1995.

[39] I. Bartolomé Rodríguez, LA INDUSTRIA ELÉCTRICA EN ESPAÑA (1890‐1936), Madrid: Banco de España, 2007.

[40] L. González Abela, La Electrificación Rural. Problema Nacional, Madrid: Gabel, 1942.

[41] Congreso de Electrificación Rural, vol. Primero, Madrid: Publicaciones de Luz y Fuerza, 1948.

[42] [Online]. Available: http://www.iies.es/. [Accessed December 2014].

[43] J. M. Marchesi y Sociats, Electrificación Rural. Ciclo de Conferencias. Conferencia Inagural, Madrid: Instituto de Ingenieros Civiles, 1932.

[44] H. V. Hunter, A Brief History of the Rural Electrification and Telephone Programs, REA, U.S. Department of Agriculture, 1982.

[45] L. Pellegrini and L. Tasciotti, “Rural electrification now and then: comparing contemporary challenges in developing countries to the USA's experience in retrospect,” Forum for Development Studies, vol. 40, no. 1, pp. 153 ‐ 176, March 2013.

[46] A. Fernández, “La electrificación rural en España. Realizaciones en los últimos años. Comparación con la de otros países,” Economía Industrial, no. 145, pp. 23 ‐ 31, 1976.

[47] A. Fernández and J. González, “Una experiencia en programación de electrificación rural,” Estudios territoriales, no. 1, pp. 187 ‐ 196, 1981.

[48] J. A. Guerra, Tesis doctoral. Análisis de los parámetros técnicos en la aplicación de los sistemas de información geográfica a la integración regional de las energías renovables en la producción descentralizada de electricidad, Madrid: Universidad Politécnica de Madrid, 2000.

Bibliography

131

[49] [Online]. Available: http://www.ine.es/varipc/index.do. [Accessed November 2014].

[50] E. Lorenzo, “Rural Electrification,” in Seventh E.C. Photovoltaic Solar Energy Conference, Commission of the European Communities, Sevilla, Spain, 27‐31 October 1986, Proceedings, 1986.

[51] J. Perlin, From Space to Earth: The Story of Solar Electricity, Harvard University Press, 1999.

[52] P. D. Maycock, “PV review: World Solar PV market continues explosive growth,” vol. 6, no. 5, pp. 18‐22, September–October 2005.

[53] IEA‐PVPS, [Online]. Available: http://www.iea‐pvps.org/. [Accessed November 2014].

[54] M. R. Vervaart and F. D. J. Nieuwenhout, Manual for the Design and Modification of Solar Home System Components, IBRD/The World Bank USA, 2001.

[55] J. M. Huacuz, J. Agredano and L. Gunaratne, “Photovoltaics and Development,” in Handbook of Photovoltaic Science and Engineering, vol. 2nd Editiom, John Wiley & Sons, Ltd, 2011, pp. 1078 ‐ 1015.

[56] E. Lorenzo, “Photovoltaic rural electrification,” Progress in Photovoltaics: Research and Applications, vol. 5, no. 1, 1997.

[57] G. M. Dobrov, “The strategy for organized technology in the light of hard‐, soft‐, and org‐ware interaction,” Long Range Planning, vol. 12, no. 4, pp. 79 ‐ 90, 1979.

[58] G. Foley, Photovoltaic applications in rural areas of the developing world, vol. 304, World Bank Publications, 1995.

[59] M. Manson, Rural electrification: a review of World Bank and USAID financed projects, Washington, DC: World Bank Backgr. Pap., 1990.

[60] Independen Evaluation Group, “The Welfare Impact of Rural Electrification: A Reassessment of the Costs and Benefits,” Washington DC, 2008.

[61] A. Sanghvi and D. Barnes, “Rural electrification: lessons learned,” World Bank, 2001.

[62] IFC, “Solar Development Capital: Lessons Learned in Financing Solar Home Systems,” no. 13, April 2007.

[63] P. Meyer, “Economic Analysis of Solar Home Systems: A Case Study for the Philippines,” 2003.

[64] E. Terrado, A. Cabraal and I. Mukherjee, “Operational guidance for World Bank Group staff : designing sustainable off‐grid rural electrification projects ‐ principles and practices,” 2008.

[65] IEA‐PVPS, “Sources of Financing for PV‐Based Rural Electrification in Developing Countries,” 2004.

132

[66] IEA‐PVPS, “16 Case Studies on the Deployment of Photovoltaic Technologies in Developing Countries,” 2003.

[67] IEA, “Comparative Study on Rural Electrification Policies in Emerging Economies. Keys to successful policies,” Paris, 2010.

[68] L. Rosenblum, W. J. Bifano, G. F. Hein and A. F. Ratajczak, “Photovoltaic power systems for rural areas of developing countries,” in Interregional Symposium on Solar Energy for Development, February 5‐10, 1979. Tokyo.

[69] S. Polgar, “Use of solar generators in Africa for broadcasting equipment,” Solar Energy, vol. 19, no. 2, pp. 201 ‐ 204, 1977.

[70] F. B. Verspieren, “The application of photovoltaics to water pumping and irrigation in Africa,” in Photovoltaic Solar Energy Conference, January 1981.

[71] W. Palz, “The French Connection: The rise of the PV water pump,” Refocus, vol. 2, no. 1, pp. 46 ‐ 47, January–February 2001.

[72] C. o. t. E. Communities, “The Units of Account as a Factor of Integration,” 1977.

[73] W. Palz, “The Photovoltaic Project of the Commission of the European Communities,” in Photovoltaic Solar Energy Conference, Luxembourg, September 27‐30, 1977, Proceedings, 1978.

[74] Solar Cell Group, “Sollar Cell Programme in India,” in Photovoltaic Solar Energy Conference, Luxembourg, September 27‐30, 1977, Proceedings, 1978.

[75] E. J. Pérez, “Photovoltaics Activity in Mexico,” in Photovoltaic Solar Energy Conference, Luxembourg, September 27‐30, 1977, Proceedings, 1978.

[76] T. Koyanagi, “Photovoltaic System in "Sunshine Project" National R & D Program in Japan,” in Photovoltaic Solar Energy Conference, Luxembourg, September 27‐30, 1977, Proceedings, 1998.

[77] A. Dollery, “Photovoltaic Activities in teh United Kingdom,” in Photovoltaic Solar Energy Conference, Luxembourg, September 27‐30, 1977, Proceedings, 1978.

[78] A. F. Ratajczak, “Photovoltaic Stand‐Alone Systems Projects,” in Medium‐Size Photovoltaic Power Plants, EEC/DOE Workshop, Sophia‐Antipolis, France, 23‐24 October 1980, Proceedings, 1981.

[79] W. J. Bifano, A. F. Ratajczak, D. M. Bahr and B. G. Garrett, “Social and economic impact of solar electricity at Schuchuli Village,” 1979.

[80] A. F. Ratajczak, “Description and status of NASA‐LeRC/DOE photovoltaic applications systems experiments,” in 13th Photovoltaic Specialists Conference, 1978.

[81] A. F. Ratajczak and W. J. Bifano, “Description of photovoltaic village power systems in the

Bibliography

133

United States and Africa,” in 2nd Photovoltaic Solar Energy Conference, 1979.

[82] Halcrow W & World Bank, “Small‐scale solar‐powered irrigation pumping systems; technical and economic review,” Sir William Halcrow and Partners, 1981.

[83] W. Palz, “Photovoltaics for Development,” in 2ndE.C. Photovoltaic Solar Energy Conference, Berlin (West), 23‐26 April 1979, Proceedings, 1979.

[84] M. R. Starr and W. Palz, “Photovoltaic Power for Europe. An Assessment Study,” Series C, Volume 2, 1983.

[85] W. B. Gillett, R. J. Hacker and W. Kaut, “PHOTOVOLTAIC DEMONSTRATION PROJECTS 2,” ELSEVIER SCIENCE PUBLISHERS LTD, London, 1989.

[86] B. Yordi, D. Stainforth, H. Edwards, V. Gerhold, G. Riesch and G. Blaesser, “The Commission of the European Communities' (EC) demonstration and thermie programmes for photovoltaic (PV) applications,” Solar Energy, vol. 59, no. 1‐3, pp. 59 ‐ 66, January–March 1997.

[87] L. Selles and B. Aubert, “Technical Coordination of the French PV Rural Electrification Programme,” in Sixth E.C. Photovoltaic Solar Energy Conference, Commission of the European Communities, London, U.K., 15‐19 April 1985, Proceedings, 1985.

[88] European Commission, “Photovoltaic rural electrification of 79 dwellings at Sierra de Segura (Jaén),” Luxembourg, 1994.

[89] M. Egido, M. Erill, R. Eyras, A. González, S. Gracia, E. Lorenzo, M. Montero and M. Sidrach, “Photovoltaic Rural Electrification of 57 Dwellings in La Sierra de Segura (Spain),” in Eight E.C. Photovoltaic Solar Energy Conference, Florence, 1988.

[90] F. C. Treble, “The CEC Photovoltaic Pilot Projects,” in Sixth E.C. Photovoltaic Solar Energy Conference, Commission of the European Communities, London, U.K., 15‐19 April 1985, Proceedings, D. Reidel Publishing Co., 1985.

[91] F. C. Treble, “Progress in International Photovoltaics Standards,” in Seventh E.C. Photovoltaic Solar Energy Conference, Commission of the European Communities, Sevilla, Spain, 27‐31 October 1986, Proceedings, 1986.

[92] J. Krell, “The Remote Home Power System Project on the Navajo Reservation,” in Fifth E.C. Photovoltaic Solar, Energy Conference, Commission of the European Communities, Kavouri (Athens), Greece, 17‐21 October 1983, Proceedings, 1983.

[93] E. Lorenzo, J. C. Miñano, F. Yebes and B. Yaici, “A Photovoltaic Installation in a Sub‐Sahel Village in Senegal,” in Seventh E.C. Photovoltaic Solar Energy Conference, Commission of the European Communities, Sevilla, Spain, 27‐31 October 1986, Proceedings, 1986.

[94] D. C. Kenkeremath and P. A. Borgo, “Egyptian PV Field Test Project,” in Sixth E.C. Photovoltaic Solar Energy Conference, Commission of the European Communities, London, U.K., 15‐19 April 1985, Proceedings, 1985.

134

[95] H. Rodríquez‐Murcia, “Telecommunication and Professional Uses of Photovoltaics,” in Seventh E.C. Photovoltaic Solar Energy Conference, Commission of the European Communities, Sevilla, Spain, 27‐31 October 1986, Proceedings, 1986.

[96] L. P. Drouot, “The French Photovoltaic Program: 1982 ‐ 1986,” in Fifth E.C. Photovoltaic Solar, Energy Conference, Commission of the European Communities, Kavouri (Athens), Greece, 17‐21 October 1983, Proceedings, 1983.

[97] P. Jourde, “French Polynesia Already in Solar Age,” in Sixth E.C. Photovoltaic Solar Energy Conference, Commission of the European Communities, London, U.K., 15‐19 April 1985, Proceedings, 1985.

[98] A. Claverie, P. Courtiade and P. Vezin, “Photovoltaic rural electrification in France,” in Photovoltaic Energy Conversion, 1994., Conference Record of the Twenty Fourth. IEEE Photovoltaic Specialists Conference‐1994, 1994 IEEE First World Conference, 1994.

[99] P. Waldner and J. Guerry, “Installation of photovoltaic generators and solar‐powered heating systems in 23 refuges in the French Alps and Pyrenees,” Luxembourg, 1991.

[100] J. S. Solera, “100 kWp Experimental PV Power Plant,” in Sixth E.C. Photovoltaic Solar Energy Conference, Commission of the European Communities, London, U.K., 15‐19 April 1985, Proceedings, 1985.

[101] A. Krenzinger and M. Montero, “Energy Consumption Patterns of the Rural Photovoltaic Market in Spain,” in Seventh E.C. Photovoltaic Solar Energy Conference, Commission of the European Communities, Sevilla, Spain, 27‐31 October 1986, Proceedings, 1986.

[102] E. Lorenzo, A. Krenzinger, M. Montero and A. Meana, “The Rural Electrification Share of the Spanish Photovoltaic Market,” in xth E.C. Photovoltaic Solar Energy Conference, Commission of the European Communities, London, U.K., 15‐19 April 1985, Proceedings, 1985.

[103] J. R. Eyras and E. Lorenzo, “An Energy Consumption Scenario for Sizing Rural Electrification PV Systems,” Progress in Photovoltaics: Research and Applications, vol. 1, pp. 145 ‐ 152, 1993.

[104] J. R. Eyras and E. Lorenzo, “Lessons from a PV Rural Electrification Projetc at "Sierra de Segura" (Spain),” in Tenth E.C. Photovoltaic Solar Energy Conference, Losbon, 1991.

[105] B. Cancino, E. Gálvez, P. Roth and A. Bonneschky, “Introducing Photovoltaic Systems in Homes in Rural Chile,” IEEE Technology and Society Magazine, vol. 20, no. 1, pp. 29 ‐ 36, 2001.

[106] E. Lorenzo, “The Importance of Sociology in Photovoltaic Projetcs,” in Seventh E.C. Photovoltaic Solar Energy Conference, Commission of the European Communities, Sevilla, Spain, 27‐31 October 1986, Proceedings, 1986.

[107] H. P. Hertlein, “Technological Issues in Photovoltaics ‐ An Overview,” in Seventh E.C. Photovoltaic Solar Energy Conference, Commission of the European Communities, Sevilla, Spain, 27‐31 October 1986, Proceedings, 1986.

Bibliography

135

[108] C. Leclercq, “Overall PV System Reliability,” in Sixth E.C. Photovoltaic Solar Energy Conference, Commission of the European Communities, London, U.K., 15‐19 April 1985, Proceedings, 1985.

[109] L. T. Slominski and D. V. Eskenazi, “Evaluation of International Photovoltaic Projects,” in Sixth E.C. Photovoltaic Solar Energy Conference, Commission of the European Communities, London, U.K., 15‐19 April 1985, Proceedings, 1985.

[110] E. Lorenzo, “Realities in the field of photovoltaic rural electrification projects,” in Conference: 3. ISES European solar congress, Copenhagen (Denmark), 2000, Jul 01.

[111] F. Nieuwenhout, A. v. Dijk, V. v. Dijk, D. Hirsch, P. Lasschuit, G. v. Roekel, H. Arriaza, M. Hankins, B. Sharma and H. Wade, “Monitoring and evaluation of Solar Home Systems. Experiences with applications of solar PV for households in developing countries,” Amsterdam, 2000.

[112] M. A. Egido, E. Lorenzo and L. Narvarte, “Universal technical standard for solar home systems,” Prog. Photovolt: Res. Appl., no. 6, p. 315 – 324, 1998.

[113] European Commission. Directorate General XVII, Energy, Universal technical standard for solar home systems. Thermie B: SUP‐995‐96, Luxembourg: Office for Official Publications of the European Communities, 1999 (updated in 2001).

[114] J. M. Huacuz, R. Flores, J. Agredano and G. Munguia, “Field performance of lead‐acid batteries in photovoltaic rural electrification kits,” Solar Energy, vol. 55, no. 4, pp. 287 ‐ 299, 1995.

[115] F. D. J. Nieuwenhout, A. v. Dijk, P. E. Lasschuit, G. v. Roekel, V. A. P. v. Dijk, D. Hirsch, H. Arriaza, M. Hankins, B. D. Sharma and H. Wade, “Experience with solar home systems in developing countries: a review,” Prog. Photovolt: Res. Appl., vol. 9, no. 6, p. 455 – 474, 2001.

[116] A. González, M. Aritio, R. Eyras and E. Ngom, “FAD Senegal: Program of Rural Photovoltaic Electrification,” in Nineteenth European Photovoltaic Solar Energy Conference, Paris, 2004.

[117] EDF, [Online]. [Accessed http://strategie.edf.com/fichiers/fckeditor/Commun/Developpement_Durable/2011/Acces_energie/2011/EDF_AccesEnergie_Senegal_vf.pdf October 2014].

[118] R. Mawhood and R. Gross, “Institutional barriers to a ‘perfect’ policy: A case study of the Senegalese Rural Electrification Plan,” Energy policy, vol. 73, pp. 480 ‐ 490, October 2014.

[119] REN 21, 2012. [Online]. Available: http://www.ren21.net/default.aspx?tabid=5434.. [Accessed October 2014].

[120] X. Lemaire, “Off‐grid electrification with solar home systems: The experience of a fee‐for‐service concession in South Africa,” Energy for Sustainable Development, vol. 15, no. 3, pp. 277 ‐ 283, September 2011.

[121] ONEE. [Online]. Available: http://www.one.org.ma/. [Accessed October 2014].

136

[122] D. Palit and A. Chaurey, “Off‐grid rural electrification experiences from South Asia: Status and best practices,” Energy for Sustinable Development, no. 15, pp. 266 ‐ 276, 2011.

[123] “Designing sustainable off‐grid rural electrification projects: principles and practices,” Washington, DC, 2008.

[124] Ministry of New and Renewable Energy (Government of India), [Online]. Available: http://mnre.gov.in/schemes/offgrid/remote‐village‐electrification/. [Accessed October 2014].

[125] J. Ondraczek, “The sun rises in the east (of Africa): A comparison of the development and status of solar energy markets in Kenya and Tanzania,” Energy Policy, vol. 56, pp. 407 ‐ 417, May 2013.

[126] A. Yadoo, Delivery Models for Decentralised Rural Electricication. Case studies in Nepal, Peru and Kenya, London for Environment and Development: International Institute, 2012.

[127] A. L. D’Agostino, B. K. Sovacool and M. J. Bambawale, “And then what happened? A retrospective appraisal of China’s Renewable Energy Development Project (REDP),” Renewable Energy, vol. 36, no. 11, pp. 3154 ‐ 3165, November 2011.

[128] H. A. Samad, S. R. Khandker, M. Asaduzzaman and M. Yunusd, “The benefits of solar home systems: an analysis from Bangladesh,” 2013.

[129] S. Komatsu, S. Kaneko, P. P. Ghosh and A. Morinaga, “Determinants of user satisfaction with solar home systems in rural Bangladesh,” Energy, vol. 61, pp. 52 ‐ 58, 2013.

[130] I. Sharif and M. Mithila, “Rural electrification using PV: the success story of Bangladesh,” Energy Procedia, vol. 33, pp. 343 ‐ 354, 2013.

[131] B. Sovacool and I. Drupady, “Summoning earth and fire: The energy development implications of Grameen Shakti (GS) in Bangladesh,” Energy, vol. 36, no. 7, pp. 4445 ‐ 4459, 2011.

[132] REN21, “Renewables 2014 Global Status Report,” 2014.

[133] R. Conor, “Bangladesh receives US$78.4 million from IDA to finance PV systems,” www.pv‐tech.org, 30 June 2014.

[134] M. Camino, Contribución al aseguramiento de la calidad en la electrificación rural con energías renovables, 2011.

[135] B. Parida, S. Iniyan and R. Goic, “A review of solar photovoltaic technologies,” Renewable and Sustainable Energy Reviews, vol. 15, no. 3, pp. 1625‐1636, April 2011.

[136] [Online]. Available: http://www.nrel.gov/ncpv/. [Accessed October 2014].

[137] M. Camino‐Villacorta, M. Á. Egido‐Aguilera and P. Díaz, “Test procedures for maximum power point tracking charge controllers characterization,” Prog. Photovolt: Res. Appl., vol.

Bibliography

137

20, p. 310 – 320, 2012.

[138] L. Narvarte, J. Muñoz and E. Lorenzo, “Testing of fluorescent DC lamps for Solar Home Systems,” Prog. Photovolt: Res. Appl., no. 9, p. 475 – 489, 2001.

[139] A. Gago Calderón, L. Narvarte, L. M. Carrasco and J. Serón, “LED bulbs technical specification and testing procedure for solar home systems,” Renewable and Sustainable Energy Reviews, vol. 41, pp. 1364 ‐ 0321, January 2015.

[140] S. R. Wenham, M. A. Green, M. E. Watt and R. Corkish, “Applied Photovoltaics,” 2nd edition ed., London, Earthscan, 2007, pp. 265 ‐ 286.

[141] B. Fahlenbock and S. Haupt, “Quality Standards for Solar Home Systems and Rural Health Power Supply,” Eschborn, 2000.

[142] IEC, TS 62257 Series ‐ Recommendations for small renewable energy and hybrid systems for rural electrification, 2.0 ed., 2013.

[143] E. Lorenzo, R. Zilles and E. Caamaño, Photovoltaic rural electrification: a fieldwork picture book, PROGENSA, 2001.

[144] E. Martinot, R. Ramankutty and F. Rittner, “The GEF solar PV portfolio: emerging experience and lessons. Monitoring and Evaluation,” 2000b.

[145] U. Hansen, M. Pedersen and I. Nygaard, “Review of Solar PV market development in East Africa,” in 1st Africa photovoltaic solar energy conference and exhibition proceedings, Durban, South Africa, 2014.

[146] S. Abdullah and A. Markandya, “Rural electrification programmes in Kenya: Policy conclusions from a valuation study,” Energy for Sustainable Development, vol. 16, no. 1, pp. 103 ‐ 110, 2012.

[147] R. Acker and D. Kammen, “The quiet (energy) revolution: analysing the dissemination of photovoltaic power systems in Kenya,” Energy Policy, vol. 24, no. 1, pp. 81 ‐ 111, 1996.

[148] ESMAP, “UGANDA: Rural Electrification Strategy Study,” 1999.

[149] IEA, “Pico Solar PV Systems for Remote Homes. A new generation of small PV systems for lighting and communication,” 2012.

[150] D. Soto, E. Adkins, M. Basinger, R. Menon, S. Rodriguez‐Sanchez, N. Owczarek, I. Willig and V. Modi, “A prepaid architecture for solar electricity delivery in rural areas,” in Proceedings of the Fifth International Conference on Information and Communication Technologies and Development, New York, 2012.

[151] A. Ellegård, A. Arvidson, M. Nordström, O. S. Kalumiana and C. Mwanza, “Rural people pay for solar: experiences from the Zambia PV‐ESCO project,” Renewable Energy, vol. 29, no. 8, pp. 1251 ‐ 1263, July 2004.

138

[152] A. Ballesteros, Keys to Achieving Universal Energy Access Series: Implementation Strategies for Renewable Energy Services in Low‐Income, Rural Areas, World Resources Institute, 2012.

[153] H. Zerriffi, “Innovative business models for the scale‐up of energy access efforts for the poorest,” Current Opinion in Environmental Sustainability, vol. 34, pp. 272 ‐ 278, 2011.

[154] M. Retnanestri and H. Outhred, “Acculturation of renewable energy technology into remote communities: lessons from Dobrov, Bourdieu, and Rogers and an Indonesian case study,” Energy, Sustainability and Society, vol. 3, no. 1, May 2013.

[155] G. Notton, M. Muselli and P. Poggi, “Costing of a stand‐alone photovoltaic system,” Energy, vol. 23, no. 4, pp. 289 ‐ 308, 1998.

[156] L. Qoaider and D. Steinbrecht, “Photovoltaic systems: A cost competitive option to supply energy to off‐grid agricultural communities in arid regions,” Applied Energy, vol. 87, no. 2, pp. 427 ‐ 435, 2010.

[157] IFC, “From gap to opportunity: business models for scaling up energy access,” Washington DC (USA), 2012.

[158] X. Lemaire, “Fee‐for‐service companies for rural electrification with photovoltaic systems: The case of Zambia,” Energy for Sustainable Development, vol. 13, no. 1, pp. 18 ‐ 23, 2009.

[159] M. Horn, “Solar photovoltaics for sustainable rural electrification in developing countries; the experiences in Peru,” in ISES Solar World Congress, Göteborg, 2003.

[160] A. Zomers, “The challenge of rural electrification,” Energy for Sustainable Development, vol. VII, no. 1, pp. 69 ‐ 76, 2003.

[161] World Bank, “The Welfare Impact of Rural Electrification: A Reassessment of the Costs and Benefits,” 2008.

[162] [Online]. Available: http://www.one.org.ma/. [Accessed November 2014].

[163] [Online]. Available: http://www.ffem.fr. [Accessed February 2015].

[164] M. Adhikari, “A Successful Public Private Partnership Model of Dissemination of Solar Home Systems in Nepal,” in 23rd European Photovoltaic Solar Energy Conference and Exhibition, Valencia (Spain), 2008.

[165] B. K. Sovacool, “Expanding renewable energy access with pro‐poor public private partnerships in the developing world,” Energy Strategy Reviews, vol. 1, no. 3, March 2013.

[166] “CEGID,” [Online]. Available: http://en.cegid.com/. [Accessed April 2015].

[167] R. Zilles, “A diagnosis on the need to establish a technical requirements protocol for home photovoltaic systems in Latin America,” Energy for Sustainable Development, vol. III, no. 2, pp. 38 ‐ 43, 1996.

Bibliography

139

[168] J. Muñoz, E. Lorenzo, F. Martínez Moreno, L. Marroyo and M. García, “An investigation into hot‐spots in two large grid‐connected PV plants,” Progress in Photovoltaics: Research and Applications, vol. 16, p. 693 – 701, 2008.

[169] P. Sánchez‐Friera, M. Piliougine, J. Peláez, J. Carretero and M. Sidrach de Cardona, “Analysis of degradation mechanisms of crystalline silicon PV modules after 12 years of operation in Southern Europe,” Progress in Photovoltaics: Research and Applications, vol. 19, no. 6, p. 658 – 666, 2011.

[170] G. Bopp, H. G. K. Preiser, D. U. Sauer and H. Schmidt, “Energy storage in photovoltaic stand‐alone energy supply systems,” Progress in Photovoltaics: Research ans Applications, vol. 6, no. 4, p. 271 – 291, 1998.

[171] Stationary lead acid batteries ‐ Vented types ‐ General requirements and methods of tests. International Standard IEC 60896‐11: 2002(E).

[172] P. Díaz, M. Egido and F. Nieuwenhout, “Dependability analysis of stand‐alone photovoltaic systems,” Progress in Photovoltaic: Research and Applications, vol. 15, p. 245 – 264, 2007.

[173] Universal Technical Standard for Solar Home Systems. Thermie B: SUP‐995‐96, EC‐DGXVII, 1998.

[174] P. O’Connor, Practical Reliability Engineering, 4th edition ed., FOURTH, 2002.

[175] J. Warleta, Fiabilidad. Bases teóricas y prácticas, INTA, “Esteban Terradas”, 1985.

[176] R. Esparza and S. Martínez, “Fiabilidad en un sistema de alimentación ininterrumpida,” Mundo Electrónico, vol. 48, 1976.

[177] M. Vázquez and I. Rey‐Stolle, “Photovoltaic Module Reliability Model Based on Field Degradation Studies,” Progress in Photovoltaic: Research and Applications, vol. 16, p. 419 –433, 2008.

[178] P. A. Tobias and D. C. Trindade, Applied Reliability, 2nd Edition ed., Chapman&Hall, 1995.

[179] B. L. Amstadter, Reliability Mathematics: Fundamentals, Practices; Procedures, McGraw‐Hill, 1971.

[180] P. R. Mishra and J. C. Joshi, “Reliability estimation for components of photovoltaic systems,” Energy Conversion Management, vol. 37, no. 9, p. 1371 – 1382, 1996.

[181] Almabat. [Online]. Available: http://www.almabat.ma/. [Accessed December 2014].

[182] Phocos. [Online]. Available: http://www.phocos.com/. [Accessed December 2014].

[183] IEC, IEC standard 60896‐11. Stationary lead‐acid batteries – Part 11: Vented types – General requirements and methods of tests, 2002.

[184] IEC, IEC standard 61427‐1. Secondary cells and batteries for renewable energy storage ‐

140

General requirements and methods of test ‐ Part 1: Photovoltaic off‐grid application, 1.0 ed., 2013.

[185] D. U. Sauer, “Electrochemical Storage for Photovoltaics,” in Handbook of Photovoltaic Science and Engineering, Chichester, UK, John Wiley & Sons, Ltd, 2015, pp. 896 ‐ 951.

[186] J. Muñoz and E. Lorenzo, “Capacitive load based on IGBTs for on‐site characterization of PV arrays,” Solar Energy, vol. 80, no. 11, pp. 1489 ‐ 1497, 2006.

[187] IEC, IEC standard 60904‐1. Photovoltaic devices ‐ Part 1: Measurement of photovoltaic current‐voltage characteristics, 2006.

[188] IEC, IEC standard 60891. Photovoltaic devices – Procedures for temperature and irradiance corrections to measured I‐V characteristics. International Electrotechnical Commission, 2009.

[189] R. Moreton, E. Lorenzo and F. Martínez‐Moreno, “Field Performance of PV Modules Quality Control Process,” in 23rd European Photovoltaic Solar Energy Conference and Exhibition, Valencia, Spain, 2008.

[190] G. Nofuentes, J. Aguilera, R. L. Santiago, L. De La Casa and L. Hontoria, “A reference‐module‐based procedure for outdoor estimation of crystalline silicon PV module peak power,” Progress in Photovoltaics: Research and Applications, vol. 14, pp. 77 ‐ 87, 2006.

[191] M. A. Muñoz‐García, J. P. Silva and F. Chenlo, “Influence of Initial Power Stabilization over PV Modules Maximum Power,” in 24th European Photovoltaic Solar Energy Conference, Hamburg, Germany, 2009.

[192] M. A. Muñoz, M. C. Alonso‐García, F. Chenlo and N. Vela, “Early degradation of silicon PV modules and guaranty conditions,” Solar Energy, vol. 85, no. 9, pp. 2264 ‐ 2274, 2011.

[193] F. J. Pern and A. W. Czanderna, “Characterization of Ethylene Vinyl Acetate (EVA) Encapsulant: Effects of Thermal Processing and Weathering Degradation on its Discoloration,” Solar Energy Materials and Solar Cells, vol. 25, pp. 3 ‐ 23, 1992.

[194] D. Berman, S. Biryukov and D. Faiman, “EVA laminate browning after 5 years in a grid‐connected, mirror‐assisted, photovoltaic system in the Negev desert: effect on module efficiency,” Solar Energy Materials and Solar Cells, vol. 36, no. 4, pp. 421 ‐ 432, 1995.

[195] E. Martinot, A. Cabraal and S. Mathur, “World Bank/GEF solar home system projects: experiences and lessons learned 1993–2000,” Renewable and Sustainable Energy Reviews, vol. 5, no. 1, pp. 39 ‐ 57, 2001.

[196] Energy & Development Research Centre, A review of international literature of ESCOs and fee‐for‐service approaches to rural electrification (solar home systems), Cape Town, 2003.

[197] A. Chaurey and T. Kandpal, “Assessment and evaluation of PV based decentralized rural electrification: An overview,” , Renewable and Sustainable Energy Reviews, vol. 14, no. 8, pp. 2266 ‐ 2278, 2010.

Bibliography

141

[198] F. v. d. Vleuten, N. Stam and R. v. d. Plas, “Putting solar home system programmes into perspective: What lessons are relevant?,” Energy Policy, vol. 35, no. 3, pp. 1439 ‐ 1451, 2007.

[199] M. Djamin, A. S. Dasuki and A. Y. Lubis, “Performance evaluation of solar home systems after more than ten years of operation in Indonesia,” in World renewable energy congress VI (WREC 2000), Brighton, 2000.

[200] S. Chowdhury, M. Mourshed, S. Kabir, M. Islam, T. Morshed, M. R. Khan and M. Patwary, “Technical appraisal of solar home systems in Bangladesh: A field investigation,” Renewable Energy, vol. 36, no. 2, p. 772 – 778, 2011.

[201] M. Schäfer, N. Kebir and K. Neumann, “Research needs for meeting the challenge of decentralized energy supply in developing countries,” Energy for Sustainable Development, vol. 15, pp. 324 ‐ 329, 2011.

[202] M. Nouni, S. Mullick and T. Kandpal, “Photovoltaic projects for decentralized power supply in India: A financial evaluation,” Energy Policy, vol. 34, pp. 3727 ‐ 3738, 2006.

[203] E. Ilskog and B. Kjellström, “And then they lived sustainably ever after?—Assessment of rural electrification cases by means of indicators,” Energy Policy, vol. 36, no. 7, pp. 2674 ‐ 2684, 2008.

[204] A. Woodruff, “An economic assessment of renewable energy options for rural electrification in Pacific Island countries,” Suva, 2007.

[205] Fraunhofer Institut Solare Energiesysteme—ISE, “Compendium of Projects on Rural Electrification and Off‐Grid Power Supply,” Freiburg, 2001.

[206] E. Rosenthal, Gams. A user's guide, Washington: Gams Development Corporation.

[207] IBM, User's manual for Cplex, IBM ILOG, 2014.

escuela tÉcnica superior de ingenierÍa y sistemas...

Documents