PREPARATORY PHASE OF THE SAMOA PHOTOVOLTAIC ELECTRIFICATION PROGRAMME Contract 2007-005 UNDP Apia, Samoa

FEASIBILITY REPORT AND RECOMMENDATIONS

FINAL VERSION

Submitted 26 December, 2008

Prepared by: Dr. Herbert Wade

Mr. Terubentau Akura

ii

ACKNOWLEDGEMENTS

The consultancy team of Dr. Herbert Wade (Team Leader), Mr. Bruce Clay (mini-grid survey), Mr. Mewang Gyeltshen (Survey Analysis) and Mr. Terubentau Akura (Institutional arrangements) wish to especially thank Mr. Thomas Jensen of UNDP for his continuing encouragement and support and Mr. Edward Langham of EPC for his outstanding management of the household surveys and his unfailing support of the project. Also the team acknowledges the high quality of effort by the survey team, by the EPC GIS team and of course the generosity of the survey respondents who gave of their time and knowledge to help make the survey a success. Inputs from the government, both national and local, were also important to the success of the survey and are acknowledged and are much appreciated. The resources generously provided by UNDP, SOPAC/PIEPSAP and EPC made the work possible. Their confidence in the team shown through their financial input is much appreciated.

iii

ACRONYMS CFL .................................................................... Compact Fluorescent Lamp DVD.................................................................................... Digital Video Disc EPC .........................................................Electric Power Company of Samoa GIS ............................................................... Geographic Information System GPS.......................................................................Global Positioning System kW .........................................................................Kilowatt (Electrical Power) Wh/day............................................... Watt Hour per Day (Daily energy flow) kWh.............................................................Kilowatt Hour (Electrical Energy) kWh/m2/day .......... Kilowatt Hours per square metre per day (Solar Energy) NGO ............................................................ Non Governmental Organisation PIC ...............................................................................Pacific Island Country PIEPSAP ........ Pacific Islands Energy Policies and Strategic Action Planning PV...................................................................................... Solar Photovoltaic SHS ................................................................................ Solar Home System SOPAC............................... Pacific Islands Applied Geoscience Commission SPSS ...................................................Statistical Package for Social Science TV/DVD ................................................................Television with DVD player UNDP ............................................ United Nations Development Programme UNESCO ... United Nations Educational, Cultural and Scientific Organization Wp......................................................Watts Peak (Solar panel output rating)

CURRENCY CONVERSIONS

All currency figures are in Samoan Tala (WST) unless specifically noted as US Dollars. The conversion used between WST and USD is 2.85 WST per 1 USD.

iv

TABLE OF CONTENTS

1 BACKGROUND ................................................................................................... 1

2 SOLAR RESOURCE............................................................................................ 1 2.1 Available measurements and solar data analysis .............................................................1

3 TECHNICAL CONSIDERATIONS........................................................................ 3 3.1 Design principles..............................................................................................................3 3.2 Solar Energy Estimate .....................................................................................................4 3.3 Sizing for load classes .....................................................................................................4 3.4 Modular approach to system installation ..........................................................................4 3.5 Component type and quality considerations.....................................................................5

3.5.1 Solar Panels ................................................................................................................................... 5 3.5.2 Battery............................................................................................................................................. 5 3.5.3 Charge/Discharge controller ......................................................................................................... 7 3.5.4 Switches ......................................................................................................................................... 7 3.5.5 Lights .............................................................................................................................................. 7 3.5.6 Voltage Converters for Radio Use ................................................................................................ 8 3.5.7 Inverters .......................................................................................................................................... 8

4 DEMAND SIDE MANAGEMENT.......................................................................... 9 4.1 Inverter and appliance policies .........................................................................................9 4.2 Appliance contract with user ..........................................................................................10 4.3 Shared large appliances ................................................................................................10

5 MINI-GRIDS........................................................................................................ 11 5.1 The Apolima Mini-Grid Installation..................................................................................11

5.1.1 The Design ................................................................................................................................... 11 5.1.2 Operation ...................................................................................................................................... 12

5.2 Mini-Grids for General Off-Grid Electrification ................................................................12 6 BIOFUEL USE FOR OFF-GRID ELECTRIFICATION........................................ 13

7 FINANCIAL CONSIDERATIONS....................................................................... 13 7.1 Cost of solar installations ...............................................................................................13

7.1.1 System sizing and costing process ............................................................................................ 14 7.1.2 Basic lighting and radio SHS system sizing............................................................................... 15 7.1.3 Lighting, radio and TV/DVD SHS................................................................................................ 16 7.1.4 Full electrification including refrigerator or freezer .................................................................... 17

7.2 Capital Costs .................................................................................................................18 7.3 Program cost .................................................................................................................21 7.4 Connection Charges and Tariffs.....................................................................................22

7.4.1 Connection Charges .................................................................................................................... 22 7.4.2 Tariff .............................................................................................................................................. 23

8 ENVIRONMENTAL CONSIDERATIONS ........................................................... 23

9 MAINTENANCE STRUCTURE .......................................................................... 24

10 SUMMARY OF RECOMMENDATIONS............................................................. 25 REFERENCES.............................................................................................................. 26

Executive Summary 1

EXECUTIVE SUMMARY Background In 2007, the first Samoa National Energy Policy was endorsed by Cabinet. In that Policy it is stated that the Vision of the policy is to “enhance the quality of life for all through access to reliable, affordable and environmentally sound energy services and supply” and that a Goal of the policy is to “to increase the contribution [of renewable energy] for energy services and supply by 20% by the year 2030.” In accordance with this policy and its own mission and goals, the Electric Power Corporation of Samoa has programmes to include hydro, biofuel, wind energy and solar energy in its generation mix and to extend services to provide access to electricity to all Samoa residents. Recent estimates indicate that as many as 95% of the population are electrified with the remaining 5% widely dispersed and largely at a distance from the grid that is too far to economically connect. In order to develop a process to make available electrical services to this currently off-grid population, EPC along with the Government of Samoa, the Secretariat of the Pacific Applied Geoscience Commission (SOPAC) and UNDP Samoa have undertaken a preparatory phase for a Samoa Photovoltaic (PV) Electrification Programme. The expected outcome of this preparatory phase is a detailed rural electrification program to provide the remaining non-electrified households in Samoa with 24-hour power. The first step of this process was to identify and survey the households that are currently off-grid (Survey Report Final Draft, Preparatory Phase Of The Samoa Photovoltaic Electrification Programme, Contract 2007-005 UNDP Apia, Samoa, 11 November, 2008). The second step was to complete a feasibility study for making available electrical services to those off-grid households that is the most economically reasonable mix of independent off-grid generation using renewable energy and extensions of the existing grid. This report provides the findings and recommendations from the feasibility study. Approximately 250 households in Samoa remain unelectrified due to the high cost of connection for households that are beyond about 60 meters from the existing grid. The households were self-identified and then approximately 200 were surveyed to determine the level of services desired, the ability and willingness to pay and their existing use of energy. For those households, the use of solar photovoltaics for individual electrification is being considered.

Photovoltaic system requirements for off-grid electrification To design stand-alone PV installations suitable for off-grid electrification it is necessary to know:

1. The amount (in kWh/m2) of solar energy received on average each day during the worst month of solar energy during an average year.

2. The losses in the system that reduce the energy available to the user. 3. The average daily load that is applied by the user.

Electrical load for the households Based on the results of the survey that indicates the need for electrical services in the off-grid households, four load classes of solar installation are proposed:

1. Class 1 to provide lighting and radio power (about 6.1 kWh/month)

Executive Summary 2

2. Class 2 to provide TV/DVD power (about 12.6 kWh/month) 3. Class 3 to provide both lighting and TV/DVD power (about 18.7

kWh/month) 4. Class 4 to provide sufficient electrification for refrigerators/freezers, fans,

lights, TV and other entertainment appliances (about 108.7 kWh/month). An additional class of installation is proposed that must be individually designed to fit the needs of an off-grid business such as an eco-tourist resort.

Based on the survey results, the great majority of the installations will be either class 1 or class 3. Since the appliance load for a class 3 installation is about double that of a class 1 installation, the same panels and batteries can be used in both but in the case of class 1 it would operate at 12V and in class 2 operation would have double the panels and batteries and be at 24V. Since the most likely upgrade in the future will be from basic lighting to lighting plus TV/DVD player, making the units modular can allow for ease in upgrades plus having the same panel and battery used in both will minimize spare parts costs.

Solar system component type and quality considerations The primary design consideration is to minimize O&M costs since the capital investment is likely to be available from external sources. Therefore high reliability components with long working lives are proposed.

Panels Few problems have been observed with solar panels from international suppliers and there need be no special considerations other than ensuring that they are indeed internationally certified to meet international manufacturing and testing standards.

Batteries The primary component that affects the O&M cost is the battery since not only is it relatively expensive, it also must be replaced periodically. For minimum O&M, the replacement period should be as long as possible. Typically open-cell batteries are chosen for their excellent benefit to cost ratio but they do require at least monthly maintenance in solar home system (SHS) installations. In Samoa the households are not concentrated into communities but dispersed over much of the two main islands. Therefore access cost is high. With open cell batteries monthly access is essential. With sealed batteries, access every two months is adequate for preventive maintenance of the system as a whole. The access cost for a two month cycle is sufficiently less to overcome the cost disadvantages of sealed batteries and for this project, sealed batteries are proposed for class 1 to 4 installations.

Controllers Charge/discharge controllers must be reliable since any failure of the controller will likely cause some damage to the battery. For sealed batteries a fairly complex controller is needed. The MorningStar PS series controller has been installed successfully in large numbers in the Pacific and is recommended for the class 1, 2 and 3 installations. For the class 4 installation, a MPPT (Maximum Power Point Tracking) type controller is recommended. The MX80 from Outback has thus far performed well in Apolima, Samoa, and is recommended.

Executive Summary 3

Switches Switches need to be low resistance, spring action toggle type switches to avoid increasing voltage drops over the long term. Lights Lights need to have high luminous efficiency and long life. Good quality screw base CFL type lights designed for 12V operation are satisfactory but probably better for SHS service in Samoa are two pin or four pin PL type bulbs plugged into an electronic ballast with an integrated light reflector such as are used in Kiribati. For night lights, 0.5W white LED lamps are optimal. DC/DC converter To operate radios, a DC/DC converter usually is needed to change the battery voltage to one that the radio is designed to use. A switching type converter such as the type assembled in Kiribati is recommended. Inverter For TV and DVD player installations, a small 150W modified sine wave inverter hard wired to the power leads of the two appliances is recommended. Also recommended is switching on the TV and DVD player with the inverter so the inverter only draws power when the TV is on. For the full AC electrification class 4 installation, the inverter must be always on. It must be a pure sine wave unit with a rated power output of at least 1.3 kW. The Outback FX series units are recommended as they are sealed against corrosion and have performed well in Apolima.

DEMAND SIDE MANAGEMENT Users need to be trained in the proper and efficient use of the solar system providing their power. On maintenance visits the technician should advise the users on how to operate the systems to get the most benefits. In particular users of full AC systems (Class 4) need to understand the limits of the system and need to manage electrical energy use efficiently.

COMMUNITY OWNED LARGE APPLIANCES It is technically practical to provide one household in a cluster of off-grid houses with a class 4 AC system and then expect other households in the cluster to share the refrigerator, washer or other large appliances installed in that house. However where such community sharing of solar powered appliances has been tried, it has not worked well due to arguments about the fair sharing of the resource and payment for services. If a group of households want to try such a plan, it is recommended that EPC not in any way participate other than by installing and maintaining the energy source.

MINI GRIDS Where a number of households are clustered into a compact group, a solar PV powered mini-grid may be more cost effective than individual SHS. In Apolima a PV powered mini-grid provides power for the village of about 100 residents. Installed is 13.76 kWp of panels with a 3000 Ah 48V battery. Outback controllers and Outback inverters provide up to 11.5 kW of 240V 50Hz power from the batteries. The system has been operational about two years with no power outages to date despite villagers using an arc welder and other high demand appliances from time to time.

Executive Summary 4

However, the opportunities for mini-grid use on Upolu and Savai’i are minimal since off-grid households are widely dispersed rather than being clustered. Some mini-grid opportunities do exist at eco-resorts with one being built at Saaga and one operating on Namua island. In both cases there is a central building with a number of separate guest fales on the grounds. EPC should design mini-grid installations on a case by case basis and tailor the installations to the needs of the business. Fees for power use should be negotiated with the user.

BIOFUEL USE FOR OFF-GRID ELECTRIFICATION Since houses are not clustered, diesel generators operating off biofuel will be more expensive to operate than individual solar for the households and are not recommended.

FINANCIAL CONSIDERATIONS System cost can be reasonably estimated as about US$8 per Wp of panel for DC installations (no inverter) and about US$9 per Wp for those with inverters.

O&M cost can be reasonably expected to consist of the cost of maintenance visits by the EPC technician plus replacement of the battery on a 5 year cycle, replacement of he controller on a 10 year cycle and replacement of any inverters on a 10 year cycle. System Sizing and costing process The first step in designing the systems is to determine the solar resource. Fortunately two pyranometers were installed over a year ago as part of an EPC/Government of Samoa/SOPAC/UNDP Upolu Wind Resource Assessment Project. By analysing the data collected by EPC from those two instruments, it was determined that the design value for solar radiation that should ensure satisfactory year around performance of stand-alone PV systems throughout Samoa is 3.97 kW/m2. That is used to determine the generation factor for sizing the solar panels. Based on Samoa conditions, a generation factor of 2.03 Wh/day per Wp of solar panel was calculated.

System losses for the basic lighting system was calculated to be about 28% and for the other systems to be about 35%.

The proposed design of the four system classes are:

Installation Type

Panel size (Wp)

Controller capacity

(Amperes)

Battery voltage and capacity in

Ah

Wh/day Appliance

Load Inverter size

Estimated installed cost

(Tala)

Basic lighting and radio 160 Wp 10 A 12V 125Ah 203 none 3,650

TV and DVD only 320 Wp 12 A 24V 125Ah 420 150 W 8,210

Lighting plus TV-DVD

160 Wp and 320 Wp

10 A and 12A

12V125Ah and

24V125Ah 623 150 W 11,860

Full AC electrification 2,746 Wp 30 A MPPT 48V,420Ah 3,623 1.3 kW 70,435

The design approach used is conservative and provides for some excess panel capacity which will provide a higher level of power reliability and lower overall O&M costs by keeping the battery at a higher average level of charge thereby increasing battery life.

Executive Summary 5

With a bimonthly visit by a preventive maintenance technician, costs associated with those four classes of installation are estimated at:

Estimated installed cost

Approximate capital cost

(Tala)

Estimated Component

Replacement Cost per Month

(Tala)

Approximate Service Cost per Month

(Tala)

Approximate Total

Monthly O&M Cost

(Tala)

kWh per month

required for the

appliances

Cost per kWh for O&M

only

(Tala)

Basic lighting and radio 3,650 15 15 30 6.1 4.92

TV and DVD only 8,210 31 15 46 12.6 3.65

Lighting plus TV-DVD 11,860 46 20 66 18.7 3.53

Full AC electrification 70,435 241 30 271 108.7 2.49

The household survey indicates that about 70% of the off-grid households are likely to opt for the basic lighting and radio unit, about 25% for the lighting plus TV/DVD unit and about 5% for the full AC electrification unit. Additionally a very small percentage may opt for an individually designed system for business use.

Although the cost of the solar electrification is higher than energy from the grid, the solar system provides lighting and radio service at lower cost as kerosene plus batteries (about 37 Tala on average) but the service from the solar is higher quality and more convenient.

Consideration should be given to grid extensions where the capital cost of the extension is equal or lower than the capital cost of the solar services desired. This is because the O&M cost for a LV grid extension up to about 1000 metres is lower than the equivalent cost solar installation. The average distance from the grid for all off-grid houses surveyed is about 407 metres which has an extension cost about the same as the cost of a Class 3 lighting plus TV/DVD installation. A Class 1 installation for lighting plus radio is about equivalent to an 80 metre extension. Therefore if funding can be arranged that allows either independent solar or a comparably priced grid extension, in many cases the grid extension makes the most financial sense. Based on the estimate of 70% in Class 1, 25% in Class 3 and 5% in Class 4 the total cost of an all-solar off-grid electrification would be around 2,382,840 Tala, and average cost of about 9,531 Tala (about US$3,345). If the installations were arranged so that grid extensions up to the equivalent solar cost are assumed, the capital investment remains the same but the overall O&M cost is lowered.

Connection Charges and Tariffs The household survey indicates that to have the majority of off-grid households agree to accept electrical services from PV, connection charges should not be greater than those charged for regular grid service connections. With about 70% of off-grid households likely to accept only basic lighting and radio services, the actual installation cost of the PV will be about the same as the actual cost of a basic grid connection (less the cost of a meter which will not be present). The larger solar installations will be greater in cost however. It is recommended that the connection fee for basic lighting and radio services be the same as for regular grid connections

Executive Summary 6

and if EPC wishes to recover some or all of the added cost of connection for larger PV installations that the monthly fee be increased to allow for their payback over a 10 year or longer period rather than charging all the fee at the time of connection. It is also recommended that the basic tariff for solar PV installations be based on the actual O&M cost of the class of installation that is made.

ENVIRONMENTAL CONSIDERATIONS Battery recycling will be important since the lead content of the batteries is an environmental hazard. It is recommended that EPC work with other organizations in Samoa to set up a recycling system for all failed lead-acid batteries. Since the number of failed batteries from automobiles is far greater than will result from EPC solar installations, the greater quantity should lower the average battery recycling cost for EPC and of course further benefit Samoa’s environment. The Kiribati recycling program operated by a local NGO may be a useful model.

MAINTENANCE STRUCTURE Preventive maintenance is essential to long battery life and long battery life is essential to minimize life cycle costs. Therefore EPC must arrange for a low-level but specialist trained technician to visit each site at least bimonthly to check for detrimental changes and make the necessary repairs. Because the sites are dispersed, it is not practical to follow the usual maintenance pattern of training a village technician to do the preventive maintenance checks. Also, turning the maintenance over to a private company is unlikely to be practical and adds problems for EPC due to the need to supervise and audit the company’s operations. Therefore to support the solar installations it is recommended that two persons at EPC be designated and trained as solar technicians. One person at a low level who is trained to just do preventive maintenance and one person at a management level who is trained in PV system design, component selection and purchasing, installation, troubleshooting and other skills necessary to support the field technician. It is expected that both persons would not be full time in their respective positions but would have other responsibilities within EPC as well.

SUMMARY OF RECOMMENDATIONS 1. That four classes of installations be established. Class 1 and Class 2

installations would use the same set of components with Class 1 providing sufficient energy to operate lights and radio at 12V and Class 2 providing sufficient energy to operate a TV/DVD player at 24V through a dedicated inverter. Class 3 would be a combination of both Class 1 and Class 2 installations. Class 4 would provide an AC supply sufficient to power a refrigerator, lights, fans, TV/DVD, and assorted small appliances.

2. That EPC work with eco-tourism developers and other off-grid businesses to provide solar power for their operations on a negotiated fee basis.

3. That the approximately 60 off-grid households who are up to 120 metres from a grid connection receive a grid connection upon payment the basic connection fee. An 80 metre connection represents about the same investment by EPC as the installation of a Class 1, basic lighting and radio PV system and the O&M cost of a grid extension appears sufficiently lower than hat of solar PV to pay the additional 40 metres as amortised over the life of the installation. This action is estimated to reduce the number of PV installations

Executive Summary 7

needed by about half and their associated O&M cost by about one-third while increasing the initial capital cost only slightly.

4. That solar connection fees be the same as those for basic grid connections. The cost of installation of a Class 1 PV system is expected to be about the same as the cost of installation of a grid extension. If EPC wishes to recover some or all of the added cost of installing the larger PV installations the cost should be recovered through a higher monthly charge, not as an initial connection fee.

5. That solar power tariffs be based on the estimated O&M cost for the class of installation provided.

6. That EPC provide all maintenance for the PV installations with one field person trained in preventive maintenance procedures and one management person trained in system sizing, component specification and purchasing, troubleshooting and repair.

7. That EPC work to establish a battery recycling programme that includes automotive batteries as well as solar batteries.

1

1 BACKGROUND In 2007, the first Samoa National Energy Policy was endorsed by Cabinet. In that Policy it is stated that the Vision of the policy is to “enhance the quality of life for all through access to reliable, affordable and environmentally sound energy services and supply” and that a Goal of the policy is to “to increase the contribution [of renewable energy] for energy services and supply by 20% by the year 2030.”

In accordance with this policy and its own mission and goals, the Electric Power Corporation of Samoa has programmes to include hydro, biofuel, wind energy and solar energy in its generation mix and to extend services to provide access to electricity to all Samoa residents. Recent estimates indicate that as many as 95% of the population are electrified with the remaining 5% widely dispersed and largely at a distance from the grid that is too far to economically connect. In order to develop a process to make available electrical services to this currently off-grid population, EPC along with the Government of Samoa, the Secretariat of the Pacific Applied Geoscience Commission (SOPAC) and UNDP Samoa have undertaken a preparatory phase for a Samoa Photovoltaic (PV) Electrification Programme. The expected outcome of this preparatory phase is a detailed rural electrification program to provide the remaining non-electrified households in Samoa with 24-hour power. The first step of this process was to identify and survey the households that are currently off-grid (Survey Report Final Version, Preparatory Phase Of The Samoa Photovoltaic Electrification Programme, Contract 2007-005 UNDP Apia, Samoa, 11 November, 2008). The second step was to complete a feasibility study for making available electrical services to those off-grid households that is the most economically reasonable mix of independent off-grid generation using renewable energy and extensions of the existing grid. This report provides the findings and recommendations from the feasibility study.

2 SOLAR RESOURCE

2.1 Available measurements and solar data analysis As Upolu and Savai’i are both mountainous islands, the solar energy can be expected to vary substantially over the islands. NASA data for Samoa are shown in Table 1

Table 1 – NASA Satellite data for the area including central Upolu

Source: http://eosweb.larc.nasa.gov/

Since the satellite data are based on about 60 km square areas, the data represent a large area average that includes some surrounding ocean. In general the solar value shown by NASA can be considered to be substantially higher than will be observed by ground stations since the mountains of the islands will cause increased cloud cover inland, particularly on the south-eastern quadrant that is the first to receive air moving in from the ocean.

2

The primary resource for solar radiation measurements from ground stations are from two 30 metre wind resource survey masts1 that include a Licor brand medium quality pyranometer with an expected accuracy range of about ±10%2. The pyranometers are mounted on a 1 metre arm extending north from the mast at about a 3 metre level. On clear days during late December and early January when the sun is at its maximum travel to the south, the pyranometer is in the shade of the mast for a short time at mid-day causing some error in the readings for those dates. Because there are few clear days during that time of the year, the error is not considered serious but it is recommended that the pyranometers be relocated to a location either very high on the mast where the shadow extension is less than one metre or ground mounted well away from the extent of the noon-time shadow cast by the mast. That is estimated to be a location about 8 metres to the north of the mast3. Since it is not reasonable to have an 8 metre arm extending from the mast, the pyranometer could be mounted on a post 8 metres north of the mast with the instrument wire extended to the existing data logger that is mounted on the mast near its base. When remounting the pyranometer, it is suggested that consideration be given to tilting the pyranometer to the north at the latitude angle as will be the case for the solar PV installations that will be sized based on the data obtained. This will allow the data to be directly used for solar PV system sizing without the need for complicated and somewhat inaccurate data conversion from the horizontal measurements to the energy received on the tilted panels. Data collected from the tilted radiometers must be clearly labelled as such so there is no confusion with other data collected with horizontally mounted pyranometers as are used for climatological and agricultural purposes. One Licor pyranometer is located on the south coast of Upolu at Aleipata and the other at Afulilo, the site of the main hydro development on Upolu. Although a full year of solar data has been received for the Afulilo site, only June to December data, mostly dry season data, has been provided for Aleipata. For that June to January period that has data available for both sites, the Afulilo site – one of the wetter areas of Samoa – shows a solar resource about 20% lower than that for Aleipata during that half-year period. Table 2 – Licor Horizontal Pyranometer data from Afulilo (December 2006 through November 2007) and that Data Converted to Estimated Solar Energy Received on a 14° Tilted Surface in Samoa.

Month of Data

Dec

2006

Jan

2007

Feb

2007

Mar

2007

Apr

2007

May

2007

Jun

2007

Jul

2007

Aug

2007

Sep

2007

Oct

2007

Nov

2007

Dec

2007

Afuilo Tilted 14° 4.24 3.12 3.89 3.83 4.45 2.54 3.82 3.44 4.26 4.34 4.43 4.35 4.14*

Aleipata Tilted 14° N/A N/A N/A N/A N/A N/A 4.66* 4.40 5.11 5.29 5.54 5.35 5.02*

*Full month not available. Number is based on extrapolating about two weeks of data.

The measurements have been made using a horizontally mounted pyranometer and need to be adjusted for the tilt of 14° to the north that will be the optimum for solar 1 These where installed as part of the EPC/Government of Samoa/SOPAC/UNDP Upolu Wind Resource Assessment Project 2 Though high accuracy thermopile type solar instruments are available from Eppley Laboratories, Kipp & Zonen and others, for PV system design purposes the accuracy of the installed instrumentation is adequate and the very much higher cost of climatological research grade instruments is not justified. 3 This is computed by the formula Sin(23.5-Latitude) times mast height which in this case provides a result of 4.95 metres minimum. Some added distance is suggested as the shadow length is longer if the mounting is not exactly due north of the mast

3

energy reception. The conversion from horizontal to tilted radiation is complex and requires both direct radiation and diffuse radiation data (as well as a number of minor parameters such as surface albedo). Therefore that cannot be directly done using the available data since only total global radiation (a constantly varying combination of both direct and diffuse radiation) is being measured. For such cases, when the site is within the tropics and therefore the tilt angle is low, although the NASA data itself is not useable due to the large area covered by the satellite measurements, using the month by month ratio of horizontal to tilted radiation seen in the NASA data provides an acceptable horizontal to tilted radiation conversion for the site. That horizontal to tilted surface conversion method has been used in this report and the results are shown in Table 2. For the PV system sizing calculations, the solar input used is 3.27 kWh/m2/day which is the average energy received during the three lowest solar months in Afulilo for a panel facing north and tilted at the latitude angle of 14°. If several years of data had been available, the average of the lowest month of each year would have been used but since only one year of data is available and the data from the month of May appear extraordinarily low and probably are anomalous, averaging the three lowest months appears reasonable. Given that Afulilo is expected to have a generally higher level of cloudiness than that for most areas of Samoa, the use of Afulilo data will be conservative and should result in an adequately sized PV system anywhere in Samoa where there are off-grid homes. For homes in areas with lower levels of cloudiness, for example those along the north west coast of Upolu, the panels provided through this assumption will be somewhat oversized and more costly than absolutely necessary. But since oversizing of panels tends to increase the battery life, fewer battery replacements will be needed over the life of the project thereby lowering the O&M cost. Therefore, the life cycle cost, including capital investment and O&M, is not likely to be significantly higher than would be the case if the panel size were reduced to fit the higher solar resource at those sites. If a donor agency provides grant funding for the panels while EPC must pay for the system O&M, then the oversizing will result in lower out-of-pocket costs for EPC.

3 TECHNICAL CONSIDERATIONS

3.1 Design principles The design of an independent solar PV installation depends on three principal factors:

1. The kWh/m2 of solar energy received on average each day during the worst month for solar energy during an average year.

2. The losses in the system that reduce the energy available to the user. This includes losses in wiring, connectors, switches, batteries, controllers and, if present, inverters and DC/DC converters.

3. The average daily load that is applied by the user. This is estimated from survey information and from experience with similar installations in other Pacific Island Countries. This is the most difficult component of the design process to determine. Also, users vary in their desire for energy from basic lighting up to full off-grid electrification that includes lighting, fans, refrigeration, TV and other appliances.

4

3.2 Solar Energy Estimate For the PV system sizing calculations, the energy received during the lowest solar months in Afulilo is used. This is used because it appears likely that the Afulilo data is reasonably representative of the solar energy availability for inhabited areas of Samoa that are the most cloudy. Although it is expected that the wet season solar data for Aleipata will be lower than the average for the dry season data already received, it is unlikely to ever go lower than the lowest month of solar input for Afulilo. For homes in areas with lower levels of cloudiness, for example those along the north-west coast of Upolu, the panels provided through the use of this assumption will be somewhat oversized and more costly than absolutely necessary. But oversizing of panels tends to increase the battery life. So the actual life cycle cost, including capital investment, is not likely to be significantly higher than would be the case if the panel size were reduced some 15% to 20% as would be allowed by the higher solar resource at those sites. For the reduced size systems, more frequent battery replacements would be required and the O&M costs higher.

3.3 Sizing for load classes In order to minimize problems of spare parts stocking, maintenance personnel training and general administration, it is not recommended that installations be each specifically designed to match the expected loads. Four general classes of solar PV installations are proposed based on the information from the user survey:

1. Basic lighting and small entertainment appliance use (radios or CD/cassette players)

2. Lighting, radio and TV/DVD use 3. Lighting, radio, TV/DVD, refrigerator / freezer and small appliance use.

4. Special usage class for tourist resorts, rural businesses and customers with special requirements.

Which class of service is accepted will be very dependent on the fees charged for that service. If a fee is charged that is comparable to lighting kerosene costs and dry battery costs for small users or generator fuel costs for large energy users, then from the analysis of survey data it is estimated that approximately 70% of off-grid users will opt for basic lighting services, 25% for lighting, radio and TV/DVD use and less than 5% will opt for class 3, full electrification. Probably no more than 1% of off-grid households will fit into the class 4 category where there are special needs for business applications.

3.4 Modular approach to system installation For the addition of a TV/DVD player, the energy use for TV/DVD use alone is typically about double that for lighting and radio use. To simplify maintenance, technician training, installation and spare parts stocking, it is recommended that the installations be modular and use the same panels and batteries. To double the available energy for the TV/DVD player, the use of two panels in series and two batteries in series for 24V operation is practical. This allows the use of the same panels and batteries as in the basic lighting system while providing double the energy available. Since the most likely future upgrade will be from basic lighting to lighting plus TV/DVD, this modular format will allow easy upgrades. Having separate systems for lighting and for TV will also prevent the loss of lighting services due to

5

excessive TV use running the battery down and of course will also avoid the loss of TV services due to excessive lighting use. These modular systems singly or in combination can be expected to supply around 95% of the off-grid customers. Unfortunately the small percentage of customers expected to opt for class 3 or 4 installations cannot be served through increasing the number of basic system modules so there will be a requirement to stock some spares just for those two classes of installation. However only five or six percent of the installations are expected to be in those two categories so the number of spares needed will be small.

3.5 Component type and quality considerations From the point of view of EPC, the key financial issue is the cost of O&M for the installations. Much, if not all, the capital cost is expected to be borne by outside sources of funding, so the designs should emphasise minimizing the O&M cost. This means modest oversizing of panel capacity, a high quality battery and a highly reliable charge-discharge control unit.

3.5.1 Solar Panels

There have been few problems with solar panels sourced from major international suppliers that comply with international testing and construction standards. For all manufacturers that meet international testing standards the performance of different panels is about the same with both single crystal and polycrystalline type solar panels from all manufacturers that meet international standards performing reliably in excess of 25 years. For stand alone installations in the Pacific Islands, thin film type panels are not recommended at this time due to uncertainties in their long term performance.

3.5.2 Battery

A major design consideration is whether to use an open cell or a sealed, valve regulated battery. Table 3 provides a comparison of the two types of battery for solar home system service. Table 3. Sealed vs. Open cell battery

Open Cell Batteries Sealed Batteries

Require monthly checks of electrolyte Significantly higher cost for comparable quality of battery

Requires access to pure (distilled) water supply for replacing lost electrolyte

Shorter operating life for a sealed battery vs. an open cell battery of comparable quality.

Easy to check each cell for early sulphation and may be possible to repair cells through an equalization charge

Cannot test for sulphation or do a cell repair through an equalization charge.

Comes dry charged so spares are easily stored without special charging facilities needed. Should be stored in an area that does not frequently exceed 35°C.

When in storage must be kept charged so a special storage arrangement for spares has to be provided so they are kept under charge. Must be kept in a cool area. Storage life is greatly diminished if stored above 35C.

Simple on-off controller is adequate and easy to make very reliable.

Require a higher quality controller because more sensitive to overcharging than open cell units. Difficult to find suitable controllers that meet the technical requirements and still provide long term reliability under Pacific Island conditions.

2 year or more storage life for dry charged batteries No acid handling or acid safety issues.

6

Open Cell Batteries Sealed Batteries

Better training for technicians is needed since battery testing and maintenance has to be included. There are some safety issues with open cell batteries due to access to acid in the battery.

Must receive batteries and initiate recharging within 4 months of manufacture or sulphation may occur.

Higher temperature storage and operation reduces life of sealed batteries more than comparable quality open cell batteries

In general, longer and more satisfactory service can be obtained from open cell batteries at a lower battery cost than for sealed batteries of equivalent capacity. The only significant advantage of using sealed batteries for solar installations is that they do not need to be serviced monthly for electrolyte checks. Open cell batteries need to have distilled water added to replace any electrolyte that has been lost though gassing and evaporation. When installations are clustered in communities, it is practical to train a member of the community to do the monthly checks at a low per-visit cost. In Samoa, the installations are not concentrated in clusters but spread over most of the two main islands, so it is impractical to have community based maintenance. With the requirement for centralized maintenance, access cost becomes significant. Without the requirement for electrolyte checks, system preventive maintenance can be carried out bimonthly instead of monthly thereby cutting access and personnel costs in half for O&M. With a probable cost of about $30 Tala labour and access cost per visit4 for maintenance, that is about $180 Tala per year in O&M savings per installation. For a five year battery life, that represents about $900 Tala in total savings, an amount that is more than the added cost of a sealed battery relative to an open cell battery of similar quality and capacity. Those savings are partly offset by the need to replace the higher cost sealed batteries somewhat more frequently than open cell batteries. However, overall it appears that the sealed batteries will incur a lower lifecycle cost (capital + O&M) as open cell batteries for general household use in Samoa. In particular if the initial capital cost is borne by a donor agency, the sealed battery becomes clearly lower in life cycle cost versus an open cell battery due to the lower O&M cost per system.

Note that for large mini-grid type installations such as might be installed for eco-tourist facilities or other off-grid businesses, open-cell batteries will provide a better life cycle cost since these large industrial grade batteries do not require checking for electrolyte level more often than every six months or so. Also for these types of installations, the business can themselves check the batteries periodically (as does the village representative in Apolima) and notify EPC if there is need for service. This approach is not workable for household type installations, however, and for those the sealed battery is preferred since with open cell batteries, expensive monthly checks by an EPC technician will be needed.

Based on the above, if EPC can provide a spare battery storage area that provides trickle or periodic charging to stored batteries and if the storage area does not exceed a temperature of 35°C, the use of sealed batteries for the off-grid household installations is recommended.

4 Assumes $90 Tala/day for the technician, 8 households visited per day and about 15 Tala per site for the cost of access.

7

3.5.3 Charge/Discharge controller

The primary operational requirement for a controller is reliability of operation. Any type of failure will almost always result in some battery damage even if repaired within a few days of noticing the problem. Therefore controllers that are complex and those that are sensitive to damage from the Pacific environment are usually not a good choice since they have a lower reliability than simple controllers specifically designed for the Pacific environment such as those used in Kiribati for rural electrification. However, the very simple, high reliability controller used in Kiribati is not accurate enough for use with sealed batteries due to the requirement of sealed batteries to maintain a very accurate and stable maximum charge voltage for strict protection against overcharging.

Another concern is lightning damage due to induced voltage surges cased by the strong electromagnetic field that is generated by lightning. The simple relay type controller used in Kiribati is highly unlikely to be damaged by lightning induced voltages but controllers using semiconductor switches are quite easily damaged by those voltages if not well protected by voltage spike suppression devices. Of the controllers that have been used in the Pacific, the MorningStar Pro-Star series of controller is known to be suitable for use with sealed batteries and has shown good reliability and stability. Additionally, it appears to be well protected against lightning damage. Based on generally good experience with the Pro-Star series and the generally poor performance seen with the other brands of semiconductor type controllers that have been tried in the Pacific, it is recommended that the MorningStar Pro-Star controller that has the appropriate current carrying capacity for the Samoa installations be considered for use in Samoa.

3.5.4 Switches

Switches need to be of a type that will not have a significantly increased voltage drop over time in Pacific Island service. Typical low cost 240VAC light switches are known to have problems with voltage drops that are excessive when used in 12V DC solar home system (SHS) circuits. Switches selected for SHS use need to be of the spring activated “toggle” type that have a strong snap action connect and disconnect action. That reduces arcing in the DC circuits of SHS and also makes a firm, clean contact that shows little change in resistance over time.

3.5.5 Lights

In remote installations that have a high access and supply cost, high reliability lights are the best economic choice. For Samoa, household members frequently visit urban areas and obtaining replacement bulbs is much less of a problem. Therefore, though high luminous efficiency remains a requirement, very long bulb and ballast life is not such a high priority as is the case with very remote installations. It will, however, be necessary to ensure that the proper lighting fixtures and bulbs are stocked by local shops. High quality, 12V DC compact fluorescent lights should be used in the SHS installations. Preferred are lights that have separate electronic ballast and bulb so that the failure of the bulb does not make it necessary to replace the ballast and vice-versa. Cleaning of lamps is also much easier with removable bulbs. Two pin and four pin PL type bulbs have been successfully used for over a decade in several Pacific Island Countries and are recommended for use in Samoa in association with a good quality

8

fixture that includes the electronic “ballast” and appropriate socket for the bulb. PL type bulbs in 7 Watt, 9 Watt and 11 Watt sizes are readily available.

Spiral type, Edison screw base 12V DC CFL bulbs with integrated bulb and ballast have also been used in PICs. Early Pacific installations saw a high failure rate with this type of bulb but the recent experience has been much better. The main problem with these bulbs is that they are fragile and difficult to clean without breakage. Another problem is that they are physically identical with most 240V AC CFL bulbs. This can cause confusion on the part of users and in some cases use of the wrong type of bulb may cause damage to lights or to other parts of the SHS installation. For these reasons, the spiral type, Edison screw base lights are not recommended.

Night lights are recommended as lighting left on all night was commonly found during the survey. Since the level of lighting need not be high (users of kerosene lamps typically turn down the wick to its minimum for night light use), it is recommended that low wattage (0.5W) LED type night lights be installed. The very long life, good luminous efficiency and robustness of LED lighting make them ideal for low level illumination. If night lights are not installed, all night usage of one of the main lights may occur and that will rapidly drain the battery and cause power outages and shorten battery life.

3.5.6 Voltage Converters for Radio Use

Most radios are not designed to operate on 12V DC. More common is 9V, 6V or 4.5V DC. Therefore a DC to DC voltage converter will usually be needed for radio operation and it will need to be adjusted to the voltage needed by the customer’s radio. Although some converters are easily changed from one setting to another, it is recommended that a converter be properly matched to the radio by the technician both with regards to the correct wiring connection and the correct voltage and the user not be able to easily change the converter settings. Inexpensive DC/DC converters that change the voltage by inserting a controllable resistance between the 12V supply and the radio input work satisfactorily but are inefficient. Switching type DC/DC converters operate much more efficiently and provide better voltage regulation than series type regulated voltage converters. The switching type of DC/DC converter is therefore recommended. The DC/DC converter for radio power as has been used in Kiribati for more than a decade is known to be reliable and technically adequate for PIC use. The converter is assembled in Kiribati and may either be purchased directly from Kiribati or arrangements for assembly in Samoa can be made.

3.5.7 Inverters

For class 2 and 3 installations, an inexpensive but good quality modified sine wave inverter of about 150 W capacity is adequate. The inverter should be hard wired to the TV and DVD player so no other equipment can be plugged in. The inverter should turn on with the TV so no current is drawn while the TV is not operating. For class 4 use, the inverter must provide a pure sine wave and should have a capacity of at least 1.3 kW. The Outback FX series is recommended because it is hermetically sealed against internal corrosion and also that model class has worked well at Apolima.

9

4 DEMAND SIDE MANAGEMENT The energy from solar is expensive and limited in quantity. Therefore for the best service, customers will need to manage the use of electricity efficiently. This means that appliances need to be efficient and also energy users need to be trained in the operation of the solar installations to make the most efficient use of the energy that is available. Energy efficiency in appliances is assured with SHS that operate only lights since the lights are provided as part of the installation. Users must, however, be educated in the proper use of the lights and they must clearly understand that the energy resources available to operate their appliances vary according to the level of cloudiness that occurs. Therefore for all SHS installations, a user training component will be needed at the time of installation. After the initial training, as a part of the maintenance process users should be interviewed at each visit by service technicians and advised regarding proper use of the system.

4.1 Inverter and appliance policies For small installations that include AC appliances such as a TV or stereo, the problems of maintaining energy efficiency become much more difficult to solve. The system design has to assume the use of the specific appliances used in the household. If an inverter is supplied that the customer is able to use for other than the intended appliance, there is a very strong likelihood that energy use will exceed the design capacity of the installation. Of the many solar electrification projects that have included an inverter for general AC appliance use, all have had serious problems with customers using the AC power for additional appliances beyond those specifically included in the system design. This has caused frequent power outages, shortened battery life and led to customer dissatisfaction. To avoid this problem, it is recommended that any AC appliance that is to be used with an EPC installed residential system served by a class 2 or class 3 system must have a “hard-wired” inverter installed. The inverter will not have an open socket for connecting the appliance but will instead be wired in line with the appliance for connection to a 12VDC power point in the building. This in essence is the same approach used for laptop computers and other DC appliances that must work off AC power points: an external converter is included that takes the 240VAC from the power point and converts it to low voltage DC at the Voltage and Amperage specifically needed by the circuits in the laptop computer. In the case of the residential SHS, an inverter would be supplied that EPC would wire into the power cord of the AC appliance that would be sized just to supply the amount of 240V 50Hz power needed by the appliance and the type of inverter would be selected to provide the AC wave shape optimised for the appliance. The appliance could not be disconnected from the inverter to allow the inverter to be used for any other appliance.

Through the use of this dedicated inverter approach, not only is the customer not able to abuse the 240VAC connection by connecting other appliances, the inverter can be selected to most efficiently supply the AC power to the appliance. In particular, an inexpensive inverter that does not provide a pure sine wave output can be used for electronic devices such as TVs and computers while a pure sine wave inverter can be designated for an appliance with a motor such as a fan.

For class 4 use, a central sine-wave type inverter is appropriate but users must be well educated in the efficient use of appliances and must clearly understand the limits of the installation.

10

For business use, where the system design is specifically fit to the needs of the user and fees are negotiated rather than set as a class, the use of a central inverter of sufficient capacity for all appliances is appropriate and recommended. The concept of dedicated inverters for each AC appliance applies only to residential SHS installations in classes 2 and 3.

4.2 Appliance contract with user As a further measure to assure that demand side management occurs, the specific appliances and their power demand requirements that are to be used with the SHS should be specified in the services supply contract and should the customer wish to change the appliance mix, written EPC notification and approval should be required. The contract should provide for system disconnection if the supply contract terms are not followed.

4.3 Shared large appliances Refrigerators and freezers are among the top four appliances desired by many households. However to provide solar power for a refrigerator – even if just sitting idle with nothing in it – requires a PV installation three to four times the size needed for basic lighting and entertainment. The value of the services provided by the solar powered appliance is modest while the operating cost is much higher than that of a grid connected appliance. The household survey indicates that many off-grid households sometimes store food in refrigerators or freezers located in the on-grid homes of relatives and some households indicates a willingness to consider sharing the use of refrigerators, freezers and washing machines.

The concept of community owned large appliances has been tried in a number of off-grid electrification projects. Unfortunately most have failed to provide the services desired and have been discontinued largely due to squabbles among users about availability and who has to pay for repairs. In one case of a community refrigerator installed in the Pacific, a disgruntled user destroyed the refrigerator to make it impossible for anyone to use it. If the multiple users have close family ties, this approach possibly can be made to work but even then shared resources often result in social problems with the person acting as host for the resource frequently blamed for any problems or accused of using more than their share of the resource. In any case, for Samoa the community ownership approach cannot generally be used since the off-grid houses are almost all dispersed and not in community clusters. This dispersion of off-grid houses also makes it unlikely that an entrepreneur could find it profitable to operate refrigerators, freezers or washing machines on a fee-for-use basis since that approach also requires that client households be able to conveniently access the appliances. With the off-grid houses dispersed widely, the customer base for the entrepreneur would not likely be large enough to keep the user fees acceptable to the off-grid households. For those few areas where there are clusters of off-grid households quite far from households with a grid connection, consideration could be given to providing a common refrigerator / freezer for that cluster but for that to work there must be a unequivocal acceptance of the terms of use by all parties and it must be understood that the large PV system needed to operate the community appliance will be removed if payment to EPC for its use is not continued.

Due to the lack of clustering in off-grid houses, it probably is not reasonable to install PV powered multi-user appliances nor is it likely that on-grid entrepreneurs will

11

develop use-for-fee appliance access for off-grid households. Therefore off-grid households will only have the options of continuing the use of friend’s and relative’s appliances or accepting a high cost PV system to operate their own refrigerator, freezer or washer. Based on the household survey results, less than 10% of off-grid households are likely to consider the high cost of solar power for refrigerator, freezer or washing machine use to be acceptable.

However it is reasonable that consideration for such an approach be given on a case by case basis and if a customer group wishes to pursue that course it is technically reasonable. But it is recommended that EPC only develop the needed power source and not in any way get involved with operating such shared resource arrangements, since, based on experience elsewhere, they are likely to be quite troublesome.

5 MINI-GRIDS

5.1 The Apolima Mini-Grid Installation The only mini-grid installation in Samoa (and one of very few in the Pacific Islands) is on the island of Apolima. The island is a volcanic caldera about 4 km across and rises abruptly from the sea in the channel between the main islands of Upolu and Savai’i. There is one village of about 100 residents that is located close to the sole access to the sea. All goods from the outside must be transported by small boat by way of a narrow passage through the rocks that is not passable when waves are high. In 2005 EPC faced having to replace the ageing and expensive to operate diesel generator with new equipment. Solar was selected as a cost effective solution for a nearly automated power system that requires no fuel. Feasibility and design funding came from UNDP, UNESCO and an NGO from Denmark: The Organization for Sustainable Energy. Hardware was tendered, purchased and installed by EPC using community development funds provided by the Government of Samoa.

5.1.1 The Design

The design used for Apolima includes protection against the high temperature, salty environment and high humidity of Apolima plus built-in redundancy to avoid total system failure should there be a component failure.

Sizing of the installation assumed the minimum energy requirement was the actual energy used from the part-time diesel and for the maximum a village survey to determine what changes in load would occur when electricity became available 24 hours a day was carried out. The design process that resulted in the success in Apolima is the same the method used in this report for sizing the proposed stand-alone installations. The total array consists of 84 modules at 160Wp each for a total array size of 13.76 kWp. Four Outback 60A MPPT controllers provide for optimal battery charging. Five sealed, Outback model 2300VA 230V 50hz inverters provide up to 11.5 kW from the 48VDC input. The Outback inverters are sealed to protect against the salt laden air of the island. The stacked inverter arrangement with associated control hub provides for continued operation at reduced capacity should one or more of the inverters or charge controllers fail. The control system also shuts down one or more of the inverters when fewer inverters are needed to meet the load requirements. The array frame was constructed using a treated pine columnar structure and Unirac aluminium mounting rails for attaching the modules. The structure was split into

12

several arrays to fit the sloping, uneven ground. Samoa sees some tropical cyclones (hurricanes/typhoons) so the mounting design was strengthened to survive high winds. The power house is a cyclone proof ventilated concrete cube with one room housing the 24 cells making up the 3000 Ah 48V battery bank and a second small room for the inverters and controls. To keep the batteries and electronics as cool as possible, a wooden shading roof is installed over the concrete building fully shading the building and providing for free air flow over the concrete roof of the power house. The high mass of the concrete combined with the shade, ventilation and its white wall surfaces makes the power house the coolest structure in the village and satisfactory for the batteries and inverters without any air conditioning or forced air ventilation. Transport of batteries was the biggest installation problem. Each battery cell weighs 155kg without electrolyte and 24 of them had to be manhandled into and out of the small aluminium boat that landed them at the beach and then hand carried 300m along an inland trail using a “stretcher” fashioned from local timber.

5.1.2 Operation

The PV mini-grid system on Apolima has provided grid quality, 24 hour power with no power interruptions for the more than two years since start-up. All the environmental problems associated with the diesel generator – oil/fuel spills, noise, and air pollution – have been eliminated. All electricity requirements of the village, including the use of an arc welder, have been satisfactorily met.

Management is by EPC but a local person checks the readings of meters and the battery specific gravity, logging the data for EPC technicians should a problem develop. During the first 18 months of operation, the system delivered to customers an average of 21 kWh per day with a daily peak demand between 4.5 and 6.5 kW and is sized to allow substantial increased demand in the future.

5.2 Mini-Grids for General Off-Grid Electrification The primary advantage of mini-grids for off-grid electrification is centralized maintenance and their ability to provide the same customer experience as a standard grid-connection. Special appliances or power converters are not needed by customers. For multiple households that each have refrigerators or freezers, some operational efficiency can be gained due to load diversity though strong evening peaks still occur in rural installations. However, where the primary use of energy is for lighting, as is expected to be the case for the great majority of off-grid households, mini-grid use has few advantages to either EPC or the customers. The total panel capacity required remains about the same whether the panels are distributed to individual houses or are in a central array but there is the added cost of 240V reticulation, technicians need a higher level of training to repair mini-grid installations, and there are substantially increased issues of safety due to the higher voltages involved. Also in some cases land issues may be faced due to the large size of the arrays. For Samoa, no additional multi-user mini-grids appear to be appropriate for installation for home electrification by EPC though it does appear appropriate to consider mini-grids for eco-resort electrification and possibly the development of other off-grid businesses.

13

For example, at Saaga there is a small cluster of off-grid houses but only one indicates a high usage. The rest indicate that basic lighting is adequate so for them individual SHS is recommended since then the cost of reticulation is avoided. For the high usage household, a single user mini-grid may be appropriate due to the development of a tourism facility by the owner (site 13-129-309 on the survey). The site is on the southern coast and close to popular surfing areas visited by tourists. The owner has constructed 5 tourist fales and renovated the main house with the intention of commencing operations as a small resort in the near future. The survey shows good income for the owner and the site is now using a generator for powering a range of appliances including refrigeration, TV, radio and lighting. The distance to the closest grid connection is 2.2km representing a cost of around $100,000 Tala for a HV + LV grid extension plus any expense that may be necessary to obtain a right-of-way from land owners along the extension path. For the Saaga site, the installation of a centralized PV system with a central inverter appears appropriate to supply the large multi-appliance load for the main building and the smaller loads for the tourist fales. The O&M cost of the solar installation can be expected to be less than half that of a diesel generator providing 24 hour power and the convenience and silence of the solar installation would be a major advantage to the site owner. It is recommended that consideration be given by EPC to contracting with the site owner for the provision of a solar powered electricity supply at a rate equal to or lower than the cost of on-site diesel generation. As this is a one of a kind commercial venture, the terms of service should be negotiated with the electricity supply cost in no way related to that of supplying solar electricity to other off-grid customers. The system design should provide full AC supply to the main building with extensions of basic services to each guest fale. It is noted that other off-grid sites for eco-tourism are likely to be developed and in fact one off-grid site, 21-193-553, on Namua Island also along the southern coast, may opt for a special mini-grid installation. Sites of this type should be specifically negotiated on an individual basis if EPC provided solar electrification is requested.

6 BIOFUEL USE FOR OFF-GRID ELECTRIFICATION Biofuel could make technical and economic sense only for mini-grid operation and the cost of operation is not expected to be significantly different from diesel fuelled generation. The dispersed nature of the off-grid homes in Samoa do not make it reasonable to consider making the investment in biofuel production as a part of the off-grid electrification process since around 95% of off-grid houses are expected to have small loads better suited to SHS.

7 FINANCIAL CONSIDERATIONS

7.1 Cost of solar installations Based on prior experience in the region, a conservative costing for a stand-alone solar installation providing DC power is about US$8 per Wp installed. For AC power, about US$9 per Wp installed is a reasonable initial estimate. The capital cost is a fairly linear function relative to the Wp capacity of the panels. The O&M cost is a function of the cost of accessing the site for preventive maintenance and the cost of replacement of failed components. For SHS that operate only with DC appliances, the components that EPC would be responsible for include

14

the panels, the controller and the battery. Repair and replacement of appliances would be the responsibility of the household. Of the three components EPC would be responsible for, by far the most costly is the replacement of the battery. In general, battery replacements can be expected on no less than a five year cycle for average quality, sealed, deep discharge batteries. The life cycle cost of a good controller for an SHS is only about $10 Tala per year.

For SHS that operate AC appliances, the replacement of the inverter must also be included in the O&M cost with a 10 year replacement cycle reasonable for good quality sine wave type inverters and a 5 year cycle for lower cost modified sine wave units. For an inverter an annual replacement cost for O&M estimation is about $250 Tala per 100 Watts of capacity per year.

7.1.1 System sizing and costing process

The basic assumption for all sites is for a solar radiation level of 3.27 kWh/m2/day for a panel tilted at 14° and facing north (section 1.1). This is the average level of solar radiation for Afulilo during the months with maximum cloudiness as based on actual on site measurements. This is considered to be the worst case situation for the off-grid sites so an installation sized for this solar input should work satisfactorily for all the sites. Since oversizing of the solar panel results in significantly lower O&M costs, the fact that some installations will be oversized is not adding to their life cycle cost.

The second assumption is that the panel size in Wp must be adjusted to take into consideration the non-standard conditions at the site. The individual adjustments assumed are:

1. Cell temperature reaches 60°C causing an output reduction of about 12%

2. The load will not generally be at the maximum power point of the panel which will cause an output reduction of about 11%

3. The energy will not strike the panel at a right angle causing increased reflections from the cover glass causing an output reduction of about 7%

4. Some dirt will accumulate between rains causing an output reduction estimated at 5%

5. The panel will reduce its output by around 10% as it ages so to provide for sufficient output from the panel throughout its life an adjustment of 10% is needed.

Combining these adjustments: 0.88 X 0.89 X 0.93 X 0.95 X 0.90 = 0.62 total adjustment. This in effect means that on average the output from the panel will be about 62% of the rated output.

Combining the adjustment factor and the amount of solar energy coming in each day provides the panel generation factor:

3.27 X 0.62 = 2.03 Wh/day per Wp of panel capacity. Using this factor, the Watt-hours per day output from any panel installed under the assumed conditions can be calculated simply by multiplying the generation factor times the Wp of the panel. Likewise, if the Wh/day that must come from the panel are known, the Wp of the panel is easily calculated by dividing the Wh/day needed by the generation factor.

15

7.1.2 Basic lighting and radio SHS system sizing

Assumptions: One 13 Watt 12VDC CFL lights operated 5 hours per night = 65 Wh/day

Two 9 Watt 12VDC CFL lights operated 4 hours per night = 72 Wh/day One 0.5 Watt LED night light operated 12 hours per night = 6 Wh/day

One 10 Watt radio operated 6 hours per day = 60 Wh/day TOTAL Wh/day = 65+72+6+60 = 203 Wh/day used by the appliances

At 203 Wh/day, this represents a monthly usage of 203 X 30 = 6.1 kWh/month The wiring, connections, and controller can be expected to lose around 5% between the panel and the battery, plus a similar 5% loss can be assumed between the battery and the load. Also, on average, the battery can only be expected to deliver about 80% of the energy that is put into it from the panels, making a total system loss of:

0.95 X .95 X 0.80 = 0.72

Thus the panel must deliver about 28% more energy than the load requires, or another way to state it is that the energy that is available to the load is about 72% of that coming from the panels. Thus to compute the energy from the panels it is reasonable to divide the load energy by 0.72.

For the basic lighting system with a calculated load of 203 Wh/day, the energy needed from the panels will be about:

203 / 0.72 = 281 Wh/day With a generation factor of 2.03, the Wp of panel to provide that energy will be:

281 / 2.03 = 139 Wp minimum panel capacity A standard panel size currently readily available from manufacturers is 160 Wp and that will provide some excess capacity with associated lower O&M cost and allow for some load growth. The battery will need to provide the 203 Wh per day to the load plus covering about 5% losses in the wiring, connections and switches between the battery and the load. So the basic daily energy delivery from the battery will be about:

203 / 0.95 = 214 Wh/day In battery sizing terms, that is 214 Wh/day / 12 Volts = 18 Ah/day at C20

5.

To provide for the maximum life for the battery, an average daily discharge depth of no more than 20% is needed. Therefore the total battery requirement is about:

18 / 0.2 = 89 Ah at 12V and C20. However, as the battery ages, approximately 20% of its capacity will gradually be lost before it finally fails. To ensure that the installation will supply adequate energy to the load throughout the life of the battery, an initial battery size about 20% larger than this will need to be installed.

5 The notation C20 means that the rate at which the battery delivers its energy is equivalent to a full discharge in 20 hours. Batteries deliver more energy at slow rates so a C100 rated battery (very slow discharge) will be physically smaller than a C20 battery for the same Ah capacity.

16

The final specification will be: 89 / 0.8 = 111 Ah at 12V and at C20 using a sealed (valve regulated) battery.

A common standard size battery is 125 Ah at C20 and that is recommended for use. To ensure that losses are minimal and there is adequate capacity to operate all loads at the same time, a 20A capacity controller for both charging and load circuits is recommended as the minimum size.

7.1.3 Lighting, radio and TV/DVD SHS

The approach that is proposed for this class of installation is to install the same basic lighting PV system described under 6.1.2 and a second installation specifically to power a TV and DVD player. The second installation would operate at 24V instead of 12V to enable the use of two of the same batteries used in the 12V lighting system. The load for the TV and DVD player will be:

One 14 inch colour television requiring 75W operating 4 hours a day. That will require 300 Wh of energy One DVD player requiring 30 Watts and operated 4 hours per day which requires 120Wh/day6

TOTAL Wh/day = 420 Wh/day

At 420 Wh/day that will represent 420 X 30 = 12.6 kWh/month. The wiring, connections, switches and controller can be expected to lose around 5% between the panel and the controller and another 5% between the battery and the load. Also the battery can only be expected to deliver about 80% of the energy that is put into it from the panels and 10% of that is expected to be lost in the inverter to run these AC appliances. This results in a total system adjustment of:

0.95 X 0.95 X 0.80 X 0.90 = 0.65 Thus the panel must deliver about 35% more energy than the load requires, or another way to state it is that the energy that is available to the load is about 65% of that coming from the panels. Thus to compute the energy from the panels it is reasonable to divide the load energy by 0.65. For the basic lighting system with a calculated load of 420 Wh/day, the energy needed from the panels will be about:

420 / 0.65 = 646 Wh/day

With a generation factor of 2.03, the Wp of panel to provide that energy will be: 646 / 2.03 = 318 Wp minimum panel capacity

This represents two of the standard 160 Wp panels in series. The battery will need to provide the 420 Wh per day to the load plus covering about 5% losses in the wiring, connections and switches between the battery and the load plus 10% losses in the inverter. So the total losses will be about:

0.95 X 0.90 = 0.86 (14% total losses)

6 This represents playing two DVDs per day on average. Weekend use typically is greater than that while weekday use is usually less. For sizing purposes, the daily average of 2 DVDs per day is used.

17

So the basic daily energy delivery from the battery will be about: 420/0.86 = 488 Wh/day

In battery sizing terms, that is 488 Wh/day / 24 Volts = 20.3 Ah/day at C20. To provide for the maximum life for the battery, an average daily discharge depth of no more than 20% is needed. Therefore the total battery requirement is about:

20.3 / 0.2 = 102 Ah at C20.

However, as the battery ages approximately 20% of its capacity will be lost before it fails. To ensure that the installation will supply adequate energy to the load throughout the life of the battery, an initial battery size about 20% larger than this will be needed and the final specification will be:

103 / 0.8 = 128 Ah at 24V and C20 of the sealed (valve regulated) type. Two standard12V batteries connected in series and rated to provide 125 Ah at C20 will be satisfactory. To ensure that losses are minimal and there is adequate capacity to operate all loads at the same time, a 12A capacity controller (at 24V) for both charging and load circuits is recommended as the minimum size with 15 A preferred. In order to make the system modular (using the same components for both lighting and TV/DVD installations), a battery capacity of around 125 Ah at C20 can be used. In the case of the lighting system it would be a single battery at 12V and in the case of the TV/DVD system it would be two batteries connected in series for 24V. Also the panels can be 160 Wp capacity for both lighting and TV/DVD installations with one used for the lighting system and two for the TV/DVD installation. A 24VDV modified sine wave inverter of 150 Watt capacity will be needed. The TV and DVD player should be hard wired to the output of the inverter and the inverter wired to turn on and off with the TV.

7.1.4 Full electrification including refrigerator or freezer

If the assumed lighting load is 203 Wh/day (3 lights plus night light plus radio) and the TV-DVD load of 420 Wh/day is to be provided plus the operation of a refrigerator/freezer and two ceiling fans, the load assumption becomes: Refrigerator/freezer = 150 Watts, 12 hours/day = 1800 Wh/day

Two ceiling fans = 75 Watts, 8 hours per day = 1200 Wh/day Plus lighting load of 203 Wh/day and TV-DVD load of 420 Wh/day =

1800 + 1200 + 420 + 203 = 3623 Wh/day At 3623 Wh/day, that represents 30 X 3623 = 108.7 kWh/month

Since this size system will include full house AC, system losses will total 35% (the same as in the TV/DVD case) so the energy needed from the panel will equal about:

3623 / 0.65 = 5574 Wh/day With a generation factor of 2.03, the panel capacity required will be:

5574 / 2.03 = 2746 Wp of solar panel The battery will need to supply the load through the 5% losses of the wiring and the 10% losses in the inverter for a total loss of 14% (also the same as in the TV/DVD case):

18

3623 / .86 = 4213 Wh/day Assuming a 48V DC battery bank, the Ah/day will be:

4213 / 48 = 88 With a 20% daily depth of discharge the total battery capacity becomes:

88 / 0.2 = 440 Ah at C1007

And with 20% added capacity to cover ageing losses:

440 / 0.8 = 549 Ah at 48V and C100. Different manufacturers choose different standard sizes though 550 Ah and 600 Ah are common sizes and both would be satisfactory. This should be a battery bank of either 24 cells of at least 549 Ah and 2V each connected in series or 8 batteries at 6V and at least 549 Ah each connected in series. No paralleled batteries should be used. The controller should be of the MPPT type with a capacity of 30A. The Outback MX80 is recommended as it has performed well in Apolima. The inverter should have a capacity of at least 1.3 kW single phase. The Outback FX unit is recommended as it is hermetically sealed and circuit board corrosion is not a problem. Also that is the type of unit used successfully at Apolima.

7.2 Capital Costs Based on the general rule of thumb of US$8 per Wp of DC system cost and US$9 per Wp of inverter equipped system cost, the three system types would cost approximately:

Basic Lighting = 160 Wp X 8 = US$1,280 per household (about 3650 Tala) Controller replacement cost (10 year cycle) about US$0.70 per month (about 2 Tala)

Battery replacement cost (5 year cycle) about US$275 = US$4.58/month (13 Tala) TV-DVD add on cost = 320 WP X $9 = $2,880 per household (about 8210 Tala) Battery replacement cost (5 year cycle) about US$550 = US$9.17/month (26 Tala) Controller replacement cost (10 year cycle) about US$40.70 per month (about 2 Tala per month) Inverter replacement cost (10 year cycle) about US$1.00 per month (about 2.85 Tala per month) Total of lighting plus TV-DVD services – US$3,560 per household (about 10,145 Tala) Battery replacement cost (5 year cycle) about US$825 = US$13.75/month (about 39 Tala) Controller replacement cost (10 year cycle) about US$1.40 per month (about 4 Tala per month)

7 A C100 rating is assumed because the load is being continuously drawn from the battery, not concentrated in the evening, so the rate of battery discharge is slower, roughly 5 days.

19

Inverter replacement cost (10 year cycle) about US$1.00 per month (about 2.85 Tala per month)

Full AC electrification cost = 2746 Wp X 9 = US$ 24,714 per household (about 70,435 Tala) Battery replacement cost (5 year cycle) about US$70 per month (about 200 Tala/month)

Controller replacement cost about US$4 per month (about 12 Tala/month) Inverter replacement cost about US$10 per month (about 29 Tala) Table 4 – Summary of Estimated Cost for Installation Types

Estimated installed cost

Approximate capital cost

(Tala)

Estimated Component

Replacement Cost per Month

(Tala)

Approximate Service Cost per Month (visit every

other month)

(Tala)

Approximate Total

Monthly O&M Cost

(Tala)

kWh per month

required for the

appliances

Cost per kWh for O&M

only

(Tala)

Basic lighting and radio 3,650 15 15 30 6.1 4.92

TV and DVD only 8,210 31 15 46 12.6 3.65

Lighting plus TV-DVD 11,860 46 20 66 18.7 3.53

Full AC electrification 70,435 241 30 271 108.7 2.49

Although the cost per kWh for basic lighting is very high relative to grid power, it is lower than the monthly cost of kerosene for equivalent lighting plus batteries for radio use which is about $39 Tala per month as shown by the household survey. Additionally, the quality of services provided is much better than kerosene and battery use. Likewise, the O&M cost of lighting, radio and TV-DVD use is much higher than the cost of equivalent grid power but is much cheaper than operating a home generator four hours a day or frequently taking a car battery for charging in order to operate the appliances. For full electrification that includes a refrigerator or freezer, fans, TV-DVD, radio and lights the O&M cost of the solar installation is comparable to the cost of grid delivered electricity. Thus there is no question that for houses that remain off-grid, using solar provides a lower cost solution for providing the electricity needed to operate the desired appliances than do off-grid fossil fuel based alternatives – provided capital costs are not borne by the customer or by EPC. However if the grid can be extended to the house at a capital cost comparable to that of the solar installation, that usually will provide a lower life-cycle cost of providing electricity services for EPC due to the lower O&M cost of a grid extension relative to solar. Additionally the customer will receive better overall services than can be provided by solar. Table 5 shows the cost that EPC currently charges for extending the grid.

20

Table 5 – Grid Extension Cost

Distance from the nearest LV line Requirement Cost for household

$ = Tala < 35m No new poles required $120 35 - 60m 1 small pole (provided by customer) $188.60 + own pole 60 - 80m 1 pole (provided by EPC) $3,500.00

60/80 - 120/180m 2 poles (provided by EPC)

$7,000 Cost to EPC, $1,000 rebate thus $6,000 cost to customer

>120/180m Every additional 60-100m span Add $3,500 Installed transformer costs: 22kV Single Phase 15 kVA Serves 10-15 customers $8,752

The household survey shows that the average distance of an off-grid house from the grid is 407 metres but that is skewed by a few households that are over 2000 metres from the grid so the median distance is 180 metres indicating that about half the households that were surveyed were closer than 180 metres. For those desiring lighting plus TV/DVD service, the cost of extending the grid up to 400 metres is about the same as the capital cost of installing solar and installing the grid has lower O&M costs. Based on the information in Table 4 and Table 5, Table 6 indicates the approximate distance that the grid can be extended to one house for the same cost as is estimated for each of the four categories of solar installation.

Table 6 – Possible distance for a grid extension for the cost of each category of solar installations

Solar Installation Category Cost in Tala Approximate distance the grid can be extended for

the cost of solar PV8

Basic lighting and Radio $3,600 80 metres

TV and DVD $8,850 200 metres

Lighting + TV and DVD $13,220 400 metres

Full electrification $76,665 1,500 metres (1000 LV, 500 HV)9

8 This assumes that this is a single house extension and that no transformer must be set. Also it assumes the entire extension is LV unless otherwise noted. This will vary considerably as in some cases several currently off-grid houses can be electrified from a single LV line extension. 9 This assumes that beyond 1,000 metres, a LV line will deliver power at an unacceptably low voltage to a house that is using a wide range of appliances. The actual distance allowable depends on the size of LV wire used and the actual load characteristics.

21

It is certainly correct that most of the customers would prefer a grid connection to a solar installation. The household survey showed that nearly 80% of the off-grid households have no knowledge of solar and of those that do know of solar, almost half consider it an unreliable source. From EPC’s perspective, grid connections are also preferable. A solar installation is limited to providing the services it was initially designed for while a grid connection is flexible in the type and size of appliances to be connected. With load growth in newly connected homes a common occurrence, the inflexibility of solar can be a problem. Also, the O&M cost of solar exceeds the cost of maintenance for a grid extension to that house by a large margin for most off-grid solar installations.

Therefore consideration should be given to establish a programme that will provide for extending the grid at least to the 80 metre point for all off-grid households. This could feasibly be covered by the EPC Power Sector Expansion Project’s ‘LV Line Extension’ sub-project planned to begin in 2009. For households demanding solar capacity for just TV and DVD, extending the grid up to 200 metres makes financial sense. For those households desiring both lighting and TV-DVD services an extension of up to 400 metres also makes good financial sense. And for those households desiring the use of a freezer or refrigerator, lights, radio, fans and TV-DVD player, an extension of up to 1,500 metres will make more financial sense than installing PV. Thus if external financing can be located that can either finance off-grid solar or grid-extensions, finance for the grid extensions is to be preferred over solar installations since the O&M cost of the solar is much higher than that of a grid extension. This is shown in the difference in kWh cost of solar vs. that of a grid connection. Since the highest grid connection tariff being charged can be assumed to include the O&M cost of power delivery (which currently is mostly fuel cost with a small amount included for grid maintenance and administration) the fact that the solar per kWh cost is several times the highest on-grid tariff indicates a much higher O&M cost for all classes of installation other than class 4 -- full electrification – in which case they are closer to being equal but solar is still higher. One problem with providing externally funded grid extensions in lieu of solar is the fact that people who have already paid a substantial sum for a grid extension beyond the basic distance may find a program that provides an essentially free extension for others unacceptable and that will make such a programme politically difficult to implement. On the other hand, it is unlikely that there will be political problems associated with a programme to install solar PV for off-grid electrification, particularly if the effective per-kWh cost is much higher than that for households on the grid.

7.3 Program cost The household survey indicates that about 70% of households currently in the off-grid category can be expected to opt for basic lighting and radio, about 25% of households currently in the off-grid category can be expected to opt for TV-DVD and basic lighting and probably not more than 5% will opt for full electrification (essentially those houses that are currently using a home generator or have a high income). Table 7 indicates the estimated cost of a programme for installing all PV based electrification of all off-grid households that do not qualify for a no cost extension.

22

Table 7 – Cost of installed PV systems for full solar electrification

Installation Category

Number of households (based

on 250 off-grid homes)

Cost per household

(Tala)

Total cost for the category (Tala)

Lights and Radio 175 3,600 630,000

Lights, radio & TV 63 13,220 832,860

Full electrification 12 76,665 919,980

Total 250 2,382,840 (~US$836,100)

However because the O&M cost for solar PV is substantially higher than that for a grid connection providing the same services, a substantial number of the off-grid households can be served with less life-cycle cost through a grid extension than through solar PV installation. Even the Class 1, most basic installation costs about the same as an 80 metre grid extension just for capital investment. This implies that there would be a net saving to EPC for those houses within 80 metres of the grid.

7.4 Connection Charges and Tariffs

7.4.1 Connection Charges

The off-grid household survey did not show that the households would be willing or able to accept significantly larger connection fees than those currently being charged for grid connection with no line extension cost. If the goal of the programme is to get as many of the currently off-grid people connected as possible, no additional connection fees are recommended. The connection of solar powered customers is recommended to be the same as that of on-grid customers and to follow the same policies regarding who pays for what components (e.g. house wiring). It is noted that EPC will not have to pay for a meter so there is some savings in that area.

For basic lighting, the installation cost of the solar is comparable to the cost of connecting an on-grid house and, based on the household survey, about 70% of the off-grid households are expected to opt for this level of service. This basic class of customer indicated in the survey that a connection cost of 100 Tala or less is expected with half saying that it should be 50 Tala or less if they are to connect. Experience has been that the acceptable cost stated to surveyors tends to be smaller than what households actually are willing to pay when faced with the decision to connect or not but still this is an indication that the connection cost cannot be very different from the standard grid connection fee if connections are to be accepted. The cost of installing larger than the basic PV installation is higher than that of a grid connection. If that is of significant concern to EPC it appears from the survey that many households would accept a slightly higher monthly charge to allow EPC to recover that extra cost over the life of the installation rather than paying a much higher connection fee.

23

7.4.2 Tariff

It is not practical to charge customers for kWh usage when the service is supplied by a solar PV system providing power to a single household. The cost of supply from solar is not dependent on the amount of energy supplied but rather on the initial cost, life and replacement cost of the components and the cost of preventive maintenance. While preventive maintenance costs are about the same for all sizes of PV installation, the battery and other component costs increase as the PV system size increases. Therefore it is recommended that the charges for solar installations be the actual cost of O&M for the class of installation provided. That is basically the replacement cost of the battery, controller and inverter per month of its estimated life plus the cost of the required bi-monthly preventive maintenance access. These estimated costs are shown in Table 4. With regards to component replacement costs, the numbers are considered to be conservative. With adequate preventive maintenance, the batteries can be expected to last longer than 5 years and a controller life in excess of 20 years has been frequently seen in the Pacific. If EPC wishes to recover some of the installation cost for the larger systems, the survey indicates that increasing the monthly fee by an amount that would allow the added cost to be recovered over at least a 10 year period should still be acceptable to most of those off-grid customers desiring a higher level of services from solar. While for most off-grid users, this will be higher than the amount paid by on-grid customers for the same level of service, it will be largely offset by greatly reduced use of kerosene for lighting and dry batteries for operating entertainment devices. For larger users, the solar electricity will be substantially cheaper than operating a generator though still higher than comparable services from the grid.

8 ENVIRONMENTAL CONSIDERATIONS In general operating solar PV systems for off-grid electrification have no negative environmental impacts. However, when batteries are replaced, the failed battery must be disposed of properly, preferably recycled by the manufacturer or a battery recycling facility. Most manufacturers of solar batteries will arrange for recycling, particularly for sealed batteries since they have little problem with shipping unlike open cell batteries that must be cleared of acid before shipping. Therefore, when purchasing batteries, it is recommended that any recycling cost be built into the initial battery cost to ensure that recycling does indeed occur. Typically the recycling cost is only on the order of 1% of the battery cost since the battery materials have substantial scrap value. The costs estimated for battery replacement include this 1% recycling charge.

Since even car battery recycling is not currently a common practice in Samoa, it is recommended that EPC work with local environmentally focused NGOs to help develop a lead-acid battery recycling programme such as has been done by Kiribati. This approach should reduce the cost of solar battery recycling for EPC plus eliminate the substantially larger environmental damage that can occur due to improper disposal of the thousands of car batteries currently in use in Samoa.

24

9 MAINTENANCE STRUCTURE It has been well documented that preventive maintenance is the key to reliable service from SHS. Without preventive maintenance to maintain good quality connections, ensure that panels are not shaded or that users do not stress the installations by adding more appliances, battery life is typically much shorter than in installations that receive good preventive maintenance. Since battery replacements are a major component of SHS O&M costs, preventive maintenance has been found to be the most important component of a maintenance programme intended to reduce life cycle costs of off-grid solar. Many approaches to solar project maintenance have been attempted in the Pacific. It is clear that user maintenance does not work well since users do not have the skills or interest necessary to provide proper preventive maintenance; indeed users are probably the main source of damage to solar installations. Also providing maintenance on call has not worked simply because this does not include preventive maintenance. By the time an actual system failure has occurred and the user calls for service, the battery probably has been irreversibly damaged and its life shortened. Preventive maintenance actions can usually find problems in their early stages and correct them before battery damage occurs.

All successful solar projects in the Pacific include a periodic preventive maintenance visit by an especially trained technician. During the visit, the technician checks all components for proper operation, checks connections and switches for excessive voltage drop and counsels the user in proper management of the system. Though these periodic checks typically take no more than 20 minutes for an installation with an open cell battery and under 15 minutes for an installation with a sealed battery, the results are an installation that has long life and provides reliable service. The preventive maintenance technician is generally backed up by a more broadly trained senior technician who is on call to solve problems the preventive maintenance technician cannot resolve.

Because solar projects in other Pacific Islands are usually community based and quite remote, the preventive maintenance technician is typically a village resident trained for the maintenance task and paid for making periodic visits to all installations for maintenance. This keeps the cost of access very low and labour rates also low while still providing effective service. In Samoa the off-grid households are very dispersed and the number of installations relatively small. Therefore is it not possible to establish community technicians since there are no communities of solar households. In Fiji, Kiribati and Tonga organizations have been set up specifically to provide maintenance for off-grid solar. The organizations recruit villagers for local technicians, have on staff a supervisory technician, manage spare parts stores and collect fees from users to pay their costs. However for this to be cost effective, around 1000 installations in concentrated clusters of 25 or more houses have to be present. In Samoa it is impractical to establish a service company specifically for the maintenance of the solar installations since the number of installations is too small to be profitable and the sites too dispersed. Therefore either EPC will need to employ and train a person to visit the sites on a bi-monthly basis for preventive maintenance or must contract with a local company specifically for that service. Because of the need to ensure proper maintenance and the proper scheduling of visits, if a contractor is hired, supervision will be needed. In the end, given the small number

25

of households involved and the relative simplicity of the tasks, it appears likely that the lowest cost option and the option that provides the best services overall will be for EPC to directly perform the necessary maintenance. Certainly, the households surveyed prefer that arrangement since they stated overwhelmingly (97.6%) that their preference for operating and maintaining their electrical supply was EPC. That will require one low-level person whose part-time responsibility is to visit each site every two months and perform preventive maintenance. Additionally a person at EPC should be trained to do basic system design, develop solar component specifications, tender for components, and do the more complex repairs that the field technician cannot handle.

10 SUMMARY OF RECOMMENDATIONS 1. That four classes of installations be established. Class 1 and Class 2

installations would use the same set of components with Class 1 providing sufficient energy to operate lights and radio at 12V and Class 2 providing sufficient energy to operate a TV/DVD player at 24V through a dedicated inverter. Class 3 would be a combination of both Class 1 and Class 2 installations. Class 4 would provide an AC supply sufficient to power a refrigerator, lights, fans, TV/DVD, and assorted small appliances.

2. That EPC work with eco-tourism developers and other off-grid businesses to provide solar power for their operations on a negotiated fee basis.

3. That the approximately 60 off-grid households who are up to 120 metres from a grid connection receive a grid connection upon payment the basic connection fee. An 80 metre connection represents about the same investment by EPC as the installation of a Class 1, basic lighting and radio PV system and the O&M cost of a grid extension appears sufficiently lower than hat of solar PV to pay the additional 40 metres as amortised over the life of the installation. This action is estimated to reduce the number of PV installations needed by about half and their associated O&M cost by about one-third while increasing the initial capital cost only slightly.

4. That solar connection fees be the same as those for basic grid connections. The cost of installation of a Class 1 PV system is expected to be about the same as the cost of installation of a grid extension. If EPC wishes to recover some or all of the added cost of installing the larger PV installations the cost should be recovered through a higher monthly charge, not as an initial connection fee.

5. That solar power tariffs be based on the estimated O&M cost for the class of installation provided.

6. That EPC provide all maintenance for the PV installations with one field person trained in preventive maintenance procedures and one management person trained in system sizing, component specification and purchasing, troubleshooting and repair.

7. That EPC work to establish a battery recycling programme that includes automotive batteries as well as solar batteries.

26

REFERENCES Access to Sustainable Power Services – Apolima Island, UNDP Project Document

Apolima Island PV Project – Final Commissioning Report, February 2007, prepared by Bruce Clay for UNDP Samoa

Apolima Photovoltaic System Design And Specifications for Preparing Requests For Quotation of Major Components, prepared by Herbert A Wade for UNDP, 31 May 2005’ Application of the RESCO Set-up and Principles on the Apolima Island PV Project, prepared by Herbert A. Wade, June 2006 Development Consent Application Form, Planning and Urban Management Agency; Draft Planning and Urban Management (Environmental Impact Assessment) Regulations, 2006, Government of Samoa

Feasibility Study – Possible Future Power Supply Options for Apolima Tai, Samoa, Final Report, March 2003, prepared by Gerhard Zieroth for UNDP Samoa & UNESCO Apia Household Income and Expenditure Survey 2002 Tabulation Report, Ministry of Finance, Statistical Services Division, Government of Samoa Lands Survey and Environment Act, Samoa Pacific Regional Energy Assessment 2004 – Regional Overview Report, Secretariat of the Pacific Regional Environment Programme (SPREP), 2005. Prepared by Mr. Hebert A. Wade, Mr. Peter Johnston, and Mr. John Vos

Samoa National Assessment Report, Volume 11 – Pacific Regional Energy Assessment 2004, Secretariat of the Pacific Regional Environment Programme (SPREP). Prepared by Mr. Hebert A. Wade, Mr. Peter Johnston, and Mr. John Vos Samoa National Energy Policy 2007, Ministry of Finance, Economic Policy and Planning Division, Government of Samoa. Samoa Photovoltaic (PV) Rural Electrification Program, Project Document, UNDP Apia, August 2007 Samoa: Preparing the Power Sector Expansion Program, prepared by John Grimston and Study Team, Tonkin and Taylor International Ltd for Ministry of Finance, Samoa Survey Report Final Version, Preparatory Phase Of The Samoa Photovoltaic Electrification Programme, Contract 2007-005 UNDP Apia, Samoa, 11 November, 2008

Tabulation Report, Population and Housing Census, 2006, Ministry of Finance Statistics Department., Government of Samoa

Taking of Land Act, Samoa UNDP, 2004b, Energy for Sustainable Development in the Asia-Pacific Region: Challenges and Lessons from UNDP Projects, pp. 83-94 UNDP/UNESCO, 2003, Feasibility Study – Possible Future Power Supply Options for Apolima Tai, Samoa, prepared by Mr. Gerhard Zieroth

27

UNESCO, 2003, Solar Photovoltaic Project Development, prepared by Mr. Herbert A. Wade

Village Profiles of Savai’i Island 2004, Ministry of Women, Community and Social Development, Government of Samoa

Village Profiles of Upolu Island 2004, Ministry of Women, Community and Social Development, Government of Samoa

World Bank, 1994, Solar Energy – Lessons from the Pacific Island Experience, prepared by Andres Libenthal, Subodh Mathur and Herbert A Wade,